Engineering Materials

Engineering Materials
Springer-Verlag Berlin Heidelberg GmbH
ONLINE LIBRARY
Engineering
http://www.springer.de/engine/
Wolfgang Grellmann· Sabine Seidler (Eds.)
Deformation and Fracture

Behaviour of Polymers
With 447 Illustrations and 51 Tables
" Springer
Prof. Dr. Wolfgang Grellmann
Department of Engineering Science
Martin-Luther-University of Halle-Wittenberg
D-06099 Halle, Germany
http://www.kunststoffdiagnostik.de
Prof. Dr. Sabine Seidler

Institute of Materials Science and Testing
Vienna University of Technology
FavoritenstraEe 9-11
A-I040 Vienna, Austria
http://www.tuwien.ac.atlE308
Lectures of a discussion conference took place at the Institute of Polymer Materials e.¥., Martin-
Luther-University of Halle-Wittenberg under the direction of Prof. Dr. W. Grellmann, Merseburg.
The respective articles of the participants in the seminar were published unchanged in content
in that version provided by the authors. Because of overall impression a uniform textual and
graphical layout of the contributions was widely realized by the editors.
Library of Congress Cataloging-in-Publication Data

Deformation and fracture behaviour of polymers 1Wolfgang Grellmann, Sabine Seidler, eds.
p. cm. -- (Engineering materials) Includes index.
ISBN 978-3-642-07453-0 ISBN 978-3-662-04556-5 (eBook)
DOI 10.1007/978-3-662-04556-5
1. Polymers--Fracture. 2. Deformations (Mechanics) I. Grellmann, Wolfgang, 1949-
II. Seidler, Sabine, 1961- III. Series.
ISBN 978-3-642-07453-0
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broad-
casting, reproduction on microfilms or in other ways, and storage in data banks. Duplication of this
publication or parts thereof is only permitted under the provisions of the German Copyright Law of
September 9, 1965, in its current version, and permission for use must always be obtained from
Springer-Verlag Berlin Heidelberg GmbH.
Violations are liable for prosecution under the German Copyright Law.
http://www.springer.de
© Springer-Verlag Berlin Heidelberg 2001

Originally published by Springer-Verlag Berlin Heidelberg New York in 2001
Softcover reprint ofthe hardcover 1st edition 2001
The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of
a specific statement, that such names are exempt from the relevant protective laws and regulations and
therefore free for general use.
Typesetting: Cameraready from editors

Cover design: de'bilk, Berlin
Printed on acid-free paper SPIN: 10760513 57/3020/M - 5 4 3 2 10
Preface
The application of fracture mechanics to polymers and composites allows the

quantitative description of the toughness behaviour by means of fracture mechan-
ics parameters and enables preventive failure analysis. In recent years this young
scientific discipline has developed rapidly, and now the experimental results are
looking for more applications in industrial practice. However, the practical appli-
cations of fracture mechanics parameters to structural-integrity assessment are
severely restricted owing to their limited transferability from specimens to com-
ponents. Indeed, geometry-independent fracture mechanics parameters are very
important for the reliable functioning of polymers and components in nearly all
industrial application fields.
These application fields include the polymer development, quality control, con-
struction and polymer-specific design of reliable components in the motor indus-
try, the electrical industry and the manufacture of household appliances, as well as
applications in information technology and medical applications.
The present status report on the deformation and fracture behaviour of polymer
materials was composed on the basis of revised lectures presented at the Merse-
burg discussion conference entitled 'Deformation and Fracture Behaviour of
Polymers' and additional single contributions.
The editors and authors have tried hard to present information about the applied
fracture mechanics of polymers and composites in the light of their current re-
search work.
The aim was to express the present standard of knowledge as comprehensively
as possible by inclusion of additional contributions referring to sets of problems
such as the assessment of toughness properties with fracture mechanics parame-
ters, approximate methods, quantification of morphology-property correlations,
and limits of application.
These discussion conferences take place every two years with the intention of
demonstrating the progress of basic and applied research in the deformation and
fracture behaviour of polymers, homopolymers, blends, copolymers, composites
and biocompatible materials. For this purpose, plenary lectures, short reports,
discussions and an exhibition of instruments in the field of destructive and non-
destructive material testing are organized.
Particular topics of these conferences are
- toughness characterization of polymers with fracture mechanics concepts
- morphology-property correlations
- hybrid methods of polymer testing and polymer diagnostics
- technological test methods for testing of components and structures
- biocompatible materials and medical prostheses
examples and limits of the application of polymers.
VI Preface
It is hoped that these conferences will make a contribution to the current under-
standing of problems in this field by specialists from universities/colleges and the
polymer industry.
We want to thank sincerely all co-authors and collaborators from the Institute
of Materials Science and Polymer Materials e.V. of the Department of Material
Sciences of the Martin-Luther-University of Halle-Wittenberg and all collabora-
tors from the Institute of Materials Science and Testing of the Vienna University
of Technology, who made possible the publication ofthis book by their readiness
for duty and their willing cooperation.
The editors sincerely thank Dr. Christian Bierligel from the Institute of Material
Sciences for the comprehensive cooperation and advice that he gave.
We also thank Dipl.-Ing. Katrin Reincke for the technical preparation of the
manuscript, as well as Springer for their forthcoming cooperation.
Merseburg, Vienna Wolfgang Grellmann
October 2000 Sabine Seidler

Contents
List of Authors.............................................................................. XIII
Nomenclature............................................................................... XV
Terminology............... ................................................................... xxv
A Characterization of Toughness
Using Fracture Mechanics Methods
A1 State of the Art and Development Trends
A 1.1 New Developments in Toughness Evaluation of Polymers

and Compounds by Fracture Mechanics
W Grellmann................................. ............ ................................... 3
A 1.2 Concepts of Fracture Mechanics for Polymers

F. Ramsteiner, W Schuster, S. Forster......................................... 27
A2 Experimental Methods
A 2.1 Influence of Specimen Geometry and Loading Conditions

on the Crack Resistance Behaviour ofPoly(vinyl chloride)
and Polypropylene
W Grellmann, S. Seidler, K. Jung, M Che, 1. Kotter...................... 51
A 2.2 Procedure for Determining the Crack Resistance Behaviour

Using the Instrumented Charpy Impact Test
W Grellmann, S. Seidler, W Hesse................................................. 71
A 2.3 Possibilities and Limits of Standards and Drafts

for J-R Curve Determination of Polymers
S. Seidler, W Grellmann........................... ...................................... 87
A 2.4 The Relationship Between the Fracture Behaviour

and Structural Parameters ofPE-HD
E. Nezbedowi, J. Kucera, Z. Salajka................................................ 95
VIII Contents
A 3 Alternative Methods
A 3.1 Application of Single-Specimen Testing Methods

for Determining J-R Curves of Polymers
S. Seidler.......................................................................................... 105
A 3.2 Application of Normalization Method for Determining

J-R Curves in the Amorphous Polymer PVC
M Che, W. Grellmann, S. Seidler.................................................... 121
A 3.3 Calculation of J-R Curves Based on Load-Deflection Diagrams

Using the Hinge Model Test Specimen
R. Steiner, W. Grellmann................................................................. 133
A 3.4 An Alternative Method Based on J-TJ and ~T6 Stability

Assessment Diagrams to Determine Instability Values
from Crack Resistance Curves
R. Lach, W. Grellmann.................................................................... 141
B Morphology-Property Correlations
B1 Homopolymers
B 1.1 Supermolecular Structure and Mechanical Behaviour

of Isotactic Polypropylene
M Raab, J. Kotek, J. Baldrian, W. Grellmann.......... ...................... 153
B 1.2 Correlation Between Structure and Toughness Behaviour

of High-Density Polyethylene under Impact Load
H. Beerbaum, W. Grellmann, S. Seidler.......................................... 161
B 1.3 Toughness and Relaxation Behaviour ofPMMA, PS and PC

W. Grellmann, R. Lach................ .................................................... 181
B 1.4 Crazing in Amorphous Polymers - Formation of Fibrillated Crazes

Near the Glass Transition Temperature
G. H. Michler. ........ .......................................................................... 193
B 1.5 Influence of Temperature and Moisture on Toughness Behaviour

of Polyamide
B. Langer, S. Seidler, W. Grellmann..... ................................... ....... 209
Contents IX
B 2 Blends
B 2.1 Relationship Between Fracture Behaviour and Morphology

in PE/PP Blends
U. Niebergall, J. Bohse, H Sturm, S. Seidler, W Grellmann.......... 229
B 2.2 Influence of Modifier Content and Temperature

on Toughness Behaviour of Polyamide
I Bethge, K. Reincke, S. Seidler, W Grellmann.............................. 243
B 2.3 Morphology and Toughness ofPP/EPR Blends

T. Koch, S. Seidler, K. Jung, W Grellmann.................................... 257
B 2.4 Morphology and Micro-Mechanics

of Phase-Separated Polyethylene Blends
R. Godehardt, W Lebek, G. H Michler.......................................... 267
B 3 Copolymers
B 3.1 Toughness Optimization of Multi-Phase Polymer Materials

Based on a PP Matrix Using Fracture Mechanics Parameters
S. Seidler, W Grellmann................................................................. 281
B 3.2 Crack Toughness Behaviour of ABS Materials

R. Lach, W Grellmann, P. Kruger.................................................. 301
B 3.3 Fracture Mechanics Characterization of ABS Materials

- Influence of Morphology and Temperature
R. Lach, W. Grellmann, Y. Han, P. Kruger..................................... 317
B 3.4 Brittle Fracture of ABS

- Investigation of the Morphology-Failure Relationship
B. Moginger, G. H Michler, H-C. Ludwig..................................... 335
C Hybrid Methods of Polymer Testing

and Polymer Diagnostics
C.I Defect-Selective Imaging

A. Dillenz, N. Krohn, R. StOj3el, G. Busse........................................ 355
C.2 Determination of Local Deformation Behaviour of Polymers

by Means of Laser Extensometry
C. Bierogel, W. Grellmann............................................... ............... 365
X Contents
C.3 Damage Analysis of Composite Materials

by Acoustic-Emission Examination
J. Bohse, T. Krietsch........................................................................ 385
D Technological Test Methods
0.1 Polymer-Based Composites for Friction and Wear Applications

K Friedrich, P. Reinicke, J. Hoffmann........................................... 405
0.2 Modification of Polymers by Means of Amorphous Carbon

for Optimization ofTribological Properties
/. Hyla, J. Myalski, W. Grellmann...................................... ............. 419
0.3 Mechanical Vibration Behaviour of a Compressor Blade

Made from a High-Performance Composite
W. Grellmann, R. Steiner, /. Kotter, M Neitzel, M Maier,
K. von Diest..................................................................................... 429
E Biocompatible Materials and Medical Prostheses
E.l Polymer Materials in Joint Surgery

J. Brandt, W. Hein................................................... ........................ 441
E.2 Material Parameters and ESEM Characterization

of Functional ENT Prostheses During Ongoing Degradation
E.-J. Haberland, A. Berghaus, M Fating,
/. Bethge, W. Grellmann.. .................................. ..................... ......... 451
E.3 Microbial Corrosion of Pharyngo-Tracheal Shunt Valves

(,Voice Prostheses')
/. 8ebowl, E.-J. Haberland, A. Stiefel.............................................. 461
E.4 Deformation Behaviour of Voice Prostheses

- Sensitivity of Mechanical Test Methods
C. Bierogel, /. Bethge, W. Grellmann, E.-J. Haberland................. 471
F Special Materials
F.l Crack Initiation, Wear and Molecular Structure

of Filled Vulcanized Materials
W. Grellmann, G. Heinrich, T. Casar.............................................. 479
Contents XI
F.2 Investigation of Crack Propagation Behaviour

of Unfilled and Filled Vulcanizates
K. Reincke, R. Lach, W. Grellmann, G. Heinrich............................ 493
F.3 Characterization of Deformation Behaviour

of Modified Polymer Concrete
H Wehner, W. Grellmann, T. Hildebrandt...................................... 505
F.4 Fracture Mechanics Testing of Modified Epoxy Resins

with Mini-Compact Tension (CT) Specimens
H Walter, C. Bierogel, W. Grellmann, M Fedtke, B. Michel......... 519
G Examples and Limits of Application
G.l Modelling of the Mechanical Behaviour

of Non-Linear Viscoelastic Materials
under a Multi-Dimensional State of Stress
E. Schmachtenberg, M Wanders, N. M Yazici............................... 533
G.2 Detergent Resistance ofPP/GF Composites

W. Grellmann, S. Seidler, C. Bierogel, R. Bischoff................ .......... 549
G.3 Material Optimization of Polypropylene-Short-Glass-Fibre

Composites
B. Langer, C. Bierogel, W. Grellmann, J. Fiebig, G. Aumayr.......... 561
G.4 Influence of Exposure on the Impact Behaviour

of Glass-Fibre-Reinforced Polymer Composites
H. Waiter, C. Bierogel, W. Grellmann, B. Rufke............................. 571
G.5 Physical Ageing and Post-Crystallization of Polypropylene

J. Fiebig, M Gahleitner..... ....................... ....... ................................ 581
Subject Index................................................................................... 593
Author Index.................................................................................... 599

List of Authors
Aumayr, Gilnther, Dipl.-Ing., Linz (Austria)

Baldrian, Josef, Dr., Prague (Czech Republic)
Beerbaum, Heike, Dr.-Ing., Halle (Germany)
Berghaus, Alexander, Prof. Dr. med. habil., Halle (Germany)
Bethge, Ines, Dipl.-Ing., Halle (Germany)
Bier6gel, Christian, Dr.-Ing., Halle (Germany)
Bischoff, Reinhard, Dr.-Ing., Berlin (Germany)
Bohse, Jilrgen, PO Dr.-Ing. habil., Berlin (Germany)
Brandt, JOrg, Dr. med., Halle (Germany)
Busse, Gerd, Prof. Dr. rer. nat. habil., Stuttgart (Germany)
Cisar, Thomas, Dipl.-Ing., Halle (Germany)
Che, Mingcheng, Dr.-Ing., Geilenkirchen (Germany)
von Diest, Konstantin, Dr.-Ing., Kaiserslautern (Germany)
Dillenz, Alexander, Dipl.-Phys., Stuttgart (Germany)
Fedtke, Manfred, Prof. Dr. rer. nat. habil., Merseburg (Germany)
Fiebig, Joachim, Dipl.-Phys., Linz (Austria)
Forster, Stephan, Ludwigshafen (Germany)
Friedrich, Klaus, Prof. Dr.-Ing., Kaiserslautern (Germany)
Fiiting, Manfred, Dr., Halle (Germany)
Gahleitner, Markus, Dr. techn., Linz (Austria)
Godehardt, Reinhold, Dr. rer. nat., Halle (Germany)
Grellmann, Wolfgang, Prof. Dr. rer. nat. habil., Halle (Germany)
Haberland, Ernst-Jilrgen, PO Dr. rer. nat. habil., Halle (Germany)
Han, Yanchun, Dr., Changchun (China)
Hein, Werner, Prof. Dr. med. habil., Halle (Germany)
Heinrich, Gert, Dr. rer. nat. habil., Hannover (Germany)
Hesse, Wolfgang, Dipl.-Phys., Halle (Germany)
Hildebrandt, Thomas, Dipl.-Ing., Rendsburg (Germany)
Hoffmann, JUrgen, Dipl.-Ing., Kaiserslautern (Germany)
Hyla, lzabella, Prof. Dr.-Ing. habil., Katowice (Poland)
Jung, Kerstin, Dr.-Ing., Merseburg (Germany)
Koch, Thomas, Dipl.-Ing., Vienna (Austria)
XIV List of Authors
Kotek, Jiri, Dr.-Ing., Prague (Czech Republic)

Kotter, Ines, Dipl.-Ing., Merseburg (Germany)
Krietsch, Torsten, Dr., Berlin (Germany)
Krohn, Nils, Dipl.-Phys., Stuttgart (Germany)
Kruger, Peter, Dr., Leverkusen (Germany)
Kucera, Jaroslav, Dr., Bmo (Czech Republic)
Lach, Ralf, Dr.-Ing., Halle (Germany)
Langer, Beate, Dr.-Ing., Merseburg (Germany)
Lebek, Werner, Dipl.-Phys., Halle (Germany)
Ludwig, Hans-Christian, Dipl.-Ing., Stuttgart, (Germany)
Maier, Martin, Prof. Dr., Kaiserslautem (Germany)
Michel, Bernd, Prof. Dr. rer. nat. habil., Berlin (Germany)
Michler, Goerg Hannes, Prof. Dr. rer. nat. habil., Halle (Germany)
Mliginger, Bernhard, Dr.-Ing., Stuttgart (Germany)
Myalski, Jerzy, Dr.-Ing., Katowice (Poland)
Neitzel, Manfred, Prof. Dr., Kaiserslautem (Germany)
Nezbedova, Eva, Dr.-Ing., Bmo (Czech Republic)
Niebergall, Ute, Dr.-Ing., Berlin (Germany)
Raab, Miroslav, Dr.-Ing. Associate Prof., Prague (Czech Republic)
Ramsteiner, Falko, Dr. rer. nat., Ludwigshafen (Germany)
Reincke, Katrin, Dipl.-Ing., Halle (Germany)
Reinicke, Petra, Dipl.-Ing., Kaiserslautem (Germany)
Rutke, Bruno, Dr.-Ing., Schkopau (Germany)
Salajka, Zdenik, Dr., Bmo (Czech Republic)
Schmachtenberg, Ernst, Prof. Dr.-Ing., Essen (Germany)
Schuster, Werner, Ludwigshafen (Germany)
Sebova, Irina, Dr. med., Halle (Germany)
Seidler, Sabine, Prof. Dr.-Ing. habil., Vienna (Austria)
Steiner, Ralf, Dr.-log., Merseburg (Germany)
Stiefel, Amd, Prof. Dr., Halle (Germany)
StoDel, Rainer, Dipl.-Ing., Stuttgart (Germany)
Sturm, Heinz, Dr. rer. nat., Berlin (Germany)
Walter, Hans, Dipl.-Ing., Merseburg (Germany)
Wanders, Martin, Dr.-Ing., Essen (Germany)
Wehner, Heidrun, Dipl.-Ing., Merseburg (Germany)
Yazici, NazifMehmet, Dipl.-Ing., Essen (Germany)
Nomenclature
Only those symbols that are used in several passages are listed here. Because of
the fmite size of the alphabets used and because of mUltiple application in
scientific usage, double use of symbols and indices was not avoidable.
a (mm) initial crack length (i.e. machined notch plus razor-

sharpened tip), the physical crack size at the start of
testing
aBS (mm) physical crack length augmented to account for crack
tip plastic deformation (fracture mirror length)
acN (kJ/m2) Charpy impact strength of notched specimen
according to ISO 179
acu (kJ/m2) Charpy impact strength of unnotched specimen
according to ISO 179
aeII (mm) effective crack length
ar (mm) final crack length used in the normalization method
a/W ratio of initial crack length to specimen width
A (J.1m) average interparticle distance
Al first amplitude considered for the calculation of the
logarithmic decrement
Ac (Nmm) complementary deformation energy used in the
J-integral evaluation method of Merkle and Corten
Acrit (J.1m) critical particle distance for brittle-to-tough
transition
Ad (J.1m) average interparticle distance, measured between
centres
Ael (Nmm) elastic part of AG
AG (Nmm) general deformation energy of test specimen
computed from the area under the load-deflection
diagram up to Fmax
AH (Nmm) nominal impact energy of pendulum hammer
An nth amplitude considered for the calculation of the
logarithmic decrement
XVI Nomenclature
ApI (Nmm) plastic part of Ao

AR (Nmm) crack propagation energy
Atot (Nmm) area under the load-deflection diagram used in the
approximate method of Schindler
b statistical segment length
B (mm) specimen thickness
C (mmIN) compliance
C), ...,9 constants of the power law for describing R-curves
Cel (mmIN) elastic compliance
CD (~m) average interparticle distance (mid point distance)
d (~m) average particle diameter
do average distance between ends of a chain segment
D maximum grain size
Do (nm) distance of fibrils
D I ,2 geometrical functions in the J-integral evaluation
method of Merkle and Corten (MC)
E (MPa) Young's modulus (modulus of elasticity)
Es (MPa) flexural modulus
Ed (MPa) dynamic flexural modulus
Edis (Nmm) dissipated energy
Ec (MPa) flexural modulus according to ISO 178
Epot (Nmm) potential energy
Espec (MPa/(kgldm3 » specific modulus of elasticity
Et (MPa) modulus of elasticity in tension
Eli (MPa) integral modulus of elasticity in tension (tensile
modulus)
Ell (MPa) local modulus of elasticity in tension
/ (mm) deflection
Is (mm) deflection of an unnotched specimen
/gy (mm) deflection at the transition from elastic
to elastic-plastic material behaviour
!K (mm) maximum deflection/max excluding the component/s
/max (mm) deflection at maximum load F max
/PI (mm) plastic component of maximum deflection
Nomenclature XVII
(mm) plastic component of maximum deflection of

ipl V-notched specimens used for the key curve method
F (N) load (force)
Fl (N) inertial load, which arises from the inertia of the part
of the test specimen accelerated after the first contact
with the striker
Fgy (N) characteristic load value corresponding to the
transition from elastic to elastic-plastic material
behaviour
FID1JX (N) maximum load
FQ' (N) pseudo-elastic load
G (MPa) shear modulus
G (N/mm) energy release rate
.
G1 (N/mm) energy release rate in mode I
G (MPa) dynamic modulus
H heterogeneity
He heterogeneity at tensile strength
HK (N/mm2) ball indentation hardness
I intensity
J (N/mm) J-integral; a mathematical expression, a line or
surface integral that encloses the crack front from one
surface to the other, used to characterize the local
stress-strain field around the crack front; fracture
mechanics parameters are calculated using methods of
evaluation of this integral
JO.2 (N/mm) technical crack initiation value for an amount of crack
growth of /),.a = 0.2 mm
~ (N/mm) J value in mode I (the index I is only used in the case
of geometry independence)
'"IdMe (N/mm) critical J value at the point of unstable crack growth,
for dynamic loading, in the geometry-independent
J-integral evaluation method of Merkle and Corten
'"IdST (N/mm) critical J value at the point of unstable crack growth,
J-integral evaluation method of Sumpter and Turner
Jdapp (N/mm) critical Jvalue at the point of unstable crack growth
determined from J-TJ stability assessment diagram,
for dynamic loading
XVIII Nomenclature
Jl L (N/mm) critical Jvalue at the point of unstable crack growth,

J-integral evaluation method of Begley and Landes
Jdc (N/mm) crack initiation value used in the key curve method
J/ (N/mm) crack initiation value used in the approximate method
of Kobayashi and Moskala
Jl (N/mm) crack initiation value used in the approximate method
of Schindler
Jg (N/mm) fracture resistance at upper limit of J-controlled crack
growth
~ (N/mm) physical crack initiation value determined from
intersection of stretch zone width and J-R curve
Jrn (N/mm) maximum available J value used in the approximate
method of Schindler
Jmax (N/mm) validity limit for J
JTJ (N/mm) energy absorption capacity of a material during
stable crack growth
leo" (mm3/Nm) wear resistance factor
K (MPamm/2) stress intensity factor
KI (MPamm I/2 ) stress intensity factor in mode I (the index I is only
used in the case of geometry independence)
K lc (MPamm Il2 ) fracture toughness, critical parameter at the point of
unstable crack growth; static loading, geometry-
independent
KId (MPamm I12 ) fracture toughness, critical parameter at the point of
unstable crack growth; dynamic loading, geometry-
independent
CTOD (MPamm/ ) K Ic and KId, calculated from C(T)OD
2
K lc,ld
E (MPamm I/2 ) K Ic and KId, calculated from equivalent-energy concept

KIc,ld
J (MPamm 1/2) K Ic and KId, calculated from J-integral

Klc,Id
KLEFM
lc,ld (MPamm/2) K Ic and KId, calculated from LEFM
Ie contour length of polymer chain between two

adjacent entanglements
L (mm) specimen length
Nomenclature XIX
Lo (mm) initial gauge length

La (run) thickness of amorphous interlayer determined
by means of Lpexp
Latbeo (run) thickness of amorphous interlayer determined by
means of Letbeo
Le (run) thickness of lamellae determined by means of Lp
Letbeo (run) thickness of lamellae calculated from melting
temperature Ts~
Lp (run) long period from small-angle X-ray diffraction
scattering
m constraint factor in relation between J and
Sconcepts
mH (kg) weight of pendulum hammer
Mo molecular weight of a monomer unit
Me molecular weight of a chain network
Mn (kg/mol) molecular weight, number average
Mw (kg/mol) molecular weight, weight average
MFR (g/10 min) melt mass-flow rate
n rotational factor
n work-hardening factor
Ne number of statistical segments of a chain
Oi surface area of craze fibrils
p (MPa) pressure
poo (MPa) Vogel pressure used in WLF equation
Q (J) quantity of heat
rK (!lm) notch radius of razor blade
Re average final distance of a chain net between two
chemical crosslinks
Re average fmal distance of a chain net between two
elastic active entanglements
Rr average final distance of a chain net between two
filler particles
s (mm) support span
s, (N/mm2) slope of blunting line used in the approximate method
of Schindler
xx Nomenclature
S2 (N/mm2) slope of crack propagation line used in the

Smax maximum of scattering
S (wt. %) percentage of fractions up to maximum particle
diameter of one fraction of the total mixture
t (s) time
tb (ms) time to brittle fracture
tB (ms) time to fracture
tan 0 mechanical loss factor
T eC) temperature
To (J/m3 ) threshold tearing energy
TaJ eC) Vogel temperature used in the WLF equation
TBn eC) brittle-to-tough transition temperature
Tg eC) glass transition temperature
TJ tearing modulus
TI2 tearing modulus determined from J-f1a curve at
f1a= 0.2 mm
Tm eC) melting point
Tmo (K) equilibrium melting temperature
TSOE (K) melting temperature determined from DSC according
toOIN 53765
Tv eC) Vicat heat resistance temperature
T/jO.2 tearing modulus determined from O-f1a curve at
f1a= 0.2 mm
U (Nmm) deformation energy
v (mm) crack-mouth-opening displacement
Vo (lis) strain rate
VF volume of fibrils in a craze
VH (mls) pendulum hammer speed
VI. (mm) load-line displacement
vp (mm) plastic component of the crack-mouth-opening
displacement
Vr (mmlmin); (mls) crosshead speed
(mm3/Nm) specific wear rate

Ws
Nomenclature XXI
Wt (Jlmlh) wear rate

W (rrun) specimen width
z (rrun) distance of knife-edge from specimen surface
al (%) degree of crystallinity obtained from density
a2 (%) degree of crystallinity obtained from melting enthalpy
a3 (%) degree of crystallinity obtained from X-ray diffraction
/3 proportionality constant of geometrical size
criterion for LEFM
/3J slope of line through origin used for determination of
J;pp; the point of unstable crack growth is the
intersection between the line through the origin and
the J- TJcurve
/30 slope of line through origin used for determination of
o;PP; the point of unstable crack growth is the
intersection between the line through the origin and
the 8-To curve
(rrun) crack-(tip)-opening displacement describing the local
strain field in front of the crack tip, calculated with the
help of the plastic-hinge model
0
. (mm/s) crack-opening-displacement rate
00.2 (rrun) technical crack-opening displacement calculated

at Ila = 0.2 rrun
0, (rrun) crack-(tip)-opening displacement in mode I
(the index I is only used in the case of geometry
independence)
Ole (rrun) critical ovalue for unstable crack growth, quasi-static
loading, geometry-independent
Old (rrun) critical ovalue for unstable crack growth, dynamic
loading, geometry-independent
OIdll (rrun) critical ovalue for unstable crack growth obtained by
using advanced plastic-hinge model, dynamic loading,
geometry-independent
Og (rrun) ovalue at upper limit of O-controlled crack growth
~ (rrun) crack-tip-opening displacement at physical crack
initiation
XXII Nomenclature
~ (mm) validity limit for 0 value

l1a (mm) amount of stable crack growth, distance between
original crack size and crack front after loading
11a", (mm) amount of stable crack growth at maximum of the
load-deflection diagram used in the
l1amax (mm) upper validity limit of l1a
l1amin (mm) lower validity limit of l1a
I1C (mm/N) variation of compliance
111 (mm) change in length
M
. (mmls) rate of change of length
!lJ (s) time difference

I1v (m/s) velocity change of pendulum hammer during the test
11(1' (MPa) excess increase of local stress
& proportionality constant of geometrical size criterion
in J-integral concept
& (%) strain
& (lIs); (o/o/min)

.
strain rate ( &= d& / dt)
&8 (%) tensile strain at break according to ISO 527
&ce (%) normal flexural strain
&01 (%) amount of uniform elongation without necking
bj (%) integral strain
&i (1/s) integral strain rate

&, (%) local strain
&, (lis) local strain rate
limax (%) maximum local strain

bimin (%) minimum local strain
&M (%) strain at tensile strength according to ISO 527
&Mi (%) integral strain at tensile strength according to ISO 527
&q (%) lateral (transverse) strain
ESE (%) critical strain at acoustic onset
Nomenclature XXIII
&y (%) yield strain according to ISO 527

GZM (%) tensile strain at maximum load according to DIN
53455
&zR (%) breaking elongation
87S (%) yield strain according to DIN 53455
T/ geometrical function
T/el;pl geometrical functions for assessment of elastic (el)
and plastic (pI) parts of deformation energy used in
the J-integral evaluation method of Sumpter and
Turner
e logarithmic decrement
e (0) scattering angle
It yield ratio of polymer chain at formation of fibrils
A.craze stretching at fracture
Amax maximum stretching
Ar relative index of damping
J.l coefficient of friction

J.l Poisson's ratio
J1c chemical-chain-knot density

f.L;. integral Poisson's ratio
J.lI local Poisson's ratio
v Poisson's ratio
; proportionality constant of geometrical size criterion
forCTOD
p (kg/m3) density
Pk density of crystalline phase at the melting point
(i (MPa) stress
(i (S/m) conductivity
(i
(MPals) stress rate
OhM (MPa) flexural strength according to DIN 53452

(is (MPa) tensile stress at break according to ISO 527
(id (MPa) yield stress determined from Charpy impact test at
specific rate
(if (MPa) flexural stress according to ISO 178
XXIV Nomenclature
O"o.s (MPa) flexural strength at peripheral strain of3.5 %

O"tM (MPa) flexural strength according to ISO 178
OJ: (MPa) yield stress: either oy or O"p = 112 ( oy+aW
OJ (MPa) integral stress
OJ (MPa) local stress
O"M (MPa) tensile strength according to ISO 527
OSpec (MPaI(kgldm3» specific flexural strength
O"SE (MPa) critical stress at onset of acoustic emission
oy (MPa) yield stress (yield point) according to ISO 527
O"zM (MPa) tensile strength according to DIN 53455
O"zR (MPa) tensile stress at break according to DIN 53455
O"zS (MPa) yield stress (yield point) according to DIN 53455
T oscillation period
qJv filler or fibre content
f/J geometrical factor
'1/ (Hz) frequency

'I/o (Hz) reference frequency used in WLF equation
tV non-dimensional constant for characterizing
J-controlled crack growth
n (Hz) upper frequency limit used in WLF equation
Terminology
ABS acrylonitrile-butadiene-styrene
AE acoustic emission
ASA acrylonitrile-styrene-acrylate
BA butyl-acrylate copolymer
BR butadiene rubber
BTT brittle-to-tough transition
CCT centre-cracked tension specimen
CF carbon fibre
CFRP carbon-fibre-reinforced polymer
CT compact tension specimen
CTOA crack-tip-opening angle
C(T)OD crack-(tip)-opening displacement
DCB double-cantilever beam
DDENT deeply double-edge-notched tension specimen
DENB double-edge-notched bend specimen
DENT double-edge-notched tension specimen
DMA dynamic-mechanical analysis
DMTA dynamic-mechanical-thermal analysis
DSC differential scanning calorimetry
EIP ethylene-propylene
EP epoxide; epoxy
EPDM ethylene-propylene-diene rubber
EPFM elastic-plastic fracture mechanics
EPM ethylene-propylene copolymer
EPR ethylene-propylene rubber
EVAC ethylene-vinyl acetate copolymer
ESCC environmental stress corrosion cracking
ESEM environmental scanning electron microscope
FEM finite-element method
GF glass fibre
GPC gel permeation chromatography
HDT heat distortion temperature
HVEM high-voltage transmission electron microscope
ICIT instrumented Charpy impact test
IFWIT instrumented falling-weight impact test
ITIT instrumented tensile impact test
LEFM linear elastic fracture mechanics
MC J-integral evaluation method of Merkle and Corten
MOPE medium-density polyethylene
MSM mUltiple-specimen R-curve method
NDT non-destructive testing
XXVI Terminology
NMR nuclear magnetic resonance

NPT notch-pipe test
NR natural rubber
PA polyamide (nylon)
PBI poly-bis-maleinimide
PBT poly(butylene terephthalate)
PC polycarbonate
PE polyethylene
PE-HD high-density polyethylene
PEEK polyetheretherketone
PEEKK polyetheretherketoneketone
PENT polyethylene notch test
PET poly(ethylene terephthalate)
PI polyimide
PMMA poly(methyl methacrylate)
PP polypropylene
PS polystyrene
PS-HI high-impact modified polystyrene
PTFE polytetrafluoroethylene
PUR polyurethane
PVC poly(vinyl chloride)
R-curve crack resistance curve
RPM J-integral evaluation method of Rice, Paris and Merkle
RT room temperature
SAN styrene-acrylonitrile
SAXS small-angle X-ray scattering
SB styrene-butadiene block copolymer
SBR styrene-butadiene rubber
SBS styrene-butadiene-styrene
SCB short-chain branching
SEM scanning electron microscope
SENB single-edge-notched bend specimen
SENT single-edge-notched tension specimen
SFM scanning force microscope
SIF stress intensity factor
SIS stepwise isothermal segregation
SSM single-specimen R-curve method
ST J-integral evaluation method of Sumpter and Turner
SZH stretch zone height
SZW stretch zone width
TDCB tapered double-cantilever beam specimen
TEM transmission electron microscope
TFA tear and fatigue analyser
TPU thermoplastic urethane
UD unidirectional
UHMWPE ultra-high-molecular-weight polyethylene
WAXS wide-angle X-ray scattering
WLF Williams-Landel~Ferry equation
A
Characterization of Toughness
Using Fracture
Mechanics Methods
A1 State of the Art and Development

Trends
A3 Alternative Methods
A 1 State of the Art and Development Trends
A 1.1 New Developments in Toughness

Evaluation of Polymers and Compounds
by Fracture Mechanics
W. Grellmann, Merseburg
1 Introduction
The application of fracture mechanics to the estimation of the failure reliability of

products made of polymers and compounds, as well as an evaluation method for
quality control and material development, requires geometry-independent pa-
rameters which react extraordinarily sensitively to structural changes in the materi-
als. An essential prerequisite for a theoretically well-based material optimization is
a knowledge about the connections of strength- and toughness-determined defor-
mation and fracture mechanisms to structural quantities.
For quantification of these connections, it is the task of fracture mechanics ma-
terial characterization to find material-specific parameters by reproducible and, if
possible, also standardized test procedures [1,2]. In applying fracture mechanics
methods to polymers, on the one hand, some basic knowledge of the methods of
evaluation of metallic materials can be used; on the other hand, the substantial time
and temperature dependence must result in further polymer-specific development of
methods (pp. 27-50, [3,4]). Progress in the technological use of fracture mechanics
material testing has yielded a new generation of material parameters [5-8], which, in
combination with morphology analyses, enables structure-related quantitative
morphology-toughness correlations [9-13].
2 Information Potential of Fracture Mechanics Material

Parameters - the Multi-Parametrical Description
of the Mechanical Behaviour of Polymers
The starting point for ascertaining fracture mechanics material parameters are
records of load-load-line displacement curves and load-crack-mouth-opening
displacement curves of notched specimens (Fig. A I ).
W. Grellmann et al. (eds.), Deformation and Fracture Behaviour of Polymers

4 W. Grellmann
Fracture Mechanics Parameters

Linear Elastic Fracture Mechanics Elastic-Plastic Fracture Mechanics
LEFM LEFM EPFM EPFM
brittle elastic-plastic unstable elastic-plastic stable
r-----~-=.,
g
~
v.f(mm) v. f(mm) v. f(mm)
Stress Intensity Factor Crack-Tip-Opening Displacement

K I ", Kid (Ku ; Kuv /)1'" /)Id
load-determined deformation-determined
J-Integral energy-determined
unstable crack growth stable / unstable crack growth stable crack growth
Fig AI. Possible ways of ascertaining fracture mechanics parameters
If we observe linear elastic material behaviour in brittle polymer materials,

filled or fibre-reinforced material systems, or specimens of great thickness or be-
low the glass transition temperature Tg, the material behaviour can be described on
the basis of linear elastic fracture mechanics (LEFM) or LEFM with small-scale
yielding, in the case where the plastic zone at the crack tip is taken into considera-
tion. As a parameter, the stress intensity factor (SIF) K describing the linear elastic
area in front of the crack tip is used. The critical value of the SIF under conditions
of plane strain is called the fracture or crack toughness K Ic (static loading), KId
(dynamic loading), Ku or Km (I, II, III refer to crack-tip-opening modes). In prac-
tice, mode I has the highest importance. For evaluating the toughness in the case of
non-negligible elastic-plastic material behaviour and extensive plastic areas in
front of the crack tip, the concepts of elastic-plastic fracture mechanics have to be
used. The most important concepts are the crack-tip-opening displacement
(CTOD) and the J-integral.
The CTOD concept is based on the assumption that, for ductile material be-
haviour, the fracture process is not controlled by stress intensity any more, but by
plastic deformation in front of the crack tip. A measure of this is the widening at
the crack tip, called the crack-tip-opening displacement or crack-opening dis-
placement 0-
Material behaviour of the kind represented in the right part of Fig. A I requires a
fracture mechanics evaluation using the crack resistance concept.
The techniques of fracture mechanics provide a vast apparatus for material
testing, but it enables one to obtain structure-related material information only
when used and applied in the right way. Unfortunately, discussions of mechanisms
and phenomena of fracture can be found over and over again in the literature
New Developments 5
Material Parameters
Stress-Stnin Curve of Tensile Test R-Curve of Elastic-Plastic Fracture Mechanics
Crack Propagat ion
Crack Initiation
Breaking Strain
Work or Deforrution Energy Dissipation

Strain (Extension) Crack Growth ru..n..
Material Panmeters Material Pan meters
• Yield Point Oy • Crack Initiation Ji
• Exponent ofWork-Hardeniog e • Crack Propagation: Tearing Modulus TJ
• Work ofDefonnation Ws • Energy Dissipation JoTJ
• Tensile Strength OM ; Breaking Strain ~B
Fig. A2. Stress-strain curve for mechanical material testing and R-curve for elastic-plastic
fracture mechanics
which take into account only the yield point and/or Young's modulus obtained
from the stress-strain curve or the stress intensity factor obtained from the load-
load-line displacement curve, for instance. In Fig. A2 an analogy between the
stress-strain curve and the crack resistance curve for elastic-plastic fracture
mechanics (EPFM) is shown.
The parameters of the stress-strain curve that are used provide basic informa-
tion about the strength- and deformation-determined mechanical behaviour of
unnotched specimens. For evaluating the fracture behaviour, the complex connec-
tion between loading, crack length and material resistance against unstable and
stable crack propagation has to be analysed. Whereas fracture mechanics material
testing of polymers to ascertain crack toughness in terms of resistance against
unstable crack propagation is already included in industrial testing practice, there
are remarkable deficits in the industrial use of parameters from the EPFM crack
resistance curve. Further interpretation of the deformation and fracture behaviour
in the sense of a structure-related interpretation is only possible on the basis of a
multi-parametrical description in connection with investigations of morphology.
In order to analyse the relevance of fracture mechanics material parameters,
crack toughness parameters can be arranged in levels of knowledge, according to
the method of determination, as shown in Table AI:
6 W. Grellmann
Table AI. Arrangement of fracture mechanics material parameters into levels of knowledge
Fracture mechanics Loading parameters Criteria for fracture Practical relevance

concept Characterization of stress and deforma- caused by cracks of material pa-
tion in the area around the crack tip rameters
Crack toughness related to resistance against unstable crack initiation
Linear elastic fracture Stress intensity factor Static crack initiation Load and stress
mechanics (LEFM) KI>Kk determined
LEFM with small- critical values: Dynamic crack initiation Fracture toughness
scale yielding } Kk,Kld
Equivalent energy mode II, III:
KIT,Km
Elastic-plastic fracture
mechanics (EPFM)
CrOD Crack-tip-opening displacement Deformation-

bj" bj.,!mode II, III, mixed-mode determined critical
crack-tip-opening
displacement
J-integral J-integral Energy-determined

J k , JI.,!mode II, III, mixed-mode J-integral
Crack toughness related to resistance against stable crack initiation and propagation
Crack resistance (R) Crack Crack propaga- Generalized Crack initiation

concept initiation tion crack 8; Jcomponent >r5~ Jj or 0.2 JTr or STir
propagation controlled stable
Crack resistance T _ d Ii !!... crack propagation
curve bi, ~.2 • - d(&l) rI, Crack propagation
J-R curve T _ dJ !!... dJ_ dJ.....,;ol
Ji. J O.2 J - d (80) CT: ~>~
J- T stability assess- Instability parameter J 50 Instability caused by Energy-determined
ment diagram advanced crack propaga- plastic instability
tion
Tcomponent> Tmaterial
Crack toughness related to resistance against rate of change of parameters
Consideration of time-dependent changing of deformation energy
Modified CrOD
and J concept 6 =~ic,ld 18
tB = time to fracture 51 >j [",Id ; j, >j ic,ld r5. or J-
controlled crack
propagation
j
k,ld
=~
18
R concept To component Energy-rate-

determined plastic
Ii -&1; J-&1 instability
TJ
component
New Developments 7
• crack toughness related to resistance against unstable crack propagation

• crack toughness related to resistance against stable crack initiation and propa-
gation
• crack toughness related to resistance against rate of change ofload parameters.
All parameters enabling the determination of crack toughness re'lated to resis-
tance against unstable crack propagation are arranged in level I of knowledge.
Thus, the LEFM concept (including LEFM with small-scale yielding and the con-
cept of equivalent energy) enables quantitative understanding of failure on the
basis of load- or stress-determined fracture toughness, as only measured values of
load or stress affect the equations describing the conditions.
In column I of Table Al the fracture mechanics concepts according to which
the material behaviour is considered are listed. In the second column characteristic
fracture mechanics parameters for characterizing the stress and deformation area
near the crack tip are shown, and in the third and fourth columns the corresponding
crack fracture criteria and the practical meanings of the material parameters are
shown. The crack-tip-opening displacement values bic and bid, determined accord-
ing to the CTOD concept of elastic-plastic fracture mechanics under static or
dynamic loading conditions, respectively, have proved to be deformation-
determined material parameters, as only deformation values are taken into account
in their calculation.
In connection with the evaluation of toughness of fibre composites, experimen-
tal methods and formalisms for interpretation for mode II, mode III and mixed-
mode loads have gained growing importance. Whereas the CTOD concept pro-
vides deformation-determined parameters and the LEFM concept provides frac-
ture-load-determined parameters, the J-integral enables an energetic interpretation
of fracture behaviour, as in that case measured values of load as well as deforma-
tion are considered. Thus, the J-integral concept holds a central position in evalu-
ating fracture behaviour.
Using the J-integral concept requires that one pays attention to the fact that in
most cases the fracture is initiated by stable crack propagation (level II of knowl-
edge). The evaluation of crack toughness related to resistance against stable crack
initiation and propagation is performed on the basis of the crack resistance (R)
concept. For constructing R-curves, the crack tip opening and the J-integral are
preferred as load parameters to the stress intensity factor. The resistance against
the actual crack initiation is called ~phYs or ~pbys and the resistance against stable
crack propagation is called TJor T6> the slope of the R-curve multiplied by E/(J"lor
E/(J"F, respectively.
Starting from the energy balance at the crack, Will and Michel [14-16) have
developed a practical model for evaluating stable crack propagation. According to
this model stable crack propagation occurs if the energy dissipated, in a material-
specific way, compensates for the surplus of available energy caused by crack
propagation. Accordingly, crack propagation is controlled by the product JTJ or
t5Tl), and it is called JTr or OI'rcontrolled stable crack propagation, respectively.
Will [14-16) showed the existence of JTrcontrolled crack propagation for sev-
eral groups of materials by means of model examples.
8 W. Grellmann
As another tough-fracture-mechanics criterion, the instability value J so has been

used by Paris and Johnson [17]. For its experimental determination, a graphical
method, explained in pp. 141-150, is used, by which J-da diagrams are trans-
formed into J-TJ stability assessment diagrams. Using such diagrams, information
about the growth instability of cracks in components becomes possible without
using a method for determining crack toughness as resistance against unstable
crack propagation.
For level III of knowledge, the determination of crack toughness in terms of
time-dependent fracture mechanics parameters, on the one hand, only limited ex-
perimental results have been gathered, but on the other hand, substantial progress
in material-related toughness evaluation is expected here [18].
The load rate is known to exert a very important influence on the fracture be-
haviour of polymer materials, whereas, in the realization of fracture mechanics
investigations, mostly the crosshead speed of the testing machine, the impact speed
of a pendulum or falling hammer or the impact speed of a projectile in an arrest
experiment is given. However, these rate values are not comparable, because dif-
ferent types of specimens produce different transformations of the load-line dis-
placement into the deformation of the area around the crack tip.
As a parameter for describing toughness related to resistance against rate of
change, the crack-tip-opening displacement rate appears appropriate because it is a
comparable parameter compensating for the influence of different specimen di-
mensions [19]. For evaluating materials in accordance with this description of
crack growth, the change in time of the load value can be used for the J-integral
and R-curve concepts as well as for the LEFM concept. Thus, the parameter.[ is to
be considered as a rate of change of energy and the crack growth is said to be J-con-
trolled.
3 Determination of Parameters of Crack Toughness

Related to Resistance Against Stable Crack
Propagation
3.1 Standardized Methods
For ascertaining J-integral values and for ascertaining parameters for resistance
against unstable crack propagation, various evaluation methods based on an ener-
getic interpretation of the J-integral are known, such as ASTM E 813-81 [20],
ASTM E 813-89 [21], ASTM E 1737-96 [22], ESIS P 2-92 [23] and ESIS TC 4
[24].
A comparison of investigations by Seidler [18,25,26] to prove the applicability
of these evaluation methods by means of an iterative solution technique [27] led,
for the materials investigated, to the statement that in the case of polymers the
evaluation method proposed in procedure (pp. 71-86),
New Developments 9
J Ael ApI (I (0.751]el -l~a) (AI)

= 1]el B(W _ a) + 1]pl B(W - a) - W- a '
was to be preferred. The main difference from the ascertainment of parameters by
known testing procedures [22-24] lies in the distinctive evaluation of the elastic
and plastic parts of the general deformation energy.
Besides different methods of evaluation of the J-integral, the testing procedures
have different requirements for the analysis of crack resistance curves, whereby in
particular the restriction on the maximum allowed amount of stable crack growth
and the requirements on specimen thickness are controlled. If the stable crack
growth is corrected during an experiment, than a restriction according to empirical
criteria is not necessary [18].
The slope criterion, which has been transferred from metallic materials [23] di-
rectly to polymers [24], has to be seen as another critical criterion:
W-a dJ
m= - J - d(L1a) ~ 10 . (A2)
Within the limits of a material-invariant model, the requirement m » I was

formulated by Will [14-16]. For materials for which the stress-strain behaviour
can be described by a power law and if the von Mises criterion is satisfied, fur-
thermore, a linear relation exists between the energy dissipation rate of non-local
damage and the fracture mechanics parameter JTJ.
In Fig. A3, some results of systematic investigations of amorphous and semi-
crystalline homopolymers, polymer blends, copolymers, and short-fibre-reinforced
and filled thermoplastics are summarized. It shows that a considerable variety of
polymer materials do not satisfy (A2), and thus an unnecessary restriction on the
amount of stable crack growth would arise.
In order to estimate the requirements on the specimen geometry regarding
specimen thickness B, initial crack length a and ligament length (W~ a), criterion
(A3) is used:
(A3)
The constant & is material-dependent and may take values in the range of 25 to
200 for metallic materials [2].
For ascertaining crack toughness related to resistance against unstable crack
propagation, reliable data on the requirements on specimen geometry exist, as
there are & values for thermoplastic materials determined in [28] as ranging be-
tween 10 and 90, and for epoxy materials determined in pp. 71-86 and [29] as
ranging from 780 up to 1220.
10 W. Grellmann
100
I
1<0
= I 8 JoT 0045 I
. J I
...- I
0
• ........ Ii""
8 10 ...n..
V •
...
/ A
v
•
~
/ 1;'"
Y ........
I
•
10 100 1000
J·TJ (N/mm)
Fig. A3. Influence of the energy dissipation capability on the slope criterion for the crack
resistance curve [18]
10000 ~~.:; ,**~~~!'l!!!"'l':'~~~~~~~~~!j1

.,
~=F=I=l::'::'::1 A HOPE 1 [30] <> HOPE 3 [30)
.... , 0 HOPE 4 [30] [] UH~iWPE [3 \) t
+--++1- + Nylon Zylel SI801 [32] • Nylon Zylel St901 [32J
1000 ~~~I!!t • ABS [32] • PS-Hl [33]
=!==F=R==iI • ADS [34J )( PVC [29]
[18J ABS [18]
! [..£79. -~ ~P I!: ".. T
w 100 ~~~~~_~~_~~_
I;-~-+~+,~
Tij--+~~~~~~~~~~~~~
0.1 10 100 1000
J1c,Id (N!mm)
Fig. A4. Requirements on specimen geometry for ascertaining J values for material resistance
against stable crack propagation
The results found in the literature concerning the influence of specimen thick-
ness on the crack resistance behaviour under static load, and the investigations that
have been performed under dynamic load of a semicrystalline PA material and an
amorphous ABS material show the relation in Fig. A4 [IS]. Including all experi-
mental data for crack toughness related to resistance against stable crack propaga-
New Developments 11
tion, a decrease of s with increasing toughness was found. The constant s was
found to be in the range between 2 and 100. The dependence of the constant son
the J value is described in pp. 71-86 as a general relation according to the power
law
s = 224 J-O.94 , (A4)
which is valid independently of the conditions of loading (static, dynamic) and the
kind of material failure (unstable, stable). The s values for PVC and PP gathered
in pp. 51-70 fit into this general dependence.
3.2 Alternative Methods
The development of new materials and the extension of the field of usage of
known materials requires novel testing and measuring procedures that can be ap-
plied with less expense. New developments have shown up in the fields of alterna-
tive methods for estimating crack resistance curves from the load-deflection be-
haviour and for determining crack instability values from R-curves (Fig. AS).
Therefore, the following developments are outlined separately in Parts A 3 'Alter-
native Methods':
• For determining crack resistance curves from complete load-deflection dia-
grams, comparing all methods described in the literature, the approximate key
curve proves to be the most suitable because the appearance of stable crack
growth typical of polymers can be reflected below the transition from 'elastic'
to 'elastic-plastic' material behaviour ([18], pp. 105-120).
• Using a modified hyperbolic tangent material law in a hinge specimen enables
one to determine important material parameters and to calculate the corre-
sponding crack length from a load-deflection curve, geometrical data and the
notch width. Thus, the possibility arises of calculating crack resistance curves
(pp. 133-140).
• The normalization method of Landes can be used successfully for determining
crack resistance curves of polymers from the load-deflection behaviour (pp.
121-132).
• Experimental methods for evaluating the safety with respect to failure of poly-
mers under limited stable crack propagation using J-TJ or 8-To stability as-
sessment diagrams have gained special practical importance. Owing to the deri-
vation of empirical parameters to defme stability limits, an estimation of load
areas or material states, for which up to now no determination of parameters has
been possible, will become feasible (pp. 141-150).
However, it has to be stated that there is the restriction that the essential inten-
tion of using these modem testing techniques, i.e. to obtain the most complex
material determinations using methods as simple as possible in a relatively short
time, can only partially be accomplished at the present state of the art. On the basis
12 W. Grellmann
of dynamic fmite-element calculations considering the time-dependent material

behaviour, a more effective application of approximate methods can be expected
[18].
Development of Alternative Methods for Evaluation of Crack Resistance Curves and

Parameters of Instability for Plastics
I
unstable ~ stable
~ stable ~ unstable
Approximate Methods J-T J and &-Tl) Stability

• Kobayashi/Moskala Assessment Diagrams
• Schindler I
• 1-TJ and 05 -Tl) stability
• Schindler II assessment diagrams ba-
• Dahl sed on crack resistance
• Key Curve (R-)curves for determina-
• Aurich tion of instability parame-
T ters (especially for descri-
Hinge Model of Steiner bing the temperature de-
pendence of toughness for
• Calculation of l-R curves polymers)
based on load-deflection
diagrams and using the hinge
model test specimen
Normalization Method
• Determination of crack resi-
stance curves by means of
normalization method
Fig. AS. Alternative methods of experimental fracture mechanics
4 Morphology-Property Correlations
4.1 Temperature Dependence of Toughness
From the practical point of view, the ascertainment of brittle-to-tough transition

temperatures and the definition of usage limits for polymer materials are of special
importance. This requires a description of the temperature dependence of tough-
New Developments 13
ness using fracture mechanics concepts that best matches the actual material prop-
erties [7].
In viscoelastic material behaviour, molecular relaxation processes play a deci-
sive role, i.e. they affect the time (rate )-temperature behaviour directly during the
fracture process [36,37]. Using concepts from fracture mechanics to evaluate the
temperature dependence of toughness properties has increasingly gained impor-
tance because of the extraordinarily sensitive reaction of fracture mechanics pa-
rameters to structural alteration [10,18,38].
As an example, the relevance of these concepts will be shown for the tempera-
ture dependence of the toughness of an unoriented and a highly oriented polypro-
pylene (orientation/x = 80 %). The investigations included both quasi-static ten-
sion loading on CT specimens and dynamic bending in an instrumented Charpy
impact test. Figure A6 shows the arrangement of the fracture mechanics specimens
in relation to the orientation achieved by cold rolling.
Fig. A6. Specimen orientation in a cold rolled PP plate (T - transverse; L -longitudinal)
Fracture toughness values K lcLEFM calculated from the LEFM concept and KlcE
calculated from the equivalent-energy concept, determined under quasi-static
loading on CT specimens of PP as a function of temperature, are represented in
Fig. A7. The increase of toughness for the highly oriented PP is interpreted as
being due to the reorientation of chain molecules in the rolling direction (a to c
texture), and owing to the loading of the main valence bonds more energy can be
dissipated. Extensive structural investigations, the influence of orientation on the
fracture behaviour and the dependence on the degree of orientation are described
in [7,39].
The fracture toughness values determined with the LEFM concept decrease
with increasing temperature for both of the materials (Fig. A7). This dependence,
contradictory to the real toughness properties of those materials, can be explained
by the fact that for T > -20°C the limit of validity of the LEFM has been ex-
ceeded, and therefore this concept should not be used any more because of the
extended plastic deformation around the crack tip. Considering the effective crack
length and using the equivalent-energy concept [40], the increase of toughness
with increasing temperature has been proved to closely follow this latter concept.
14 W. Grellmann
Parameters
E Condition of test specimen
K lc
s 300 • • PP highly oriented
~ • ... PP unoriented
~ •
•
CTspecimen
vT = 8.33 . IO-s ms- l
lOO'~---r---r---?--~--~----~--~~
-90 -60 -30 o 30
TeC)
Fig. A7. Temperature dependence of fracture toughness values KlcLEFM and Klc E for unoriented
and highly oriented PP under quasi-static loading conditions
The parameters of the CTOD and J integral concepts shown in Fig. AS prove
appropriate as well, while, independently of the material parameters of EPFM
used, a decrease of the brittle-to-tough transition temperature (pp. 243-256)
by ~ 10K for PP matrix material could be found.
2.5r--------------"T""1
CTspecimen b
8.33 . lO-s mg- l
VT =
2
highl oriented
0.5
O~--'--'r-........---r--"---r-........ --I
-90 -60 -30 o 30
TeC)
Fig. AS. Temperature dependence of critical crack-tip-opening displacement (a) and critical
J values (b) for unoriented and highly oriented PP under quasi-static loading condi-
tions
Under dynamic loading, the set of problems of describing the temperature de-
pendence becomes especially obvious. An investigation of the temperature de-
pendence of dynamic fracture toughness was performed on SENB specimens over
the temperature range of -150°C to +20 °C. Calculated according to the LEFM
concept, the dynamic fracture toughness decreases over the whole temperature
range and cannot reflect the real toughness behaviour. In contrast to the results
New Developments 15
under static loading, a qualitative description of the toughness behaviour using the
equivalent-energy concept is not possible. For temperatures T> -50 DC a small
increase can be observed in the experiments, but the toughness level of -150 DC
cannot be reached again. Obviously, the equivalent-energy concept takes insuffi-
cient account of the increasing deformation capability of the material under dy-
namic loading (Fig. A9a).
10~------------'-'b
SENB specimen a SENB specimen
• Kld(a)
16 vH = I mil vH = I ms· 1
• K~d (a.IT ) 8
RPM: Rice, Paris, Merkle
highly oriented
2
unoriented
-150 -100 -50 o 50 -150 -100 -50 o 50

T(DC) T (DC)
Fig. A9. Temperature dependence of fracture toughness according to LEFM and equivalent-
energy concept (a) and of critical J values (b) for unoriented and oriented PP under
dynamic loading conditions
Using the CTOD and the J-integral concepts (Fig. A9b) results in a temperature
dependence of toughness relevant to the material. For unoriented PP, the set of
problems described above are illustrated in Figs. A7 and A9a.
In order to evaluate the fracture safety within the range of validity of the LEFM,
J 1c and J ld values can be transformed into dynamic fracture toughness values ac-
cording to (A5) [2]:
K J . KJ _ (Jlc;Jld)E
Ie, Id - 2' (A5)
I-v
In contrast to the toughness behaviour under static loading (Fig. AIOa), a char-
acteristic temperature dependence can be observed under dynamic loading (Fig.
AIOb), which is characterized by an inverse behaviour of Young's modulus and of
the extent of the plastic zone, as well as by a deformation capability increasing with
increasing temperature.
This leads to the conclusion that in evaluating the temperature dependence of
toughness, the fracture mechanics parameters J 1c and J 1d as well as bic and bid are
appropriate, whereas dynamic fracture toughness values have to be rated as being
unsuitable, independent of the method of calculation (e.g. (A5».
This statement is supported by some other material examples, especially those
described in Part B, 'Morphology-Property Correlations'. So, using the tempera-
ture dependence of the toughness parameters of EPM-modified PA blends, Bethge
16 W. Grellmann
et al. (pp. 243-256) have shown that the process of deformation in specific tem-
perature ranges is load and/or deformation determined, from which the most ad-
vantageous way of using of the J-integral concept can be derived. Simultaneously,
it is vigorously stated that a description of the temperature dependence of tough-
ness using the LEFM concept will not provide material-equivalent results.
CTspecimen
40
VT = 8.33 . IO. S ms"
IOO+---......--r---.-..,--....--r--...--I
-90 -60 -30 o
TeC)
Fig. AIO. Temperature dependence of dynamic fracture toughness K[/ (a) and Ki (b) for
unoriented and highly oriented PP under static and dynamic loading conditions
4.2 Structural Sensitivity of Fracture Mechanics Material Parameters
So far, fracture mechanics parameters for polymer materials have mainly been
used in the field of material development. Thus, one of the most important aims is
rmding quantitative morphology-toughness correlations. However, this is only
possible if the influence of the structure on the crack initiation and crack propaga-
tion behaviour is known. In quantifying these influential quantities, many questions
still remain unanswered.
Using as an example a copolymer with a polypropylene (PP) matrix, the differ-
ent influences of the structure on crack initiation and crack propagation will be
described. PP is a member of the group of semicrystalline polymers and is widely
used as a standard polymer and, to some extent, also as an engineering plastic.
Research projects centred on similar materials within the DFG program named
'Innovationskolleg': 'New polymer materials by purposeful modification of inter-
face structures/properties in heterogeneous systems' at the Martin-Luther-
University Halle-Wittenberg are described by Seidler [10,42], Koch (pp. 257-
266), Beerbaum (pp. 161-180, [41,43]), Casar [44], Lach [45], Han [12,46-48]
and Jung [49].
The main problem with using PP as an engineering plastic consists in its rela-
tively high glass transition temperature, around 4 to 12°C. Below their glass tran-
sition temperatures, semicrystalline polymer materials show insufficient toughness
properties, as the mobility of the molecular chains is very limited. In order to im-
prove the toughness there are strategies, depicted schematically in Fig. All, com-
prising the embedding of a second, 'soft' phase into the PP matrix by compound-
A 1 State of the Art and Development Trends
A 1.1 New Developments in Toughness

Evaluation of Polymers and Compounds
by Fracture Mechanics
W. Grellmann, Merseburg
1 Introduction
The application of fracture mechanics to the estimation of the failure reliability of

products made of polymers and compounds, as well as an evaluation method for
quality control and material development, requires geometry-independent pa-
rameters which react extraordinarily sensitively to structural changes in the materi-
als. An essential prerequisite for a theoretically well-based material optimization is
a knowledge about the connections of strength- and toughness-determined defor-
mation and fracture mechanisms to structural quantities.
For quantification of these connections, it is the task of fracture mechanics ma-
terial characterization to find material-specific parameters by reproducible and, if
possible, also standardized test procedures [1,2]. In applying fracture mechanics
methods to polymers, on the one hand, some basic knowledge of the methods of
evaluation of metallic materials can be used; on the other hand, the substantial time
and temperature dependence must result in further polymer-specific development of
methods (pp. 27-50, [3,4]). Progress in the technological use of fracture mechanics
material testing has yielded a new generation of material parameters [5-8], which, in
combination with morphology analyses, enables structure-related quantitative
morphology-toughness correlations [9-13].
2 Information Potential of Fracture Mechanics Material

Parameters - the Multi-Parametrical Description
of the Mechanical Behaviour of Polymers
The starting point for ascertaining fracture mechanics material parameters are
records of load-load-line displacement curves and load-crack-mouth-opening
displacement curves of notched specimens (Fig. AI).
18 W. Grellmann
these results, it could be stated that in the range of stable crack growth there is an
influence of the interparticle distance on the crack initiation behaviour described
by Odko.2 and J1do. 2 as well as on the energy dissipation capability, while crack ini-
tiation values are, essentially, less influenced than JTJ parameters. This is true for
parameters from J-lla as well as from O-lla curves. The different influences on
energy-determined and deformation-determined parameters show that for very fme
particle morphology the stress component responds more sensitively to changes in
interparticle distance than does the deformation component.
120,-~~~~~~----~--------------~
CD ~~ Id ITJ
(11m) (11m) (N/mm) (N/mm)
i [] 4.1 26
o 3.3 33 1.3
1.3 5
-J
90 7
'?
o ~ 3.0 33 1.5 10
' -'
<> 2.0 33 1.9 21
60 <><> <>
30
O~----~-----r----------~----------~
o 0.4 0.8 1.2
M(mm)
Fig. All. ~Aa curves of a PP copolymer model system studied in order to investigate the
influence of interparticle distance on toughness behaviour
In order to explain the differences in toughness behaviour of the model system

in comparison with standard copolymers, in-situ tensile tests were performed at
semi-microsections [42,50].
Standard copolymers, compared with the model system, showed obvious differ-
ences in deformation and fracture behaviour. Whereas in the model system failure
occurs at the PPIEPR interface in connection with shear deformation, the standard
copolymers show a craze-like failure. This can be traced back to the fact that the
matrices of the standard copolymers are homopolymers with higher crystallinity
and stiffiless than the copolymer matrix of the model system and that the particles
in the standard copolymers are substantially larger than in the model system. Both
of these factors limit the capability of the material to fail by shear yielding. In the
standard copolymers the particles themselves are much more involved in the fail-
ure process. At the start of loading decohesion appears at the EPRIPE interface in
the first place, and with increasing load crazing occurs in the EPR, i.e. the rubber
New Developments 19
disintegrates into fibrils. So far, it is not clear whether this disintegration process is
connected with a reduction of the PE particle diameter. It can be proved that the
particles themselves are not removed directly from the matrix, but rather they are
attached to the matrix by the EPR phase that has disintegrated into fibrils. The
process of crazing in EPR is obviously an additional dissipation mechanism, de-
pending on the thickness of the EPR shell as well as on the dimensions of the em-
bedded PE particles.
Consideration of the rate of change of loading parameters is of special interest
in describing the fracture mechanics behaviour (Fig. AI3).
The crack-tip-opening displacement rate for copolymers reaches a constant
value of around 0.08 m s-', independent of interparticle distance (Fig. Al3a).
After the interparticle distance has fallen below a critical value, a significant
change in the dependence of the J values on the crack-tip-opening displacement
rate appears (Fig. Al3b). For an interparticle distance of2 Ilm, stable crack growth
will start with a higher crack-tip-opening displacement rate at a significantly higher
energy level than for copolymers with a greater interparticle distance.
0.12 6
a b
.CD=4.llIm
..
.CD = 3.3 11m
;.4' ........ t/..• •••

"CD =3.0 11m
-;;- 0.08 4 .CD =2.0 11m
I
! ... t: .... I
-0
• <0 ••• • ...,
0.04 2
o+-----~------~----~ O+-----~~~~~----~
o 0.4 0.8 1.2 0 0.04 • 0.08 0.12
Aa Cmm) !) dkCm/s)
Fig. A13. Dependence of crack-tip-opening displacement rate on crack propagation (a) and
loading parameter (b) for various interparticle centre distances
The crack-tip-opening displacement rate itself, as well as the connection be-

tween the crack-tip-opening displacement rate and the J values, is influenced by
the deformation mechanisms in the material, which are determined by the particle
diameter and interparticle distance again (pp. 281-300).
Considering the results obtained for the amorphous polymer/polymer systems
PC/ABS and TPU/ABS [18], for other copolymers and blends with a PP matrix
(pp. 257-266) and for impact-modified PA materials (pp. 243-256), it can be
proved that for polymer multi-phase systems, changes in morphology affect much
more the energy dissipation capability than the crack initiation behaviour.
The correlations of the fracture mechanics parameters with the main (ex) and
secondary (~) relaxation processes of PMMA, PS and PC described in [36,37],
20 W. Grellmann
showing an extraordinary influence on the temperature dependence of toughness

(Fig. A14), also give evidence for the structural sensitivity of fracture mechanics
parameters. If micro-morphological fracture surface phenomena are included,
material-related statements about microscopic deformation processes are possible.
Figure A14 shows that PC has the greatest relative height of J3 relaxation as well
as a very good separation of ex and J3 relaxation phenomena, which can be ex-
plained by a shear-yielding process affecting a relatively large volume of the
specimen. For PMMA, besides shear, tensile deformation mechanisms are also
responsible for the deformation (pp. 181-192).
5 5
PC PS
00
.§
0.5
00
.§
0.5
~JJ
~
1-39% _ ___
00
.§
O. \=5%
0.05 0.05 0.05
0.005 0.005 0.005

-200 -100 0 100 :ro -200 -100 0 100 nJ -200 -100 0 100 :ro
T("C) T("C) T("C)
40 40 40
21 21 21
~
SIO SIO .1J = 1.25 N'nrn(33 %) SIO
.1J =ON'nrn(O%)
=4
~ e.E 4 b= 4
..., 2
--4
~:9 2 t;:s ...,
t;;g 2 ~
.1J = 3.75N'nrn(100%)
0.4 0.4 0.4
-200 -100 0 100 :ro -200 -100 0 100 nJ -200 -100 0 100 nJ
T("C) T("C) T("C)
Fig. A14. Connection between the height of the p relaxation maximum, the kinds of deforma-
tion observed for impact loading and the height of the p-stimulated increase in
toughness for PC, PS and PMMA
The relatively small height of the J3 relaxation peak for PMMA, being only
39 % of the height of the PC peak, as well as the significantly less well-separated
ex and J3 relaxations compared with PC, can be assumed to be the cause of this
coexistence of shear bands and crazes. PS shows only a weak J3 maximum of
around 5 % of the corresponding height for PC, which is hardly separated from the
ex maximum. For PS this results in athermal fracture behaviour over the whole
temperature range investigated so that there are no maxima of toughness in the
relation of J = f(1). A linear relation can be found between the J3-stimulated in-
crease of toughness M and the relative heights of the relaxation maxima.
New Developments 21
5 Outlook
Fracture mechanics and micro-mechanics have been established in international

research on polymer materials. Over a period of only a few years, a new generation
of fracture mechanics parameters and experimental methods of fracture mechanics
material testing has been created.
Figure A 15 indicates, using as an example two commercial polypropylene stan-
dard copolymers (Borealis AG, Linz, Austria), that the conventional impact tough-
ness does not allow a differentiation of toughness behaviour, whereas fracture
mechanics parameters indicate significant differences in crack initiation and, espe-
cially, in crack propagation behaviour, whereby differences in behaviour in use
can often be explained. It can be seen that a slight rise of molecular weight results
in a tripling of the material resistance against technical crack initiation. The mate-
rial resistance against crack propagation, as well as the part played by energy dis-
sipation processes, is increased even more.
Conventional Impact T oughness Crack Initiation
3.0
2.4
1.8
1.2
0.6
Crack Toughness as Resistance against Stable

Crack Propagation
Crack Initiation and CrackPropagation
E
(R-Curve)
.E
e-. KSC 1012A
M", = 346 kg/mol
5
J = 10.3 t;a0.9S
4
3
MSC 1012
2 Mw = 316kg/mol
J = 3.9 /ja0.97
0~~-4--T-~~--+--r~ __T-~
o 0. \ 0.2 0.3 0.4 0.5 0.6 0 .7 /ja (mm)
Fig. AIS. Impact toughness and crack resistance behaviour of two standard copolymers, KSC
1012 A (Mw = 346 kg/mol) and MSC 1012 (Mw = 316 kg/mol), and schematic rep-
resentation of differences regarding crack initiation and crack propagation behaviour
Important progress in the development of materials and assessment of products

can be expected if it becomes possible to use polymer-specific evaluation methods
in technical testing practice more, as a consequence. Breakdowns in the chemical
22 W. Grellmann
and energy-producing industries resulting from failure of polymer materials indi-

cate that too often only a partial characterization of the materials has been per-
formed, and the safety and quality control systems developed up to now are not yet
sufficient.
Further investigations should cover searching for quantitative morphology-
toughness correlations using damage and micro-mechanics models in order to
enable the effective development of materials (Fig. A16).
Moisture I Tempe:rarure Volume Dilatometry

Time/Rate Dielectrometry
Laser Extensometry
Notch Extension
Load-Line Displacement
Load ! Stre..
Morphology - Property - Correlation
Material Optimization
Fig. AI6. Detennination of quantitative morphology-property correlations by hybrid methods

of testing and diagnostics of polymers
Further methodical investigations concerning the development of hybrid meth-

ods of material testing are necessary to accomplish this. Because of the distinctive
selectivity of the NDT methods listed in the 'Sensors' block, such as acoustic and
thermal emission and videothermography, and because of the choice of variations
within the 'Parameters' block, the aim has to be the determination of structure-related
fracture mechanics material parameters that will enable one to obtain systematic
information concerning the deformation and fracture mechanisms occurring. In
order to understand the deformation processes of heterogeneous polymers, a local
measurement of strain is necessary; this is becoming possible by means of laser
extensometry systems, described in pp. 365-384 and in [51].
The results concerning crack initiation and crack propagation behaviour for the
various amorphous and semicrystalline polymers collected within this book do not
allow a universally valid representation of the physical causes of the ongoing pro-
New Developments 23
cesses yet. Therefore, it has to be the aim of further research to find a theoretically
well-established explanation for the crack initiation and crack propagation proc-
esses using modem concepts of elastic-plastic fracture mechanics, by defming
morphology-toughness correlations. For this, further investigations regarding the
influence of the degree of crystallinity, the molecular weight and the formation of
supermolecular structures on fracture mechanics material parameters are necessary
(pp. 281-300 and 161-180, [43]). Property-determining structure parameters for
stable crack growth are to be found. The evaluation of material-physics processes
within the crack-blunting and crack initiation processes, with special consideration
of stretch zone forming and of in-situ SEM investigations using miniature test
specimens will expand knowledge in this field.
For investigation of fracture toughness of polymers and assessment of the
structural integrity of components, the fracture mechanics of polymers has been
intensively researched during recent years with the aim of developing the field of
application of fracture mechanics parameters. Several fracture mechanics parame-
ters have been proposed for the evaluation of the crack resistance behaviour. In the
first step, lasting until 1985, the determination of fracture mechanics behaviour
was accomplished on the basis of the LEFM and EPFM concepts, and fracture
mechanics parameters related to unstable crack growth were determined. In the
second step, the fracture mechanics assessment was accomplished on the basis of
crack resistance curves. All scientists work all over the world working in this field
now work on this level. The quantitative integration of the crack growth behaviour
is indispensable for an effective application of the fracture mechanics parameters
to material development. Only a few workers consider the time dependence of the
fracture mechanics assessment. A quantity of instinct and spirit will be necessary
to achieve an improvement and an increase of knowledge in the coming years. A
greater use of damage and pre-damage will be an important point.
References
I. Anderson T. L. (1995): Fracture Mechanics. Fundamentals and Application. CRC Press,

Boca Raton
2. Blumenauer H., Pusch G. (1993): Technische Brucbmechanik. Deutscher Verlag fUr Grund-
stoffindustrie, Leipzig
3. Friedrich K. (1989): Application of Fracture Mechanics to Composite Materials. Elsevier,
Amsterdam
4. Akay M. (1999): Fracture mechanics properties. In: Brown R. (Ed.) Handbook of Polymer
Testing. Marcel Dekker Inc., New York: 533-588
5. Kausch H. H. (1987): Polymer Fracture. Springer, Berlin, Heidelberg
6. Williams J. G. (1978): Failure in Polymers: Advances in Polymer Science 27, Springer,
Berlin, Heidelberg
7. Grellmann W., Che M. (1997): Assessment of temperature-dependent fracture behavior with
different fracture mechanics concepts on example of unoriented and cold-rolled polypropyl-
ene. J. Appl. Polym. Sci. 66: 1237-1249
24 W. Grellmann
8. Seidler S., Grellmann W. (1995): Application of the instrumented Charpy impact test to the
toughness characterization of high impact thermoplastics. Polym. Test. 14: 453-469
9. Seidler S., Grellmann W. (1999): Application of crack resistance concept to the toughness
characterization of high impact thermoplastics. In: Kahovec J. (Ed.) Mechanical Behaviour
of Polymeric Materials. Wiley-VCH, Weinheim: 63-71
10. Grellmann W., Seidler S., Jung K., Kotter I. (2001): Crack-resistance behavior of polypro-
pylene copolymers. J. Appl. Polym. Sci. 79: 2317-2325
11. Niebergall U., Bohse J., Seidler S., Grellmann W., Schtlrmann B. L. (1999): Relationship of
fracture behaviour and morphology in polyolefin blends. Polym. Engng. Sci. 39: 1109-1118
12. Han Y., Lach R., Grellmann W. (1999): Effects of rubber content and temperature on ductile
tearing stability in ABS materials. Appl. Macromol. Chern. Phys. 270: 5-12
13. Raab M., Kotek J., Baldrian 1., Grellmann W. (1998): Toughness profile of injection-molded
polypropylene specimens: the effect of the p-modification. 1. Appl. Polym. Sci. 69: 2255-
2259
14. Will P., Michel B., Schaper M. (1990): Justification of nonlinear J-resistance curves. Engng.
Fract. Mech. 37, 2: 275-281
15. Will P. (1994): R-curves of energy dissipative materials. 1. Mat. Sci. 29: 2335-2340
16. Will P., Helbig S. (1997): A local cumulative damage measure for composites. In: Proceed-
ings of MicroMat '97, Berlin, Germany, April 16-18: 348-351
17. Paris P. C., Johnson R. E.·(1983): Fracture Resistance Curves and Engineering Applications.
ASTM STP 803 Vol. II: 5
18. Seidler S. (1998): Anwendung des RiBwiderstandskonzeptes zur Ermittlung strukturbezo-
gener bruchmechanischer WerkstoftkenngroBen bei dynamischer Beanspruchung.
Fortschritt-Berichte, VDI-Series 18: MechanikIBruchmechanik No. 231, VDI, Dusseldorf
19. Grellmann W., Seidler S., Lach R. (2000): Expressiveness of crack toughness as resistance
against the intrinsic rate of fracture mechanics parameters for polymeric materials. In: Pro-
ceedings of the 3rd International Conference on Mechanics of Time Dependent Materials,
Erlangen, Germany, September 17-20: 226-228
20. ASTM E 813-81 (1981): Standard Test Method for J.c. a Measure of Fracture Toughness.
Annual Book of ASTM Standards, Philadelphia, Vol. 03.01
21. ASTM E 813-89 (1989): Standard Test Method for J(c, a Measure of Fracture Toughness.
22. ASTM E 1737-96 (1996): Standard Test Method for J-Integral Characterization of Fracture
Toughness. Annual Book of ASTM Standards, Philadelphia, Vol. 03.01
23. ESIS P2-92 (1992): Procedure for Determining the Fracture Behaviour of Materials.
24. Standard Draft ESIS TC 4 (1995): A Testing Protocol for Conducting J-Crack Growth
Resistance Curve Tests on Plastics.
25. Seidler S., Grellmann W. (1996): Instrumented impact test: determination of crack resistance
curves of polymers. In: Proceedings of the Eleventh Conference on Fracture, Poitiers-
Futuroscope, France, September 3-6, Vol. Ill: 1839-1844
26. Grellmann W., Seidler S. (1999): Possibilities and limits of standards and drafts for JR-curve
determination on polymers. In: Winkler T., Schubert A. (Eds.) Material Mechanics - Frac-
ture Mechanics - Micromechanics. DDP Goldenbogen, Dresden, Germany: 336-341
27. ASTM E 1152 (1987): Test Method for Determining J-R Curves. Annual Book of ASTM
Standards, Philadelphia, Vol. 03.01
28. Grellmann W., Seidler S. (1994): Determination of geometry-independent fracture mechan-
ics values of Polymers. Int. J. Fracture 68: RI9-R22
New Developments 25
29. Seidler S., Grellmann W. (1999): Determination of geometry independent J-integral values
of tough polymers. Int. J. Fracture, Lett. Fracture. Micromech. 96: L17-L22
30. Barry D. B., Delatycki o. (1989): The strain rate dependency of fracture in polyethylene:
fracture initiation. J. Appl. Polym. Sci. 38: 339-350
31. Rimnac C. M., Wright T. M., Klein R. W. (1988): J integral measurements of ultra high
molecular weight polyethylene. Polym. Engng. Sci. 28: 1586-1589
32. Huang D. D. (1989): A comparison of multispecimen J-integral methods as applied to
toughened polymers. In: Proceedings of ICF 7, University of Houston, TX: 2725-2732
33. Lee C.-B., Lu M.-L., Chang F. C. (1992): A new approach in J-integral method based on
hysteresis on ductile polymeric materials. Polym. Mater. Sci. Engng. 669: 510--511
34. Bernal C. R., Frontini P. M. (1992): Fracture toughness determination of ABS polymers
using the J-method. Polym. Test. 11: 271-288
35. Reese E. D. (1996): Zur Anwendung des R-Kurven-Verfahrens der elastisch-plastischen
Bruchmechanik aufPolymere. Herbert Utz Verlag Wissenschaft, MUnchen
36. Lach R, Grellmann W., SchrOder K., Donth E. (1999): Temperature dependence of dynamic
yield stress in amorphous polymers as an indicator for the dynamic glass transition at nega-
tive pressure. Polymer 40: 1481-1485
37. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour of poly (methyl meth-
acrylate), polystyrene and polycarbonate. Appl. Macromol. Chem. Phys. 253: 27-49
38. Seidler S., Grellmann W. (1997): Fracture behaviour and morphology of polymers. In:
Proceedings of the 'Ninth International Conference on Fracture', ICF 9, Sydney, Australia,
April 1-5, Vol. 2: 1021-1027
39. Hoffinann H., Grellmann W., Hille E. (1987): Explorations of the influence of material
orientation of the fracture behaviour of polymers. In: Proceedings of Plastco, Gottwaldov,
Poland, April 7-9: L14/1-L14/6
40. WittF.l., MagerT. R. (1971): Nucl. Engng. Design. 17: 91
41. Beerbaum H., Seidler S., Grellmann W. (1999): Microstructure and toughness behaviour of
high density polyethylene under impact load. In: Proceedings of Technomer '99, Chemnitz,
Germany, November 11-13: P4211-8
42. Starke J. U., Michler G. H., Grellmann W., Seidler S., Gahleitner M., Fiebig I., Nezbedova E.
(1998): Fracture toughness of polypropylene copolymers: influence of interparticle distance and
temperature. Polymer 39: 75-82
43. Beerbaum, H. Grellmann, W. (1999): The influence of morphology and structure on the
crack growth of linear polyethylene. In: Williams J. G., Pavan A. (Eds.) Fracture of Poly-
mers, Composites and Adhesives. ESIS Publication 27, Elsevier Science Ltd, Oxford:
163-174
44. C!isar T., Seidler S., Grellmann W. (1998): Bruchmechanische ZlIhigkeitsbewertung des
RiBinitiierungsverhaltens von Ethylen-Propylen-Random-Copolymerisaten. In: Grellmann
W., Seidler S. (Eds.) Deformation und Bruchverhalten von Kunststoffen. Springer, Berlin,
Heidelberg: 271-284
45. Lach R., Grellmann W., Weidisch R., Altstadt V., Kirschnick T., Ott H., Stadler R,
Mehler C. (2000): Poly(styrene-block-butadiene-block-methylmethacrylat) copolymers as
compatibilizers in PPO/SAN blends, I. Morphology and fracture. J. Appl. Polym. Sci. 78:
2037-2045
46. Han Y., Lach R, Grellmann W. (1999): The Charpy impact fracture behaviour in ABS
materials. Appl. Macromol. Chern. Phys. [Angew. Makromol. Chern.] 270: 13-21
47. Han Y., Lach R., Grellmann W. (1999): Effects of rubber content and temperature on dy-
namic fracture toughness of ABS materials. 1. Appl. Polym. Sci. 75: 1605-1614
26 W. Grellmann
48. Han Y., Lach R., Grellmann W. (2001): Effects of rubber content and temperature on unsta-
ble fracture behaviour in ABS materials with different particle sizes. J. Appl. Polym. Sci.
79:9-20
49. Grellmann W., Jung K., Kotter I. (2000): Zusammenhang zwischen Morphologie und Ziihig-
keit von Polyolefinblends unter besonderer Beriicksichtigung des Rissinitiierungsprozesses.
In: Michler G. H. (Ed.) Innovationskolleg 'New polymer materials by purposeful modifica-
tion of interface structures/properties in heterogeneous systems': 105-117
50. Kim G.-M., Michler G. H., Gahleitner H., Fiebig J. (1996): Relationship between morphol-
ogy and micromechanical toughening mechanisms in modified polypropylenes. J. Appl.
Polym. Sci. 60: 1391-1403
51. Grellmann, W., Bierl)gel, C. (1999): Verfahren, Lichtextensometer und PrUfkl)rper zur
berUhrungslosen Bestimmung der Querkontraktion. German Patent P19936249.1 (31.07.99)
A 1.2 Concepts of Fracture Mechanics
for Polymers
F. Ramsteiner, W. Schuster and S. Forster, Ludwigshafen
1 Introduction
The concepts of fracture mechanics received increasing interest during the last 60
years, when large objects such as ships, tanks and jets suddenly fractured in serv-
ice, although on the basis of the conventional data for stiffuess and strength this
failure should not have happened [1]. Microdefects and flaws in the materials have
been identified as the origin of this unexpected failure. These defects, which in
nearly every material grow to an over-critical size during use of the item, finally
become unstable and initiate a catastrophic fracture. Consequently, it has become
urgent to study the instability of cracks, which is the subject of fracture mechanics.
The basis of this branch of materials science was developed in the 1920s by
Griffith [2]. However, this development scarcely had an effect in engineering sci-
ence until the actual situation called for a change. Against this background, it is
not surprising that the standards of fracture mechanics have been developed
mainly for metals.
A few years ago the European Structural Integrity Society (ESIS) was asked to
found a task group with the aim of studying the application of fracture mechanics
to polymers. This group, called Technical Committee 4 (TC4), cooperated inten-
sively with the institutions for standards in the United States (ASTM) and Japan
(JIS), as soon as an ISO standard was believed to be feasible.
In the field of polymer systems, the first progress was made successfully in
long-fibre-reinforced composites, where the methods of fracture mechanics are
routinely applied today. These standards are not treated in the present chapter.
The focal point of the present chapter is to give a brief survey of fracture me-
chanics for polymers, especially of methods which are being studied at the moment
within ESIS TC4. These measurements are complemented by some of our own
measurements. For details, the ESIS protocols or the publications cited here must
be consulted.
2 Outline of Principles of Fracture Mechanics
In fracture mechanics, the natural undefmed micro-flaw is simulated by a macro-

scopically machined, well-defmed notch in the specimen. The instability of this
notch is characterized during deformation (Fig. AI7).

28 F. Ramsteiner, W. Schuster, S. Forster
Crack Instability
J-Integral (non-linear-elastlc range) Crack-Opening Displacement

J ~------------~~ 0
Fig. A17. Outline of fracture mechanics
Within linear elastic fracture mechanics (LEFM), the tensile stress increases
linearly with the strain, followed by an abrupt drop of the load at the instant of
crack growth initiation. The beginning of failure of the notch can be described
equivalently either by the stress at failure or by the deformation energy up to this
point.
If we consider the path of the deformation energy, we use the original idea of
Griffith. The notched specimen can sustain only deformation energy which is not
able to extend the notch by energy release. The maximum critical energy at crack
instability is called the critical energy release rate, Gc:
2
G =_.!. dU =~ dC =~ (A6)
c B da 2B da BW,p'
where U is the deformation energy, B is the thickness of the specimen, a is the
notch length, Wis the width of the specimen, C is the compliance of the specimen,
Fi is the force at failure and ,p is a geometrical factor (values for this factor are
given in [3,4]).
When characterizing the crack instability by the stress, the possibility of sepa-
rating analytically the total stress field ahead of the tip of a sharp notch into a pure
geometrical term [3] and a solely stress-dependent term is used. The stress-de-
pendent term ahead of the notch is given by the stress intensity factor K:
Concepts of Fracture Mechanics for Polymers 29
K =CTj fa, (A7)

where Ol is the external stress. This is called the critical stress intensity factor Ke in
the case of failure. Since these values are evaluated by using fmite specimens, a
geometrical factor ttalW) must be included, leading to the fmal relation
K = F j ttalW) (AS)
e B../W
A list of the values of the geometrical factors ttalW) is given in the literature
[3-5]. For example, the most commonly used specimens in fracture mechanics
investigation start with an initial notch, the depth of which is half the width of the
specimen (aIW= 0.5). In this case,
ttO.5) = 10.65 (A9a)
for the single-edge-notched bend (SENB) specimen and
ttO.5) = 9.66 (A9b)
for the compact tension (CT) specimen.
The descriptions of crack instability by stress and energy are physically
equivalent in linear elastic fracture mechanics. For example, in the case of plane
strain and in crack-opening mode I, the following relation holds:
2
Gle = K~e (1-v2) (AIO)
where Gle is the critical energy release rate in crack-opening mode I, Kle is the
critical stress intensity factor in crack-opening mode I, E is the Young's modulus
and vis the Poisson's ratio.
Since even small non-linear contributions make this relation invalid, this cross
check of the two fracture mechanics values Gle and Kle can only be used as a rough
estimate to judge the consistency.
In the case of pronounced non-linear behaviour, the route given by the energy
release rate G in linear elastic fracture mechanics must be substituted by [6] a
method based on the J-integral,
J=J(WdY-T.:ds), (All)
r
where w is the work done by the load per unit volume, r is the path of the loop
integral connecting the lower to the upper crack surface, enclosing crack tip, T is
the vector of the stress on the contour path, ds is the increment on the contour path,
u is the displacement vector at ds and x, y are rectangular coordinates.
The value of this integral describes, in the end, the transport of energy into the
region surrounded by the integration line minus the stored part of the energy. The
released energy is dissipated during crack propagation and its precursor. The de-
termination of the critical value for crack instability follows from the relation
J=-! dU (A12)
B da '
where U is the deformation energy, equal to the area beneath the force-displace-
ment line. In the case of linear behaviour the J value coincides with the energy
release rate G.
The alternative to the stress intensity factor, which is appropriate to linear be-
haviour, is given by a strain criterion for non-linear materials. In this case the ulti-
mate deformation of the material at the tip of the crack determines the crack insta-
bility. This maximum stretching of the material, in form of fibrils at the notch tip,
can be observed in transparent materials by the expensive technique of optical
interference, as described by Weidmann and Dt>ll [7] in their pioneering work on
PMMA. Usually the value of the ultimate deformation is measured macroscopi-
cally in an indirect way from the crack-opening displacement (COD). At the mo-
ment, this procedure is being considered by the ESIS TC4 group.
3 Linear Elastic Fracture Mechanics
The following drafts or publications on fracture mechanics for linear behaviour of

polymers have been or are being produced by ESIS TC4:
• Plastics. Determination of Fracture Toughness (Gc and Kc) - Linear Elastic
Fracture Mechanics (LEFM) Approach - ISOIDIS 13586.2 (2000-03) [4]
• Williams, J. G., Cawood, M. J. (1990): European Group on Fracture: Kc and
Gc Methods for Polymers. Polym. Test. 9: 15-26 [8]
• Plastics. Determination of Fracture Toughness (Klc and G\c) at High Loading
Rates - Draft
• A Protocol for the Measurement of Toughness for Injection Moulded Discon-
tinuous Fibre Reinforced Composites - Draft
• Plastics - Determination of Tension Fatigue Crack Propagation - Linear
Elastic Fracture Mechanics (LEFM) Approach - ISOIDIS 15850 [9].
The advantage of fracture mechanics is its possibility to characterize crack in-
stability in a way independent of the shape of the specimen. Real material proper-
ties are measured. For demonstration, in Table A2 the measured strengths of two
commonly used types of specimens in fracture mechanics investigations are com-
piled for four different materials and compared with the predictions of fracture
mechanics. For example, a 6 mm thick CT specimen of epoxy breaks at a force
F = 96.7 N, leading to a critical stress intensity factor Klc = 0.66 MPa mIl2 accord-
ing to (A8). With this value, the failure of the SENB specimen is predicted to
occur at Fcal = 52.4 N, which corresponds rather well to the measured value Fexp
= 51.6 N. In the same way, this correspondence is apparent for PI, and with some
reservations also for PA 66. However, for the pronouncedly non-linearly deform-
ing PVC, a critical stress intensity factor cannot be given within the framework of
linear fracture mechanics.
Table Al. Prediction and experimental results for the failure of differently shaped specimens
Material
tJ R B
(mm)
F
(N)
Klc
(MPam I12)
Fcal
(N)
Fexp
(N)
Epoxy resin EP 6 96.7 0.66 52.4 51.6

Polyimide PI 4 81.0 0.81 40.5 38.0
Polyamide PA 66 12 895.0 4.07 634.0 661.0
Poly(vinyl chloride) PVC 10 2220.0 5.39 1113.0 890.0
;~'I!'
...
~
q~
~~
r.
~
10
Cyclic Stress Intensity Factor 8K (MPa m 112)
Fig. AIS. Crack propagation in PS-lll during fatigue test
The dynamic crack propagation during a vibration test is usually followed in CT

specimens, according to the standard for fatigue [9]. In Fig. AIS, the incremental
crack propagation speed, as a function of increasing amplitude of stress intensity
factor, is plotted for PS-HI. The test was performed at room temperature at a fre-
quency of 10Hz. Within the time of observation, no crack propagation was ob-
served up to a stress intensity amplitude of 0.4 MPa mll2. Above 2 MPa ml12 un-
stable crack propagation occurred. The fracture surfaces shown in Fig. A19 reveal
a change in failure mechanism. For slow crack growth the individual rubber parti-
cles are seen; they are broken or disentangled, as well as the matrix. At high crack
growth only an intensely deformed fracture surface is observed with PS-HI; the
molecules have no time to disentangle. The advantage of this dynamic fracture
mechanics test compared with the classical, time-consuming WOhler curves is its
possibility of obtaining information very quickly, within a few hours, on the notch
sensitivity of a material.
Fig. AI9. Fracture surface of PS-HI (test shown in Fig. A18) (a) slow crack propagation, (b)
fast crack propagation
14
-------- • • •
12 •
10 L annealed
8 • j
f •
~ 6
i
IV I:
4
~ ~
~
2 injection moulded
0
o 2 4 6 8 10
a (mm)
Fig. A20. Notched impact strength of ASA with different initial notch depths
a (mm)
876 5 4 3 2
350
annealed
300
250
., 200
S.
:;)
150
100
50
0
0 4 8 12 16 20 24
BWIlI (mnf)
Fig. A21. Results of Fig. A20 plotted according to the method of data reduction used in frac-
ture mechanics
The methods of fracture mechanics are also used to characterize the impact be-
haviour of a polymer in a material-specific way, either by energy or by stress. In
Fig. A20 the specific fracture energies of different deep-notched specimens of
ASA, measured with the Charpy arrangement, are plotted as a function of crack
length. Clearly, these values are not independent of the initial notch length. No
pure material property can be defined. However, if the same results are plotted
according to (A6) within the framework of fracture mechanics (Fig. A21), the
behaviour becomes clearer. The slopes of the straight lines in Fig. A21 are given
by the energy release rate. It follows from the results in Fig. A21 that the energy
release rate of this material is higher in the case of annealed specimens than in
specimens after injection moulding, with a higher orientation of the molecules.
Also, long notches with short ligaments of material below the notch tip yield
higher energy release rates than specimens with short crack lengths. The latter
result is known from the literature [10]. The deformation curves of the annealed
ASA specimens reveal (Fig. A22) that the specimen with the longer crack length
of 5 mm deforms much more after the peak of stress than the specimen with the
short initial crack length of 3 Mm. If only the deformation energy up to the peak
force is used for the data reduction by the methods of fracture mechanics, a single
energy release rate ~c = 11.7 kJ/m2 is obtained over the whole range of crack
lengths (Fig. A23). In the force-deflection diagrams a small non-linear region
cannot be excluded. Owing to the high oscillation of the force signal in impact
testing, the limit of validity of this method cannot be determined unambiguously,

because the criterion for validity, which corresponds to the relation
FmaxIFs% offset < 1.1, (A 13)
where Fmax is the peak force and Fs% offset is the force at 5% offset, cannot be
checked when the force signal shows scatter. If this condition of 10% non-linearity
is violated, the test should be rejected. Nevertheless, in such cases the methods are
very often used as a rough assessment. However, the results should be considered
as an approximation.
250 a =3mm a =5mm

BW+ = 15.5 mmz BW+ = 9.52 mmz
UTe.., = 196 mJ UTotol = 257 mJ
200 UP••k = 139 mJ
g 150
~ 100
50
3 4 6
Deflection (mm)
Fig. Al2. Force-deflection diagrams of 3 mm and 5 mm deeply notched specimens of ASA

(annealed)
300~-------------------------.
GQ = 11.7 kJ/m 2
250+---------------
o '
........ 200
...., --------------.r:tf.--
.-
E
......
...• 150 8 .-
...•
:::>
.:~
100
50 .- 0
.- '
0
0 4 8 12 16 20 24
BW<ll (mm 2)
Fig. A23. Fracture mechanics plot for ASA (annealed) taking into consideration only the
deformation energy up to the peak force
If it is intended to characterize the impact behaviour not by the energy as de-

scribed above but by the stress intensity factor, the peak stress must be clearly
defined. Therefore, the unwanted vibration superimposed on the record of the
force should be damped by a thin viscoelastic foil between the falling hammer in
the test equipment and the specimen. Figure A24 shows the influence of the thick-
ness of this damping pad on the force-deflection record during testing of a rubber-
modified PMMA. In contrast to the energy, the maximum force scarcely seems to
be changed by the foil. This procedure facilitates the measurement of the peak
stress, and consequently, according to (A8), the calculation of the stress intensity
factor. In conclusion, the characterization of crack instability by the energy release
rate starts with the undamped force-deflection line; alternatively, the contribution
of the deforming damping foil should be taken into consideration, according to the
proposed procedure given in the ESIS protocol. For the calculation of stress inten-
sity factors of brittle materials, damping to reduce the oscillation is recommended.
0.78mm
1------1
o 0.3 0.45 0.65 0.9 1.1

Thickness of the Pad (mm)
Fig. Al4. Force--detlection diagrams in impact test (3 mls) of rubber modified PMMA with
different thicknesses of damping pads
t
a
t Interface
tEJ .
•
~
L--1J
o +
Fracture I
Surface
:=O~----11"l lnte+rfacego
~L-0____~~
'r:-r------,~ •
Fig. AlS. Procedure (schematic) for preparing specimens for measurement of interfacial stress
after welding
The methods of fracture mechanics are also often applied in characterizing the
interfacial strength of interfaces between polymers after welding. In this way, not
only the role of the welding conditions but also the influence of modifiers such as
compatibilizers or coating agents on the welding process can be investigated. In
relation to this topic, only one example, about the compatibility of polymers, is
described in this chapter for demonstration. The final CT specimen is formed (Fig.
A25) from two half pieces of the two polymers which are to be tested. These two
pieces are welded together at a temperature above the melting point or above the
glass transition temperature in the case of amorphous polymers, to make interdif-
fusion of the molecules possible [II]. After welding, the interfacial strength is
measured. This residual strength characterizes the intensity of interdiffusion of the
molecules, and in the end the compatibility of both polymers. The experimental
results are given in Fig. A26. The numbers between the large circles corresponding
to the various polymers are the interfacial strength. Within the circle for each
polymer, the left-hand number is the original strength of the polymer, and the
right-hand number gives the strength after annealing of the fracture surface of the
same polymer. For example, PS and SAN are not compatible after annealing at
160°C for two hours; the residual strength is zero. In contrast to this pair of poly-
mers, some compatibility is observed between SAN and PMMA, where a residual
strength of21 N/mm 2 is measured.
Going to the limits of fracture mechanics, some information can also be derived
on the morphology of foams. In the case of plane stress, the equation correspond-
ing to (AI0) for plane strain is
Material
Interfacial Stnlngth
after Welding
Origi1a1 Strength
Fig. A26. Interfacial strengths (N/mm 2) between different polymers after annealing in contact
for 2 hours at 160°C
Concepts of Fradure Mechanics for Polymers 37
Kr = JE r Gr , (A14)
where Kf is the stress intensity factor, Ef is the Young's modulus and Gf is the
energy release rate; the index f refers to foam.
According to Gibson and Ashby's model [12], the Young's modulus of the
foam Ef correlates with the modulus of the matrix Em and the ratio of the densities
of the foam and the matrix. In the case of a closed-cell structure, when only la-
mellae are present and are stretched, it follows that
Er Pr (AlS)
Em = Pm'
where P is the density; the indices f and m refer to foam and matrix, respectively.
For open-cell foams with nothing but struts,
Er = (pr )2 (AI6)
Em Pm
is derived.
With the assumption that the energy release rates of the foam Gf and of the ma-
trix Gm are given by the density ratio
Gr=G m pr, (AI7)

Pm
it follows, with (AlS), that for the apparent stress intensity factor KQ in the case of
a closed-cell foam is given by
KQ = JEm G m (;:} (A18)
for the open-cell structure, with (A 16),
KQ =JEm G m (;: r/2

(A19)
The term 'apparent stress intensity factor', with the index Q, is chosen because
the numerous approximations and the non-linear deformation behaviour of foams
up to fracture in the SENB test are not in line with linear fracture mechanics. In
Fig. A27, apparent stress intensity factors calculated for closed-cell polystyrene
foams and an open-cell foam of melamine are plotted as a function of foam den-
sity. Additionally, the correlations according to (A18) and (A19) are shown in the
figure. The experimental value for the pure polystyrene matrix Kc = (Em Gm>1I2 =
1.2 MPa m112 has been used as a parameter in these equations and, owing to a lack
of measurements on pure melamine, this value for PS has also been adopted for the
melamine matrix. As can be seen in Fig. A27, all the closed polystyrene foams
follow roughly the prediction for closed foams, and the open-cell melamine foam
meets the prediction for open-cell foams.
38 F. Ramsteiner, W. Schuster, S.Forster
• S-Foam
0.10 • S-Foam Extruded
• Foam of Melamine Resin
0.08
0.02
o 10 20 30 40 50 60 70 80
Fig. A27. Apparent stress intensity factor of closed-cell PS foams and open-cell melamine
foam; the predictions of(A18) and (A19) are also shown
The application of fracture mechanics to short-fibre-reinforced polymers is dif-

ficult because the material is anisotropic. The orientation of short fibres in injec-
tion-moulded specimens is not isotropic, and therefore it is unlikely that a single
toughness value exists. In the centre of the specimen, the fibres are preferentially
oriented perpendicular to the direction of injection, but near the surface of the
specimen they are more or less parallel to it [13]. The strength of a notched speci-
men depends on the spot where the tip of the notch is situated. To address this
difficulty, the ESIS protocol for short-fibre-reinforced polymers, mentioned at the
beginning of this chapter, requires at least two different thicknesses of injection-
moulded specimens. The stress intensity factor is measured with specimens paral-
lel and perpendicular to the injection direction as usual. These values are plotted
as a function of the plane part of the fracture surface area. The critical stress inten-
sity factor is now defmed as the value where 50 % of the fracture surface is
smooth; this means that the crack has propagated over 50 % of the fracture surface
within a region parallel to the fibre orientation. Figure A28 shows, using short-
fibre-reinforced P A as an example, the results from an ESIS round robin test.
All the examples given in this chapter are related to mode I, which means the
crack-opening mode (Fig. A29). This mode is the most important in practice since
it is the most sensitive to crack instability. In mode II, in which the specimen is
sheared macroscopically along the plane of the notch, failure in the plane of the
notch is not observed in polymers, in contrast to composites, where the crack
propagation is guided by the long fibres. In polymers the crack propagates at an
inclination of 70° to the initial notch surface, which means that the crack extends
in the plane where the normal stress near the tip of the notch is a maximum [14].
Therefore, on the microscopic level, the failure occurs locally more or less in mo-
de I. If crack propagation is forced to occur in the plane in front of the initial notch
by side grooves, yield strength, not a fracture mechanics parameter, dominates
failure [15]. At the moment, there are no activities within the ESIS group TC4 in
the field of mode II or III for polymers. In this chapter, therefore, the index I indi-
cating mode I is omitted in most relations.
12
10
C!~
-e 8 •
!U 0
a.. 6 0
~
':i:." 4
•
2
0
0.0 0.2 0.4 0.6 0.8 1 .0
Smooth Fraction
Fig. A2S. Kc plotted against smooth fraction of the fracture surface are of 2 (.) and 5 (0) mm
thick injection-moulded specimens of glass-fIbre-reinforced PA, measured parallel
and perpendicular to the injection-moulding direction (example from the ESIS proto-
col)
Mode I Mode II
Mode II
Fig. A29. Failure modes in fracture mechanics
4 Non-linear Fracture Mechanics
The relevant ESIS Protocols are as follows:

• A Testing Protocol for Conducting J-Crack Growth Resistance Curve Tests
on Plastics. Draft.
• Crack Growth Resistance Curve Tests Under Impact Loading. Draft.
The non-linear deformation behaviour is caused by the extended plastic defor-

mation of polymers at the crack tip before crack propagation, and by the visco-
elasticity of the material. An idealized illustration of this type of crack propagation
within those materials is shown in Fig. A30 [16]. In the first step of deformation
the crack tip is blunted, whereby the stress intensity at the tip is significantly re-
duced by yielding of the material. Therefore, in this region the J values are likely
to follow the 'blunting line'
J=2oyfla, (A20)
where fla is the amount of stable crack growth and oy is the yield stress, succeeded
by real crack propagation with only a minor increase of the J value with Il.a.
Rearrangement of (A12) and inclusion of the analytical expression for the com-
pliance lead to
J = TJU , (A21)
B(W -ao)
where TJ is a geometrical function (equal to 2 for an SENB specimen, and
2 + 0.522(1 - aolw) for a CT specimen) and ao is the initial crack length.
J / ' Blunting Line
J c ----- •/
~~4-4-------~a
~ I rlnitial Crack
~ f-- Crack-Tip Blunting and
A~ I Stretch Zone Formation
i f
~- Stable Crack Propagation
Fig. A30. Idealized illustration of crack propagation in non-linearly deforming materials [16]
Owing to the high plastic deformation around the notch tip, the whole external
deformation energy U is no longer transformed into crack propagation energy
alone as in the case of elastic behaviour. To obtain a critical quantity for crack
instability in practice, the whole deformation energy is used to evaluate the J val-
ues according to (A21), and these values are plotted as a function of crack propa-
gation for each step. In most cases, and in the ESIS protocol, the J values as a
function of the total increase of crack length are approached not by two straight
lines, as shown in Fig. A30, but more realistically by a power law,
(A22)
c) and C2 are fitting parameters. A crack initiation value JO.2 is defined which
measures the fracture resistance at 0.2 mm of total crack growth.
The main experimental difficulty of this method is to measure exactly the small
increase in crack length during deformation. If the multiple-specimen method, as
described in the ESIS protocol, is used, various geometrically identical specimens
are deformed up to different strains. By integration of the force-deflection curve,
the energy is evaluated. The specimens are then broken after the deformation, e.g.
in liquid nitrogen if the measurements were performed at room temperature. The
different conditions for fracture and the preceding deformation make it easy to
differentiate, on the fracture surface in the microscope, between the initial notch
front, the line of the increased crack length after deformation and the post-fracture
surface of the specimen.
60 Blunting Line (-20°C) Blunting Line (23°C)
50 •
.......... 40 .0
~
-,
~
-, 30
20
10
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
da (mm)
Fig. A31. J-lla curves ofMDPE at-20 °C and 23°C
Figure A31 shows some data from an experiment on MDPE. The J values ob-
tained according to (A2I) are plotted as a function of the increase of crack length
at two test temperatures. For simplicity, the experimental data points are approxi-
mated by straight lines. Also, the blunting lines are inserted in the figure for both
temperatures.
At -20 °C the J values increase with crack growth at low deformation mainly
according to the blunting line, up to J"" 23 kJ/ml. At higher deformation some real
cracking seems to occur with smaller J values. Figure A32a shows a side view of
this specimen, with the crack partly bridged by fibrils.
For 23 °C, however, the experimental data points fit the corresponding blunting
line over the whole range of experiments. Obviously, at this temperature, no crack
propagation occurs in the sense of fracture mechanics. The deformation is deter-
mined by the yield strength. In the side view of this specimen which is shown in
Fig. 32b, clearly only stretching is seen, with some crazes. No clear crack is dis-
cernible.
Notch
Fig. A32. Side view of a notch tip ofMDPE after deformation at (a) -20°C and (b) 23°C
The mUltiple-specimen method described above is very time-consuming owing

to the numerous specimens which must be tested and analysed. More economic
methods are required. ASTM E 1152 [17] describes, for metals, a single-specimen
method by which the crack propagation is evaluated from the compliance of the
specimen when unloaded and reloaded. This method seems, however, doubtful for
polymers, with their strong relaxation behaviour. Zhou et al. [18] propose for the
single-specimen technique an analytical expression with three parameters L, M and
N for the compliance, from which the crack propagation is derived. Within this so-
called LMN technique, the analytical expression must be fitted at three points,
involving some iteration, which also entails some new problems. An example of
the application of this method to amorphous PVC is given by Che et al. (pp. 121-
132, [19]).
J values can be measured in impact for example with the instrumented falling-
weight impact test (IFWIT). A piston falling from the same height (constant im-
pact velocity) with variable mass (variable deformation energy) on an SENB
specimen deforms it [20]. The kinetic energy of the falling piston must be lower
than the fracture energy. A schematic illustration of the equipment required is
shown in Fig. A33. The data analysis is the same as described above for the case
of low defonnation rates. The energy is given by the kinetic energy; the crack
propagation is measured on the fracture surface in the microscope. The results of
some experiments on ABS are shown in Fig. A34. The thin lines in the diagram
delimit the validity region of experiments according to the ESIS protocol. The
critical value for crack instability yields JO.2 = 6 kJ/m 2 • The crack propagation is
shown in side view and on the fracture surface on the left-hand side of Fig. A34.
Masses
Inductive Linear
Position Transducer
Falling Weight
I...."r----_,--In-' Specimen
o Q. Load Cell
Fig. All. Falling-weight apparatus (schematic) for measuring JO.2 values in impact
In a different, very popular method the Charpy assembly is used. By changing

the height :from which the hammer falls, the defonnation energy can be varied;
however, the defonnation speed is not constant in this case. To overcome this
shortcoming, the 'stop block' method has been developed [21], where the defor-
mation is limited by stopping the falling hammer near the specimen with a metallic
bar which can be shifted slightly relative to the specimen. In all the methods for
impact testing mentioned and described above, the mUltiple-specimen method is
applied.
m =100 g; v =2 mls ; J =7.9 kJ/m'; Aa =0.35 mm

M40:1
20
18
16 1st Impact
14
1,0
12
o
8
0+i--~~~~~L-~--~~--~--4
o 0.2 0.4 0.6 0 .8 1.2 1.4
Crack front Aa (mm)
Fig. A34. J-lla curves of ABS after impact, with a side view and a view of the fracture surface
of a specimen
5 Current Activities in ESIS TC4
The following activities are still in a preliminary state.
5.1 Essential Work
• ESIS Protocol: Test Protocol for Essential Work of Fracture. Draft.

The fracture mechanical procedures as described above should preferably be ap-
plied to thick specimens, which are in a plane strain state. The dimensions of the
specimens should be large compared with the extent of the yield zone, which is
described by Dugdale [22]. The geometrical restrictions on the specimen which
must be observed in order to satisfy this condition have not been discussed in this
chapter, because they are widely given in the protocols. In very thin specimens like
foils, the plane stress state must be dealt with [23].
One method that can also handle thin specimens like foils within the framework
of fracture mechanics is the essential-work procedure [24-29]. The measurements
are performed with deeply double-edge-notched tension (DDENT) specimens
(Fig. A35a). Specimens with various ligament lengths are fractured. The fracture
energy is plotted as a function of the ligament length (Fig. A35b). The data reduc-
tion method is based on the assumption that the whole fracture energy can be di-
vided into two parts. One part is proportional to the ligament length and describes
the energy required to fracture the corresponding plane. This portion is called the
essential work. The other part is proportional to the second power of the ligament
length and refers to the deformation in the extended plastic zone surrounding the
ligament length:
(A23)
where Wf is the fracture energy, We is the essential work, Wp is the plastic work, I is
the ligament length, B is the specimen thickness and P is the shape factor of the
yield zone. Rearrangement of (A23) leads to:
(A24)
If the fracture energies are plotted in a linear way according to (A24), the
crossover of the straight line with the ordinate yields the value of the essential
work We. The slope of the straight line is given by the energy dissipation WpP in
the plastic zone. As an example (Fig. A35b), Karger-Kocsis [24] has shown with
this method that J3-nucleated PP is much tougher, owing to higher dissipated en-
ergy in the yield zone, than the normal uPP, whereas the essential work, which
means the fracture energy for breaking the ligament, is identical in both materials.
The background for this different behaviour is given by Raab et al. (pp. 153-160,
[30]).
oL----L----L---~--~----~--~
o 5 10 15 20 25 30
Ligament Length I (mm)
Fig. A3S. DDENT specimen used to evaluate essential work data for foils (a); data for a- and
J3-nucleated pp (b) [24]
The extrapolation has some shortcomings in its precision. By rearranging (A23)

according to
(A25)
the essential work is given by the slope when Wf //2B is plotted as a function of W .
But in this case the physical meaning of the terms is a little bit hidden at first glance.
5.2 Energy of Adhesion
• ESIS Protocol: Determination of Mode I Adhesive Fracture Energy G1c of

Structural Adhesives Using the Double-Cantilever Beam (DCB) and the Ta-
pered Double-Cantilever Beam (TDCB) Specimens. Draft.
Load F
h
Adhesive
/
a /'
/
Load F /
/
/
h
Fig. A36. DCB (top) and TDCB (bottom) specimens used to measure the strength of adhesives
1600~------------------------------------------~
Toes Specimen
1200
~ 800
u.
Pre crack
Growth
400
O~~----'--------r------~------~------~
o 0.5 1.5 2 2.5
I (mm)
Fig. A37. Force-extension diagram of an adhesive-joint TDCB specimen

This method described in this protocol should allow one to characterize the
fracture resistance of a structural adhesive joint in a material-specific way. The
tapered double-cantilever beam (TDCB) specimen has, in contrast to the standard
double-cantilever beam (DCB) specimen (Fig. A36), the advantage that during
deformation the stress intensity factor and the energy release rate do not become
dependent on the crack propagation [3\]. A geometrical influence on these frac-
ture mechanical quantities can be excluded. The first round-robin test with these
methods has just started. Therefore, the evaluation of this method has not finished
yet. Figure A37 shows as an example the adhesive-force-deflection diagram of a
metal TDCB specimen bonded with an adhesive. The force required to propagate
the crack is independent ofthe crack length [3\].
5.3 Peel Test
• ESIS Protocol: A Protocol for Determination of the Interfacial Work of Frac-

ture by Peel Testing of Flexible Laminates. Draft.
Fig. A3S. Peel test jig to change the angle of the draw direction
The method described in this protocol is being developed to measure the adhe-
sion between adjacent layers within laminates of foils [32]. The standard peel test
does not take into consideration that part of the deformation energy which is dissi-
pated within the curved part of the strip near the root of the propagating crack
where one layer is drawn off from the rest of the laminate. This portion of the
measured peel energy has nothing to do with the adhesion between the layers
within the laminates. Therefore, the measured peel energy becomes dependent on
the peel angle and is no longer given only by the adhesion. The ESIS protocol
intends to describe a data reduction method to avoid this deficiency. To test this
procedure some equipment (Fig. A38) to change the peel angle has been recom-
mended, so as to make it possible to prove the independence of the peel energy
from the peel angle. One part of the laminate is fixed on a table, which can be
turned relative to the tensile direction of a tensile-testing machine to change the
draw angle. The other part of the laminate is pulled. The preliminary results have
shown that the energies are roughly independent of the angle, except at very small
or very large angles. Further theoretical and experimental work is still to be done.
6 Conclusion
The aim of fracture mechanics is to develop methods for characterizing the initial
instability of macroscopic cracks and to defme quantities which are only depend-
ent on the material properties and not on the geometry of the specimen. ESIS has
tried to test and modify those methods for polymeric systems. Within the linear
and non-linear range of deformation, parameters such as the stress intensity factor,
energy release rate and J-integral have proved to meet this goal when applied in
the range of validity. For each of these tests, the ESIS protocols are mentioned in
this chapter and should be consulted for details. Some additional experiments have
been added to show the wide application of fracture mechanics. Some activities
have not been fmished yet.
Acknowledgement
The authors gratefully acknowledge helpful discussions with and scientific input
from the members ofESIS TC4.
References
1. Marder M., Fineberg J. (1996): How things break. Phys. Today, September 1996: 25
2. Griffith A. A. (1920): The phenomena of rupture and flow in solids. Phil. Trans. R. Soc.
A221: 163-198
3. Williams J. G. (1980): Stress Analysis of Polymers. Ellis Horwood
4. ISOIDIS 13586.2 (1998): Plastics - Determination of Fracture Toughness (Gc and Kc) -
Linear Elastic Fracture Mechanics (LEFM) Approach
5. Hahn H. G. (1970): Spannungsverteilung an Rissen in festen Korpern. VDI-Forschungsheft
542: 1-46
6. Rice J. R. (1968): A path independent integral and the approximation analysis of strain
concentration by notches and cracks. J. Appl. Mech. 35: 379-386
7. Weidmann G., Doll W. (1976): Interferenzoptische Vermessung der Craze-Zonen vor der
RiBspitze in PMMA. Colloid Polym. Sci. 254: 205-214
8. Williams 1. G., Cawood M. J. (1990): European group on fracture: Ko and Go methods for
polymers. Polym. Test. 9: 15-26
9. ISOIDIS 15850 (1999): Plastics - Detennination of Tension Fatigue Crack Propagation -
10. Newmann L. V., Williams J. G. (1978): The impact behaviour of ABS over a range oftem-
peratures. In: Proceedings of the International Conference on Toughening of Plastics, Lon-
don, July 1978: 18.1-18.8
11. Ramsteiner F. (1996): Method to evaluate the strength of interfaces in polymeric systems.
Polym. Test. 15: 573-584
12. Gibson L. J., Ashby M. F. (1988): Cellular Solids. Pergamon, Oxford.
13. Heckmann W., Ramsteiner F., Theysohn R. (1980): Effects influencing the fibre concentra-
tion dependence of tensile strength of unidirectional, short-glass fibre reinforced thenno-
plastics. In: Proceedings ofICCM 3, Advances in Composite Materials, Paris (France): 95-
103
14. Erdogan F., Sih G. C. (1963): On the crack extension in plates under plane loading and
transverse shear. J. Basic Engng.: 519-527
15. Ramsteiner F. (1993): An approach towards understanding mode II failure in PMMA. Poly-
mer 34: 312-317
16. Hashemi S., Williams 1. G. (1986): A fracture toughness study on low density and linear low
density polyethylene. Polymer 27: 384-392
17. ASTM E 1152 (1987): Standard Test Method for DetenniningJ.-R Curves
18. Zhou Z., Landes J. D., Huang D. D. (1994): J-R curve calculation with the normalization
method for toughened polymers. Polym. Engng. Sci. 34: 128-134
19. Che M., Grellmann W., Seidler S., Landes J. D. (1997): Application of a normalization
method for determining J-R curves in glassy polymer PVC at different crosshead speeds.
Fatigue Fract. Engng. Mater. Struct. 20: 119-127
20. Ramsteiner F. (1999): JO.2-values by impact testing. Polym. Test. 18: 641-647
21. Seidler S., Grellmann W. (1995): Application of the instrumented impact test to the tough-
ness characterization of high impact thermoplastics. In: Williams J. G., Pavan A. (Eds.) Im-
pact and Dynamic Fracture of Polymers and Composites, ESIS Publication 19, Mechanical
Engineering Publications, London: 171-178
22. Dugdale D. S. (1960): Yielding of steel sheets containing slits. J. Mech. Phys. 8: 100--104
23. Mai Y. W., Cotterell B. (1991): On the essential work of ductile fracture in polymers. J.
Polym. Sci., Polym. Phys. 29: 785-793
24. Karger-Kocsis J. (1996): How does 'phase transformation toughening' work in semicrystal-
line polymers? Polym. Engng. Sci. 36: 203-210
25. Clutton E. Q. (2000): ESIS TC4 experience with the essential work of fracture method
applied to plastics films. In: Williams J. G., Pavan A. (Eds.) Composites and Adhesives.
ESIS Publication, Elsevier Science Ltd, Oxford: 187-199
26. Wong S. C., Hai Y. W. (1999): Essential fracture work of short fiber reinforced polymer
blends. Polym. Engng. Sci. 39: 356--364
27. Casellas J. J., Frontini P. M., Carella J. M. (1999): Fracture characterization of low-density
polyethylenes by the essential work of fracture: changes induced by thermal treatments and
testing temperature. 1. Appl. Polym. Sci. 74: 781-796
28. Maspoch M. L., Ferrer D., Gordillo A., Santana O. 0., Martinez A. B. (1999): Effect of
specimen dimensions and the test speed on the fracture toughness of iPP by the essential
work of fracture (EWF) method. J. Appl. Polym. Sci. 73: 177-187
29. Karger-Kocsis J. (1999): Toward understanding the morphology-related crack initiation and
propagation behaviour in polypropylene systems as assessed by the essential work of frac-
ture approach. J. Macromol. Sci. Phys. B (5 & 6) 38: 635-646
30. Raab M., Kotek J., Baldrian J., Grellmann W. (1998): Toughness profile of injection-molded
polypropylene: The effect of the p-modification. J. Appl. Polym. Sci. 69: 2255-2259
31. ASTM D 3433 (1999): Standard Test Method for Fracture Strength in Cleavage of Adhe-
sives in Bonded Joints
32. Kinloch A. L., Lau C. C., Williams J. W. (1994): The peeling of flexible laminates. Int. J. of
Fract. 66: 45-70
A 2.1 Influence of Specimen Geometry

and Loading Conditions
on the Crack Resistance Behaviour
of Poly(vinyl chloride) and Polypropylene
W. Grellmann, S. Seidler, K. Jung, M. Che
and 1. Kotter, Merseburg
1 Introduction
The concepts of elastic-plastic fracture mechanics have proved their worth in to

characterizing the toughness of thermoplastic polymers. Two different parameters
can be used for the characterization of the elastic-plastic material behaviour.
These are the energy-determined J-integral and the deformation-determined criti-
cal crack-opening displacement.
The geometrical independence of values determined from specimens is a fun-
damental requirement for the application of fracture mechanics parameters for the
evaluation of toughness of components. Whereas the requirements regarding
specimen geometry for unstable crack propagation can be quite reliably assessed,
this cannot be done in a simple way for stable crack propagation. By considering
the effects of test temperature and crosshead speed on the crack initiation and
crack propagation behaviour, a basis for quantitative morphology-property corre-
lations needs to be created.
2 Experimental
Specimens were prepared directly from commercially extruded PVC and PP sheets
(thicknesses 4, 6,8, 10, 15 and 20 mm).
Compact tension (CT) specimens with dimensions W = 40 mm, B = 4 to 20 mm
and a/W= 0.55, and single-edge-notched bend (SENB) specimens with B = 10 mm,
W = 20 mm, s/W = 4 and a/W = 0.55 were used for the tests.
Fracture mechanics studies and tensile tests were performed on a Zwick and an
Instron (equipped with a temperature chamber) testing machine, respectively. The
load-line displacement was measured with the help of a laser scanner system (pp.
365-384, [1]). The functioning of the laser double scanner, working in transmis-

52 W. Grellmann, S. Seidler, K. Jung, M. Che, I. Kotter
si.on mode, is shown in Fig. A39. A helium-neon laser with a constant power of 4
to 6 mW was used.
The effects of specimen thickness, specimen configuration and temperature
were checked at a crosshead speed of 1 mmlmin. The effect of crosshead speed on
the crack resistance behaviour of PVC was studied in the range from 0.01 mmlmin
to 100 mmlmin. PVC material was selected because of its clearly measurable sta-
ble crack growth on the fracture surface.
.-
Junction Diode
HeINe-Laser
Deflection
Mirror
,"
"
~~
ll~) Start
se!conductor - '® scanner

p' l1'l .
nsm Stop '
~\ Evaluation and Rotating Mirror
Graphics
Fig. A39. Laser double scanner system for measurement of load-line displacement and crack-
opening displacement
Two different evaluation methods for determining the J values were used. If the
experiments are evaluated according to ESIS TC4 [2], the J values are calculated
from (A26):
(0.751] -I)/la ).
J = Jo(1 (W-a)
(A26)
with J0-- 1]-40 (A27)

B(W -a)
and 1] =2 + 0.522(1- a/W). (A28)
It is also possible to determine J-values following an evaluation method of

Sumpter and Turner [3], which is preferred when determining R-curves under
dynamic loading (pp. 71- 86 and pp. 87-94). The energy under the load versus
load-line displacement curve is separated into an elastic and a plastic part, and the
plastic part includes a crack length correction term (A29):
Influence of Specimen Geometry and Loading Conditions 53
J = l1elAeI + l1plApI (I (0. 7511 -I)Aa), (A29)

B(W -a) B(W -a) (W -a)
with
l1eJ = 11.30 - 68.42(a/ W) + 198.75(a/ W)2 - 258.33(a/W)3 + 125(a/ W)4 (A30)
and l1pJ == 11 (A28).

To characterize the resistance of the material against stable crack propagation,
the tearing modulus TJ at Aa = 0.2 rom was determined (pp. 71-86).
The evaluation of the t5-Aa curves was performed according to ESIS P2-92 [4].
The crack-opening displacement 0 was separated into an elastic and a plastic part
according to (A31):
K2(l-V 2 ) v
0- + p (A31)
- 2Eu y I+n[(a+z)/(W -a)]'
withn=2.5.
The crack length correction was calculated from (A32):
~ K2(l-v 2 ) 0.6Aa+0.4(W -a)
D = + vp ' (A32)
2Euy 0.6(a+Aa)+0.4W +z
F
K = 112 f(a / W) . (A33)
BW
3 Results and Discussion
3.1 Effect of Specimen Thickness
3.1.1 Poly(vinyl chloride) (PVC)
The determination of the J-Aa curves for the PVC material was performed flrst
according to ESIS TC4 [2] and a flt curve was drawn using a simple power law.
The J-Aa curves determined, taking account of the validity limits Aamin, Aamax ,
J max and tV (see [2]), are shown in Fig. A40. All data points which are situated
outside the window of validity are represented by open symbols. The J max criterion
results from the validity criterion on the J-integral. The restriction on the amount
of stable crack growth was applied in [2] for two reasons:
• Stable crack growth Aa ~ 0.05 rom (Aamin) is an artefact and is not considered,
in order to avoid subjective measurement error since small amounts of stable
crack growth are difficult to measure and hence are subject to error.
• Stable crack growth Aa ~ 0.1 (W - a) (AOmax) ensures J-controlled crack growth.
Besides the effect of specimen thickness B on the J-R curve, the effect of B on
the 8--R curve was also considered. The determination of the 8--R curves was
based on ESIS P2-92 [4] and ASTM [5] because ESIS TC4 does not cover the cal-
culation of 8--R curves. Figure A41 shows the 8--R curves of PVC with the limits
of validity. The J-Aa and 8--Aa curves behave similarly; the toughness behaviour
is both load and deformation determined.
50 J mo (B=20 mm)
A8 min 15mm .c18 max

40 6,
-
6,
0
--
E 30 ~ ~~
E
z 8 (mm)
-, 20 --.--..
....•
.-
4
6
- ... 8
10 -"-"-10
___ -15
_..._20
0.5 1 1.5 2
Aa (mm)
Fig. A40. Effects of specimen thickness on J-tuJ curves of PVC, evaluated according to ESIS
TC4 (filled symbols, valid test points; open symbols, invalid test points)
Fig. A41. Effects of specimen thickness on Ii-tuJ curves of PVC (filled symbols, valid test
points; open symbols, invalid test points)
In agreement with the literature [6], for PVC for a specimen thickness of B = 10
mm, geometry-independent initiation values were determined.
According to Zerbst and Will [7], a restriction on the allowable crack growth is
only necessary if the stable crack growth has not taken into account in the calculation
of the Jvalues. When (A26) and (A27) are used, the l1amax criterion becomes unnec-
essary.
Another critical point is the criterion (i) ~ 10, which implies a criterion on the
slope of the J-!:J.a curve. Anderson and Stricklin [8] have demonstrated, using
tough copper alloys as an example, that (i) ~ 10 can be an unnecessary restriction
on the validity of the J-R curve. For the materials investigated in [8], (i) values of 7
were found. Therefore, the J-!:J.a values were additionally analysed as follows:
1. calculate the J values using (A29)
2. include all experimental J-!:J.a values which fulfil the condition (i) » 1 [9]
3. check the validity of the resulting JO.2 values using (A34):
J
B~&-, with & =25. (A34)
O'y
The J-R curve determined according to this procedure is shown in Fig. A42.
The advantage of an analysis including the condition (i) » 1 [9] is that the rela-
tionship between the loading parameter and the crack propagation can be quanti-
fied over a greater range of !:J.a and so a better toughness evaluation is possible.
40
8 (mm)
e4mm *10mm
.. Smm -15mm
.Smm *20mm
00 0.5 1 1.5 2
~a (mm)
Fig. A42. J-~a curve of PVC, calculated according to (A29)

3.1.2 Polypropylene (PP)
A different result was obtained for the PP material. If ESIS TC4 is consistently
used, no definite conclusion about the effect of specimen thickness is obtained,
because under these conditions valid J-R curves can only be determined for
specimen thicknesses of B = 15 mm and B = 20 mm (Fig. A43).
Therefore, the procedure described above using the example of PVC, including
all experimental values, was used (Fig. A44). It is evident that the curves for
B = 6 mm and B = 10 mm show a much steeper slope than all other curves.
70
60
"
!lin ML max
*
-
50 0
~*
--.,
E 40 0
E tr i3 0
Z 30 0 ~ 0
0 0 0
(II:!!'
20 B(mm)
10
•
... 4mm
Smm
* 10mm
• 15mm
• 8mm • 20mm
0.5 1 1.5 2
AS (mm)
Fig. A43. J-t:.a curves of PP, evaluated according to ESIS TC4 (filled symbols, valid test
points; open symbols, invalid test points)
70
60
50
E
~40
-, 30
20 B (nm)
..... 4nm ""*10nm
""" '6 nm ........ 15 nm
·..··Snm ...... 20nm
0.5 1 1.5 2
~a (mm)
Fig. A44. J-t:.a curves ofPP, calculated according to (A29)
To clarify the reason for this behaviour, light-microscope investigations (Fig.

A45) were carried out. It was found that the specimens of 6 and 10 mm thickness
contained f3 crystallites, whereas in all other specimens no f3 crystallites were de-
tected. So the effect of specimen thickness is superimposed on the effect of crys-
talline morphology. The change in the polypropylene structure first causes an
increase of material resistance against stable crack propagation.
The crack initiation values show no dependence on the specimen thickness but
show a greater scatter than for PVC, attributed to the stronger effect of B on the
slope of the J-R curve. Excluding the PP materials with f3 crystallites from consid-
eration, a decrease in the TJ values with increasing specimen thickness results.
Fig. A45. Light-microscope micrographs of spherulite fonnation in PP: (a) specimen thick-
ness 8 mm, (b) specimen thickness 10 mm
•
-
1.5 i
-
E
E
tQ
0.5
...... 4mm -*- 10mm
• ••. .·Smm
......... 8 mm
•.....·15mm
...... 20 mm
0.5 1 1.5 2
~a (mm)
Fig. A46. a-b.a curves ofPP
If the 8-R curves are calculated according to ESIS P2-92 [4], no valid 8-l1a
curves can be determined. For this reason, all experimental data points were in-
cluded in the estimation (Fig. A46). The 8-values were calculated from (A32).
The experimentally determined thickness dependence can be used to calculate
the constant E of the geometrical criterion used with the J-integral according to
(A34). The values of Edetermined are E= 31.8 for the PVC material and E= 10 for
the PP material.
Both values fit the general dependence E = f (.I) (pp. 3-26 and pp. 71-86). The
same assertion applies to the relationship ~ = f (0).
3.2 Effect of Crosshead Speed
In the literature, differing conclusions about the time-dependent crack resistance

behaviour of polymers have been reported. A reduction [10] as well an increase
[11-13] of toughness parameters was found with increasing loading rate. Bernal
and Frontini [14] have observed no influence of loading rates.
J-l1a curves for PVC, determined at different loading rates and with a specimen
thickness of B = 10 mm, are shown in Fig. A47. The evaluation of the J-R curves
followed the procedure ofESIS TC4 [2].
40r-----------------------------------.
-:.
J max "'tl/i::, (> *
30~F==================~~~==~
10 VT(nm'ri'I)
-;1- 10
--+-0.01 --+--.. 1 .-+-- 50
-...A- 0.1 ----..-. 5 -*- 100
O~-J~---L--~----~--~--~----~~
o 0.5 1 1.5 2
Aa (mm)
Fig. A47. J-AtJ curves of PVC at different crosshead speeds
Figure A48a shows the dependence of the JO•2 and TJ values as a function of the
crosshead speed. For JO.2, a maximum at a crosshead speed of 5 mmlmin is ob-
served, while the tearing modulus remains constant up to 5 mmlmin, then de-
creases with increasing crosshead speed.
10 50 0.2 b 40
a
..102 00.2
/
•
~~~ 40
... ,
... ~...... ", 30
8 ,,
E ,,
I
, 30
- -- - '-'" -~:- - -+,
~ '" , N
0..,
T]2 0.1 20~
~
; +\
f-
...,'"d 6
20
10
10
CT CT
4 o .........................
-'-~"'-- ..................--'-.. . . . . "'" 0
0.01 0.1 1 10 0.01 0.1 1 10 100
Vr(rm1nin) Vr(rrmnin)
Fig. A48. Influence of crosshead speed on fracture mechanics parameters: (a) J02 and TJ ;
(b) ~2 and To
Using the concept of crack-opening displacement (COD) (Fig. A48b), the ini-
tiation value ~.2 remains constant up to 5 mmlmin, and then decreases with in-
creasing crosshead speed. The tearing modulus To behaves approximately in the
same way. The 8-Aa curves are shown in Fig. A49. These curves were determined
using (A32) using a procedure analogous to that described in sect. 3.1. From Fig.
A48, it can be concluded that for the PVC tested the time-dependent crack initia-
tion resistance should be evaluated by using both JO.2 and ~2' while the time-
dependent crack growth resistance can be well characterized by either TJ or To.
0.8
...-
Yr(mmlmln)
0.6
.......
.-.....
0.01 - 10
-g ........ 0.1 -...···50

1 4-100
E --*_. 5
0.4
c.o
L\a max
0.2
0.5 1 1.5 2
Aa (mm)
Fig. A49. 8-!::"a curves of PVC at different crosshead speeds
3.3 Effect of Test Temperature
Typicalload-deflection curves for PVC at different temperatures are shown in Fig.

A50. With increasing temperature, the slope of the load-deflection curve de-
creases. To produce the same stable crack extension, more deflection needs to be
used at a higher temperature. The maximum load reaches a maximum value at
room temperature.
~~----------------------------------~
-1.48nm
see
----z
sIl:!Ije
600
1.4
"C
cu
o
...J
0.59
300 T (OC)
-+- -4Q°C -+- 23°C
-+-
---
-20°C ~ 4Q°C
O°C -e- 60°C
O~----~----~----~----L---~~--~
o 2 4 6
Deflection (mm)
Fig. ASO. Load versus deflection curves of PVC as a function of test temperature
40r-------------~._--------------------,
30
-
E
E 20
Z
--- --.-----'f-
.--+"""-......
--- ...J"
T ("C I
--...L----:.
.. _.-------
..-+-- -.- 60
;.---- ..-+-.-
..
40
10 ... ----* -A- 23
._-+--- o
--.....--- .._- _...
-20
___ -40
O~~--~~--~--~~--~~--~--~~--~
o 0.5 1.5 2 2.5 3

!l. a (mm)
Fig. ASl. J-tla curves of PVC at different test temperatures

The J-R and the 8-R curves are shown in Figs. A51 and A52 as a function of
the test temperature. If the procedures recommended in ESIS TC 4 [2] and ESIS P2-
92 [4] are strictly followed, there is an insufficient number of valid data points to
construct crack resistance curves. So the procedure with an extended validity range
discussed above is advantageous.
The values of JO.2 and ~J.2, as well as their corresponding tearing moduli, are
shown in Fig. A53. ~.2 and Tt5 show similar behaviour, whereas the JO.2 and TJ
values show different behaviour in their dependence on temperature.
1.2 . - - - - - - - - - . = - - - - - - - - - - - - - - - ,
T (OC)
• ____ 60
--+--
•
40
0.9
• ........
_._+_.-
23
0
----:"
-_..... - .. - -20
.sE 0.6
.~£:~:/
. .-(f;:.:~~-+: . -.-.
t.O
0.3
o-=~-~-~~-~-~~-~-~~-~~
o 0.5 1.5 2 2.5 3

L\a (mm)
Fig. AS2. t5-1':la curves of PVC at different test temperatures
15 200 0.4 100

SENB a SENB
...
"E 1O E
.E I
I
",+', ,
'+-- • 100~""
E
0.2
N
~ ~I
I
~ 50~«>
,-'- *
I
N
'" ,
.-
c::i
-
...f5 00
..
--' ~-/-
,.-
...--- ---- ~
0 0 0 0
-40 -20 0 20 40 60 -40 -20 0 20 40 60
T (OC) T (0C)
Fig. AS3. Influence of temperature on fracture mechanics parameters: (a) J02 and TJ , (b) ~2
and Ta
Obviously, the fracture mechanics parameter COD related to the amount of de-
formation represents the temperature dependence of the crack-tip deformability
only. Therefore, it is not appropriate to use the single parameter COD to evaluate
the effect of temperature on the crack initiation toughness. In contrast, the J-
integral concept, which is related to the elastic-plastic energy and takes account of
loading capacity and crack-tip deformability, seems more suitable. The tearing
moduli obtained from these two parameters show similar behaviour in their de-
pendence on temperature (Fig. A53). So it can be concluded that the temperature-
dependent crack initiation behaviour can be well characterized only by the com-
bined application of the concepts of the J-integral and the COD. The temperature-
dependent crack propagation behaviour can, in contrast, be described by using the
tearing modulus of either the J-integral or the COD.
3.4 Effect of Specimen Configuration
The effect of specimen configuration on the crack resistance behaviour was stud-
ied by using CT and SENB specimens. Load versus load-line displacement curves
are shown in Fig. A54. The maximum load is the same for both specimens, but the
SENB specimens exhibit a greater load-line displacement. However, the S-da
curves are independent of the specimen configuration (Fig. A55). The J-da curves
for the SENB specimens lie higher than those for the CT specimens (Fig. A56), i.e.
a higher JO.2 value and a steeper slope of the crack propagation curve were ob-
tained for the SENB specimens.
1200 CT
900
g SENB
Aa= 2.24 mm
"t:J
~
..J
600
B = 10mm
300 W =40mm
aNV= 0.55
0
0 1 234 5
Load-Line Displacement (mm)
Fig. A54. Comparison of load versus load-line displacement curves of CT and SENB specimens
(PVC)
-• cr
SENB
0.8
~0.6
E
.§.
r.o 0.4
0.5 1 1.5 2 2.5

As (mm)
Fig. ASS. Effects of specimen configuration on t5-ALl curves of PVC
50
-cr
.SENB
40
E 30
.E
~
..., 20
10
0.5 1 1.5 2 2.5

Aa (mm)
Fig. AS6. Effects of specimen configuration on J-ALl curves of PVC
Similar effects were fOlUld in modified PVC [15]. In several other polymers,
such as LDPE, MDPE, HDPE and PS, no influence of specimen geometry was
observed. Results related to this set of problems also can be found in [16].
Under the assumption of deformation-determined fracture behaviour, the same
crack resistance curves were obtained for both specimen configurations. This
means that the critical crack-opening displacement for this material is a geometry-
independent material parameter. On the other hand, the quantification of the crack
resistance behaviour in terms of the energy-determined J-integral results in ge-
ometry-dependent parameters, even though the approximate solutions proved to be
suitable in principle. The influence of geometry is apparently superimposed on a
material effect which is not included in the calculation of the J-l1a data points.
64 W. Grellmann, S. Seidler, K. Jung, M. Che, L Kotter
In view of these aspects, the application of the CTOD concept is more appro-
priate for compare fracture mechanics parameters determined with different
specimen configurations.
3.5 Simulation of Crack Growth by Finite-Element Method
The experimentally estimated results can be verified numerically with the help of
the finite-element method, and the complex stress and strain distributions at the
crack tip can be investigated. If it is presumed that continuum mechanics is appli-
cable to polymers, the von Mises yield condition may be applied as a material law.
The distinctive features that appear in the deformation behaviour of polymers in
comparison with metals, such as viscoelasticity and viscoplasticity, crazing, com-
pression-sensitive yield behaviour and so on, are not used as input in the form of
material constants. For numerical simulation, the input of the realload-extension
behaviour as a multi-linear expression is necessary. The determination of yield
curves for polymers requires a high expenditure of effort in the measuring proce-
dure. Because of the many parameters which influence the material behaviour,
these curves are difficult to reproduce. For that reason, stress-strain curves from
the literature were used for the input to the FEM. For PVC, the a-& curve by
Cayard [17], measured at a crosshead speed of 0.1 mmlmin, was used. The stress-
strain curves for other crosshead speeds were converted with the help of a rate-
sensitive factor m by G'Sell and Jonas [18].
The numerical simulation of stable crack propagation was carried out by node
release in the ligament, where a maximum crack extension Aamax = 2.5 mm had to
be realized. The meshing of the CT specimen for two-dimensional calculations
(under the condition of large deformations) was done with isoparametric eight-
node elements as shown in Fig. A57.
Because the CT specimen has a symmetric geometry and it is also symmetri-
cally loaded, only one half of the specimen has to be meshed. The area of the
ligament in which the stable crack propagation was to be simulated was meshed
more finely in comparison with the remaining area.
Investigations by Oliveira and Kaiser [19] and Brocks and Yuan [20] on the
numerical simulation of crack propagation show that the r- l12 singularity in front of
the crack tip can be re~ected sufficiently well by this finer distribution of elements
in the vicinity of the crack tip. A maximum element length ofO.Ol(W - a) is rec-
ommended.
The FEM calculations described here were carried out with the help of the
ANSYS software package on a Silicon Graphics work station.
If the crack propagation is simulated for other specimen configurations (SENB,
SENT, CCT, DENT), the discretization of the crack-tip zone remains unchanged
but the residual network is modified to correspond to the specimen configurations.
Experimentally determined curves of displacement (VII) versus crack extension
(Aa), and the crack-tip-opening angle (CTOA) served as parameters for simulating
the crack propagation.
/ /
IL r--.
I\, V -
-I-
):: b"
r
V
" '" ""-

"-
I
~ IT"
/
1'1- r-! 1- r-! 1- r-! \- ~
In M'
/
1l"TT
I
/
Fig. AS7. Finite-element network for a CT specimen used to calculate fracture mechanics
parameters and to simulate crack propagation
For CTOA control, the crack-tip nodes were released by one crack propagation
unit if the opening angle of the first edge node behind the crack tip at the moment
reached a critical value. Loading was always represented as displacement at the
load-line displacement point.
The J-!J.a curve, relevant to the energy, and the 8-!J.a curve, relevant to the de-
formation, were calculated. The J-integral was calculated according to (A35):
J = J(W dy - O"ijnj : ds ). (A35)

R
For PVC, only one control curve should be necessary, as a consequence of the
experimentally estimated geometry independence (see sect. 3.1). To demonstrate
this geometry independence numerically, for every specimen thickness the corre-
sponding vlI-!J.a curve was taken as a basis for the control curve of the crack
propagation simulation. The numerically determined J-!J.a curves were found to lie
in the scattering range of the experimentally determined J-!J.a curve. This is an
indication of the independence of the curve from the thickness in the thickness
range investigated. Because of the free deformability in the direction of the speci-
men thickness, a plane stress state occurs at the edge of the specimen, but it can be
neglected and does not affect the crack propagation behaviour of PVC. This fact
could be confirmed with three-dimensional simulation calculations. The numerical
investigations on PVC were carried out by using the 045 definition [21] for the
crack opening t5. The FEM results obtained taking account of the plane strain state
agree well with the experimentally determined results. The numerically estimated
66 W. GreUmann, S. Seidler, K Jung, M. Che, I. Kotter
crack-opening values for the plane stress state are greater than the experimental val-
ues, and because of the free defonnability in the direction of the specimen thickness,
a more ductile material behaviour is simulated.
While the crack propagation simulation controlled by the vll-Lla curve always
presumes the experimental determination of crack resistance curves, the use of the
critical crack-opening angle CT0Acnt as a control parameter for the crack propa-
gation simulation is advantageous. In this case, previous experimental work is not
necessary.
For polymers in particular, the critical crack-opening angle CTOAcrit depends
on the specimen geometry and the environmental and testing conditions. This fact
can be illustrated using the crack resistance behaviour of PVC as a ftmction of the
crosshead speed (Fig. A58).
While the experimental crack resistance curves estimated at a crosshead speed
of VT = 1 mmlmin with CTOAcnt = 40° could be numerically simulated satisfacto-
rily, for a crosshead speed of VT = 100 mmlmin the best conformity between ex-
periment and numerical simulation was obtained at CT0Acnt = 30° (Fig. A58). The
experimental data are marked by crosses and the regression curve is represented by
a continuous line.
a) VT = 1 mmlmin b) VT = 100 mmlmin

--~--------~-----------~-
------
50 CTOA +
+
+ CTOA
+
40 • 20· + .. 30·
..• • •
+
.. 30· +
+ ·40·
~
E • 40· +
E 30 +
•
~
20
+ 50·
• ••
10
::
+
.............. .
&"
•
• •"
&
•
CTOA +
+ CTOA
+
0.8 .20· + .. 30·
.. 30· +
+
• 40· ••
••
•
E 0.6 • 40·
• ••
.s + 50·
+ ••
••
0.4 +
• .•
+ ..............
0.2
x
+ .........
:~ •••••
• ••••••••
.
•
• &
&
•
o 0.5 1 1.5 2 0 0.5 1 1.5 2
M(mm) ~a (mm)
Fig. ASS. R-curves of PVC as a function of the crack-tip-opening angle CTOA at different
crosshead speeds, at room temperature (crosses - experimental data): VT = 1 mmlmin
(a) and VT = 100 mmlmin (b)
According to the RDS model of Rice, Drugan and Sham [22], the crack propa-
gates with a constant crack-tip-opening displacement 80 recorded at a distance rm
behind the crack tip. This crack-growth criterion corresponds to crack propagation
at a constant crack-tip-opening angle. Investigations of the correlation between the
numerical simulation and experiment, especially for the crack initiation values,
provide satisfactory results only at low testing speeds, because at high testing
speeds the local adiabatic heating in front of the crack tip is not negligible.
The experimental estimation of the critical crack-tip-opening angle CT0Acnt is
complicated for ductile polymers because of their typical relaxation behaviour, but
this characteristic value can be determined by in situ investigations of the crack
initiation behaviour.
3.6 In Situ Experiments on the Crack Blunting Process
In the literature, different crack-tip-deformation processes for semicrystalline and

amorphous polymers are described [23,24]. In both cases blunting of the crack tip
occurs. For semicrystalline polymers it is supposed that the subsequent process of
crack initiation occurs by opening of the crack edges. The opening of the crack
edges takes place after a critical amount of crack blunting is reached. For amor-
phous polymers, on the other hand, the crack initiation occurs by translation of the
whole crack front.
Dependent on the crack initiation process, the stretch zone can be formed in dif-
ferent ways. For semicrystalline polymers whose crack initiation process is
equivalent to that of metals, formation of the stretch zone at the end of the razor
blade notch is expected. A stretch zone at the end of stable crack growth can also
be observed for semicrystalline polymers. Which of these crack initiation proc-
esses appears seems dependent on the morphology.
In situ experiments were carried out in an atmospheric scanning electron micro-
scope. This microscope permits the examination of polymers without a sputtered
coating. Thereby, such problems as the appearance of cracks in the sputtered
coating are avoidable.
SENB specimens with dimensions of 4 x 4 x 20 mm3 were loaded into the
scanning electron microscope. A bending load appeared to be better than a tensile
load, which was also possible, since in the CT specimen bending occurs predomi-
nantly [25]. The image sequences (Figs. A59 and A60) show the crack-tip-defor-
mation processes ofPP and PVC at different loading stages.
The behaviour of PVC corresponds to the expected crack initiation process for
amorphous polymers. The crack propagation occurs by translation of the whole
crack front. Indeed, the presence of a stretch zone could not be proved at the end
of stable crack propagation.
68 W. GreUmann, S. Seidler, K Jung, M. Che, I. Kotter
Fig. A59. Crack-tip deformation of PP: (a) initial state, (b) crack-tip blunting, (c) incre
sing crack extension, (d) opening of the crack edges
Fig. A60. Crack-tip deformation of PVC: (a) initial state, (b) crack-tip blunting, (c) increasing
crack extension, (d) opening of the crack edges
4 Summary
A comparison of different procedures for the evaluation of J-Aa curves has shown
that by following all limit criteria regarding crack length and slope according to
ESIS, only a few experimentally estimated data points may be used in an evalua-
tion. Therefore, a procedure based on the J-integral approach of Sumpter and
Turner [3] was favoured. The results of FEM calculations have shown that this
method is justified.
In their dependence on specimen thickness and crosshead speed, the J-Aa and
8--/ia curves show similar behaviour. This fact demonstrates that the crack tough-
ness behaviour is load-determined and likewise deformation-determined in this
case. The geometrical constants & determined from the dependence of the J-Aa
curve on the specimen thickness can be integrated very well into the general de-
pendence & = f (J).
Using PVC as an example, the crosshead speed dependence of crack initiation
and crack propagation behaviour was demonstrated. Whereas the crack initiation
values decrease only at VT ~ 5 mmlmin, the tearing moduli decrease steadily in the
crosshead speed range investigated, independent of the EPFM concept used.
The in situ investigations described here make statements about the effect of
morphology on the crack initiation process possible.
Acknowledgements
The authors would like to thank the German Research Foundation (DFG) for the
financial support of this study, 'Characterization of the crack initiation behaviour
of amorphous and semicrystalline polymers'. They also thank the FhG Institutes
for 'Materials Mechanics', Halle, and 'Microintegration and Reliability', Berlin,
for scientific cooperation and the opportunity to carry out in situ experiments.
References
1. Che M., Grellmann W., Seidler S. (1997): Crack resistance behavior of polyvinyl chloride. J.
Appl. Polym. Sci. 64: 1079-1090
Resistance Curve Tests on Plastics
3. Sumpter J. G. D., Turner C. E. (1976): Cracks and fracture. In: ASTM STP 601: 3-18
4. ESIS P2-92 (1992): Procedure for Detennining the Fracture Behaviour of Materials
5. ASTM E 1820-96 (1996): Standard Test Method for Measurement of Fracture Toughness.
American Society for Testing and Materials, Philadelphia
toughened polymers. In: Proceedings of the International Conference on Fracture, ICF7,
University of Houston, Texas, March 20-24: 2725-2732
70 W. Grellmann, S. Seidler, K Jung, M. Che, I. Kotter
7. Zerbst U., Will P. (1988): Ermittlung von elastisch-plastischen RiBwiderstandskurven mit-

tels kleiner Proben. Neue Hiitte 33: 269-272
8. Anderson T. L., Stricklin L. L. (1989): A re-evaluation of criteria for J-controlled crack
growth. Int. J. Fract. 41: R65-R70
9. Will P., Michel B., Schaper M. (1990): Justification of nonlinear J-resistance curves. Engng.
Fract. Mech. 37: 275-281
10. Huang D. D., Williams J. G. (1987): Jtesting of toughened nylons. J. Mat. Sci. 22: 2503-
2508
11. Crouch B. A., Huang D. D. (1994): The J-integral technique applied to toughened nylons
under impact loading. J. Mat. Sci. 29: 861-864
12. Otterson D. M., Kim B. H., Lavengood R. E. (1991): Effect of testing speed on the crack
growth resistance properties of nylon 6/ABS polymer blends. J. Mat. Sci. 26: 4855-4860
13. Barry D. B., Delatycki O. (1989): The strain rate dependence of fracture in polyethylene:
fracture initiation. J. Appl. Polym. Sci. 38: 339-350
14. Bernal C. R., Frontini P. M. (1992): Fracture toughness detennination of ABS polymers
15. Hashemi S., Williams J. G. (1986): The effects of specimen configuration and notch tip
radius on the fracture toughness of polymers using Jc. Plast. Rubber Process. Appl. 6:
363-375
16. Reese E. D. (1996): Zur Anwendung des R-Kurven-Verfahrens der elastisch-plastischen
Bruchmechanik aufPolymere. Herbert Utz Verlag Wissenschaft, Miinchen
17. Cayard M. (1990): Fracture Toughness Testing of Polymeric Materials. PhD dissertation,
Texas A&M University, College Station, TX
18. G'Sell C., Jonas J. J. (1979): Determination of the plastic behaviour of solid polymers at
constant true strain rate. J. Mat. Sci. 14: 583-591
19. Oliveira R., Kaiser S. (1987): Modelling the HRR-singularity with akin and conventional
singular elements. Engng. Fract. Mech. 27(2): 199-204
20. Brocks W., Yuan H. (1991): ESISIEGF 9, Numerical Studies on Stable Crack Growth.
Mechanical Engineering Publications, London: 19-33
21. Blumenauer H., Pusch G. (1993): Technische Bruchmechanik. Deutscher Verlag fUr Grund-
stoffindustrie, Leipzig, Stuttgart
22. Rice J. R., Drugan W. J., Sham T. L. (1980): Elastic-plastic analysis of growing cracks. In:
ASTM STP 700: 189-221
23. Narisawa I., Takemori M. T. (1989): Fracture toughness of impact-modified polymers based
on the J-integral. Polym. Engng. Sci. 29: 672-678
24. Kishimoto K., Notomi M., Matsui M., Koizumi T. (1990): Evaluation on low-temperature
strength of plastic film for balloon. In: Proceedings of Toughness, Fracture and Fatigue of
Polymers and Composites, Yamagata, Japan, October 8-11: 160-165
25. Brocks W., Fricke S. (1992): Analyse von Geometrieeffekten bei der Beschreibung von
duktilem RiBwiderstandsverhalten. In: DFG Berichtskolloquium FlieBbruchmechanik:
39-48
A 2.2 Procedure for Determining the Crack
Resistance Behaviour Using
the Instrumented Charpy Impact Test
W. Grellmann, S. Seidler and W. Hesse, Merseburg
1 Scope
The instrumented Charpy impact test is used for determining properties related to
the impact strength of plastics. It is an addition to the conventional pendulum
impact test described in ISO 179 [1]; it is carried out on razor-blade-notched
specimens. The plastics on which it can be used range from brittle thermosets to
high-impact polymer blends. Both the load and the deflection signals are recorded,
and the impact energy is divided into an elastic and a plastic part. If the require-
ments on the specimen size and on the notch are satisfied, geometry-independent
material parameters can be calculated. These parameters can be used for control
and for quality assurance as well as for research and development.
2 Short Description of the Procedure
The testing of tough polymers is carried out by a pendulum Charpy impact tester
of type PSW 4, with 4 J work capacity at maximum fall height. A single-edge-
notched square specimen is broken by the impact of the pendulum hammer. The
striker, which has an edge on its impact side, is fastened to a tubular pendulum
arm. After release, it moves in a circle and transfers part of its kinetic energy to
the specimen at the lowest point of its trajectory.
The registration of the load signal is carried out by strain gauges positioned di-
rectly on the striker edge and arranged as a Wheatstone bridge. The output is am-
plified by a two-stage amplifier.
The measuring device is able to register either load-time (F-t) diagrams or
load-deflection (F-j) diagrams. In the case of load-time diagrams, the deflection
can be calculated from Newton's second law. In a first integration step (A36), the
velocity can be determined, and in a second integration step (A37), the deflection/
of the specimen as a function of time:
1I
v(t)=vo--f F(-r)dT, (A36)
rno
I
let) = fVeT) dT (A37)
o
An optical device is used for the measurement of the deflection. A diaphragm
fastened to the pendulum hammer, in the path of a beam of light, moves with the

72 W. Grellmann, S. Seidler, W. Hesse
motion of the pendulum. Depending on the movement, a greater or lesser intensity

of light reaches a phototransistor. The resulting signal is amplified for further
analysis.
Both the load signal and the deflection signal are transferred to a storage oscil-
loscope. The oscilloscope is used as an analogue to digital converter and as a
monitor for an initial visual interpretation of the load-time and load-deflection
curves. For further analysis, the digital signals are transferred to a personal com-
puter. The characteristic values Fmax, Fgy,/max, and/gy , as shown in Fig. A61, are
used for interpretation.
unnotched specimen notched specimen
deflection
Fig. A61. Characteristic load-deflection diagram with crack propagation energy
3 Determination of Characteristic Fracture

Mechanics Parameters for Resistance against
Unstable Crack Propagation
3.1 Summary of the Testing Method
The aim of this testing method is to determine characteristic fracture mechanics

parameters which quantify the resistance of the material to unstable crack propa-
gation. This method is only valid for single-edge-notched bend (SENB) specimens
of polymers with a sharp crack. Starting from load-deflection diagrams, stress
intensity factors KId, J values JId and critical crack-tip-opening displacement val-
ues bid can be calculated.
3.2 Specimens
As stated in ISO 179 [1], specimens of thickness B = 4 mm, width W= 10 mm and

length L = 80 mm are preferred. The notching is carried out on the narrow side of
Procedure for Determining the Crack Resistance Behaviour 73
the specimen up to an initial crack length of 2 mm by a pneumatic notching de-

vice. The support distance is s = 40 mm. The specimens must not be twisted and
both pairs of vertical surfaces should be parallel. The surfaces must be free of
scratches, craters, depressions and flashes .
Before testing the specimens, the thickness and width must be measured with a
precision of 0.01 mm. The results must be recorded.
At least 10 specimens should be tested, after storage in the testing room for 12 h.
3.3 Testing
The test should be carried out under standard conditions; if deviations occur or if
variation from the standard conditions is necessary, this must be reported. The
specimen must be arranged on the supports so that the striker edge hits the middle
of the specimen. The notch must lie in the plane in which the pendulum swings,
on the opposite side of the specimen from the striker edge. The arrangement is
shown in Fig. A62. The pendulum velocity should be 1.0 or 1.5 mls.
Charpy Impact Tester

IKBV-4J
Striker
Analysis ofF-fDiagram
Amplifier - Checking of Experimental Conditions
Digital • Fl < Fmu
Oscilloscope • AH> JAG
• t a > J,
- SEM Analysis of Fracture Surface
- Fracture Mechanics Concept
- Checking of Geometry Independence of
Photooptical Fracture Mechanics Values
Transducer
~ KId ;lild;Jld
Fig. A62. Fracture mechanics testing device for the instrumented impact test
The recording of the load-deflection curves is carried out as described in the

operating instructions.
The determination of the dynamic flexural modulus Ed and the dynamic yield
stress ad are carried out on at least five unnotched specimens. For this purpose, the
linear part of the load-deflection diagram is used, i.e., after fmding the yield point,
Ed and CTd can be calculated from (A38) and (A39).
F s3
Ed = gy , (A38)
4BW 3 fgy
3Fgy s
CTd = - - 2 ' (A39)
2BW
3.4 Utilization
3.4.1 General
For utilization of the test data, the loads and deflections shown in Fig. A61 are
determined using the 'Instrumented Charpy Impact Test' software. In principle,
the energy, i.e. both AG and AR and also (AG + A0, can be determined. After input-
ting the geometrical details of the specimens, including the initial crack length a,
the physical crack length augmented to account for crack tip plastic deformation
(the fracture mirror length) aBS, the dynamic yield stress oy and the dynamic flex-
ural modulus Ed, the fracture mechanics parameters KQd, JQd and ~d can be cal-
culated. Both the input values and the calculated values are printed and added to
the testing protocol; the F-f diagrams can also be plotted.
3.4.2 Determination of Fracture Mechanics Parameters
For the impact toughness evaluation of polymers by this procedure, the following
values are preferred.
1. Dynamic Stress Intensity Factor KQd ill.
Fmax s
KQd = 3/2 f (a/W ) , (A40)
BW
where
f(.!!...) = ~(.!!...)112(1.99-a/W(1-a/W)~.5-3.93a/W +2.7(a/W)2 lJ (A41)

W 2 W (1 + 2a/W)(1-a/W)3/2
and
f(; ) 2.9 (a/W)ll2 -4.6 (a/W)312 + 21.8 (a/W)SI2

=
(A42)
-37.6 (a/W)112 +38.7 (a/W)9/2 .
2. JValue JQd •
Following this procedure, Jvalues can be evaluated by two methods.
• Evaluation method of Sumpter and Turner - J ST [3]:
J ST Ael ApI W - aeff (A43)

Qd = llel B(W -a) +l1pl B(W -a) W -a '
where
(A44)
and
2- (l-aIWXO.892-4.476aIW)
llpl = (A45)
1. 125+0.892aIW -2.238(aIW)2
• Evaluation method of Merkle and Corten - JMC [4]:
J~C = G1+ B(i-a) [DtAo + ~AK - (Dt + ~lAel]' (A46)
where
G1 = K
E
~ (1- v 2 ) for plane strain state, (A47)
Dt - l+r (A48)
- l+r2 '
D2 - r(1- 2r-r 2 )
-'-_'-;----'---=-=---L_
(A49)
- (l+r2) ,
1.456(W -a)
r= s
, (A50)
AK = Fmaxfmax -Ao· (A51)
3. Critical Crack-Opening Displacement bQd~.
1 (W -)4
b'Qd =- a-- f max (A52)
n s
and
(A53)
where
(A54)
The determination of further characteristic values depends on technical re-

quirements. These values must be reported in the testing protocol.
3.4.3 Validity Check of the Characteristic

Fracture Mechanics Parameters
3.4.3.1 Check of the Experimental Conditions
For the determination of characteristic fracture mechanics parameters using the

Charpy impact test, the following experimental conditions must be checked [7-9].
• Time to fracture tB.
The time to fracture tB must be greater than or equal to 2.3 to 3 times the inertial
oscillation period l' of the specimen to ensure a quasi-static state:
tB ~ 2.3-31' (A55)
Any further decrease in tB causes problems in evaluation and must be avoided.
• Energy absorption (AG + A0.
The checking of the energy absorption is carried out in accordance with (A56),
according to which the maximum energy of the pendulum must be greater than
three times the absorbed energy of the specimen:
(A56)
• Inertial peak.
The inertial load F\ must be smaller than the maximum impact load Fmax:
F\ < Fmax. (A57)
3.4.3.2 Requirements on the Specimen Geometry
For the determination of geometry-independent characteristic fracture mechanics

parameters, a plane strain state must be ensured, i.e. the geometrical values of
thickness B, notch depth a and ligament length (W - a) must be bigger than a
certain value. For this to be the case, the following relationships must be fulfilled.
• Dynamic fracture toughness.
B; a; (W -a)~p( :: )', (A5S)
where
p = 3466 K I t 73
(A59)
[10].
• Jvalue.
B; a; (W - a) ~ c J 1d ,
ad
(A60)
where
c = 224 JidO. 94 (A61)
[10-13].
• Critical crack-tip-opening displacement.
B;a;(W -a)~~OId (A62)

and
(A63)
where
(A64)
[10].
The functions P= f(KId), c = f(JId) and ~ = f( bidk) have been determined experi-
mentally and are shown in Figs. A63-A65.
100
10 I~ &
13 J73
3466 K I
"'
& r: &
'"
& & &
0.1
10 '" 1000
Fig. A63. Requirements on the specimen geometry for the estimation of the dynamic fracture
toughness KId
10000
- 224 J-0.94 I
Id J
1000 r. E-
100
"" i.
- ........
A experimental results :~ ........
== •
10 ~ results from the literature
1
0.1 1 10 100
J Id (N/mm)
Fig. A64. Requirements on the specimen geometry for the estimation of the J value Jld
100
'\
'\. ~ = 3.6 8 {J.S3
'\
" "\
'\.
"\
10
O.oI 0.1
1'\ ...
3Idk (mm)
Fig. A65. Requirements on the specimen geometry for the estimation of the critical crack-
opening displacement
4 Determination of Characteristic Fracture Mechanics

Parameters for Resistance against Stable Crack
Propagation
4.1 Summary of the Testing Method
The aim of this testing procedure is the determination of fracture mechanics cha-
racteristics that quantify the initial process of crack growth and the energy dissi-
pation of the material. This method is only valid for SENB specimens with a sharp
crack, and the mUltiple-specimen method must be used. Starting from load--de-
flection curves, J and crack-tip-opening displacement values can be determined.
These values are plotted versus the crack growth Aa. The data points characterize
the resistance of the material against stable crack initiation and propagation [14].
By a suitable fitting of the data points, a crack initiation value can be determined
and energy-dissipating processes can be quantified.
4.2 Specimens
For the determination of fracture mechanics parameters for resistance against

stable crack propagation, the preferred specimen geometry is the same as for the
determination of characteristic values for unstable crack propagation; this means
the dimensions are thickness B = 4 mm, width W = 10 mm and length L = 80 mm.
The notch is made on the narrow side up to an initial crack length of a = 4.5 to
5 mm by a pneumatic device. The support distance is s = 40 mm, except when
using the support distance method.
Before testing the specimen, the thickness and the width must be measured with
a precision of 0.01 mm. The results must be recorded.
For the determination of one dynamic crack resistance curve using the multiple-
specimen method, 15 to 20 identical specimens are necessary. They should have
been stored for 12 h in the testing room before testing.
4.3 Testing
The test should be carried out under standard conditions; if deviations occur or if
variation from the standard conditions is necessary, this must be reported.
The specimen must be arranged on the supports as described in Sect. 3.3.
Before the determination of the J-Aa data points themselves, at least one load-
deflection diagram must be recorded under the following conditions:
• specimen geometry: L =80mm
B = 4mm
W= lOmm
a = 4.5 to 5.0 mm
• testing conditions: s =40mm

vH = 1 and 1.5 mls
mH = 0.955 kg.
This load--deflection diagram is used as a starting diagram for the actual meas-
urements. After recording this diagram, stable crack growth is produced using one
of the following methods, and load--deflection diagrams are recorded in parallel
with this.
1. Energy method.
Changing the pendulum hammer energy by varying the pendulum hammer
weight or velocity (low-blow technique).
2. Support distance method.
Changing the ratio of support span to specimen width.
3. Specimen length method [15].
Reducing the specimen length L to a length = support span + 0.1 to 0.2 mm
oversize.
4. Stop block method [16.171.
Changing the deflection by using hardened steel bars or by catching the pen-
dulum hammer (Fig. A66).
Charpy Impact Tester

Load-OeOect ion Diagram
IKBV-4J
Striker
Dynamic Crack Resislance C urve
Fig. A66. Fracture mechanics testing device for the instrumented impact test with stop block
device
A continuous sequence of measurements from smaller to larger crack growth

l1a or in reverse order is useful. The measurements are carried out as long as F ::;
Fmax of the load-deflection diagram recorded previously.
The recording of the load-deflection curves is carried out as described in the
operating instructions.
The [mal breaking of the specimens is carried out at low temperatures and/or at
high testing velocities. After this is done, the amount of stable crack growth is
measured by using a light microscope.
The determination of the dynamic flexural modulus Ed and the dynamic yield
stress O'd is carried out in accordance with (A38) and (A39).
4.4 Utilization
4.4.1 General
For the utilization of the test data, the loads and deflections shown in Fig. A61 are
determined using the 'Instrumented Charpy Impact Test' software as in Sect.
3.4.1. The energy required for the crack growth can be determined from the load-
deflection diagrams.
After input of the specimen dimensions, the initial crack length a, the amount
of stable crack growth l1a, the dynamic yield stress O'y and the dynamic flexural
modulus Ed, the loading parameters J d and 6d can be calculated. Both the input
values and the calculated values are printed and added to the testing protocol; the
F-f diagrams can be plotted, as well as exported as ASCII files.
4.4.2 Determination of the Loading Parameters
For the determination of the loading parameters (J, 8) of the crack resistance curve
the following equations are used.
1. JValues.
Ae1 ApI ( (0.757]el-I)l1a)
J d =7]el B(W-a)+7] PI B(W-a) 1 (W-a) , (A65)
where
7] el = 0.5 + 5.5 (a/W)- 5 (a/W)2 . (A66)
For the determination of 7].1. (A44) can also be used. The geometrical function
7]pl can be determined according to (A45).
2. Critical Crack-Opening Displacement.

The determination of the critical crack-opening displacements 6d and 6dk is carried
out using the relations (A52) and (A53).
4.4.3 Determination of Valid Crack Resistance Curves
For the determination of the crack initiation values (JO.2, ~.2)' the following
evaluation algorithm is used in accordance with ESIS TC4, 'A Testing Protocol
for Conducting J-Crack Growth Resistance Curve Tests on Plastics' [18] (see also
[19].
1. Plot the functions J and 8= f(Aa).
2. Draw lines parallel to the J and 8 axis at Aa = 0.05 mm (Aamin in Fig. A67).
3. Determine Aamax using Eq. (A67):
Aa = O.I(W - a). (A67)
4. Draw second line parallel to the J and 8 axis at Aamax .

5. Check the data point distribution; the data points should be spaced as shown in
Fig. A67.
~a."in =0.05 mm ~amax =0.1 (W-a)
J=Cl~aC2
1\= Cl~aCz
10.2
1\0.2
~a=0.2mm
Fig. A67. Validity limits, data point distribution and detennination of the crack initiation value
6. Curve fitting.
A fit curve is plotted through the data points falling between the 0.5 mm offset
exclusion line and the Aamax exclusion line, using a power law «A68) or
(A69»:
J and 8 = CI Aa C2 , (A68)
J and 8 = C 1(C 3 + Aa)C2 • (A69)

7. Checking of the validity of the J and the 8 values.

An upper limit to the validity range (Jmax or JG ) within which crack growth is
considered to be J-controlled must be determined. As seen in Fig. A68, this oc-
curs where the fit curve from (A68) or (A69) intersects either the f'!..a exclusion
line (JG) or a line drawn parallel to the f'!..a axis through J max • The value of Jmax
is the smaller of the two values in (A70):
J max = (W -a}ad and J max = Bad (A70)

20 20
The way in which the ~f'!..a curve is checked is analogous (Fig. A68), (A 71):
W-a B (A71)
8 max = - - and omax =-
50 50
8. Finally, the checking of the slope of the J-f'!..a curve is carried out at the point
Jmax or JG (see Fig. A68) according to (A 72):
W -a dJ I
{O = -J- d(f'!..a} ~ 10. (A72)
J max orG
When this condition is fulfilled, the J-f'!..a data points represent a material-
specific, geometry-independent crack resistance curve within the area delimited
by f'!..amin, f'!..a max , Jmax and JG • If {O < 10 the allowed maximum crack growth
f'!..amax reduces according to (A73):
(A73)
t case 1
""-;
!
I !
···T············································· ..........
lG
0G
I
i; !l.amin
!l.a- !l.a-
Fig. A68. Validity limits of J-controlled crack growth

The ~!l.a curve also meets all the requirements for a geometry-independent
crack resistance curve within the limits !l.arnax , !l.amin, ~ax, and Oc..
9. Determination of crack initiation.
According to this procedure, the crack initiation value is defmed as that value of
that J or 15 which is obtained from the point of intersection between the crack re-
sistance curve and a line parallel to the J or 15 axis, respectively, at !l.a = 0.2 mm.
These values of JO.2 and ~.2 are valid J and t5values under the following condi-
tions:
• There must be at least one J-!l.a or ~!l.a data point between 0.2 and 0.4
mm crack growth.
• JO.2 must be smaller than or equal to Jrnax; ~.2 must be smaller than or
equal to ~ax'
• The J O.2 value must meet the size criterion for validity of the concept of
the J-integral (A 74):
25J
B;a;(W -a) >~. (A74)
O"d
• The ~.2 value must meet the size criterion for validity of the concept of the
CTOD (A75):
B;a;(W -a» 2515 0.2 , (A75)
• The slope of the J-!l.a curve at !l.a = 0.2 mm must be smaller than the
yield stress O"y ofthe material (O"y = O"d is used).
• None of the specimens should exhibit brittle cleavage fracture before
maximum load.
4.4.4 Determination of JTJ
The quantification of the energy dissipation processes is carried out on the basis of
the JTJ concept of Will and Michel [20,21] using the following formalism:
1. Plotting the data points.
2. Fitting the J-!l.a data points to a function like the following one:
J=JC 4 +Cs!l.a. (A76)
3. Checking the validity of the crack resistance curve:

(a) Checking of the slope criterion at the maximum amount of stable crack
growth determined experimentally !l.aexp (A 77):
W-a dJ I
OJ = - J - d(!l.a) !!.a »1. (A77)
exp
(b) If there are differences between the data points and the regression function
determined at higher values of crack growth, !l.arnax and wexp must be deter-
•
J'=C.+CsAa
Fig. A69. Experimental determination of Ii)
mined with the help of a J 2-l1a diagram (Fig. A69). After doing this, a new fit-
ting of the data points must be carried out for l1a:;; l1amax according to (A76).
4. Determination of the value of JTJ.
The determination of the JTJ value follows (A 78). This parameter is a geome-
try-independent fracture mechanics parameter which allows a direct compari-
son of crack resistance curves in relation to the resistance of the material
against crack growth, presupposing that the resistance curve is valid according
to (A77):
1 Ed (A78)
JTJ =-Cs -
2 (}'2
d
The tearing modulus TJ [22] is calculated in accordance with (A79), and J = f(l1a)
according to (A76):
(A79)
References
I. DIN EN ISO 179: Plastics - Determination of Charpy Impact Properties - Part 1 (2000):
Non-Instrumented Impact Test. Part 2 (2000): Instrumented Impact Test
2. ASTM E 399 (1990): Standard Test Method for Plane-Strain Fracture Toughness of Metallic
Materials. Annual Book of ASTM Standards, Philadelphia Vo!. 03.01
3. Sumpter J. D. G., Turner C. E. (1976): Cracks and fracture. In: ASTM STP 601: 3-18
4. Merkle J. G., Corten H. T. (1974): J-integral analysis for the compact specimen, considering
axial force as well as bending effects. J. Pressure Vessel Techno!. 96, 4: 286-292
5. Hoffinann H., Grellmann W., Zilvar V. (1986): Instrumented impact studies of some ther-
moplastic composites. In: 28th Microsymposium on Macromolecules and Polymer Compos-
ites, Prague, July 1985. Walter de Gruyter, Berlin, New York: 233-242
6. Grellmann W., Jungbluth M. (1987): Application of the crack-opening displacement concept
for determining geometry-independent fracture mechanical parameters in the instrumented
Charpy impact Test. In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC)
Series No. 37, Institut fUr Mechanik, Berlin, Chemnitz: 186-192
7. Grellmann W. (1982): Problems and results of instrumented Charpy impact test of polymers.
In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Series, No.3, Institut
flir Mechanik, Berlin, Chemnitz: 102-111
8. Grellmann W. (1982): Problems to investigate fracture loads and inertial forces for the
determination of dynamic fracture toughness of polymers. In: Fracture Mechanics, Micro-
mechanics and Coupled Fields (FMC) Series, No.3, Institut fUr Mechanik, Berlin, Chem-
nitz: 142-151
9. Grellmann W., Sommer J.-P., Hoffinann H., Michel B. (1987): Application of different J-
integral evaluation methods for a description of toughness properties of polymers. In: Pro-
ceedings of the 1st Conference on Mechanics, Prague, Czech, June 29-July 3, Vol. 5: 129-
133
11. Grellmann W., Seidler S. (1991): Application of ESIS-procedure for determining the fracture
resistance of plastics. In: Proceedings of the 10th Congress on Material Testing, Budapest,
Hungary, October 7-11, Vol. II: 385-390
on tough polymers. Int. J. Fracture, Lett. Fract. Micromech. 96: L17-L22
13. Grellmann W., Seidler S., Bierogel C. (1997): Geometry-independent fracture mechanics
values as a requirement for toughness optimization of polymers. In: Proceedings of the 9th
International Conference on Fracture (ICF 9), Sydney, Australia, April 1-5, Vol. 2: 1013-
1020
ness characterization of high impact thermoplastics. Polym. Test. 14: 453-469
15. Sunderland P., Kausch H. H. (1988): The application of fracture mechanics to the impact
behaviour of rubber-toughened polyamides. Macromol. Chern., Macromol. Symp. 16:
365-378
16. Savadori A., Bramuzzo M., Marega C. (1984): J integral analysis of ductile fracture of
PP/EP rubber blends. Polym. Test. 4: 73-89
17. Grellmann W., Seidler S., Schierjott U., Rutke B. (1988): Anordnung zur Bestimmung des
J1J-gesteuerten RiBwachstums bei schlagartiger Beanspruchung. German Patent DD 275 113
19. Pavan A. (1998): ESIS TC 4 activity on high rate testing of plastics. In: Proceedings of the
12th European Conference on Fracture, September 14-18, Sheffield, U.K., Vol. ill: 1363-1368
20. Will P., Michel B., Schaper M. (1990): Justification of non linear J-resistance curves. Engng.
Fract. Mech. 37, 2: 275-281
21. Will P. (1994): R-curves of energy dissipative materials. J. Mat. Sci. 29: 2335-2340
22. Paris P. C., Tada H., Zahoor A., Ernst H. (1977): The theory of instability of tearing mode of
elastic-plastic crack growth. In: ASTM STP 668: Elastic plastic fracture: 5-36
A 2.3 Possibilities and Limits
of Standards and Drafts for J-R Curve
Determination of Polymers
S. Seidler, Vienna, and W. Grellmann, Merseburg
1 Introduction
The fracture mechanics behaviour of polymer materials has been intensively re-
searched in recent years [1-4]. However, the test results have not been fully un-
derstood yet. Furthermore, the practical applications of fracture mechanics pa-
rameters to assessment of structural integrity are very restricted owing to their
limited transferability from specimens to components. Moreover, with the excep-
tion of the ESIS TC4 recommendation for plastics [4,5], no standardized test
procedures for carrying out fracture toughness testing on tough polymer materials
exist, and hence methods developed for metallic materials have simply been
adopted. All of this has induced great efforts to seek geometry-independent frac-
ture mechanics parameters.
Several fracture mechanics parameters have been proposed for the evaluation
of crack resistance behaviour. Among these, the J-integral and COD (crack-
opening displacement) have mostly been used to study polymer materials which
exhibit significant crack tip plastic deformation during the process of fracture. A
combined application of the J-integral and COD concepts has been introduced by
our working group on thermoplastics, particle-reinforced composites and fibre-
reinforced composites under impact loading. The J-integral describes an energy-
determined fracture process and the COD a deformation-determined fracture
process. The combined application of these parameters contributes to better under-
standing of the fracture behaviour [6,7].

The determination of fracture mechanics parameters for the quantification of
stable crack growth requires a knowledge of the connection between a loading
parameter and the amount of stable crack growth Aa. For elastic-plastic material
behaviour, the concepts of yield fracture mechanics are used, i.e. the J-integral or
the CTOD is the loading parameter.
The determination of the J values on the basis of an energetic definition may be
performed by several approaches. Some of these different approaches are used in
the standards and drafts for determination of J-R curves, as listed below.

88 S. Seidler, W. Grellmann
• ASTM 813-81 [81.
J1= 1]Ao
B(W -a)
(1- [f(a/W)-I]Aa),
W-a
(A80)
with 1] = 2 for SENB; a/W> 0.45.

• ASTM 813-89 [91.
J = K2(I-v 2
2 E
L 2Apl
B(W -a)'
(A81)
with
(A82)
and
f(a/ W) = ~ (a/ WY12[1.99-(a/ W)(I- a/ W) g.15-3.93a/ W +2.7(a/ wf 1]

2 (1+2a/W)(I-a/WY12'
(A83)
following [10]
f(a/W) =2.9(a/W)1/2 -4.6(a/W)312 +21.8(a/W)5/2

(A84)
-37.6(a/wf/ 2 +38.7(a/W)912
• DVM-Merkblatt 002 [11].

-
J 3- 1]AG
(A85)
B(W -a)'
• ESIS P2-92 D [121. ESIS TC4 Testing Protocol [51.
J E =J3(1- (O.751]-I)Aa). (A86)

W-a
A comparison of some experimental methods for J-integral determination using

the material resistance against unstable crack growth with the numerically com-
puted J-integral shows that it is necessary to divide the general deformation en-
ergy AG into its elastic and plastic parts [13,14]. Such investigations have not been
performed for a stable crack propagation process. Some other authors [15,16] have
pointed out the necessity for such a procedure. Therefore, (A86) was modified as
follows:
Possibilities and Limits of Standards and Drafts 89
J - Aci ApI (1 (O.7S l1 -1)ila) (A87)

4 -l1cl B(W -a) +l1pl B(W -a) - W-a '
with
l1cl =0.S+S.5(a/W)-S(a/W)2. (A88)
Here l1el and l1pl are geometrical functions; l1pl = 2. Ael and ApI are the elastic and
the plastic parts of the general deformation energy up to the maximum impact
load.
The experimental values which are necessary for J-integral determination are
summarized in Fig. A70.
The specified approaches lead to different J values because of the different
ways of considering the elastic and plastic parts of the general deformation energy
and the different correction functions for stable crack growth.
Since the crack length changes continuously during a J-R curve test, the J-
integral must be calculated incrementally by means of an iterative solution:
J. EJ.
It 1+1
=JCI+l1(A:!I-ATlt[JPI+l1(Ai~I-Arl)l[l_yai+I-ai]
I B(W-a) B(W-a)
I W-ai'
(A89)
where 11 = 2 and y= 1 for SENB specimens [17]. This iterative solution was used
for checking the different J-integral approaches.
Figure A71 shows the results of a comparison between the various J-integral
approaches and the iterative solution using results from nylon and ABS resin
[18,19]. Over the whole range of stable crack growth, the best agreement with the
iterative solution was achieved by applying (A87).
PC/ABS Blend
Fmox .
~I --~~d-~~~-·-on--~~~--~~~..------ O.Smm
Fig. A70. Typical load-deflection diagram for elastic-plastic material behaviour and stable
crack propagation; the experimentally measured quantities used for determining
fracture mechanics parameters are shown
3.0
Nylon
2.5
2.0
:::a
.., 1.5 -:::::- J 3/JII
1.0
J 1/J 11
7"" "' J 4 /J It
0.5
0 0.2 0.4 0.6 0.8
~ (mm)
2.0-r-_ _ _ _ _ _ _ _ _ _ _ _ _--,
..,':2.- 1.0
0.5 f------=-t:------:-!::-----....,..
0.6 1.2 1.8
~ (mm)
Fig. A 71. J values obtained from the different estimation approaches, nonnalized to J;\
In addition to the different approaches to the J-integral, the standards and drafts
place different requirements on the construction of valid crack resistance curves.
There exist different limits for the stable-crack-growth values and for the J values.
Since small amounts of crack growth are difficult to measure and hence are sub-
ject to error, exclusion lines are used to determine Aamin. ASTM E813 [8,9] rec-
ommends fixed offsets for the minimum and maximum crack growth. In ESIS P2-
92, for ductile materials [12], the limit of Aamax is fixed at 0.06(W-a). For plastics,
this limit is too restrictive, and therefore the maximum-crack-growth criterion is
fixed at 10 % of the remaining ligament, i.e. Aamax = O.I(W-a). Above this limit,
J-controlled crack growth cannot be guaranteed. The J values are valid if they
satisfy the following criterion:
cr
J max = B; (W - a)-.l.. , (A90)
E
where E= 25 [8,9,11,12] or &= 20 [5].

In addition to ASTM E813-81 [8], a power law ofthe form
(A91)
can be used for fitting the data points falling between the exclusion lines J rn • x ,
~amin and ~arnax.
Figure A72 shows the application of the different standards and drafts to the
crack resistance curve of an ABS resin. Only the valid data points are considered.
All J values obey (A90).
Crack initiation values were determined according to the instructions in the
various standards. The different J-integral approaches and the different crack
growth limits lead to different J-~a data points. This produces differences in the
crack initiation values obtained (Table A3). The fixed offsets for the minimum
and the maximum crack growth are especially disadvantageous. Since the liga-
ment can be of any size, the fixed offsets will not always guarantee the validity of
the J-R curves (Fig. A73). Therefore, only crack growth validity limits which
consider the remaining ligament are useful.
Another critical requirement is expressed as follows (A92):
W-a dJ
m=----~lO. (A92)
J d(~a)
10 10
~mln .a.amln
8 8
I
E 6 E 6
~
~ 4
.....E
~ 4
I
-;- -;- J = 5.4 .a.aO.8 /
J = 2.3 + 3.6~
2 2
0 0
0 0.6 1.2 1.8 0 0.6 1.2 1.8
~ (mm) .a.a (mm)
10 10
.a.amin .a.amax .a.amin .a.amax
8 8
E 6 E 6
~ J =6.7 ~0.6 ~ 4
~ 4 ~ J=5.9~0.7
-;- 2 -;- 2
0
0 0.6 1.2 1.8 0.6 1.2 1.8
.a.a mm .a.a(mm)
Fig. A72. Comparison of standards using ABS resin as an example

In several copper and aluminium alloys and in steel, OJ values between 4 and 14
were determined [20,21]. This means that the restriction OJ ~ 10 may also be an
unnecessary restriction on the range of validity of the J-R curves for metals. In
[22] this requirement was modified to
W-a dJ
OJ = -J- d(Aa) »1. (A93)
a 40 40
4&mln
30 30
E E
!....,
_ 10
20 Jrnax
=-2.4 + 56.84&
E 20
~
Jrnax
J
...., 10
0.6 1.8 1.2 1.2
4& (mm)
c 40 40
4&max 4&min 4amax
30 J =50.4 4&0.97 30 J =48.4 4&0.98
E E
.E 20
! !O
~
....,_ 10
....,_10
0 0
0 0.6 1.8 1.2 0 0.6 1.8 1.2
4& (mm) 4& (mm)
Fig. A73. Comparison of standards using nylon as an example
From these results, it can be seen that methodical investigations of the require-
ments on the slope of the J-R curves of polymers are necessary if a useful mate-
rial-specific criterion is to be obtained.
Table A3. Results of the comparison of standards
J, O.2Ib1 J ldo. 2
Jld Id
Material OJ
(N/mm) (N/mm) (N/mm)
[8] [9] [11] [5] [11] [5]

ABSresin 2.4 1.5 3.0 2.0 5.5 7.7
Nylon 15 10 17.6 17.6
As can be seen from Table A3, for some polymers under impact-loading condi-
tions, only the method of evaluation following ESIS TC4 [5] enables the determi-
nation of crack initiation values.
3 Conclusions
For a polymer-specific evaluation of crack initiation values, the following proce-

dure can be suggested:
1. determination of J values following (AS7)
2. application of the b.a limits corresponding to the ESIS TC4 testing protocol [5]
3. verification of the slope using (A93)
4. determination of the crack initiation value as a technical crack initiation value
where the onset is taken to occur at b.a = 0.2 mm.
References
1. Anderson T. L. (1995): Fracture Mechanics. Fundamentals and Application. CRC Press,

Boca Raton
2. Kausch H. H. (1993): Advanced Thermoplastic Composites. Characterization and Process-
ing. Hanser, Munich
4. Pavan A. (1998): ESIS TC4 activity on high rate testing of plastics. In: Proceedings of the
12th Conference on Fracture, Sheffield, U.K., September 14-18, Vol. III: 1363-1368
5. Standard Draft ESIS TC4 (1995): A Testing Protocol for Conducting J-Crack Growth Re-
sistance Curve Tests on Plastics
Appl. Polym. Sci. 66: 1079-1090
different fracture mechanics concepts on examples of unoriented and cold-rolled polypropyl-
8. ASTM E 813 (1981): The Determination OfJIc, a Measure of Fracture Toughness
9. ASTM E 813 (1989): Standard Test Method for J Ic, A Measure of Fracture Toughness
10. Tada H., Paris P. C., Irwin G. R. (1985): The Stress Analysis of Cracks Handbook. Paris
Production Inc., Del Research Corporation, 2nd ed., St. Louis
11. DVM Merkblatt 002:1987 (1987): Ermittlung von RiBinitiierungswerten und RiBwider-
standskurven bei Anwendung des J-Integrals
12. ESIS-Procedure P2-92 (1992): ESIS Procedure for Determining the Fracture Behavior of
Materials
13. Grellmann W., Sommer J.-P. (1986): Toughness properties description of polymers using the
J-integral concept. In: Fracture Mechanics, Micromechanics, Coupled Fields (FMC) Series,
No. 17 Institut rur Mechanik, Berlin, Chemnitz: 48-72
14. Grellmann W., Sommer J.-P., Hoffinann H., Michel B. (1987): Applications of different J-
ceedings of the 1st Conference on Mechanics, Prague, Czech, June 29- July 3, Vo!' 5:
129-133
16. Hashemi S., Williams J. G. (1986): Fracture characterisation of tough polymers using the J-
integral method. Polym. Engng. Sci. 28: 760-767
17. ASTM E 1152:1987 (1987): Standard Test Method for Determining J-R-Curves
gener bruchmechanischer WerkstoftkenngroBen bei dynamischer Beanspruchung. Fortschritt-
Berichte, VDI Series 18: MechaniklBruchmechanik, VDI, DUsseldorf: No. 231
determination on polymers. In: Winkler T., Schubert, A. (Eds.) Material Mechanics - Fracture
Mechanics - Micromechanics, DDP Goldenbogen Dresden, Germany: 336-341
20. Anderson T. L., Stricklin L. L. (1989) A re-evaluation of criteria for J-controlled crack
growth. Int. J. Fract. 41: R65-R70
21. Hellmann D., Schwalbe K.-H. (1984): Geometry and size effects on J-R and 8-R curves
under plane stress conditions. In: ASTM STP 833, Philadelphia: 577-f>05
22. Will, P., Michel B., Schaper M. (1990): Justification of nonlinear J-resistance curves. Engng.
Fract. Mech. 37: 275-281
A 2.4 The Relationship Between the Fracture
Behaviour and Structural Parameters
of PE-HD
E. Nezbedova, J. Kucera and Z. Salajka, Bm6
1 Introduction
Structural and morphological parameters, together with processing conditions,

determine the type of fracture behaviour that occurs in polyethylene. Long-time
brittle failure, which occurs under low stresses and at room temperature, limits the
lifetime of polyethylene pipes used for water and gas distribution. In order to
optimize the lifetime of such pipes, it is necessary to apply methods aimed at a
comprehensive understanding of their structure and their fracture properties [1,2].
The lifetime of PE-HD is influenced by a large number of structural and mor-
phological parameters. The primary parameter is the structure of the chains:
• the molecular weight and molecular-weight distribution
• the number, type and distribution of short and long chain branches
• the type and number of unsaturated bonds.
The molecular weight and the molecular-weight distribution are usually esti-
mated through GPC (gel permeation chromatography) analysis. The average den-
sity of chain branches can be estimated on the basis ofDSC (differential scanning
calorimetry) records after an SIS (stepwise isothermal segregation) procedure [3].
The structure of the polymer chains affects the fracture behaviour ofPE-HD either
directly (by intercrystallinks) or indirectly (through the morphology). Further, the
structure and morphology of the polymer influence significantly the processes
taking place during processing (palletizing, pipe extrusion). The structure and
morphology of PE-HD, together with the service or test conditions (temperature,
rate, environment), determine the type offracture.
Polyethylene specimens fail at room temperature under low stress in a brittle
mode after a period of time. The failure mode is called 'long-time brittle failure'
[4]. A PE-HD pipe undergoes the same mode of the fracture after a long time in
service, which limits its lifetime. The crack initiation stage and the rate of slow
crack growth play an important role in determining the lifetime.
It is clear that applications in pressurized pipes require polymers with very long
lifetimes, but the problem is to devise a test which will estimate the lifetime of PE-
HD resins in a relatively short time.
Some accelerated tests utilizing fracture mechanics have been developed in the
last ten years [5~8]. The polyethylene notch test (PENT) [5,6], developed by Brown

96 E. Nezbedova, J. Kucera, Z. Salajka
and his co-workers, produces the same type of brittle fracture in test specimens as
that which occurs in pipes after a long time in service. The further tests utilizing
fracture mechanics are the notch-pipe test (NPT) [7] and the full-scale test (PST)
[8], which employ a part of the pipe as a test specimen.
PENT tests, together with structural tests, were used to determine the correla-
tion between fracture and structural parameters and to assess the quality of pipes
in service.
2 Experimental
A set of 206 samples from pipes which had been in service since 1968 was ana-
lysed by the SISIDSC procedure. The samples were taken from the pipe surface.
The pipes had an outside diameter of 40 to 225 mm. Additionally, 15 samples of
the same pipes were taken, from which PENT specimens were machined. All the
samples were supplied by the Czech Gas Company. As standards, four PE-HD
blends with different average densities of short chain branches were prepared by
mixing two types ofPE-HD. These polymers were prepared on a silica-supported,
Cr-based catalyst (UCC-S-2) using a pilot reactor. Compression-moulded plaques
(130 x 130 x 10 mm3) were prepared from these standard materials according to
the method described in [9].
2.1 Structural Characteristics
The structure of the PE-HD samples was characterized by the SISIDSC method
[3,9]. GPC analysis was also applied to 65 samples. The reason for doing so was
to exclude any content of bimodal materials, because the SISIDSC record of such
materials resembles the record of a homopolymer with a monomodal distribution.
The PE-HD samples for the SIS procedure were placed in aluminium cases be-
fore performing DSC. The cases were sealed into a glass ampoule under an inert
atmosphere of ultra-pure nitrogen, and the samples were fractionated by means of
a selected temperature programme. The contents of the separated fractions after
fractionation were evaluated with the aid of DSC (Perkin-Elmer DSC 7, scanning
rate 10 Klmin). The DSC record gives several peaks depending on the number of
temperature halts (Fig. A74). The locations and areas of these peaks correspond to
the average density of short chain branches (SCBs) in each individual fraction and
to the representation of this fraction in the polymer.
The density of SCBs in an individual fraction can be estimated for a butene co-
polymer (without carbon black) according to the following equation [3]:
SCB = (136 - TpeaIJ 11.6 (A94)
Relationship Between Fracture Behaviour and Structural Parameters 97
Peak I, with the highest melting temperature, corresponds to the homopolymer

fraction of the copolymer, and the area of this peak increases after processing. The
content of homopolymer must be in balance with the copolymer fraction because
it influences significantly the stiffness and fracture behaviour of the resin (too
small a content of the homopolymer fraction decreases the stiffness; too high a con-
12000 r---------------------------,
II
Scanning Rate: 10 Klmin 132.0 I
I
Sample Weight 4.03 mg
III
IV
116.7
I
128.9
120.9
o ~----~-----~-----~----~
100 110 120 130 140
Temperature eC)
Fig. A74. DSC record after SIS, using as an example a pipe containing carbon black
12000
133.3 134.2
'C'
j 8000
.~
.!!
g
~
£
1;j
4000 without
Carbon Black
~
Carbon Black
0
110 llS 120 125 130 135 140
Temperature eC)
Fig. A75. DSC records after SIS - comparison between samples with and without carbon black
tent of the homopolymer fraction supports brittle fracture). Peak II, at a slightly
lower temperature, corresponds to a copolymer fraction with very low content of
SCBs and relatively high molecular weight. Peak III corresponds to a copolymer
fraction with a higher content of SCBs and a medium molecular weight. Fraction
IV, which crystallizes after cooling the sample to room temperature, consists
mostly of low-molecular-weight chains with a high content of comonomer.
Figure A75 shows that Tpeak generally decreases in the presence of carbon
black. Thus, we derived (A95) for the evaluation of the SCB density in pipes con-
taining carbon black. The result of this type of calculation shows a good correla-
tion with IR determinations of the SCB density:
SCB = (133.2 - Tpeak ) 11.6 (A95)
2.2 Fracture Behaviour
For evaluation of 'long-time brittle failure', the methods described in [5] were
applied, using SENT (single-edge-notched tension) specimens. The shapes and
dimensions of two specimen types are shown in Fig. A76.
(a)
3.5
Axial Specimen Tangential Specimen (b)
I 25.0
I
LJ 25.0
Fig. A 76. Schematic illustrations of specimens prepared from (a) moulded plaque and (b) pipe
(dimensions in millimetres)
The notch depth was chosen to minimize the failure time but not to produce ex-
cessive creep in the remaining ligament. The notch was made by pressing a fresh
razor blade into the specimen at a constant speed of330 ~mmin-l. The PENT tests
were carried out at 80°C and at a constant nominal stress of 2 MPa.
3.1 SIS/DSC Analysis
The DSC records of four standard blends are shown in Fig. A77. The standards
had the following average densities of short chain branches:
• standard A: SCB = 1.2
• standard B: SCB = 2.2
• standard C: SCB = 3.2
• standard D: SCB = 4.9.
The DSC records of 206 samples were compared with these standards accord-
ing to the values of the enthalpy of fusion between 120 and 140°C (dH120-140) and
according to the average density of short chain branches. It was possible to sort all
of these samples into four groups, each of which corresponded to a particular
standard blend. The DSC records of representative samples are presented in
Fig. A78.
The GPC analysis confirmed that none of the samples analysed was bimodal.
According to the results of structural analysis 15 pipes were chosen for PENT
tests.
12000
1i'
or>
e
z::I 8000
~
.~
g
~
0
li: 4000
'iii
~
o~----------~------------~------------~
100 120 140 160
Temperature ("C)
Fig. A77. DSC records for standards

1 2 0 0 0 - - - - -- - --------------.
Group 2
I ",Group 1
o L------~------~-----~
100 120 140 160
Temperature (oq
Fig. A78. DSC records for representative samples from pipes
3.2 Fracture Behaviour
The PENT tests were carried out on compression-moulded SENT specimens of

standard blends. A set of specimens from 15 pipes was prepared. The specimens
were prepared parallel and perpendicular to the extrusion direction (Fig. A76b),
taking account of the recommendation of [6]. An example of the PENT test results
is given in Fig. A79.
The times to fracture of the specimens of group 4 varied from 4300 min to
34 000 min. The specimens which had the shortest time to fracture at 80°C were
also tested at 40 °C. The measurement at two temperatures enables us to extrapo-
late the time to fracture to other temperatures. A method based on the Arrhenius
equation (A96) was used for this extrapolation:
tf = A4 e Q / [R(T+273)]. (A96)
Here, A4 is a constant which includes the effect ofbtanch density [10], Q is the
activation energy, T is the temperature and R is the gas constant. The temperature
40°C was chosen as a reference temperature. Equation (A96) was applied to the
experimental data at 80°C and 40 °C. In contradistinction to [10,11], the experi-
mental data did not give a constant value of Q. The dependence of Q on the time
to fracture te was approximated by (A97):
Q = 120 - ae- btr . (A97)
3.5 r-----------------------,
Group 2
3.0
!
8' 2.5 Group 1
-8
.13 2.0 /
g
...c::1.5
~;:.:::q
___. . . -.-
lJ
Group 4
~
I
Group3 \
~ 1.0
-e-a-<.
_ -_ -_ _
0.5
0.0 ' - - - - - ' - - - - - - ' " - - - " " " ' - - - -.........- - - - ' - - - - - - '
10- 1 102
Timet (min)
Fig. A79. Results of PENT tests for pipes
As a value of Q oflimited validity [10], Q = 120 kJmor l was chosen. The cal-
culated values of the parameters a and b were a = 68.39 kJmor l and b =
1.53 X 10-4 min-I.
The calculated values of Q and the values of If for 40°C (see (A96)) were used
for estimating the dependence of A4 on If. The values of Q and A4 were used in
(A96) for extrapolation to 0, 20, 40, 60 and 80°C (Fig. A80), and the experimen-
tal data at 80°C were compared with the extrapolated data. A good correlation
over the range of times measured (short failure times) was found.
, ...... tcll80 -.-. trll60 -00-trlI 40 --- tcll20 -tcllO • tf 80 I

)0 10
------
'2
'.....,s 106
-----=::::::----- ---::.-.-~ ----------_.,-
-~
~
102
..
:~:.:=- ::~.:.:.:--:::~~.:::.-.- -.-.-.-.~:~:~:~:::~:~ :..........
102 104
tl140 (min)
Fig. A80. Extrapolation of time to fracture

4 Correlation Between Structural

and Fracture Parameters
The basic structural classification was done according to the average density of
short chain branches (SCBs) calculated from (A95). The samples with SCB < 3.7
exhibited low values of the time to fracture (Fig. AS1).
5 ~----------------------------------------------~
4 Specimens from Plaques
~3
)
I Sp",;m,n, (,om P;p"
2 r-------·-··---~
I0 2 3 4 5 6
Density of SCB
Fig. AS1. Correlation between the density of SCBs and the time to fracture for polyethylene-
butene copolymers
Although all samples in group 4 have SCB > 3.7, their times to fracture differ
pronouncedly. Therefore, we had to take into the consideration the further pa-
rameter dH1/dH2 , where dHl is the enthalpy of fusion of the highest-melting-
temperature peak and dH2 is the enthalpy of fusion of the second highest peak. If
dHl / dH2 < 1, the time to fracture is high. If, however, dHl / dH2 > 1, a high propor-
tion of the homopolymer fraction is present, and the time to fracture is lower (10 2
to 103 min).
It is clear from Fig. ASI that the time tfll for the pipe specimens is about one or-
der lower than for the standard blends, and the pipe specimens have SCB values
mainly between 1 and 3. The reason for the shift in time to fracture is the different
stress state at the notch tip and the higher ratio of dH1/dH2 for pipe specimens
compared with the standard specimens.
5 Conclusions
Two structural parameters were chosen for PE-HD: the average density of short
chain branches (SCB) and the value of the ratio dH'/dH2 (determined from
SISIDSC analysis). These parameters show a good correlation with the fracture
measurement. The evaluation of these parameters enables us to assess the quality
of pipe grade PE-HD as well as the quality of new pipes and pipes that have been
in service for a long time. The procedure for this assessment consists of the fol-
lowing steps:
• For monomodal material, an SISIDSC analysis must be carried out and the
parameters SCB and dH'/dH2 have to be calculated.
• If SCB < 4, the material is declared to be unsuitable.
• IfSCB > 4, the parameter dH'/dH2 is evaluated.
• For dH'/dH2 > 1, the material is declared to be unsuitable.
• For dH'/dH2 < 1, a PENT test has to be performed.
The extrapolation of times to fracture to a temperature of 80°C resulted in a
good correlation (in the short-lifetime region) with the observed experimental
data.
Acknowledgement
The authors thank the Czech gas industry and the chemical company Chemope-
trol, Litvinov, Czech Republic, for sponsorship of this project.
References
1. Egan B. J., Delatycki O. (1995): The morphology, chain structure and fracture behaviour of
high-density polyethylene: part I. Fracture at a constant rate of deflection. J. Mat. Sci. 30:
3307-3318
2. Egan B. 1., Delatycki O. (1995): The morphology, chain structure and fracture behaviour of
hjgh-density polyethylene: part II. Static fatigue fracture testing. J. Mat. Sci. 30: 3351-3357
3. Adisson E., Ribeiro M., Deffieux A., Fontanille M. (1992): Evaluation of the heterogeneity
in linear low-density polyethylene comonomer unit distribution by differential scanning ca-
lorimetry characterization of thermally treated sample. Polymer 33: 4337-4342
4. Huang Y.-L., Brown N. (1988): The effect of molecular weight on slow crack growth in
linear polyethylene homoploymers. J. Mat. Sci. 23: 3648-3655
5. Lu X., Brown N. (1992): A test for slow crack growth failure in polyethylene under a con-
stant load. Polym. Test. 11: 309-319
6. ISOrrC138/SC4 N 144: Notch Tensile Test to Measure the Resistance to Slow Crack
Growth of Polyethylene Resins (PENT). First version of ISO document
7. Draft International Standard ISOIDIS 13 479 (1994): Polyethylene (PE) Pipes for the Con-
veyance of Fluids - Resistance to Crack Propagation - Method of Test for Slow Crack
Growth on Notched Pipes (Notch Test)
8. Draft International Standard ISOIDIS 13 478: Polyethylene (PE) Pipes for the Supply of
Natural Gas - Resistance to Crack Propagation - Determination of the Critical Hoop Stress
for Rapid Crack Propagation (Full-Scale Test)
9. Nezbedova E., Salajka Z., Kucera J. (1997): Relation between toughness and structural
parameters of PE-copolymer. In: Proceedings of the 1st International Conference on Weld-
ing Technology, Materials and Materials Testing, Fracture Mechanics and Quality Manage-
ment, Vienna University of Technology, Vienna, Austria, September 18-22, Vol. 2:
507-516
10. Brown N. and Lu X. (1989): The Kinetics and Microscopic Processes of Long Temi Fracture
in Polyethylene Piping Materials. Annual Report 90 GRl 90/0104, University of Pennsylva-
nia, Philadelphia, PA: 19104-6272
11. Triinkner T., Hedequist M., Gedde U. W. (1996): Structure and crack growth in gas pipes of
medium-density and high-density polyethylene. Polym. Engng. Sci. 36: 2069--2076
A 3 Alternative Methods
A 3.1 Application of Single-Specimen

Testing Methods for Determining
J-R Curves of Polymers
S. Seidler, Vienna
1 Introduction
Fracture mechanics materials testing allows the determination of characteristic

values which can be adapted to the behaviour of structures under real loading
conditions, and it can ensure adequate safety with respect to the different possible
modes of failure. Depending on the material behaviour, different fracture me-
chanics parameters can be used; the concept of crack resistance has been shown to
provide a quantitative description of the fracture behaviour of ductile thermosets
[1]. The critical crack initiation values determined on the basis of the various stan-
dards and drafts [2-6] are of special practical value.
The experimental investigation of R-curves may be performed by either the
single-specimen method [4,5] or the multiple-specimen technique [2,3,5,6]. Both
experimental techniques require a large experimental effort, and the multiple-
specimen technique requires a large number of specimens. In addition to the large
experimental effort required to ensure reproducibility of the results, there is a
relatively large subjective element in the determination of the amount of stable
crack growth lla in polymer materials. For these reasons, we obviously would like
to fmd simpler methods for the determination of crack resistance curves. In the
literature, various authors [7-14] present suggestions for the determination of J-R
curves, i.e. crack resistance curves related to the loading parameter J. However,
the biggest difficulty lies in the determination of the crack initiation point in the
load-displacement diagram.
The starting point of the considerations published in [7-13] is a smoothed load-
displacement diagram of a ductile material. Smoothing, i.e. removal of system
oscillations, is especially necessary if the load-displacement diagram is determined
in the instrumented Charpy impact test.
A further possibility for the investigation of J-R curves is provided by the
combination of the determination of crack toughness as resistance against crack
instability as a function of test temperature with microfractographic investigations.
For that purpose, a procedure was proposed [14] for the determination of J-R
curves from the dependence of the Charpy impact strength on the test temperature.

106 S. Seidler
In the following, selected approximate methods for determination of J-R curves

[S-II,14] are considered regarding their applicability to polymer materials. The
possibilities and limits of the normalization method [7] and of the calculation on
the basis of the hinge specimen [14] are discussed extensively in pp. 121-132 and
pp. 133-140.
2 Single-Specimen Methods
2.1 Method of Schindler [8,9]
2.1.1 Approximation of the Crack Resistance Behaviour Following

ASTM 813-81 (Schindler I [8J)
In the approximate method of Schindler, the crack resistance behaviour of a duc-

tile material is estimated from its load-deflection behaviour in the instrumented
Charpy impact test (ICIT).
The starting point of this approximation method is a load-deflection (F-j) dia-
gram without oscillations (Fig. AS2). The area under this curve represents the
deformation energy A tot • A tot can be divided formally into an elastic component A ...
a plastic component ApI and, after reaching the maximum load F max, a component
corresponding to the crack propagation energy A R •
Fmax - --a
FlO'
Deflection Deflection
Fig. A82. Load-deflection diagrams for elastic-plastic material behaviour with crack propa-
gation energy shown, illustrating the values measured for determination of fracture
mechanics parameters (a) experimental, (b) smoothed
The considerations developed by Schindler and Morf in [S] assume that it is

sufficient in many cases to consider the crack initiation and crack propagation
processes as approximately linear. From this assumption, a J-R curve can be de-
termined, and it can be evaluated following ASTM ES13-S1 [15]. The character-
Application of Single-Specimen Testing Methods 107
istic points required are detennined by the basic connections between the energy
components of the F-fdiagram represented in Fig. A82 (A98):
(A98)
At the maximum load F max, the maximum 1 value 1m is detennined by (A99):
1m =1 mel + 1 mpl ' (A99)
with
(AlOO)
(AlOl)
and
1 = --,-2_A--,--p_I'"
mpl B(W -a)' (Al02)
The geometrical function f(alW) for a single-edge-notched bend (SENB)

specimen with s/W= 4 can be calculated according to [16] from (AlO3).
f(a IW)= 2.9(aIW)l/2 - 4.6 (aIW)3/2 + 21.8(aIW)5/2 - 37.7 (alw)1/2

(AlO3)
+ 38.7 (aIW)9/2.
The amount of stable crack growth at the maximum of the load-deflection dia-
gram, Llam , which corresponds to 1m, can be calculated following (AlO4):
Lla m = nC2(W -a) (AI04)

(2+nC 2 ) ,
where n is the work-hardening factor; n = In (1 - eGl), where eGl is the unifonn

elongation without necking. C2 is the exponent of a power-law graph of the l-R
curve. The exponent C2 takes values between 0.3 and 0.75 [8].
The slopes SI of the blunting line 11 = f(Lla) and S2 of the crack propagation line
12 = f(Lla) are abtained from the following equations:
11 = sl Lla, (AlO5)
sl = 2 O'F, (AlO6)
12 = 10 + s2 Lla, (AlO?)
4 (A tot - ApI)
s2 = (AlO8)
B(W _a)2
108 S. Seidler
By this procedure, a crack initiation value Jl can be detennined from the point
of intersection between the blunting line and the crack propagation line:
{Atot - ApI )Llarn }

2 Frnax 8{ J rnp1 -4 ( )2
S B W-a (AI09)
Jd = () + J rne1
2 Frnax 8 - 4c Atot - ApI
The constant c takes the values c = 1 under the condition of a plane stress state
and c = 1.46 under the condition of a the plane strain state. On the basis of the
ductile material behaviour during stable crack propagation, the plane stress state
can be assumed, i.e. c = 1. Using two straight lines with the slopes 81 and 82, a J-R
curve can be estimated.
2.1.2 Approximation of the Crack Resistance Behaviour

Taking Account of the Non-Linearity
of the Crack Propagation Process - Schindler /I [9J
In [9] an extension of the approximate method presented in sect. 2.1.1 is de-

scribed. In this extension it is considered that a linear fitting reflects the real crack
propagation process only partly. Following the known standards and drafts [2,4-
6J, the crack resistance behaviour is described by a power law, presented in pp.
71-86. The detennination of the crack initiation values is carried out at Lla = 0.2
mm. From this, it follows that
(All 0)
The constant C1 is estimated from the ratio of Jm to the corresponding amount

of stable crack growth Llam:
(AI 11)
The calculation of Llam follows (AI04) here; the detennination of nand C2 fol-
lows (AI 12) and (AI 13):
n= ApI , (AI12)
Atot
C=~(I+~rl
2 (AI13)
2.2 Method of Kobayashi [10]
The starting point for the approximate detennination of the crack resistance curve
following Kobayashi et a1. [10] is also an F-f diagram recorded in the instru-
mented Charpy impact test with the oscillations removed (Fig. A82b). The crack
initiation point is determined from the compliance changing rate AC/Ce1 ; the
method is illustrated in Fig. A83 using an example of an acrylonitrile-butadiene-
styrene (ABS) material. The crack initiation point is defmed as the point where the
slope changes in the dependence AC/Ce1 = f(f). The corresponding crack initiation
value is determined from the area under the load-deflection diagram up to this
point following (Al14):
J= 7lA (1_(O.75 71 -1)Aa). (Al14)

B(W-a) W-a
The compliance changing rate is given by (AI15):
AC C - Cel
(All 5)
Cel Cel
with
C= df (AI 16)
dF'
dfgy
Cel = dF. . (All 7)
gy
10O r - - - - - - - - - - - - - - - - - - - " T 0 . 8
ADS
80 Crack Initiation -
0.6
60
~
Po.
0.4 ~ IJ
40
0.2
~-----L+_-----+_----__40
2 3
f(mm)
Fig. AS3. Detennination of the crack initiation point from the compliance changing rate !lC/Cel
Starting from the crack initiation point, J values are determined from the energy
under the load-deflection diagram up to the maximum load. The determination of
II 0 S. Seidler
the corresponding amounts of stable crack growth following the method in [10]
leads to physically unreasonable results for the polymers investigated. Therefore,
the determination of l1a follows (AIlS) [17]:
(AIlS)
2.4 Key Curve Method [11,12]
A description of the complete crack resistance curve is also provided by the use of
the key curve method [11,12]. The crack resistance curve is again determined from
the non-linear part of the load-<leflection diagram. The amount of stable crack
growth l1a is determined by application of a material-specific, general relation
between the impact load F and the plastic component of the deflection ,{plv (F, =
f(f/)) of a V-notched specimen without crack propagation. The investigation of
this relation for the determination of stable crack growth (AI 19) can be carried
both experimentally and numerically. Using this method for J-R curve determina-
tion, a technical crack initiation value at l1a = 0.2 mm can also be determined. The
equation is
(All 9)
3 Check of the Applicability of the Methods
3.1 Amorphous Polymers
A check of the methods was performed using, as examples, an ABS resin and a PC
(polycarbonate)/ABS blend containing 70 % PC (PC/ABS 70/30). The results for
both materials are given in Figs. AS4 and AS5 in comparison with the experimen-
tally determined J-R curve. Table A4 includes a summary of the characteristic
values. The fitting of the J-l1a data points followed, except for the method of
KobayashiIMoskala [10,17], a power law. Using the method of Kobaya-
shiIMoskala for the data point fitting, a square root function is practicable. In all
cases, validity limits were not considered.
For ABS (Fig. AS4), all methods used apart from the method of Kobaya-
shiIMoskala [10,17], lead to a good agreement with experiment. The crack initia-
tion value determined from the compliance changing rate (Fig. AS3) [10] is too
high and the slope of the J-R curve is too low.
Application of Single-Specimen Testing Methods III
ABS
J = 2.5 + 13.2 Ll.a
J = 13.5 M0 53
J= 131.8Ll.a
8 I
J = 11.6 M051
........
/ .
.,...............
J = 10.6 M052
•.•• ,.- •. -:;;;~;.~' J = (13.2 + 56 3 M)05
.;'<."'"
.-:<.'" - - Experiment
.:~>.... -------------- Schindler I [8]
~/' ...................... Schindler II [9]
---.-.----- Key Curve (11,12]
--.-.-.-.-.- KobayashiIMoskaIa (10,17]
o 0.2 0.4 o.
ila (mm)
Fig. A84. Comparison of an experimentally determined J-R curve for ABS with J-R curves
determined by application of various approximate methods
15r-------------------------------------~
.........
Pc/ABS 70/30 ...........
...........
J""'- - J/"'/:':_/?:p.-
12 •'->,>~.~'?::~:-::.<.-. -
I
2:; Jd ....... 'J = 19.7 D.a0 56
. ./ -- ... - ...-- ... -
Schindler 11 [9]
Key Curve [11,12)
-, 6 I ,. . . // -'-"--'-"-'" KObayashilMoskala [10,17)
.././.... J = 7.4 D.ao."
./
3 /1/
J = 7.0 D.aO.•9
o 0.2 0.4 0.6

Ll.a (mm)
Fig_ A8S. Experimentally and approximately determined J-R curves for a PCIABS 70/30 blend
By way of contrast, for the PCIABS 70/30 blend only the key curve method
leads to a good correlation with experiment (Fig_ A85)_ This result is due to the
fact that stable crack growth starts with loading (Fig_ A86). However, all models
are based on the assumption that stable crack growth starts above the general yield
112 S. Seidler
point F gy/fgy. This discrepancy leads to considerable overestimation of the energy

which is consumed by the material in crack initiation and crack propagation. How-
ever, taking account of this fact by assuming the whole F-f diagram to be non-
linear, leads to good agreement with experimental results.
In the ABS resin, on the other hand, the experimentally determined start of sta-
ble crack growth is near the point F gy/fgy (Fig. A87), and therefore experiment and
the approximate method agree very well.
The determination of the crack initiation point from a complete load-deflection
diagram is impossible with the methods used because the starting point of stable
crack growth is below the transition from 'elastic' to 'elastic-plastic' material be-
haviour.
For an approximate description of the J-R curve of polymers, the key curve
method seems to be the best. Such a crack resistance curve enables the determina-
tion of a technical crack initiation value. However, the experimental determination
of the key curve does not allow any reduction in the experimental effort, which is
an essential aim of the approximate methods. An effective use of this method is
only possible on the basis of dynamic finite-element calculations taking account of
strain-rate-dependent material behaviour.
A further possibility for the investigation of J-R curves is provided by the combi-
nation of determination of crack toughness as resistance against crack instability as a
function of test temperature with microfractographic investigations. For this purpose,
a procedure has been proposed [14] that was developed for the determination of J-R
curves from the dependence of the Charpy impact strength on test temperature.
Table A4. Summary of the results from comparison of experimental results and approximate
methods
Characteristic value ABS PC/ABS 70/30
JO.2 (N/mm) 4.6 2.3

Experiment
dJ/d(~) I0.2 (N/mm2) 11.9 8.4
Jl (N/mm) 2.8
Schindler I
JO.2 S (N/mm) 5.1
[8]
dJ/d(~) I0.2 (N/mm2) 13.2
Schindler II JO.2S (N/mm) 5.8 8.0
[9] dJ/d(~a) I0.2 (N/mm2) 15.2 22.4
J/ (N/mm) 3.6 6.8
KobayashilMoskala (N/mm)
Jo.l 4.9 9.6
[10,17]
dJ/d(~) I0.2 (N/mm 2) 5.7 12.1
Key curve JO.2C (N/mm) 5.1 2.3
[11,12] dJ/d(~a) I0.2 (N/mm2) 13.0 8.0
250~-----------------------------------' 2.5
POABS70130
200 2.0
150 1.5 E
§
g 100 1.0
os
<l
"'"
50 0.5
o 0
o 2
f(nun)
Fig. A86. Load-deflection behaviour of the PC/ABS 70/30 blend and experimentally deter-
mined amount of stable crack growth Ila versus deflection
120 1.2
ADS
90 0.9
60 0.6
g E
§
os
"'" 30 0.3
<l
0 0
0 2 3
f(nun)
Fig. A87. Load-deflection diagram of ABS and dependence of measured amount of stable
crack growth Ila on deflection
Figure A88 shows a comparison of experimental results obtained from both

methods for characterization of crack initiation and crack propagation behaviour,
using as an example the 60/40 PC/ABS blend at a test temperature T= -30 DC. All
results of this study are summarized in Table A5. The temperature T = -30 DC is
the limit at which no unstable crack growth occurs in the blends [18]. The method
114 S. Seidler
is suitable in principle. Its importance lies in the increase of the number of load-
deflection diagrams while, simultaneously, the experimental effort is minimized.
The differences in the technical crack initiation values and in the slopes of the
J-R curves can be attributed to the strong temperature dependence of the mechani-
cal properties of polymers.
Because, when using the method of [14], the data points result from experi-
ments at different test temperatures, a number of different yield stresses are con-
sidered in the obtaining results. This leads, in comparison with the crack resistance
behaviour at constant test temperature, to overestimation of the energies necessary
for the generation of small stable crack extensions at lower test temperatures. In
[14] a correction is given, but it is not practicable for polymers. Another aspect is
the possibility of a change of fracture mechanism in a small temperature range
[21], which leads to an unfavourable data point distribution.
PC/ABS 60/40 (T = -30°C)
....., J = 6.7 llaO. 41
4 1-----0.'"
--- Experiment
--0-- [14]
o 0.5 1.5
Ila(mm)
Fig. ASS. J-R curves of the PC/ABS 60/40 blend
Table AS. Fracture mechanics parameters of PC/ABS blends, detennined using all J-tuJ data
points
Jl 2 (N/mm) dJ/d(tuJ) I0.2 (N/mm2)

Material Experiment [14] Experiment [14]
PC/ABS 45/55 2.6 3.2 7.4 6.4

PC/ABS 60/40 1.9 3.5 8.0 7.1
PC/ABS 70/30 1.1 1.9 4.4 4.8
3.2 Short-Fibre Composites
The methods of Schindler [8,9] and Kobayashi [10] were investigated additionally
for their suitability for the description of the crack resistance behaviour of short-
fibre-reinforced polyamide (PA). Figure A89 shows the experimentally deter-
mined J-R curves as a function of fibre volume fraction ({Jv obtained using the
multiple-specimen stop block technique. The experimental amount of stable crack
growth is very small in these materials. With increasing fibre volume content, the
slope of the crack resistance curves increases, which indicates an increase of the
energy dissipation mechanisms in the material.
Using as an example the composite PA 6/GF, with ({Jv = 0.2, Fig. A90 shows a
comparison of experimentally and approximately (Schindler I [8], Schindler II [9]
and Kobayashi [10]) determined J-R curves. In Table A6, the results for the
whole concentration range investigated are summarized.
All models need the work-hardening factor n for estimation of the material be-
haviour. However, the determination of n is very difficult for polymers. For the
PA/GF composites investigated, the determination of strain-hardening behaviour
from the elongation without necking l'Gl is suitable. The elongation without neck-
ing l'Gl is determined from the stress-strain curve on the assumption of a homoge-
neous deformation of the specimen.
116 S. Seidler
5n.----------------------------------------~
PA 6/GF (</lv = 0.2)

...........
4 ................
J=2Aacry .........
3 .......
........
.......
.....
____-;f".;;.--:..,'-------------_--_-:-::-.:.------:-:;::::'::.-:.=====
. . . -' -
2 _" __,:'f":-::"'_"_"_"""-- -- _.
.-..-..- - - - - - Experiment
........
.............. Schindler I [8]
Schindler II [9]
................... KobayashilMoskala [10,17]
o 0.05 0.1 0.15 0.2 0.25
~a(mm)
Fig. A90. Comparison of experimentally and approximately determined crack resistance cW"Ves
Despite the deviations between the experimentally and approximately deter-

mined J-R curves, the models of Schindler [8,9] are in principle suitable for a
qualitative description of the crack resistance behaviour of the PA 6/GF compos-
ites as a function of fibre volume content.
Both procedures suggested by Schindler [8,9] enable the fitting of the J-R
curve with a straight line or a power law. Judging by the values of Aarnax and Jrnax,
the differences from experiment become lower if fitting by a straight line is used
(Table A6).
The application of the method of Kobayashi [lO] to the determination of the
crack growth from load-deflection diagrams leads, for the composites also, to
physically unreasonable results. For this reason the amount of stable crack growth
was determined by the method presented by Moskala [17]. The J-Aa curves de-
termined in this way have much higher J values than the experimentally deter-
mined curves (see Table A6).
The reference values for the comparison of the experiment with the model are
those of the crack initiation values Jd and the material resistance against crack
propagation TJ .
For the experimentally determined R-curves, physical crack initiation values
were ascertained from the point of intersection of the amount of stable crack
growth Aaszw with the crack growth curve. The amount of stable crack growth
Aaszw was determined by scanning electron microscopy. On this basis, technical
crack initiation values were determined at Aa = 0.01 mm (Fig. A91).
TableA6. Comparison of the experimental results with results of the Schindler I [8] and II [9]
methods
tpv=O tpv = 0.06 tpv = 0.12 tpv = 0.20

n 0.19 0.06 0.10 0.11
Experiment
&!max (mm) 0.23 0.15 0.19 0.12
Jmax (N/mm) 1.54 4.44 6.69 12.44
Schindler I
&!max (mm) 0.30 0.10 0.17 0.18
Jmax (N/mm) 2.60 2.8 8.3 23.5
Jl (N/mm) 2.30 2.7 7.4 21.1
Schindler II
&!max (mm) 0.30 0.10 0.17 0.18
Jmax (N/mm) 4.0 6.6 15.9 42.3
Ja.l s (N/mm) 2.0 6.4 11.4 28.8
KobayashilMoskala
6amax (mm) 0.21 0.20 0.24 0.23
Jmax (N/mm) 2.3 2.3 7.0 24.2
Ja.IK (N/mm) 1.8 2.0 5.7 20.5
The slopes of the crack resistance curves were determined uniformly from the
derivative of the fitted crack growth curve at Aa = 0.2 mm. The graph in Fig. A92
refers only to the non-linear fitting.
The normalized diagram (Fig. A91) shows that the approximate approaches
lead in principle to overestimation of the crack initiation toughness. Among other
possibilities, this could be explained on the assumption that stable crack growth
starts above the formally defmed transition from linear elastic to elastic-plastic
material behaviour. As already mentioned, in polymers stable crack growth occurs
below this transition point.
In contrast to the approach of KobayashiIMoskala [10,17], the approach of
Schindler [9] uses a power law for fitting the crack growth curve, but it also leads
to overestimation of the material resistance against crack propagation (Fig. A92).
The linear fitting by the Schindler I method [8] is not considered here because of
the different mathematical description of the crack growth curve.
While the Schindler II method [9] enables in particular the description of the
range of small amounts of stable crack growth in these materials, the combined
approach of Kobayashi and Moskala [10,17] is suitable for the description of the
stable crack propagation behaviour with increasing crack growth. Since material
evaluation follows the crack initiation behaviour, the Schindler II method [9] is
more suitable. The special advantage of this method is the unproblematic handling
of the data.
118 S. Seidler
20~--------------------------------------~
PA6/GF
16
Schindler I [8]
12
Schindler II [9]
___
, KobayashilMoskala [10,17]
. ------ ------
8 '--'" _------------------
"'," ~--- ..... _._._._.-..
4~-=-
o 0.05 0.1 0.15 0.2
<jlv
Fig. A91. Approximately determined crack initiation values compared with the experimentally
determined values as a function of fibre volume content
15r-------------------------------------~
PA6IGF
10
-..... ... .....-.......~~....~-..--.-............-.................-........................_........-
...................... Schindler II [9]
. ............. " "
-.............................--.... ----
5
...•............. KobayashiIMoskala [10,17]
~ 0 ..-._..-..-..-._..-..-..-..-..-..-..-..-._.-._.-..-..-..-.. _.-..-..------._._..---._..-..-._--._--..-.-..-..-.. -..-..--
o 0.05 0.1 0.15 0.2

<jlv
Fig. A92. Slope of the J-R curves at!la = 0.2 mm versus fibre volume content
A better agreement between experiment and approximate methods could be

achieved by a fundamental analysis of the strain-hardening behaviour of the model
materials, because all approaches discussed include the strain-hardening beha-
viour. In addition, fundamental examination of the defmition of the transition

point from linear elastic to elastic-plastic material behaviour is necessary.
4 Summary
The approximate methods described above have proved to be suitable in principle

for a qualitative description of the crack resistance behaviour of polymers. With
the methods presented, the determination of the crack initiation point from a com-
plete load-deflection diagram is impossible because in polymers stable crack
growth occurs below the formally defmed transition point from linear elastic to
elastic-plastic material behaviour (Fgy/fgy).
For an approximate description of crack resistance curves of polymers, the key
curve method seems to be most suitable. However, effective application of this
method requires dynamic fmite-element calculations on the basis of a strain-rate-
dependent material behaviour.
References
I. Grellmann W., Seidler S. (1992): J-integral analysis of fibre-reinforced injection moulded

thermoplastics. J. Polym. Engng. 11: 71-101
2. ASTM E813 - 89 (1989): Standard Test Method for J(c, A Measure of Fracture Touglmess
3. ASTM E1290 - 93 (1993): Crack-Tip Opening Displacement (CTOD) Fracture Touglmess
Measurement
4. ASTM El152 - 87 (1987): Standard Test Method for Determining J-R Curves
5. ESIS Procedure P2-92 (1992): Procedure for Determining the Fracture Behaviour of Materi-
als
7. Landes J. D., Zhou Z., Lee K., Herrera R. (1991): Normalization method for developing
J-R-curves with the LMN function. J. Test. Evaluat. 19: 305-311
8. Schindler H. J., Morf J. (1991): On estimation of fracture touglmess from instrumented and
standard Charpy V-notched tests. In: Proceedings of the 10th Congress on Material Testing,
Budapest, Hungary, 1991, Vol. 2: 172-177
9. Bohme W., Schindler H. 1. (1996): Single specimen methods to evaluate J-R-curves from
instrumented Charpy impact test: results of a round robin. In: Proceedings of the 11th Euro-
pean Conference on Fracture (ECFll): Mechanisms and Mechanics of Damage and Failure,
Poitiers-Futuroscope, France, September 3-6, Vol. 3: 1977-1982
10. Kobayashi T., Yamamoto I., Niinomi M. (1993): Introduction of a new dynamic fracture
touglmess evaluation system. J. Test. Evaluat. 21: 145-153
II. Ernst H. A., Paris P. C., Rossow M., Hutchinson J. W. (1979): Analysis of load-displace-
ment relationship to determine J-R curve and tearing instability material properties. In:
ASTM STP 677: 581-599
12. Ernst H. A., Paris P. C., Landes J. D. (1981): Estimation on J-integral and tearing modulus T
from a single specimen test record. In: Roberts R. (Ed.) ASTM STP 743: 476-502
13. Steiner R. (1997): Berechnung von J-R-Kurven aus Kraft-Durchbiegungs-Diagrammen auf
Basis des Gelenkpriltkorpers. Fortschritt-Berichte, VOl Series 18, No. 208, VOl, DUsseldorf
120 S. Seidler
14. Aurich D. (1987): Ennittlung des Initiierungsverhaltens und des RiBwiderstandes bei sta-
tisch und dynarnisch beanspruchten ISO-V-Proben zur Verbesserung der Aussagefiiliigkeit
des Kerbschlagbiegeversuchs. In: BAM-Forschungsbericht 137: 136--158
15. ASTM E813-81 (1981): Standard Test Method for JJc, A Measure of Fracture Toughness
16. Tada W., Paris P. C., Irwin G. R. (1973): The Stress Analysis of Cracks Handbook. Del.
Research Corporation, Hellertown, PA
17. Moskala E. J. (1992): Fracture behavior of rubber-toughened polymer blends. J. Mat. Sci.
27: 4883---4889
18. Seidler S., Grellmann W. (1993): Fracture behaviour and morphology of PC lABS blends. J.
Mat. Sci. 28: 4078--4084
A 3.2 Application of Normalization Method
for Determining J-R Curves
in the Amorphous Polymer PVC
M. Che, w. Grellmann and S. Seidler, Merseburg
1 Introduction
The increased industrial application of thermoplastics as engineering plastics has

resulted in an increase of fracture mechanics research on polymer materials [1-5].
The fracture of polymers may become a decisive factor in material selection and
their crack resistance, especially against stable crack growth, determines their
usefulness in many applications (pp. 3-26, [6]). Therefore, much work has been
done to develop effective evaluation methods for polymers and composites. For
tough polymers that exhibit strong viscoelastic behaviour or significant plastic
crack tip deformation during the process of fracture, elastic-plastic fracture me-
chanics (EPFM) methods have been successfully used to describe the fracture
behaviour under quasi-static loading [7,8] and under dynamic loading [9,10].
For metals, the J-I'la curve is experimentally determined using the multiple-
specimen testing method or the single-specimen testing method (elastic-compli-
ance technique or potential drop technique) [11,12]. In the multiple-specimen
testing method, several identical specimens are loaded to obtain different amounts
of crack growth. The crack lengths are then measured on the fracture surfaces after
the specimens have been broken [13]. The resulting J-R curve represents the
mean behaviour of all specimens tested, which makes it impossible to study the
influence of variables such as inhomogeneity of microstructure or of chemical
composition on the fracture behaviour. Also, the method is quite time- and mate-
rial-intensive. In contrast, although the conventional single-specimen method
needs only one specimen, additional equipment is required for on-line monitoring
of the crack growth. Chung and Williams [14] and Wert, Saxena and Ernst [15]
used the partial-elastic-compliance method, while Atkins et al. [16] applied the
potential method. However, it is difficult to make an accurate and reliable meas-
urement of the amount of crack growth in polymers, and therefore it is rather
questionable whether these methods are really applicable to polymer materials.
Owing to viscoelasticity and/or inelastic micromechanisms resulting from local
damage in polymer materials in the form of crazing, cavitation, debonding, shear
yielding and so on, there occur hysteresis loops during the loading-unloading
cycles which are needed to measure the crack growth. Additionally, the crosshead
speed is severely restricted when these partial unloadings are performed. To deal
with the electrical non-conductivity of polymers, a metal layer with good electrical
conductivity could be coated onto the specimen surface. But with this method only
the crack length at the specimen surface is measured. With all of these difficulties

122 M. Che, W. Grellmann, S. Seidler
it was obvious that a new method was needed with which one could easily, relia-
bly and reproducibly determine the crack resistance behaviour in polymer materi-
als. In one method [17] the hysteresis energy was used to determine the crack
initiation value J[e, while Strebel and Moet [18] applied a fatigue method which
enabled them to test fracture mechanics specimens of tough polymers with typical
component thickness values. These thickness values of polymer products are nor-
mally much smaller than those required by the thickness criterion for standard J
testing. However, with these techniques, only information on .h.e was obtained.
Much more effort was expended to fmd new methods other than the multiple-
specimen method to measure the whole crack resistance curve. Under the assump-
tion that the non-linearity of the load versus the load-line displacement is due
solely to crack growth, Moskala [19] found that the estimated crack length agreed
well with that from the multiple-specimen testing method only at the initial stage.
With increasing crack growth, the crack length was measurably overestimated.
Joyce, Ernst, and Paris [20] introduced the key curve method, where a universal
key curve was used for a given material and no details of individual test records
were taken into consideration. Owing to variations in deformation properties, this
method did not give satisfactory results.
Recently, Landes and co-workers [21] developed a new method based on the
conventional key curve method, called the normalization method, and have suc-
ceeded in determining J-R curves in metallic materials [22] as well as, more re-
cently, in polymer materials (SENB specimens of polycarbonate [23] and CT
specimens of rubber-modified nylon 6.6 [24]).
2 Normalization Method and Calibration
2.1 Normalization Method
In contrast to the conventional key curve approach, the normalization method

develops individual normalized calibration curves for each specimen based on the
test record itself. The method is based on the principle of load separation devel-
oped by Ernst, Paris, and Landes [25]. The technique has been demonstrated for
several specimen dimensions and for various materials with different strength
levels by Sharobeam and Landes for a stationary crack [26] as well as for a grow-
ing crack [27]. According to the assumption of load separation, the load P can be
represented by separable multiplicative functions of the crack length a and the
plastic displacement Vpl:
P = G(aIW)H(vpl IW), (A120)
where W is the specimen width. A normalized load PN can be then defmed to be a

function of only the plastic displacement Vpl:
Application of Normalization Method 123
PN = P /G(a/W) = H(vpl /W). (AI21)
The geometrical calibration function G(a/W) is dependent on the specimen ge-

ometry and can be determined from the J calibration [26,27]:
G(a/W) = BW(b/W)"P' , (AI22)
where the uncracked ligament length is b = W - a, and 1]pl is a geometrical func-

tion which depends only weakly on material properties. Landes and Zhou [23]
reported that 1]pl determined using the assumption of load separation for polycar-
bonate compact tension specimens was close to the values from metallic compact
tension specimens. Further, the plastic displacement can be expressed as
Vpl =v-vel =v-PC(a/W), (AI23)
where C(a/W) is the elastic compliance calibration, which is a function of the

current crack length. If the deformation function H(Vpl/W) is known, (A12I),
(AI22) and (AI23) can be solved to give the relationship between the load, load-
line displacement and crack length at every point, and therefore the current crack
length can then be evaluated.
Several kinds of material deformation function H(vplw) have been proposed,
including a power law with two unknown constants [28,29], a combination of a
power law and a straight line with three unknown constants [23,30], and the LMN
function with three unknown constants [21,22,24]. The fIrst two kinds of function
have not proven to have suffIcient accuracy, especially for materials which un-
dergo extensive plastic deformation [23,27], such as some amorphous polymers.
More recently, Landes et al. [21] demonstrated that the LMN function, proposed
by Orange [31], could describe the whole material deformation behaviour very
well. The LMN function is given by
L+M(vpl/W)
PN = (Vpl/W), (AI24)
N +vpl /W
where L, M and N are unknown fItting constants that must be determined from the
test records.
2.2 Calibration Procedures
The determination of the unknown constants L, M and N from the test records is
the key task in applying the normalization method with the LMN function. The
procedure for the determination of crack resistance curves with the normalization
method is described in Fig. A93. Information about the material tensile properties,
the Young's modulus E, the Poisson's ratio v and the yield strength O"y when pos-
sible, as well as the ultimate strength O"m is needed. The method uses directly the
Fig. A93. Procedure for determination of crack resistance curves using the normalization
method for an SENB specimen of PVC (room temperature, VT = 1 mm1min): (a)
load-deflection diagram, (b) fitting constants L, M, N, (c) J-!:o.a and c5-t:..a curves
(. multiple-specimen method, 0 normalization method according to ESIS; 0 nor-
malization method according to ASTM)
recorded individual load versus load-line displacement curves, the physically

measured initial crack length ao and the physically measured fmal crack length af
for the calibration. As previously mentioned, an on-line monitoring of crack
growth is not needed.
To determine these three unknown constants L, M and N, three independent
calibration points are necessary where the load, displacement and crack length are
simultaneously known. The first is the single point corresponding to the fmal
crack length af, and the second is a group of points defmed by forced blunting:
J = maf!la, (A125)
where af is the flexural stress and m is the crack tip constraint factor. The third is
an intermediate calibration point, which is variable and is determined so as to
minimize the fitting error [21,22].
Once the calibration curve, i.e. the constants L, M and N, has been determined,
the current crack length can be evaluated directly from the load versus load-line
displacement records at every point. The J value can be estimated from the load,
displacement and crack length. For purposes of comparison with values obtained
from the multiple-specimen testing method, the J values were determined using
the ESIS standard calculation with crack length correction [32]. This is given by
(AI26):
J = Jo (l- (0. 7517 -1)aa).

W-ao
(AI26)
The crack resistance Jo without crack growth is given by (AI27):
-
J 0- 17AG , (AI27)
B(W - ao)
and the geometrical correction function is
17 = 2 + 0.522ao1W (AI2S)
for CT specimens. The current amount of crack growth aa is taken at the relevant
loading level, B is the specimen thickness and AG is the area under the load versus
load-line displacement curve. Then J-R curves are constructed by plotting the
values of J as a function of the amount of crack growth aa.
In the present work, the normalization method with the LMN function was used
to construct J-R curves directly from the recorded load versus load-line displace-
ment curves. Its applicability was studied in a sample of the amorphous polymer
PVC at different crosshead speeds, ranging from 0.01 to 50 mmlmin. In the amor-
phous PVC a quite different blunting behaviour was observed compared with that
observed in metallic materials and in semicrystalline polymer materials [33,34].
3 Experimental Procedures
The material used in this study was a commercial extruded PVC sheet (PVC-
CAW, Kunststoffwerke Kim, SIMONA AG, Germany) with a thickness of 10
mm. Tensile tests were conducted using an INSTRON machine at different
crosshead speeds. The multiple-specimen J-integral tests were conducted with a
ZWICK machine using compact tension specimens with B = 10 mm and W = 40
mm at crosshead speeds ranging from 0.01 to 50 mmlmin. The test temperature
was room temperature. Specimens were cut from the 10 mm thick PVC sheet with
the crack growth direction along the extrusion direction and were notched with a
razor blade, which had a notch tip radius of 0.2 ~. The ratio aolW was kept con-
stant (0.55) throughout all the tests. The load-line displacement was measured
with a non-contact laser scanning system which was developed especially for
polymer materials and has been demonstrated to be very suitable [35]. The crack
lengths were determined from an average of nine optically measured points after
breakage at a much higher crosshead speed or at liquid-nitrogen temperature. The

detailed experimental procedures and the time-dependent fracture behaviour of the
PVC have been presented elsewhere [36].
4 Results
The method of normalization with the LMN calibration function, as described

above, was used to develop the J-R curves directly from the load versus load-line
displacement records at different crosshead speeds, ranging from 0.01 to
50 mmlmin. The geometrical function 7]pl and the crack tip constraint factor m
were assumed to be independent of the crosshead speed and to have the values
7]pl = 2.15 for the CT specimen used and m = 2 for the amorphous PVC. Owing to
the inevitable early occurrence of necking and subsequent stabilized neck propa-
gation in the amorphous PVC, it is impossible to obtain a reasonable flow curve
with conventional dumb-bell tensile specimens by using conventional extensome-
ters. Therefore, the yield stress <Ty was simply taken to be the flow stress <Tf in the
forced blunting (A125). The Young's modulus and the yield strength are shown in
Fig. A94 as a function of the crosshead speed. They both showed a positive de-
pendence on the crosshead speed, as is usually observed. The Poisson's ratio v
was 0.35 for the amorphous PVC tested.
To examine the reproducibility of the novel single-specimen technique, two
separate specimens with different crack extensions, one with af - ao = 2.06 mm
(open circles in Fig. A95b) and the other with af- ao= 2.58 mm (diamonds), were
used to determine J-R curves at Vr = 0.1 mmlmin. As can be seen in Fig. A95b,
good reproducibility of the J-R curves was achieved.
4000 90
3000
•~•
60
i 2000 't;'
..,:::E
'-'
~
30 b
;.,
1000
• Young's Modulus E
• Yield Strength cry
0
0.01 0.1 10 100
vT{mmlmin)
Fig. A94. Tensile properties as a function of the crosshead speed in the amorphous polymer PVC
30,-----------------------"
~
VT= 0.01 mm/min
40
20
o
0 0
~ ~oo
~IO J~oo:ooO 8 o Nonnalization Method (Spec. I)

o NonnaHzation Method o Nonnalization Method (Spec. 2)
• Multiple--Specimen Method • Multiple·Specimen Method
O~&~I~~--~--~~--~
o 0.5 1.0 1.5 0 2 3
t.a (mm) t.a (mm)
50.----------------------,~~c 50
l.<!
VT= 5 mm/min
o
0 ce VT= 10 mm/min
40 40
o
o •
30 o· 30
o.
~o
o
fP
eO
o
10 ~o o Normalization Method o Normalization Method
I~
• Multiple-Specimen Method • Multiple·Specimen Method
O~L-~--~~--~--~----~--~
o 2 3 0.5 1.0 1.5 2.0 2.5

t.a (mm) t.a (mm)
50.----------------------T~e~
VT= 50 mm/min
40
30
o Normalization Method
• Multiple-Specimen Method
2 3
t.a(mm)
Fig. A95. Comparisons between J-/1a curves obtained from the normalization method and
from the multiple-specimen method for PVC at different crosshead speeds: (a) 0.01
mm/min, (b) 0.1 mmlmin, (c) 5 mmlmin, (d) 10 mmlmin and (e) 50 mmlmin
J-R curves determined by the normalization method with the LMN calibration
function are presented in Figs. A95a-A95e, where they are compared with curves
determined by the mUltiple-specimen testing method at the same crosshead speeds.
Information on the individual specimens used for the normalization method, includ-
ing the final amounts of crack growth ar - ao and the corresponding fitting constants
L, M and N, is listed in Table A7 as a function of the crosshead speed. The J-R
curves determined by these two techniques agree fairly well with each other
throughout the entire curves at all the crosshead speeds used. The application of
the LMN function has also improved the accuracy of the J-R curve for short crack
growth, in comparison with the results of a normalization method based on a
power law or on power law plus straight-line functions, which gave reasonable
J-R curves only for large amounts of crack propagation, but were not satisfacto-
rily accurate at the early stage of crack propagation [21] where initiation fracture
toughness values such as J 1e are evaluated.
The most important fracture mechanics parameter for the crack resistance be-
haviour is the crack initiation value J"e, marking initiation of stable cracking. The
J"e values were evaluated in accordance with the ESIS recommendation for plas-
tics [32], i.e. a J value at a stable crack growth of 0.2 mm, denoted as JO.2, was
determined. The JO.2 values are shown in Fig. A96 and given in Table AS as a
function of the crosshead speed.
TableA7. Information about the specimens used for the normalization method and the corre-
sponding fitting constants L, M and N as a function of the crosshead speed
VI" ao W B W-ao ar-ao Fitting constants

(mmlmin) (mm) (mm) (mm) (mm) (mm) L M Nx 103
0.01 22.2 39.6 9.9 17.3 1.27 13.75 160.31 0.75

0.1 (Spec. 1) 22.2 40.0 10.0 17.8 2.06 17.39 38.47 1.54
0.1 (Spec. 2) 22.0 39.7 10.0 17.8 2.58 18.55 33.54 3.36
5 22.3 40.1 9.9 17.8 2.56 19.34 81.60 1.88
10 22.0 39.8 10.0 17.8 2.07 18.73 169.76 0.61
50 22.1 40.0 10.0 17.8 3.22 26.08 64.69 5.55
The JO.2 values obtained from the normalization method were all somewhat
smaller than those from the mUltiple-specimen method in the crosshead speed
range considered. However, the difference decreases with increasing crosshead
speed. Moreover, a two-regime dependence of the crack initiation value on the
crosshead speed is discernible, i.e. at VT ~ 5 mmlmin the crack initiation resistance
increases with increasing crosshead speed, and at VT> 5 mmlmin it decreases. By
combined application of the J-integral and the crack-opening displacement [36], it
was found that at lower crosshead speeds the local fracture was due to plasticity
exhaustion. This positive dependence can be predicted well by using the concept
of local fracture on the basis of a constant COD, whilst the negative dependence at
higher crosshead speeds was attributed to reduced plastic blunting.
12
PVC
8
e
e.€
.....,;
N
• Normalization Method
• Multiple-Specimen Method
00.01 0.1 10 50
vT(mm/min)
Fig. A96. Crack initiation J values, defined as JO.2 according to the ESIS TC4 test protocol for
plastics [32], as a function of the crosshead speed
Table AS. JO.2 values determined by the normalization and multiple-specimen methods
Initiation values
Vr
(mmlmin) Normalization, Multiple-specimen,
JO.2(N/mm) JO.2(N/mm)
1 X 10-2 5.8 7.3

I X 10-1 6.8 8.4
5 x 10° 8.1 8.9
1 X 101 7.2 7.9
5 X 101 5.9 6.5
5 Discussion
The normalization method using a constant plastic geometry factor llpb works
fairly well for constructing J-R curves for the amorphous polymer PVC at differ-
ent crosshead speeds without any automatic crack-growth-monitoring equipment.
The J-R curve results compare well with results from the mUltiple-specimen
method. The smaller JO.2 values determined from the normalization method may
be due to the assumption of a forced-blunting line at the initial stage, where the
constraint factor m was assumed to be independent of the crosshead speed and to
have a constant value of2. The assumed blunting behaviour may be unsuitable for
some polymer materials, as observed for some amorphous polymers by Seidler
and Grellmann [33,34]. These authors have found that the blunting zone translates
with stable crack growth in amorphous polymers, in contrast to what is observed
in semicrystalline polymers as well as in metals. Before the normalization method
can be applied as a general method for polymer materials, more detailed work
needs to be done over a wider material spectrum (amorphous and semicrystalline
polymers, and blends) and under various testing conditions (temperature, medium
and strain rate). Moreover, detailed micromechanical research on crack initiation
and propagation behaviour in polymer materials will contribute to developing a
more suitable deformation behaviour function for various polymer materials under
different testing conditions.
With the successful application of the normalization method to the amorphous
polymer PVC at different crosshead speeds, it is to be expected that the normali-
zation method can also be applied to construct J-R curves under dynamic loading.
The new single-specimen testing method offers an easy way to determine a J-R
curve. Additionally, it can provide more information on fracture behaviour than
can the mUltiple-specimen method, e.g. about the crack growth rate, which is an
important parameter for describing the time-dependent crack resistance behaviour
in polymer materials. Therefore, the normalization method is not only an alterna-
tive, economic testing method for polymer materials, but also is more powerful.
With the help of the normalization technique it should be possible to investigate
the influence of material inhomogeneity, e.g. composition and microstructure, on
fracture mechanics behaviour.
6 Conclusions
1. The normalization method, with the material deformation behaviour expressed

in the form of the LMN function, has been successfully applied to compact ten-
sion specimens of the amorphous polymer PVC to determine J-R curves at dif-
ferent crosshead speeds, ranging from 0.01 to 50 mmlmin.
2. The J-R curves determined by the normalization method are in good agreement
with results obtained by the traditional multiple-specimen testing method, al-
though the JO.2 values estimated by the normalization method were somewhat
smaller than those from the mUltiple-specimen method. The normalization
method with the LMN function appears to offer a promising method of investi-
gating time-dependent fracture mechanics behaviour in polymer materials.
3. It is reasonable to take TlpJ as constant; it appears to be independent of the
crosshead speed within the crosshead speed range investigated.
4. The crack initiation value JO.2 for the amorphous polymer studied showed a
two-regime behaviour in its dependence on the crosshead speed. At lower
crosshead speed, JO.2 increased with increasing crosshead speed; the maximum
value was reached at a crosshead speed of 5 mmlmin and then it decreased with
increasing crosshead speed.
Acknowledgement
The authors from the Martin-Luther-University of Halle-Wittenberg, Halle (Ger-

many) gratefully acknowledge the fmancial support provided by the German Re-
search Foundation (DFG).
References
1. Swallowe G. M. (Ed.) (1999): Mechanical Properties and Testing of Polymers - An A-Z

Reference. Polymer Science and Technology Series. Brewis D., Briggs D. (Eds.), Chapman &
Hall, London
2. Choi N. S., Takamashi K., Oschmann, D., Karger-Kocsis 1., Friedrich K. (1998): Influence
of weathering on unreinforced and short glass fibre reinforced thennoplastic polyester. 1.
Mat. Sci. 33: 2529-2532
3. Karger-Kocsis J., Putmoki I., SchOpf A. (1997): Radial perforation impact behaviour of p-
polypropylene homopolymer pipes and polypropylene random copolymer pipes. Plast. Rub-
ber Compos. Process. Appl. 26 (8): 372-375
4. ISOIDIS 13479 (1994): Polyethylene (PE) Pipes for the Conveyance of Fluids - Resistance
to Crack Propagation - Method of Test for Slow Crack Growth on Notched Pipes (Notch
Test)
5. Beguelin P., Fond C., Kausch H. H. (1998): The influence of inertial effects on the fracture
of rapidly loaded compact tension specimens, part A: loading and fracture initiation. Intern.
J. Fract. 89: 85-102
6. Akay M. (1999): Fracture mechanics properties. In: Brown R. (Ed.) Handbook of Polymer
Testing. Marcel Dekker, Inc. New York: 533-588
7. Hashemi S., Williams J. G. (1986): Fracture characterization of tough polymers using the J
method. Polym. Engng. Sci. 26: 760-767
8. Huang D. D., Williams 1. G. (1987): J testing of toughened nylons. J. Mat. Sci. 22: 2503-
2508
9. Grellmann W., Seidler S. (1992): J-integral analysis of fibre-reinforced injection-moulded
thennoplastics. J. Polym. Engng. 11: 71-101
10. Grellmann W., Seidler S., Lauke B. (1991): Application of J-integral concept for the de-
scription of toughness properties of fibre reinforced polyethylene composites. Polym. Com-
pos.12:320-326
11. Anderson T. L. (1995): Fracture Mechanics. Fundamentals and Applications. CRC Press,
Boca Raton
12. ASTM E 1737 (1996): Standard Test Method for J-Integral Characterization of Fracture
Toughness. Annual Book of ASTM Standards, Philadelphia, Vol. 03.01
ness characterization of high impact thennoplastics. Polym. Test. 14: 453-469
14. Chung W. N., Williams 1. G. (1991): Detennination of Jle of polymers using the single
specimen method. In: Joyce J. A. (Ed.) Elastic-Plastic Fracture Test Methods: The User's
Experience. ASTM STP 1141: 320-339
15. Wert M. J., Saxena A., Ernst H. A. (1990): Applicability of modified J as a fracture pa-
rameter for polycarbonate. J. Test. Evaluat. 18: 1-13
16. Atkins A. G., Lee C. S., Caddell R. M. (1975): Time-temperature dependent fracture tough-
ness ofPMMA. J. Mat. Sci. 10: 1381-1393
17. Lu M. L., Chang F. C. (1995): Fracture toughness of PCIPBT blend based on J-integral
methods. J. Appl. Polym. Sci. 56: 1065-1075
18. Strebel J. J., Moet A. (1993): Determining fracture toughness of polyethylene from fatigue.
J. Mat. Sci. 28: 2981-2988
19. Moskala E. J. (1992): Fracture behavior of rubber-toughened polymer blends. J. Mat. Sci.
27:4883-4889
20. Joyce J. A., Ernst H. A., Paris P. C. (1980): Direct evaluation of J-resistance curves from
load displacement records. In: Paris P. C. (Ed.) Fracture Mechanics. Proceedings of the 12th
National Symposium on Fracture Mechanics, ASTM STP 700: 222-236
21. Landes J. D., Zhou Z., Lee K., Herrera R. (1991): Normalization method for developing J-R
curves with the LMNfunction. J. Test. Evaluat. 19: 305-311
22. Zhou Z., Lee K., Herrera R., Landes J. D. (1991): Normalization: an experimental method
for developing J-R curves. In: Joyce J. A. (Ed.) Elastic-Plastic Fracture Test Methods: The
User's Experience. ASTM STP 1114: 42-56
23. Landes J. D., Zhou Z. (1993): Application of load separation and normalization methods for
polycarbonate materials. Int. 1. Fract. 63: 383-393
24. Zhou Z., Landes J. D., Huang D. D. (1994): J-R curve with the normalization method for
toughened polymers. Polym. Engng. Sci. 34: 128-134
25. Ernst H. A., Paris P. C., Landes J. D. (1981): Estimation on J-integral and tearing modulus T
from a single specimen test record. In: Roberts R. (Ed.) ASTM STP 743: 476-502.
26. Sharobeam M. H., Landes J. D. (1991): The load separation criterion and methodology in
ductile fracture mechanics. Int. J. Fract. 47: 81-104
27. Sharobeam M. H., Landes J. D. (1993): The load separation and 77pJ development in pre-
cracked specimen test records. Int. J. Fract. 59: 213-226
28. Herrera R., Landes J. D. (1988): A direct J-R curve analysis of fracture toughness tests. J.
Test. Evaluat. 16: 427-449
29. Bernal C. R., Cassanelli A. N., Frontini P. M. (1995): A simple method for J-R curve deter-
mination in ABS polymers. Polym. Test. 14: 85-96
30. Herrera R., Landes J. D. (1990): Direct J-R curve analysis: a guide to methodology. In:
Gudas J. P., Joyce J. A., Hackett E. M. (Eds.) Fracture Mechanics: 21st Symposium, ASTM
STP 1074: 24-43
31. Orange T. W. (1990): Methods and models for R-curve instability calculations. In: Gudas
J. P., Joyce J. A., Hackett E. M. (Eds.) Fracture Mechanics: 21st Symposium, ASTM STP
1074:545-559
ness characterization of high impact thermoplastics. In: Williams J. G., Pavan A. (Eds.) Im-
pact and Dynamic Fracture of Polymers and Composites. ESIS Publication 19: 171-179
34. Seidler S., Grellmann W. (1994): Crack initiation behaviour of polymers. In: Proceedings of
Deformation, Yield and Fracture of Polymers, Cambridge, U.K., April 11-14: 1081-1084
35. Grellmann W., Bier6gel C. (1998): Laser extensometry applied. Materialpriifung - Mater.
Test. 40, 11-12: 452-459
36. Che M., Grellmann W., Seidler S. (1997): Crack resistance behaviour of polyvinylchloride.
J. Appl. Polym. Sci. 64: 1079-1090
A 3.3 Calculation of J-R Curves
Based on Load-Deflection Diagrams
Using the Hinge Model Test Specimen
R. Steiner and W. Grellmann, Merseburg
1 Introduction
One of the main fields of application of fracture mechanics is materials develop-

ment. In order to avoid problems in the behaviour of components arising from the
materials, theoretically founded material optimization can be performed with the
help of fracture mechanics concepts combined with morphological investigations.
The various characteristic toughness parameters have proved, in materials science
practice, to be sensitive indicators for the selection or development of materials.
In general engineering practice, one mostly operates with toughness values
which characterize the resistance against unstable crack initiation and propagation.
Modem tough materials should also be characterized, however, regarding their
resistance against stable crack initiation and propagation.
For the determination of such toughness values as the resistance against initia-
tion and propagation of stable cracks and to obtain a large amount of information,
an experimental determination of the amount of stable crack growth is additionally
necessary. This is directly determinable only with complex methods (e.g. the mul-
tiple-specimen technique) or hybrid experimental methods (the potential method).
This circumstance makes substantial automation of the testing method difficult and
thus hinders the widespread use of such methods. A requirement of materials de-
velopment is to make a reliable statement about the toughness behaviour, particu-
larly regarding qualitative differences, with methods as simple as possible. There-
fore, a target of such work is the development of a procedure for the calculation of
toughness values related to resistance against initiation and propagation of stable
cracks for the material optimization of plastics.
A possibility for achieving this target is the indirect determination of the amount
of crack growth from measured values obtained from test equipment according to a
standard (Fig. A97), by the application of a local computing technique. The
method presented here, however, uses other methods for the description of the
crack resistance behaviour because of the indirect calculation of characteristic
values, in this case the amount of crack growth (pp. 105-132, pp. 141-150, [1,2]).
In this connection we can name the modem normalization method (pp. 121-132,
[1,3]) for the determination of crack resistance curves, which permits using a di-
rectly determined value for the final crack growth with mathematical methods to
ascertain the complete process of crack propagation, as the compliance of the test
specimen is divided into a specimen-geometry-dependent term and a crack-
growth-dependent term.

134 R. Steiner, W. Grellmann
/
a==0.815
~J J==f(F; y; a)
/
Fig. A97. Principle of the determination of J-R curves by indirect crack growth ascertainment
For the implementation of the method, a mathematical description of the defor-

mation behaviour of SENB specimens is needed. Within the area of the crack, high
loading of the material occurs. Therefore, because of the mechanical properties of
polymers, non-linear material behaviour in the stress--elongation relationship must
be considered. A suitable analytic description of the load-deflection curve is pro-
vided by a function of the amount of crack growth. Determination of the appropri-
ate 'effective' crack growth from pairs of values of the load and deflection meas-
ured by the instrumentation is always possible using this functkm. The amount of
crack growth determined in such a manner permits the determination of toughness
values from the strength and deflection in the analysis described following. Even a
single load-deflection curve is sufficient for this calculation.
By the use of a computer program for the automatic analysis of load-deflection
curves, the determination of J-R curves is possible in order to obtain toughness
values in terms of resistance against stable crack initiation and propagation. This
contributes to the efficient material optimization of polymers. A mathematical
description of the test specimen permits further analytical and numerical calcula-
tions for the study of various effects which are difficult or impossible to access in
an experiment.
Calculation of J-R Curves Based on Load-Deflection Diagrams 135
2 Theoretical Derivation of the Model

of the Hinge Test Specimen
An accurate analytical description of the deformation behaviour of a notched

specimen is not possible under high loading if non-linear material behaviour is
considered. For the indirect determination of the amount of crack growth, how-
ever, an analytic function is needed as simple as possible which takes account of
the crack growth in the description of the deformation behaviour. This function
depends also on material and geometrical characteristics, which can be regarded as
parameters. Suitable functions have been derived theoretically. A hyperbolic func-
tion is suitable for the description of the non-linear stiffness behaviour. Thereby,
the specific deformation behaviour of a polymer can be included in the model. It
has also been shown that a modified hyperbolic material law, as represented in Fig.
A98, is suitable.
_E~_
/".:;;;r,r._--..Fracture
TANH
Fig. A98. Modified hyperbolic material law
Additional characteristics of the strength and failure behaviour result in a modi-

fied hyperbolic material law with four parameters (the basic parameters). By this
means, a non-linear function for the description of the stiffness behaviour with a
yield limit and a failure limit is obtained.
The deformation behaviour of a small hinge area in a two-dimensional model
under the influence of a normal load N and a bending moment M can be defmed as
a function of the basic parameters (Fig. A99). In the special case of plane bending,
the normalized bending moment can be described by a suitable approximate solu-
tion as a simple analytic function.
Using the equations of this joint area for the plastically deformed section of the
test specimen, a model of the hinge test specimen can be developed, as shown in
Fig. Al 00, using solutions from flexible-beam theory and the concepts of linear
elastic fracture mechanics (LEFM) and elastic-plastic fracture mechanics (EPFM).
136 R. Steiner, W. GreUmann
Fig. A99. Two-dimensional model of a hinge area
Fig. AIOO. Model of the hinge test specimen
The mathematical analysis of the defonnation behaviour of a hinge test speci-

men with a constant amount of crack growth loaded in a bending test leads to the
finish function. The resulting analytic function, called the key curve, can be ex-
pressed in the standardized representation as
(A129)
3 Experimental Application of the Model
The key curve defines a strictly monotonically increasing function, which is

specified by three parameters. Therefore, an experimentally determined load-
deflection curve can be represented, in a standardized representation, by a key
curve if the three parameters are chosen appropriately.
Considering the stability of the function with respect to perturbations of the
measured values according to a likelihood estimation shows that a favourable
method of determining the parameters is provided by a least-squares fit with equi-
distant deflection values. A rapidly converging adjustment of the key curve to the
measured values can be achieved on the basis of a non-linear approximation pro-
cedure. The basis parameters of the hinge test specimen can be calculated from the
parameters of the key curve obtained in this way. These basis parameters, in con-
junction with further solutions for the hinge test specimen and further pairs of
measured load-deflection values, permit determination of the 'effective' amount of
crack growth. Thus, in particular, the calculation of crack resistance curves and the
determination of toughness values in terms of resistance to stable crack initiation
and propagation are possible, as shown in Fig. AI0l.
This method was programmed in the macro language MATHCAD™ as a pro-
gram for off-line analysis of dynamically and quasi-statically stressed SENB
specimens of polymers [4]; the possible use of this program for extending the use
of automated testing methods was also borne in mind [5]. Optimization of this
procedure was performed with the aim of obtaining reliable results for a range of
tough material systems, test specimen dimensions and loading conditions, and the
results were compared with results obtained by the FEM.
100 -'---""'---Expe-ri-rne-ntal-Load--""
Deflection Curve
1.0
50 Key-Curve
0.5 Matching
-----+
'(OJ,'
O~--~----~---r--~ O~~-----r------~~
0.5 1.0
o '. '~o
15-.---------------..., 10-.----------------...,
Detennined Crack Length based
on Hinge Test Specimen
Calculated J-R Cuve end

Fracture Mechenics vaJues
O~----...,-----...,-----~ O;----r---...,----~--~
o 0.5 1.0 1.5 o 2 4 6 8
&a(mm) y(mm)
Fig. AIOI. Calculation of J-R curves with the help of the key curve adjustment
The calculated crack resistance curves are qualitatively, and approximately

quantitatively, equivalent to curves calculated by other methods [1], as shown in
Fig. Al02. The remaining quantitative differences can be explained mainly by the
differences between the calculated 'effective' and optically determined amounts of
crack growth. In cases where the method has been applied to material optimiza-
tion, the calculated J-R curves show the same qualitative behaviour.
l38 R. Steiner, W. Grellmann
30,-----------------------~--------~
3-Point-Flexural Test
E
~ 20
~
c:
.l!!
Ul
'iii
8!. 10
tl R-Curve Procedure
~
() Normalization Method
• Multiple-Specimen Technique
O~--------~r--L-------r--------~
o 0.5 1.0 1.5
Crack Growth aR (mm)
Fig. AI02. Comparison of J-R curves ofPP detennined with different methods
400,----------------------------,
Key Curve 120-1--------------+-1
Experimental Curve
/ 1 80-l----------;--+---If-----+-1
~
~200
u..
~ 40-t------r-<-----L...-------I
o~----_,------,_----_,--~~
a 0.5 1.5 2.0
4 12 20 28 36 4 1'2 2~ 2~ ~6
Modifier Content (%) Modifier Content (%)
Fig. AI03. Characteristic values under impact loading as a function of the modifier content of
SAN graft copolymers; ten test specimens were used for each value of modifier
content: the minimum and maximum values are indicated
In addition to characteristic toughness values, other characteristic mechanical

parameters of a test specimen can be determined at the same time, such that the
further expenditure required reduces substantially during a series of tests. In the
minimal case, material optimization can be performed with one test specimen of a
particular material system because the 'automatic R-curve procedure' [4] calcu-
lates, in addition to the toughness values, characteristic values that describe the
behaviour of the stiffness and strength. However, an analysis of several test speci-
mens is preferred in order to increase the statistical accuracy of the characteristic
values for dynamic loading, as shown in Fig. Al 03 for a case in which ten speci-
mens were analysed for each value of modifier content.
4 Further Application of the Model
The 'automatic R-curve procedure' represents only one specific application of the
hinge test specimen. The equations derived for the description of the deformation
behaviour of this type of specimen also form a basis for further analytical descrip-
tions of the bending test or for its simulation [6], as illustrated in Fig. AlO4 by a
calculation of the fracture mirror length on the basis of a heterogeneous energy
balance. Two-dimensional simulation calculations of the bending test lead to the
development of a criterion for' forced' crack growth. A prediction of this criterion
is that a larger crack to width ratio favours stable crack growth; this justifies theo-
retically an experimental observation which is applied practically in the fracture
mechanics testing of materials.
Phase I Phase III

10
4~
,,
E
g
,,
~~ .~ .
8 .:
: 5~
.{::oOJ/
; "
-. -~ ..
,1
.... ~:
:
cQ)
~6
I..L.
&,,'
if '
~/
........ 6:;
u
til
"0 ~'b/ +-__~__,--+~ ~____- r____-+ 7U
~ 4
...J
~/
"',
,
0;;;/
Simulation Edge
,
I
2
,
,,
I
,
o~------~-------r------~-------'-------'
o 2 3 4 5
Denection y (mm)
Fig. AI04. Analytical description of phase I by means of the key curve and numeric simulation
of phase II in comparison with the realload-deflection curve (- - - - 2D plane strain
state, _._.- 2D plane stress state)
Simulation calculations of the bending test in which the temperature depend-

ence of the modulus of elasticity of the polymer is considered lead, in conjunction
with a firmness distribution, to a strong rise in temperature within the softening
area of the specimen, which leads to a microscopic plastic failure . Therefore, the
140 R. Steiner, W. Grellmann
material. behaviour can be interpreted as stable crack growth with microscopic

melting, which has been demonstrated by infrared measurements of the fracture
process [6].
An energetic consideration of the crack initiation process predicts a 'delayed'
crack propagation. Analytical, experimental and numerical investigations confirm
the existence of such a process, which is comparable to the blunting process. A
functional relationship between the toughness and the basic parameters can be
derived using the hinge model equations.
5 Summary
The model of the hinge test specimen presented above is particularly suitable for
use in the 'automatic R-curve procedure', for an efficient determination of charac-
teristic toughness values to be used in the material development of polymers. The
equations derived permit one to summarize the material behaviour by means of
analytical and numerical methods.
References
1. Che M., GrelImann W., Seidler S., Landes J. D. (1997): Application of a nonnalization
method for determining J-R curves in glassy polymer PVC at different crosshead speeds.
Fatigue Fract. Engng. Mat. Struct. 20 119-127
2. Grellmann W., Seidler S., Jung K., Kotter I. (2001): Crack resistance behaviour ofpolypro-
pylene copolymers. 1. Appl. Polym. Sci., accepted
3. Che M., Grellmann W., Seidler S. (1997): Crack resistance behaviour of polyvinyl chloride.
1. Appl. Polym. Sci. 64: 1079-1090
4. Steiner R. (1996): Prozedur zur automatischen Ennittlung von RiBwiderstandskurven aus
Kraft-Durchbiegungs-Diagrammen. MPK-Prozedur. Martin-Luther-University Halle-Wit-
tenberg, Merseburg
5. Steiner R. (1996): Verfahren und Anordnung zur automatischen Ennittlung von J-R-Kurven
sowie Zlihigkeits-, Festigkeits- und Steifigkeitskennwerten unter dynamischer. Bean-
spruchung von SENB-Priltk6rpem. Gennan Patent DE 196 35 968
Basis des Gelenkpriltk6rpers. Fortschritt-Berichte, VDI-Series 18: MechanikIBruchme-
chanik, No. 208, VDI, Dusseldorf
A 3.4 An Alternative Method Based on J- TJ and
t5- To Stability Assessment Diagrams
to Determine Instability Values
from Crack Resistance Curves
R. Lach and W. Grellmann, Merseburg
1 Introduction
Recently, the improvement of stiffness and strength and the optimization of tough-
ness behaviour have become an important focus of engineering science because of
the expanding fields of application of polymeric materials not only for common,
but also for constructional purposes [1]. Measurement of crack resistance curves
(R-curves) using the stop block method with the multiple-specimen technique and
analysis of R-curve data by means of various procedures are well established for
the evaluation of toughness levels, especially those expressed as resistance against
initiation and propagation of stable cracks [2-4]. However, the relatively large
amount of time required for measuring and analysis and the expensive personnel
involved work against considerations of economy. Hence, many research groups
have been working intensively on developing approximate R-curve determination
methods and on their application to polymeric materials [2,5-7]. Because of limits
on the specimen geometry in standard fracture mechanics specimens and because
of the loading conditions used, unstable crack propagation can often not be ob-
served. In such cases it is also of interest to calculate material parameters describ-
ing the crack instability point [8].
2 Fracture safety Criteria

of Elastic-Plastic Fracture Mechanics
The crack resistance curves of elastic-plastic fracture mechanics describe the

dependence of a loading parameter (for example the crack-(tip-)opening displace-
ment 8 or the J-integral) on the amount of stable crack growth !:ia. Therefore,
quantification of the different states of crack propagation (blunting of the initial
notch, stable crack initiation and propagation, beginning of unstable crack propa-
gation) is possible. Various materials science approaches to the interpretation ofR-
curves exist because of the normally non-linear behaviour of measured R-curves [4].
As an example, a crack resistance curve theoretically derived by Will [9] on basis of
a stochastic damage model can be mathematically described, for small crack growth
values !:ia, by (pp. 71-86):
(Al30)

142 R. Lach, W. Grellmann
Various fracture safety criteria using J-Aa crack resistance curves (J-R curves)
have been formulated by Blumenauer and Pusch [10], which can be used if greater
parts of the stable crack growth corresponding to the overall fracture process can
be detected (ductile material behaviour), as listed below.
1. Material resistance against stable crack initiation.
Jcomponent ~ J i or Jcomponent ~ J 0.2 . (AI31)
The loading parameter J component can be determined either numerically or by

evaluation methods.
2. Material resistance against stable crack propagation.
dJ component dJ material
----<---- (Al32)
d(Aa) d(Aa)
Material parameters on the right-hand side of the inequalities (Al31) and

(AI32) can be quantified by analysing the experimentally determined J-Aa curves.
Crack initiation values .l;. or JO.2, corresponding to (AI31), can be determined ei-
ther at the physical crack initiation point (Aa = stretch zone width) or at the techni-
cal initiation point for defmed amount of crack growth (Aa = 0.2 rom).
3. J- TJ stability assessment diagrams.
Another method to estimate failure safety for limited stable crack growth, i.e. to
evaluate the resistance against the start of unstable crack propagation, is the deter-
mination of J-TJ stability assessment diagrams [10], which should be used for the
polymeric materials described in this chapter.
On the basis of the tearing modulus of Paris et al. [II],
T _~Ed (Al33)
J - d(Aa) eTJ '
an instability criterion [12]
T component < Tmaterial (Al34)

J - J
can be formulated. This makes it possible to avoid an instability if a crack is

growing.
The loading parameter T/omponent is derived from plastic load limit conditions.
Thus, unstable crack propagation can be observed if the tearing modulus of the
component is greater than that of the material [10].
Starting from the energetic interpretation of the J-integral, critical J values
can be determined from measured load-deflection diagrams of unstable, broken
specimens using different evaluation methods (for example, according to Sumpter
J- TJ and 0-T,5 Stability Assessment Diagrams 143
and Turner (ST) [13] or Merkle and Corten [14]). These methods are compared
with others in [15]. By using the instability criterion (A134), approximate Jvalues
J/PP related to resistance against unstable crack propagation can be determined if
no unstable crack propagation is observable owing to the limited specimen ge-
ometry, i.e. only stable crack propagation behaviour can be found.
For the evaluation of components, J-Aa curves (Fig. AI05a) are transformed
into J versus TJ plots, the so-called J-TJ stability assessment diagrams (Fig.
A105b). Following an idea by Paris and Johnson [16], a material-specific parame-
ter PJ is introduced. PJ is the slope of the line from the origin to the intersection of
the load line and the J-TJ curve (also plotted in Fig. AI05b):
(A135)
The intersection point of these two curves is named the instability point, and TJd
is the tearing instability [17]. The definition of an empirical fixed value J so (with
PJ= 50 psi in = 8.8 N/mm) [16] as an instability parameter Jdapp has been proved,
for polymeric materials, to be a conservative assessment criterion. Therefore, J/PP
values are calculated from
(A 136)
using (A 13 0) and (A133).
J b
lla lj
Fig. AI05. Transformation of a J-Aa crack resistance curve (a) into a J-TJ stability assessment
diagram (b)
From J d values measured experimentally for a specific material or material sys-

tem, material-dependent PJ values can be estimated.
Introducing the parameter Plio the instability parameter 8/PP for 8-To stability
assessment diagrams can be analogously described by
8 dapp -c
- 8 Pd Ed , (A 137)
O"d
where a linear function between the crack-opening displacement 8 and the amount
of stable crack growth lla is assumed (8 = C s lla + C9 , in accordance with ASTM
E813-81 [18]).
3 Examples of Use
• Polycarbonate.
Polycarbonate (PC) is a thermoplastic, amorphous polymeric material showing
pseudo-ductile fracture behaviour. This is the main cause of the transition from
dominantly unstable to stable crack propagation, which occurs at temperatures
higher than 50°C. This was demonstrated by investigations of the temperature-
dependent toughness behaviour of PC [17,19] using a loading rate of 1.5 mls and a
defmed specimen geometry.
2Or---------------------------------. 250
200
15
E 1SO
E
E
~ bor-
b 10
100 - 12
<.()
5
50
o 0
-250 -200 -150 -100 -SO 0 50 100 150 200
Temperature ("C)
J
Fig. AI06. Temperature dependence of the value J"rand the crack-opening displacement bid
for polycarbonate (conventionally detennined: JldST, 0; bid, L1; detennined from J- TJ
and 8--T6 stability assessment diagrams: J"r,0, bid, \7)
J-TJ and 8-T6 Stability Assessment Diagrams 145
Therefore, for the approximation of the toughness level at temperatures higher

than 50°C, data for both JIlT and bid versus T were determined using an alterna-
tive method based on J-TJ and 8-T6 stability assessment diagrams, which has been
described in Chap. 2 (Fig. A 106). The empirical parameters PJ and PM listed in
Table A9, were measured by transformation of J-Aa and 8-Aa curves, determined
at T= 20°C and T= 40°C, respectively.
• PPIEPRIPE reactor blends.
The experimental determination of instability values by means of J-TJ and 8-T6
stability assessment diagrams is very easy to apply to multi-phase PP materials
containing an EPRIPE phase because of their especially high structural variety and
broad spectrum of mechanical behaviour, which includes brittle behaviour and
plastic collapse too. The investigations were performed on model systems pro-
vided by Borealis AG, Linz, Austria. The basis materials were RAHECO® materi-
als, where an EPRIPE phase is dispersed into a random copolymer matrix [20,21].
The influence of concentration and temperature on the toughness behaviour of the
RAHECO® I [2,22-24] and RAHECO® II material systems [2,3,23,24] was inves-
tigated.
Because the empirically observed relationship between the J and 8 values and
the amount of stable crack growth Aa can be described by using a power law of the
form
(A138)
the approximate J or 8values were obtained from the following equations, rather
than from (A136) and (A137):
J;pp = C{Pl !: f2 , (A 139)
c5;PP = C{Po !: f2 (AI40)
It can be seen from Figs. AI07 and AI08 that a useful analysis of the concen-
tration and temperature dependences of the toughness in relation to resistance
against unstable crack propagation is not possible without application of J-TJ and
8-T6 stability assessment diagrams. It is especially important to emphasize that the
concept of a brittle-to-tough transition temperature has no experimental basis here,
because only the upper shelf of toughness is covered by conventionally obtained J
values (pp. 281-300, [2,3,17,19,22-24]). For this reason, the parameters PJ and P6
were estimated separately for each test temperature (T= 10,23,30 and 40°C), and
the values of PJ were found to range from PJ = 1.6 N/mm (T = 10°C) to
PJ = 2.2 N/mm (T = 40°C).
100
80
I-- 60
~
Ii; "0 40
-)
20
o 20 40 60 80 100
RAHECO Content (wt.%)
Fig. AI07. JliT as a function of RAHECO® II content and temperature for heterogeneous PP
materials (-20°C (0), 10°C (0), 23°C (d), 30°C (V), 40°C (0); conventionally
determined values, filled symbols; values determined from J-TJ stability assessment
diagrams, open symbols) [2,3,23,24]
800
600
E
E
b 400
:s.><
:!2
00
200
0
0 20 40 60 80 100
RAHECO Content (wt.%)
Fig. AIOS. bid as a function of RAHECO® II content and temperature for heterogeneous PP
materials (-20°C (0), 10 °C (0), 23°C (d), 30°C (V), 40°C (0); conventionally
determined values, filled symbols; values determined from c5--T" stability assessment
diagrams, open symbols)
J- TJ and ~ T6 Stability Assessment Diagrams 147
Table A9. Empirical parameters PJ and P6 defining stability limits of crack propagation, sum-
marized for some polymeric materials
Material Loading conditions and Values of empirical parameters Reference

range of validity PJ and P6
pJ (N/mm) P6 <mm)
Polycarbonate y= 1.5 mls 31 2.7 x 10-3 [17,19]

T= 20 °C-90 °C
PPIEPRIPE y= 1.5 mls 2.6 [2,22-24]

reactor blends T=20°C
RAHECO®I
RAHECO®n y= 1.5m1s 1.6-2.2 17-57 x 10-3 [2,3,23,24]

T= 10 °C-40 °C
HECO y= 1 mls 0.9 pp.281-300

T=20°C
PPIEPR y= 1 mls 1.1 pp. 257-266,

blends T=20°C [25]
TPU/ABS y= 1 mls 3.0 [26]

T=20°C 3.1
0% ABS 2.0
50% ABS
80% ABS
PC/ABS y= 1 mls [27]

T=20°C
0% ABS 0.06
5% ABS 0.12
10% ABS 0.14
20% ABS 0.46
30% ABS 2.8
90% ABS 4.2
100% ABS 26
Quantification of the toughness behaviour and the fonnulation of relationships

between the structure and material properties, of which an example is shown in
Fig. AI09 in the fonn of the J-integral versus the interparticle distance, are neces-
sary for the systematic optimization of mechanical properties. As in the situation
shown in Figs. AI07 and A108, the detennination of the brittle-to-tough transition
related to the critical interparticle distance of the PP materials investigated here is
only possible if the alternative assessment method described above is used.
80
60
E
--E 40
~
ti:E
-, 0
20
0
0.5 1.0 1.5 2.0 2.5
Interparticle Distance (IJm)
Fig. AI09. Toughness parameters as a function of interparticle distance for PP materials with
different RAHECO® I contents (-20 °C (0), 10 °C (0), 23 °C (~), 30 °C (V), 40 °C
(0); conventionally determined values of JId S\ filled symbols; values of JIlT deter-
mined from J-TJ stability assessment diagrams, open symbols) [2,24]
The empirical parameters PJ and p(j for the RAHECO® materials and some
other PP materials (HECO (pp. 281-300) and PPIEPR blends (pp. 257-266), [25])
are summarized in Table A9. For purposes of comparison, TPU/ABS [26] and
PCIABS [27] blends are also listed. The parameter PJ varies over several orders of
magnitude with values from 0.06 Nlmm to 31 N/mm. This material dependence
can be observed in a concentration series (see the PC/ABS blends) as well as by
comparison of different material systems.
4 Summary
For determination of the toughness behaviour of polymeric materials, the concepts

of elastic-plastic fracture mechanics are used because of the viscoelastic-plastic
behaviour of polymers, and J values related to resistance against unstable crack
propagation are determined. Besides the determination of resistance against stable
crack initiation and propagation, J-TJ and 8-T(j stability assessment diagrams can
be applied as another method to evaluate the failure safety of polymeric materials.
This evaluation method has been discussed, using the examples of the temperature-
dependent toughness behaviour of polycarbonate and of the toughness behaviour
as a function of concentration and temperature of multi-phase random copolymers
based on polypropylene. Furthermore, a comparison of empirically estimated pa-
J-TJ and t5-Ts Stability Assessment Diagrams 149
rameters defining stability limits has been made for several materials. The use of
this procedure also makes possible the determination of fracture mechanics mate-
rial parameters related to resistance against unstable crack initiation in cases where
no unstable crack propagation can be observed because of the material behaviour
under the loading conditions used and with the selected specimen sizes.
Acknowledgement
The authors acknowledge the fmancial support of the German Research Founda-
tion (OFG) within the context of the graduate course 'Heterogeneous Polymers
(Polymer Science)' and the innovation course 'New Polymer Materials'.
References
1. Swallowe G. M. (Ed.) (1999): Mechanical Properties and Testing of Polymers. An A-Z

Reference. Polymer Science and Technology Series, Brewis D., Briggs D. (Eds.), Chapman
& Hall, London
2. Grellmann W., Seidler S., Jung K., Kotter I. (2001): Crack-resistance behavior of propylene
copolymers. J. Appl. Polym. Sci. 79: 2317-2325
3. Seidler S., Grellmann W. (1998): Application of crack resistance concept to the toughness
characterization of high impact thermoplastics. In: Proceedings of the 18th Discussion Con-
ference: Mechanical Behaviour of High Impact Materials, Prague, Czech, July 20-23: 63-71
4. Seidler S., Grellmann W. (1995): Application of the instrumented impact test of the tough-
5. Fasce L., Pettarin V., Bernal C., Frontini P. M. (1999): Mechanical evaluation of propylene
polymers under static and dynamic loading conditions. J. Appl. Polym. Sci. 74: 2681-2693
6. Che M., Grellmann W., Seidler S., Landes J. D. (1997): Application of normalization in
determining J-R-curves in glassy polymer PVC at different crosshead speeds. Fatigue Fract.
Engng. Mater. Struct. 20: 119-127
7. Bernal C. R., Montemartini P. E., Frontini P. M. (1996): The use of load separation and
normalization method in ductile fracture characterization of thermoplastics polymers. J.
tearing stability in ABS materials. Appl. Macromol. Chern. Phys. 270: 5-12
9. Will P. (1994): R Curves of energy dissipative materials. J. Mat. Sci. 29: 2335-2340
10. Blumenauer H., Pusch G. (1993): Technische Bruchmechanik. Deutscher Verlag fUr Grund-
stoffindustrie, Leipzig
11. Paris P. C., Tada H., Zahoor A., Ernst H. (1977): The theory of instability of the tearing
mode of elastic-plastic crack growth. In: ASTM STP 668: Elastic-Plastic Fracture: 5-36
12. Tada H., Paris P. C., Irwin G. R. (1985): The Stress Analysis of Cracks Handbook. Paris
Production Inc., Del. Research Corporation, 2nd ed., St. Louis
14. Merkle J. G., Corten H. T. (1974): J integral analysis for the compact specimen, considering
axial forces as well as bending effects. J. Pressure Vessel Techno\. 96, 4: 286-292
15. Grellmann W., Sommer J.-P., Hoffmann H., Michel B. (1987): Application of different J-
ceedings of the 1st Conference on Mechanics, Prague, Czech, June 29-July 3, Vol. 5:
129-133
16. Paris P. C., Johnson R. E. (1983): Fracture resistance curves and engineering applications.
In: ASTM STP 803 Vol. II: 5ff
17. Grellmann W., Lach R. (1996): Influence of temperature on toughness, fracture surface
morphology and molecular relaxation behaviour of polycarbonate. Appl. Macromol. Chern.
Phys.273: 191-208
18. ASTM E 813-81 (1981): Standard Test Method for J Ie, a Measure of Fracture Toughness.
19. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour of poly(methylmeth-
acrylate), polystyrene and polycarbonate. Appl. Macromol. Chem.Phys. 253: 27-49
20. Paulik C., Gahleitner M., NeiBI W. (1996): Flexible, tough and resilient PP copolymers.
Kunstst. PlastEurope 86: 1144-1147
21. Kim G. M., Michler G. H., Gahleitner M., Fiebig 1. (1996): Relationship between morphol-
ogy and micromechanical toughening mechanisms in modified polypropylenes. J. Appl.
Polym. Sci. 60: 1391-1403
22. Starke 1. D., Michler G. H., Grellmann W., Seidler S., Gahleitner M., Fiebig J., Nezbe-
dova E. (1997): Fracture toughness of polypropylene copolymers: influence of interparticle
distance and temperature. Polymer 39: 75-82
23. Grellmann W., Seidler S., Kotter I. (2000): Brittle-to-tough transition in toughened polypro-
pylene copolymers. In: Proceedings of the 13th European Conference on Fracture, ECF 13,
San Sebastian, Spain, September 6-9, CD-ROM 'Polymer and Composites' No. 17: 1-8
24. Grellmann W., Michler G. H., Seidler S., lung K., Kotter I. (1998): Fracture behaviour of
propylene copolymers - influence of interparticle distance and temperature. In: Proceedings
of the 12th European Conference on Fracture, ECF 12, Sheffield, DK, September 14-18,
Vol. III: 1441-1446
25. Koch T., Grellmann W., Seidler S. (1997): Morphologie und Ziihigkeit von Blends auf der
Basis von Polypropylen. In: Proceedings of the International Conference Welding Technol-
ogy, Materials and Materials Testing, Fracture Mechanics and Quality Management, Vienna,
Austria, September 22-24: 497-506
26. Schierjott D. (1990): Bruchrnechanische Bewertung der Ziihigkeitseigenschaften von Poly-
merkombinationen auf der Basis Thermoplastischer Polyurethane. PhD thesis, TH Leuna-
Merseburg,Merseburg
27. Grellmann W., Seidler S. (1991): unpublished
B
Morphology-Property
Correlations
B1 Homopolymers
B2 Blends
B3 Copolymers
B1 Homopolymers
B 1.1 Supermolecular Structure

and Mechanical Behaviour
of Isotactic Polypropylene
M. Raab, J. Kotek and J. Baldrian, Prague,
and W. Grellmann, Merseburg
1 Structural Hierarchy in Isotactic Polypropylene
The kinetic theory of rubber elasticity has successfully explained the macroscopic
mechanical behaviour of polymer networks starting from the thermodynamics of
individual molecules [1]. This approach, however, cannot be applied to semicrys-
talline polymers, where several levels of structural hierarchy can be distinguished,
starting from molecular architecture and going up to supermolecular morphology
and the macroscopic geometry of a body. Four basic levels of structural hierarchy
in isotactic polypropylene are schematically shown in Fig. Bl [2]. In the case of
rubber-modified polypropylene, the embedded rubber particles form an additional
structural level. The individual structural levels reflect not only the composition
but also the thermal and deformation history of the material.
O.1Srm 10nm 10".1m O.11llT1

!------l 1---1
Fig. HI. Structural hierarchy ofisotactic polypropylene

154 M. Raab, J. Kotek, J. Baldrian, W. Grellmann
Under mechanical loading, the complex heterogeneous structure is manifested

by various localized processes. Thus, a corresponding hierarchy in the deformation
and fracture phenomena develops. In most cases, it is the highest structural level
which plays a decisive role in determining the ultimate mechanical behaviour. In
particular, the polymorphism of isotactic polypropylene (the fraction of the ex and
13 crystalline modifications) influences the deformation and fracture behaviour. In
the case of rubber-modified polypropylene, the macroscopic mechanical behaviour
is affected both by the supermolecular crystalline structure and by the embedded
rubber particles.
The processes in the individual structural levels are interrelated. Some of these
effects can be described in terms of feedback mechanisms [3]. For example, the
geometry of the neck shoulder during the cold drawing of a rubber-modified poly-
propylene depends on the rubber concentration. On the other hand, the shape of
the neck, in turn, affects the intensity of the structural transformation on the mo-
lecular scale [3].
In this chapter, an investigation of the fracture toughness of injection-moulded
isotactic polypropylene is presented for notches located at different distances from
the moulding gate. In parallel, the crystalline structure at the same locations has
been monitored by wide-angle X-ray diffraction. Thus, the relationship between
the fracture toughness and the crystalline structure of the material could be estab-
lished.
2 The Elyashevich Hypothesis

Polymers with flexible chains can form two types of crystallites. The first type (A)
involves lamellae, with chains folded regularly and aligned perpendicular to the
long axes of the crystalline lamellae. In this case, an individual molecule typically
forms a single lamella. The second type (B) consists of extended-chain crystallites
with molecular chains aligned along the crystallite length. In this case many mo-
lecular chains are incorporated in several crystallites and interconnect them. It is
the change of the Gibbs energy during crystallization which decides which type of
crystallites is preferred.
The theoretical analysis of Elyashevich et al. [4,5] has shown that for polymers
with flexible chains such as polyethylene or polypropylene, two temperature 'win-
dows' exist where there is an increased probability p of the formation of extended-
chain crystals of type B. These temperature intervals are located just below the
melting temperature Tm and in a certain region above the glass transition tempera-
ture Tg• Between these two temperature intervals, virtually only crystallites of type
A are formed, and the crystallization rate shows a maximum in the same region.
This is schematically shown in Fig. B2.
According to the Elyashevich hypothesis, in the case of isotactic polypropylene
the morphology with extended-chain crystallites corresponds to the 13 phase.
Supermolecular Structure and Mechanical Behaviour 155
0.5
Fig. B2. Probability of the formation of extended-chain crystallites as a function of temperature

(schematic illustration, after Elyashevich et aI. [3,4])
3 Experimental
3.1 Starting Materials and Sample Preparation
The material used in this study was a commercial grade of the isotactic polypro-
pylene homopolymer Mosten 58.412, supplied by Chemopetrol Litvinov, Czech
Republic. The manufacturer characterizes the material by a melt flow index of
3 g/IO min (21.2 N, 230°C) and a weight-average molecular mass Mw of about
170000. Some of this polymer was blended with 7.6 wt.% of Royalene 1M 7565
(Uniroyal, USA), which consists of EPDM rubber and PE-HD (65/35). From pel-
lets of these materials, standard bars for impact testing with dimensions
4 x 10 x 120 mm 3 were injection-moulded at the Polymer Institute, Bmo. A Bat-
tenfeld BA 7501200 injection-moulding machine was used for the production of
the bars.
3.2 Test Specimens
Single-edge-notched bend specimens for impact testing (length L = 60 mm, width

W = 10 mm, thickness B = 4 mm) were prepared from the injection-moulded bars
of original length of 120 mm. The notches, with a depth of 2 mm and a tip radius
of 0.2 /lm, were introduced with a razor blade. In the case of the neat polypropy-
lene specimens, the location of notches along the edge of the original moulding
varied as shown in Fig. B3. Seven locations of the notch in the original moulding
were selected, the distances from the moulding gate being 30, 40, 50, 60, 70, 80
and 90 mm. In the case of the rubber-modified polypropylene specimens, the
original mouldings were cut in half and each half was provided with a razor-blade
notch in its centre. Thus, the distances between the notch and the moulding gate
were 30 and 90 mm; these locations are also depicted in Fig. B3.
{ 120mm
90
30
I
[
120mm
Fig. B3. Location of notches in injection-moulded bars of neat polypropylene (top) and rubber-
modified polypropylene (bottom)
3.3 J-Integral Determination
The J-integral was selected as a toughness characteristic [6]. An instrumented

impact pendulum with support span s = 40 mm was used to perform the experi-
ments. The experimental procedure has been described in detail elsewhere [7]. The
tests were repeated under identical conditions several times. An average value and
standard deviation of the J-integral were calculated from the test results.
In the impact tests, the load-deflection (F-j) diagrams were recorded. The gen-
eral deformation energy up to the maximum impact load (AG) was separated into
elastic (A. I ) and plastic (ApI) parts. The evaluation procedure described in (pp. 71-
86) was used with the Sumpter and Turner approximation. As shown in [6], this
method is particularly suitable for polymer materials.
3.4 X-Ray Diffraction
The crystalline structure of the materials at various distances from the moulding
gate (corresponding to the notch locations) was characterized by wide-angle X-ray
diffraction. An HZG wide-angle powder diffractometer (Prazisionsmechanik,
Freiberg) using CuI<.. radiation was used for the measurement. Profile analysis was
then applied to the diffraction patterns obtained. The relative amount K of the
crystalline p modification was calculated [8] from (Bl):
I~
K= , (Bl)
10.1 + 10.2 + 10.3 + I ~
where 1['1 is the integral intensity of the peak for the (300) diffracting plane of the p
phase, and I al , Ia2 and Ia3 are the integral intensities of the (110), (040) and (130)
planes of the a phase.
4.1 Neat Polypropylene
The results of the instrumented impact tests, expressed as the J-integral, together
with relative amount of the ~ phase, characterized by the K value (B 1), are plotted
in Fig. B4. A characteristic decrease in the fracture toughness with increasing
distance from the moulding gate can be seen. The relative concentration of the ~
phase shows the same trend, thus offering a structural explanation for the tough-
ness changes. Both characteristic values show the steepest gradient in the central
part of the injection-moulded bar (gate-notch distance L = 60 mm). It is interesting
to note that a relatively small amount of the ~ phase causes such a dramatic effect
on the toughness. This fact must be incorporated in a realistic structural model.
5 0.20
o J 1d
----K
2
4 0.15
E
~3 Q 0.10 ~
Q Q
,-u
2
Q 22 0.05
1 L-~ __ ~~ __ ~ __ ~~ __ ~ __ ~-L __ ~~ __ ~
o 30 60 90 120
L(mm)
Fig. B4. Fracture toughness, expressed as the J-integral, and p phase concentration K of neat
polypropylene as a function of distance from the moulding gate
4.2 Rubber-Modified Polypropylene
In the case of the rubber-modified polypropylene, the J-integral and the ~ phase
concentration K were measured at two different distances from the moulding gate
(30 and 90 mm). The results are reported in Table B 1. Even though the results are
similar to those reported for neat polypropylene, the difference between the two
notch locations is less pronounced. Correspondingly, the ~ phase concentration
closer to the gate is lower than for the neat polypropylene.
Table B1. J values and 13 phase concentration of rubber-modified polypropylene for two gate-
notch distances L
L(mm) J1d(N/mm) K
30 4.5 ±0.3 0.04

90 3.0 ± 0.2 0.0
5 Structural Model
Several structural models have been suggested to explain the observed relation
between the p phase concentration and the toughness of isotactic polypropylene.
Karger-Kocsis [9] has explained the higher toughness of the p phase by a phase
transformation, i.e. by a mechanism similar to that which improves the toughness
of zirconium ceramics. Indeed, a stress-induced transformation from a less dense
(hexagonal p phase) to a more dense (monoclinic a phase) crystalline structure
was demonstrated [10]. Nevertheless, the assessment of its energy contribution at
the crack tip is difficult. Besides, the p~a phase transformation in polypropylene
is associated with a volume decrease, in contrast to the volume increase in zirco-
nium ceramics.
Fig. B5. A schematic model of the structure of polypropylene, with a crystallites (left) and p
crystallites (right)
It might be suggested that the higher toughness of the f3 phase of isotactic poly-
propylene, compared with the a modification, is an inherent property of the f3
phase itself. For example, it might reflect the sheaf-like spherulitic structure, as
suggested by Tjong [11]. The theory of Elyashevich and co-workers [4,5] offers
another possibility: if the f3 phase consists basically of extended-chain crystallites
with a high content of continuous chains, this could also cause better continuity in
the amorphous part of the material. It is important to note that in this case, a rela-
tively small fraction of the f3 phase can exert a long-range effect. The polypropyl-
ene structure according to this model, with a and f3 crystallites, is schematically
illustrated in Fig. B5.
6 Conclusions
Isotactic polypropylene is polymorphic, i.e. it can crystallize in three crystalline

modifications, the a and f3 modifications being of practical importance. The pres-
ence of the f3 modification can be detected by wide-angle X-ray diffraction and by
DSC. In the present study it has been demonstrated, for injection-moulded test
specimens, that an increase in the f3 phase content is reflected in a corresponding
toughness increase.
Even though the effects of the morphology of isotactic polypropylene on the
macroscopic properties have been studied experimentally by several authors, no
unambiguous structural model has been accepted. This contribution brings a new
approach to this problem, derived from the Elyashevich theory. It has been sug-
gested that the presence of the f3 crystallites has a long-range effect on the amor-
phous phase, inducing a high content of continuous chains.
Clearly, many questions still remain to be resolved. Of these, in particular, the
problem of the stability and time dependence of the morphology should be studied.
Moreover, the effects of the f3 phase are associated with practical problems of
testing and application of polypropylene-derived materials [12-14].
Acknowledgements
The authors are indebted to the Grant Agency of the Czech Republic (Projects
106198/0718 and 106199/PO 11), the Grant Agency of the Academy of Sciences of
the Czech Republic (Project A4050904) and the Martin-Luther-University of
Halle-Wittenberg for their support. The assistance of W. Hesse in the fracture
testing and the evaluation of results is also gratefully acknowledged.
References
1. Schurz J. (1974): Physikalische Chemie der Hochpolymeren. Springer, Berlin

2. Slizova M., Raab M., Soya M. (1994): Importance of polymer morphology in understanding
technological processes. Int. Polym. Sci. Techno!. 21: T66-T74
3. Raab M., Kotek J., Baldrian J. (1996): Strukturebenen der Polymerplastizitat und ihre
Wechselwirkungen. In: Tagungsband Polymerwerkstoffe '96, Merseburg, Germany, Sep-
tember 18-20: 57-60
4. Elyashevich G. K., Poddubny V. I., Raab M. (1990): Relationships between orientation and
crystallization of flexible chain polymers. Plasty Kaucuk. 27: 353-357
5. Poddubny V. I., Elyashevich G. K., Baranov V. G., Frenkel S. Va. (1980): Topomorphic
states and phase transitions connected with the crystallization of polymers. Polym. Engng.
Sci. 20: 206-211
J-integral concept. In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Se-
ries. Institut fUr Mechanik, Berlin, Chernnitz, No. 17: 48-72
8. Tjong S. c., Li R. K. Y., Cheung T. (1997): Mechanical behavior of CaC0 3 particulate-
filled p-crystalline phase polypropylene composites. Polym. Engng. Sci. 37: 166-172
9. Karger-Kocsis J. (1996): How does 'phase transformation toughening' work in semicrystal-
line polymers. Polym. Engng. Sci. 36: 203-210
10. Varga J. (1992): Supermolecular structure of isotactic polypropylene. J. Mat. Sci. 27:
2557-2579
11. Tjong S. c., Cheung T., Li R. K. Y. (1996): Mechanical behavior of injection molded
p-crystalline phase polypropylene. Polym. Engng. Sci. 36: 100-105
12. Raab M., Kotek 1., Baldrian J., Grellmann W. (1998): Toughness profile in injection molded
polypropylene. J. App!. Polym. Sci. 69: 2255-2259
13. Varga J., Ehrenstein G. W. (1998): Beta modification ofisotactic polypropylene. In: Karger-
Kocsis J. (Ed.) Polypropylene: A-Z Reference. Kluwer Academic Publishers: 51-59
14. Kotek 1., Raab M., Baldrian J., Grellmann W. (2001): The effect of specific p-nucleation on
morphology and mechanical behaviour of isotactic polypropylene. J. App!. Polym. Sci.,
submitted
B 1.2 Correlation Between Structure
and Toughness Behaviour
of High-Density Polyethylene
under Impact Load
H. Beerbaum and W. Grellmann, Merseburg,
and S. Seidler, Vienna
1 Introduction
The crack initiation and crack propagation mechanisms of semicrystalline poly-

mers have not been completely resolved. The blunting process in polymers is in-
fluenced by many factors determined by the structure [1,2]. The various ideas
about the actual blunting process in polymers are described in [3]. A decisive
improvement is expected if structural aspects can be included in the examination
of micro-mechanical processes and a correlation between morphology and tough-
ness can be established.
On the basis of fracture mechanics experiments under static load by Egan and
Delatycki [4], the molecular weight has been shown to have a crucial influence on
the fracture toughness of high-density polyethylene (PE-HD). The rate of disen-
tanglement varies inversely with the chain length and the number of tie molecules
[5]. If the difference in molecular weight is small, short-chain branching and there-
fore crystal thickness play an important role in determining the fracture behaviour
[4]. The time to failure under static load, as established by Brown and Zhou [6],
depends on crystallinity, on crystal thickness and, most importantly, on the intrin-
sic resistance to slow crack growth of the crystal network, as determined by the
density and branching of its tie molecules. The influence of crystallinity on the
strength of the crystal-tie molecule network can be superimposed on the influence
of molecular weight. However, the importance of the crystal thickness in deter-
mining the strength of the crystal-tie molecule network is not influenced by mo-
lecular weight [7].
The thickness of the lamellae, the side chain concentration and the degree of
order in the crystalline regions have been shown by Barry and Delatycki [8] to
influence the resistance of PE-HD against crack initiation. Thicker and the more
ordered crystalline regions imply a greater resistance against crack initiation in the
material. The tearing modulus appears to be influenced mainly by changes of the
morphology that are due to side chain branches [8].
As the crystallinity of the polymer decreases, which means the crystal thickness
decreases and the thickness of amorphous regions increases, the polymer becomes
more compliant and the crack tip becomes blunter for a given applied stress. The
blunting of the notch reduces the stress concentration so that, for example, the time

162 H. Beerbaum, W. Grellmann, S. Seidler
to failure under static load increases [6]. Also, the fracture toughness and J-
integral, detennined as critical parameters at the point of unstable crack growth
under static load, increase with decreasing crystallinity and decreasing thickness of
lamellae [4]. On the other hand, as the crystallinity of the polymer increases at a
constant molecular weight, which means the tie molecule density decreases, the
resistance against stable crack propagation increases [8,9]. This is a consequence
of the greater strength of crystals with a higher crystallinity. The results of Lu et al.
[9] indicate that, for slow crack growth, the strength of the crystals is more impor-
tant than the number of tie molecules. In contrast, the critical parameters at the
point of stable crack initiation are not influenced by the crystallinity, but are influ-
enced by the molecular weight and the predominant stress conditions [8].
A model of the defonnation of semicrystalline polymers was introduced by
Friedrich [10]. It can be concluded from this model that at the start of defonnation
the strain is accommodated almost entirely in the interlamellar amorphous regions.
Then void and craze fonnation occur, and most of the energy is absorbed by chain
sliding and unfolding of chain segments in the fibrils. The fracture process of
polyethylene has been described by models by Brown et al. [11] and Dekker and
Bakker [12]. These models indicate that the failure of fibrils is the important
mechanism of stable crack growth.
In the work described below, various high-density polyethylenes were examined
by fracture mechanics methods to study the quantitative connection of fracture
mechanics values describing crack processes with molecular weight, crystallinity
and the various morphologies that result. On the basis of the elastic-plastic mate-
rial behaviour of high-density polyethylene, the concepts of EPFM were used for
the assessment of the load-deflection diagrams. In addition, the crack propagation
process was characterized using investigations of the fracture surfaces by means of
light microscopy and scanning electron microscopy.
2 Materials
For the fracture mechanics tests, two sets of high-density polyethylenes, one set
with a constant molecular weight and different densities and one set with a con-
stant density and varying molecular weight, were selected with the aim of achiev-
ing morphology-toughness correlations. The high-density polyethylenes examined
and their properties are shown in Table B2.
The melting temperature Tm was ascertained by differential scanning calorime-
try (DSC) measurements according to DIN 51004 at a heating rate of 10 Klmin.
The test mass was 11 mg. The melt flow rate MFR was detennined according to
ISO 1133 at 190°C under a load of 21.6 kg. For PE-HD D90 only, the load was
5 kg. The measurement of the modulus of elasticity in tension Et and the yield
stress oy was perfonned with a static tension test according to ISO 527-1 at a test
velocity of 100 mmlmin. For PE-HD M36, the test velocity was 500 mmlmin.
Structure and Toughness Behaviour of High-Density Polyethylene 163
In addition to the characterization of mechanical properties, the crystallinity and

the structural parameters on the nanometre scale were determined.
Table B2. Characteristic and mechanical properties of the ofPE-HD materials investigated
PE-HD {J Mw SCB Tm MFR E. oy

material (kg/m') (kg/mol) (branches/ (DC) (g/10 min) (MPa) (MPa)
1000 C atoms)
M36 936 220 4-5 115 14 480" 20"

M48 948 220 1.8-2.3 114 16 1103 26
M60 960 220 1.5 120 17 1380 31
D90 952 90 2-2.5 122 16.2b 1172 28

D360 944 360 2.5-3.5 120 5.8 940 24
D550 954 550 0.8-1.3 124 1.3 1250 30
a lIT = 500 mmlmin; b MFR 5 kg/190DC
The crystallinity a and structural parameters such as the long period L p, thick-
ness oflamellae Lc and calculated tie molecule density T(M2) (see next section) are
listed in Table B3. The crystalline fraction of each high-density polyethylene was
measured using DSC. To do this, the melting enthalpy of the crystalline peak was
determined and compared with the ideal melting enthalpy of a 100 % crystalline
polyethylene [13]. The first heating cycle was used for this determination, in order
to agree with the measurements on the additional samples processed at various
cooling rates (see below; Table B4). For these annealed samples, the first heating
cycle is the decisive cycle for the characterization of their crystalline morphology
after processing.
Table B3. Structural parameters of the PE-HD materials investigated
PE-HD (J a Lp Lcth<o" LcTEM ~M2)

material (kg/m') (%) (nm) (nm) (nm) (nm- 1)
M36 936 55 20.0 10.5 9.6 0.69

M48 948 64 23.5 13.9 10.1 0.57
M60 960 75 29.5 17.5 12.7 0.40
D90 952 69 23.0 14.2 9.2 0.16

D360 944 61 22.5 12.8 9.0 1.01
D550 954 66 28.5 16.5 8.7 1.22
"Calculated from Lcth<o= (0.625 nm x 414 K)/(414 K - Tm) [14]

Small-angle X-ray scattering (SAX:S) was used to determine the long period.
The long period is the sum of the thicknesses of the crystalline and amorphous
domains, including the boundary domains. The theoretical thickness of the lamel-
lae Lctheo was calculated from the melting temperature Tm in the first heating cycle
of the DSC measurements using an equation due to Illers and Hendus [14] (Ta-
ble B3). TEM investigations of the lamellar structure were also performed. The
measured thickness of the lamellae LcTEM was determined from the TEM images
with the help of an image analysis system. The calculation of the tie molecule
density T(M) will be explained separately in the next section.
The molecular weight of the materials of type M is constant, and for the materi-
als of type D the density is almost constant, as can be seen from Table B2. The
material PE-HD M48 can also be classed as type D because of its density of
948 kglm3• As usual, the crystallinity rises with increasing density, as can be seen
from Table B3. As expected, the elastic modulus and yield stress increase with
increasing crystallinity, as can be seen from Tables B2 and B3.
The materials listed in Table B2 were compression moulded at a temperature of
180°C and a pressure of 80 bar. The cooling rate vc was 15 Klmin, in accordance
with DIN 16770 (procedure B). The range of crystallinity was expanded to deter-
mine the influence of crystallinity on the toughness behaviour and the crack
growth in detail. To do this, the materials in Table B2 were processed at other
cooling rates. However, the other compression-moulding parameters were main-
tained constant. Some materials were quenched at 55 Klmin, and other materials
were slowly cooled at a cooling rate of 1 Klmin. These differently cooled materials
are listed with their structural parameters in Table B4.
Table B4. Structural parameters ofPE-HD materials processed at different cooling rates
PE-HD VC U a Lp Lc1heoa LcTEM T(M2)

material (Klmin) (kglm3) (%) (nm) (nm) (nm) (nm- 1)
M36K 55 934 53 11.6

M48K 55 947 65 22.0 14.7 0.62
M48L 1 954 71 28.5 17.3 12.6 0.43
b
M60K 55 954 70 24.5 20.0 0.54
M60L 1 969 85 35.0 22.6 12.1 0.28
a Calculated from Lctbeo = (0.625 nm x 414 K)/(414 K - Tm) [1]

b Not determined
The relationship between the structural parameters, molecular weight and crys-
tallinity for all materials investigated is shown in Fig. B6. It can be seen that the
structural parameters are not particularly influenced by the molecular weight. Only
for the materials having nearly the same density does this dependence on molecu-
lar weight need to be considered.
60
40
•ffi Lp
E 20 .--~ • • L
ctheo
.s '" L
*
~
~ ~ f!.
cTEM
:!: 40
w 0
I-
0 50 300 450
...J
- 0
G>
Mw (kg/mol)
~
...J'd
20 20
~
0..
...J
0 0
55 60 65 70 75 80 85
a (%)
Fig. B6. Structural parameters as a function of molecular weight Mw and crystallinity a
The long period and the thickness of the lamellae increase with increasing
crystallinity, independently of the molecular weight, whereas the amorphous thick-
ness decreases slightly with increasing crystallinity. This dependence on crystal-
linity applies to all materials considered here. The increase of the thickness of the
lamellae LcTEM is not so pronounced as for Lctheo '
3 Calculation of Number of Tie Molecules
Tie molecules are molecules that extend into two adjacent crystalline domains and
hold these together. In combination with branches, they have a great influence on
the disentanglement rate of the chains during deformation. There are two types of
effective tie molecules. The 'conventional' tie molecules connect two adjacent
crystals, with an amorphous region in between. If two molecules from adjacent
crystals join together in a strong random entanglement, they can act as a tie mole-
cule too. These tie molecules are called the 'entangled' tie molecules.
The significance of the number of tie molecules for the toughness behaviour of
polyethylene has been established by various authors [5,6,15]. A calculation of the
number of tie molecules was carried out to complete the determination of struc-
tural parameters on the nanometre scale. The number of tie molecules depends on
the molecular weight, and especially on the chain length and on the long period,
which means on the distance between the lamellae.
3.1 Basics
There are some experimental methods, such as TEM, NMR and neutron scattering,
for determining the number of tie molecules, but they are not suitable for solid
specimens. A method has been introduced by Brown and Ward [16] to calculate
the fraction of tie molecules from the brittle fracture stress at very low tempera-
tures. However, the sensitivity of this method is not sufficient for materials with
similar numbers of tie molecules. For this reason, many models for the calculation
of the number of tie molecules, based on a model by Huang and Brown [5], exist
in the literature.
The basis of this calculational model is the random-coil model of Flory, where
the probability of a given end-to-end distance p(r) is given by (B2), where r is the
end-to-end distance of the random coil in the melt:
(B2)
where
(B3)
and
b2 =3.. _1 . (B4)
2 ;:2
This probability p(r) depends on the root mean square value ;:2 of the end-to-
end distance of a random coil. The value of ;:2 depends on the molecular weight
and is given by (B5) for the liquid state, where the constant C is 6.8, the link
length I is 0.153 nm and x is the number oflinks [13]:
;;2 =CxP. (B5)
It is basically assumed that if the end-to-end distance of the random coil in the
melt r is equal to or greater than a critical distance between adjoining lamellar
crystals, a tie molecule will be formed. This distance depends mainly on the value
of the long period. The probability of forming a tie molecule P in a monodisperse
system can be calculated from (B6), according to Huang and Brown [5]. In this
calculation only the conventional type of tie molecules is considered. The equation
is
I r2 e-b2r 2 dr
00
P = 1 _LT"--_ __
(B6)
3 ooI 2 _b 2r2
r e dr
o
The factor 113 is introduced because the two other dimensions of the lamellar
crystals are much larger than the long period [5].
The critical distance between the lamellar crystals LT is defmed by Huang and
Brown [5] as the sum of twice the thickness of the lamellae and the thickness of
the amorphous layer, i.e.
(B7)
The absolute value of the calculated number of tie molecules depends on the
choice of the critical distance, but the relative numbers of tie molecules are much
less sensitive to the specific choice of this distance.
3.2 Tie Molecule Density
The number of conventional tie molecules formed per chain in a mono disperse
polymer T1(M) was calculated from (BS), given by Yeh and Runt [17]. The term
I(r, Lp) is an integer function i(x) which takes account of the increasing radius of
gyration (B9). The critical distance is defmed as 2.45L p . This means a tie molecule
can only be formed if the end-to-end distance of the random coil in the melt r is
equal to or greater than 2.45L p:
(BS)
with
I(r,Lp)=i( r
l.22Lp
-1). (B9)
The number of entangled tie molecules formed per chain in a monodisperse

polymer is described by the tie molecule density T(Mz), given by (BID) according
to Strebel and Moet [15]. This calculation considers the number of entanglements
N per chain after crystallization. The number of entanglements is assumed to be
proportional to the ratio of the hydrodynamic volume of a chain having a given
molecular weight to that of a chain with the critical molecular weight. The number
of entanglements N was calculated from the integer part of this quotient. If the two
limits of integration, the critical distance for the lower limit and the extended chain
length (instead of infmity) for the upper limit, are within the lamellae, a tie mole-
cule will not form. Thus, a factor should be introduced to take account of the pos-
sibility that both endpoints are in the lamellae [15]. The critical distance is defmed
as 1.22 Lp. The resulting equation is
r(M 2 )= r
1.22Lp
dN r(I-~) e-b2r2 dr.
l.33 r
(BIO)
The tie molecule density T(Mz) is not an absolute value of the number of tie
molecules but is related to the dimensions of the lamellae.
Bulk polymers are not generally monodisperse. Huang and Brown [5] have
shown that the probability of forming a tie molecule for a monodisperse polymer is
similar to the probability of fonning a tie molecule for a polymer with a molecular-
weight distribution. In the further considerations below, the tie molecule density
T(M2) is used because the number of tie molecules is detennined mainly by the
number of entangled tie molecules for the polyethylenes investigated.
The dependences of the tie molecule density T(M2) on the molecular weight,
crystallinity and structural parameters are shown in Fig. B7. At a nearly constant
crystallinity, the tie molecule density increases if the molecular weight is in-
creased.
However, at constant molecular weight the tie molecule density decreases if the
crystallinity, long period or lamellar thickness increase. Above a lamellar thickness
of 12 nm, the tie molecule density does not vary (Fig. B7). Nevertheless, the tie
molecule density decreases continuously with increasing crystallinity or long period.
1.5 1.5
1; 1.0
1; 1.0
.s .s
.....
::.i:
~'"
::.i:
~
j:::" j:::"
0.5 0.5
0.0 0.0
150 300 450 60 70 80
Mw{kg/mol) a{%)
1.5 1.5
~
.s 1.0 i 1.0
~'" ~'"
::.i: ::.i:
~
j:::"
0.5 j:::" 0.5
0.0 0.0
20 25 30 35 9 10 11 12 13
Lp (nm) LCTEM (nm)
Fig. B7. Effect of structural parameters on the tie molecule density T(M2)
4 Fracture Mechanics Analysis
For the fracture mechanics experiments, a Charpy pendulum tester with a maxi-
mum impact energy of 4 J was used. The test conditions and the analysis followed
the procedure described by Grellmann et al. in (pp. 71-86). The pendulum ham-
mer speed was 1.5 mis, and SENB specimens were used. The initial crack length
was 2 mm; the crack was made using a razor blade with a tip radius of 0.2 11m.
The polyethylenes exhibited different material behaviours in the instrumented
Charpy impact test. At a molecular weight of 90 kg/mol, a linear elastic material
behaviour occurred (Fig. B8). However, elastic-plastic material behaviour ap-

peared at medium crystallinity and molecular weight, and at higher crystallinity
and molecular weight, stable crack propagation without fracture of the specimen
occurred (Figs. B8 and B 11).
Side grooves were cut into the specimens according to ESIS TC4 [18] to com-
pare the results for the different polyethylenes. The side grooves ensure unstable
crack propagation, and the crack propagation energy is reduced. For tough poly-
mers, side grooves are used to obtain a straight, ductile crack front and to minimize
the area of plane stress. The side grooves should have an included angle of 45° ± 5°
and a root radius of 0.25 mm ± 0.05 mm. The total reduction in thickness should not
exceed 20 % of the specimen thickness. Side grooves of20 % reduction in thickness
were used for the determination of crack resistance against unstable crack growth.
Crack resistance curves were determined using the multiple-specimen technique
according to A2.2 (pp. 71-86), at the same pendulum hammer speed, to identify
the crack initiation and crack propagation behaviour during stable crack growth.
Pre-cracking of the specimens was performed with a fly cutter and subsequently
the notch was sharpened by sliding a razor blade across the machined notch tip.
The initial crack length was 4.5 mm. SENB specimens without side grooves were
used. Load-deflection (F-j) diagrams at different amounts of stable crack growth
were recorded. The actual amounts of stable crack growth were determined with
the help of microscopic fracture surface analysis.
Various fracture mechanics methods were applied to determine geometry-
independent fracture mechanics parameters related to resistance against unstable
and stable crack initiation and to resistance against stable crack growth. Other
fracture mechanics parameters, characterizing the energy absorption capacity and
the deformation absorption capacity of the material, were defmed as the product of
the J-integralloading parameter and the crack-opening displacement, respectively,
and the resistance against stable crack growth (tearing modulus).
5 Results
Fracture mechanics tests under impact loading were performed to characterize the
effect of molecular weight, crystallinity and structural parameters on the toughness
behaviour of the polyethylenes during unstable and stable crack growth. To do
this, the instrumented Charpy impact test described in A2.2 (pp. 71-86) and the
test procedure explained above were used.
The separate examination of the individual effects on the fracture mechanics pa-
rameters of parameters such as molecular weight, crystallinity and structural pa-
rameters on the nanometre scale is just as important as the clarification of the joint
influence of all parameters on the toughness behaviour. It is necessary to disregard
the fact that all such parameters are interdependent and related to each other.
On the one hand, a relatively constant crystallinity was chosen for the examina-
tion of the dependences on molecular weight and, on the other hand, a constant
molecular weight was selected for the assessment of the dependences on crystal-
linity. For the determination of the effects of tie molecule density on the toughness,
all polyethylenes were examined, independently of their molecular weight and
crystallinity .
5.1 Dependence on Molecular Weight
Load-deflection diagrams give information about the strength and deformation

behaviour and on the kind of crack growth under the chosen test conditions. A
linear elastic deflection behaviour with low crack propagation energy occurred at a
molecular weight of 90 kg/mol (Fig. B8). If the molecular weight is increased, the
material behaviour changes from elastic-plastic material behaviour to stable crack
propagation without fracture (Fig. B8). The ductility increases if the molecular
weight is increased.
300 090 300

Mw=90 kg/mol
0550
~ 200 200 Mw=550 kg/mol
~
u.
no complete
diagram available
2 3 4 2 3 4
300 300
200 200
2 3 4 2 3 4
f(mm)
Fig. B8. Effect of molecular weight Mw on the load-deflection (F-f) diagrams (a/W= 0.2)
without side grooves and with side grooves (SG)
With side grooves, the material behaviour of stable crack growth at higher mo-
lecular weights changes to a material behaviour with a high crack propagation
energy AR (Fig. B8). Simultaneously, the measured impact loads, deflections and
crack propagation energies decrease when side grooves are used.
For the examination of the fracture mechanics parameters for unstable crack
propagation described below, specimens with side grooves were used.
The toughness behaviour for unstable crack propagation as a function of mo-
lecular weight is determined by energy. The Jvalue J dST increases if the molecular
weight is increased, and also the fracture toughness Kd and critical crack-opening
displacement bid increase (Fig. B9). At higher molecular weight, the chains are
longer and the disentanglement rate decreases. More energy can be dissipated by
the entanglement of the chains.
This can also be illustrated by fracture mechanics parameters determined from
the crack resistance curves as a function of the molecular weight [19-21]. The
technical crack initiation value JO.2 and the energy absorption capacity JTJ for stable
crack growth increase at a molecular weight of 550 kg/mol (Fig. B 10). Because of
the linear elastic material behaviour of the material with a molecular weight of
90 kg/mol (Fig. B8), a crack resistance curve cannot be determined for this material.
150
400
C!~
20
~
~~100 300 E E
(11 '"I
a.. 200 0
6 50 :::. 15
>::t:." 100 ""Y.
E
0 0 ~
150 300 450 10 ~
Mw (kg/mol) Iii-c
-,
5
~
0
100 200 300 400 500
Mw (kg/mol)
Fig. B9. Fracture mechanics parameters for unstable crack growth as a function of molecular
weight Mw (20 % SG, a/W = 0.2)
4 60
E E
E
~2 40 ~
N
~
....,c:i 1---'
-,
o
. 20
~~~--~~--~~~--~~--~~o
100 200 300 400 500

Mw(kg/mol)
Fig. BI0. Effect of molecular weight Mw on the fracture mechanics parameters for stable
crack growth (a/W = 0.45)
5.2 Dependence on Crystallinity
With increasing crystallinity, the material behaviour also changes from elastic-
plastic material behaviour with a medium to high crack propagation energy to a
material behaviour of stable crack propagation without fracture (Fig. Bll) [19-
21]. The impact loads increase whereas the deflection values decrease with
increasing crystallinity.
The side grooves lead to unstable crack propagation at higher crystallinity, with
a relatively high crack propagation energy. With side grooves the impact loads, the
deflection values and the crack propagation energy decrease.
300 M36 300 M60

,,=55% ,,=75 %
~ 200 200
~
u..
no complete
diagram available
4 2 3 4
300 M36SG 300 M60SG
200 200
AR
2 3 4 2 3 4
f(mm)
Fig. B11. Effect of crystallinity a on the load-deflection (F-f) diagrams (a/W= 0.2) without
side grooves and with side grooves (SG)
The difference in the fracture surfaces of specimens without and with side
grooves can be seen in Fig. B 12. Arrest lines indicate a varying crack growth rate;
crack stops are not observed for the specimens with side grooves (Fig. B12b).
Instead, a straight ductile crack front appears for the specimens with side grooves.
The toughness behaviour was determined from the deflection for the range of
crystallinity at a constant molecular weight of 220 kg/mol. The toughness for un-
stable crack propagation, characterized by the J value J dST , decreases with in-
creasing crystallinity because deformation constraint increases (Fig. B l3). How-
ever, if the crystallinity increases, the fracture toughness Kd increases and the
critical crack-opening displacement bid decreases (Fig. Bl3). The homogeneity of
the structure increases with increasing crystallinity, which means a higher strength
of the crystals. However, this also means that the semicrystalline phase system is
less deformable during unstable crack propagation because the amorphous thick-
ness decreases.
Fig. B12. Fracture surfaces of specimens without (a) and with (b) side grooves for a polyethylene
material with low crystallinity (PE-HD M36; arrow - arrest line)
150
S--' 400 20
~100 300 E
co
a.. ~ "E
200 .....
0
650 ~- ..,
~ 15
100 '-0- E
~..,
o 60 70 80
0
10 -
~
E
a (%) ...,
~"O
,.-
y--
'" -f 5
0
55 60 65 70 75 80 85
a (%)
Fig. B13. Fracture mechanics parameters for unstable crack growth as a function of crystal-
linity a (20 % SG, a/W = 0.2)
For the examination of the toughness behaviour for stable crack propagation,
the materials processed at different cooling rates (see Table B4) were studied. At a
constant molecular weight of 220 kg/mol, the technical crack initiation values are
lower at 85 % than at 53 % crystallinity, whereas they are not influenced in the

range of65 % to 75 % crystallinity (Fig. BI4). The amorphous thickness is higher
for lower crystallinity, so that void growth in the amorphous region can dissipate
more energy at the beginning of stable crack growth.
4 150
.l,*~
E E
E 2 100 ;
~ 0
:s
....,'"
° '"
~
r.o0
0 50
0
55 60 65 70 75 80 85
0. (%)
Fig. B14. Effect of crystallinity a on the technical crack initiation parameters for stable crack
growth (a/W = 0.45)
150 8
E
; 100 6
0 • •• '"
°
:s
1-'"
r.o
• • • 4
1-'"
50
0 0 0
0-- 0 2
0
0
55 60 65 70 75 80 85
0. (%)
Fig. B15. Deformation absorption capacity oro and tearing modulus ToO. 2 for stable crack
growth as a function of crystallinity a (a/W = 0.45)
Stable crack growth under a high impact load appears at a crystallinity of ap-
proximately 85 %, resulting in a high resistance against stable crack growth (tear-
ing modulus) (Fig. B 15). The strength of the crystals is of importance for the re-
sistance against stable crack propagation. The strength of the fibrils increases with
increasing crystallinity; this leads to an enhanced toughness for stable crack
growth. The deformation absorption capacity is not influenced by changes of the
crystallinity in the range from 55 % to 75 % (Fig. B 15).
5.3 Effect of Tie Molecules and Lamellar Structure
The tie molecule density is decisive, in terms of mechanisms on the nanometre

scale, for the toughness behaviour of polyethylene. The toughness described by the
J-integral increases with increasing tie molecule density, especially above a tie
molecule density of 0.9 nm-I, because the stretched polymer chains can dissipate
more energy under mechanical load if a higher number of tie molecules is avail-
able (Fig. B16) [19-21]. It can also be seen from Fig. B16 that the critical crack-
opening displacement increases continuously with increasing tie molecule density.
It can be concluded that the toughness behaviour is stress-determined.
The technical crack initiation values for resistance against stable crack initiation
were less influenced by structural changes (Figs. B 10, B 14, and B 17). This could
be caused by' the problem in defming the actual crack initiation point from the
crack resistance curves.
150 450
8/ A - 300 ~
"6
20
~ 1~) 00-
15
E
o 0.4 0.8 1.2
0 .E
10 ~
T{M 2) (nm-')
....,
t;"
'1'-
~-- 5
0
0.0 0.4 0.8 1.2
T(~) (nm-')
Fig. B16. Effect of tie molecule density 1tM2) on fracture mechanics parameters for unstable
crack growth (20 % SG, a/W = 0.2)
In Fig. B 17 it can be seen that the energy absorption capacity increases with in-
creasing tie molecule density, especially above a tie molecule density of 1.0 nm-I.
The disentanglement rate decreases owing to the stretched polymer fibrils, which
can dissipate more energy at a higher tie molecule density if simultaneously the
strength of the crystals is high enough.
4 60
I
rW-
E
-
E
~
N
c:i
2 •
T
I
40
E
.E
Z
........
...,
:- ....,
I-
0 20
~--~--~--~--L---~--L-~O
0.0 0.4 0.8 1.2
T(M 2) (nm-1)
Fig. B17. Dependence of technical crack initiation parameter J02 and energy absorption ca-
pacity JTJon the tie molecule density T(M2) (a/W= 0.45)
150 8
E
.f 100 6
.....
0
• ••
N
• •
c:i
........
r-
<0
"' • 4
I- "'
50
0
0
2
0
0
20 25 30 35
Lp (nm)
Fig. B18. Effect of long period Lp on deformation absorption capacity oro and tearing
modulus TP at stable crack growth (a/W = 0.45)
From the examinations performed as a function of the crystallinity, it follows

that the tearing modulus is suitable for the characterization of toughness behaviour
for stable crack growth. This is also reflected in the increase of the resistance
against stable crack growth, determined from the crack-opening displacement as a
loading parameter, with increasing long period (Fig. B 18). Here, one must keep in
mind that the tearing modulus is a geometry-dependent fracture mechanics parame-
ter. As in the case of the dependence on the crystallinity, the deformation absorption
capacity is not influenced by the structural parameter the long period (Fig. B 18).
5.4 Crack Initiation and Propagation in High-Density Polyethylene
The fracture surfaces were investigated using a scanning electron microscope with
the aim of fmding a morphology-toughness correlation and to characterize the
processes of crack initiation and propagation. The results of these examinations
allow the crack process in high-density polyethylene to be described. The fIrst part
of the crack process appears to consist of blunting, void formation, void growth
and formation of fIbrils. An important conclusion is that the crack front moves
forward during the blunting process (Figs. B19c, d). Depending on the material,
for example at higher molecular weight, shear deformation appears at the boundary
of the craze zone. As a result, the amount of shear deformation can exceed the
craze thickness.
e. craze formation
a. notching
a
-----~
b. relaxation
f. crack initiation by
rupture of fibrils
c. blunting
d. forming of voids
a BS
Fig. B19. Scheme of the crack process in high-density polyethylene (a - initial crack length,
stable crack growth, aBS - fracture mirror length, aSMZ- smooth zone)
/:;.a -
In the next stage, the crack initiation is characterized by failure of fIbrils

(Fig. B 19t). The stable crack growth is influenced by crack opening at the crack
tip and stretching of the fIbrils. Depending on the intrinsic strength of the fIbrils,
which is determined by the crystallinity, and the craze length, which is influenced
by the molecular weight, the failure of the fIbrils characterizes a different type of
stable crack growth.
Finally, a critical crack-tip opening and a specifIc amount of failure of the fI-
brils lead to unstable crack growth (Fig. B 199). The materials with lower Crystalli-
Fig. B20. Fracture surfaces for stable crack growth of PE-HD with a low crystallinity (a-
M36, !!.a = 0.15 rom) and a high crystallinity (b - M60, !!.a = 0.2 rom) and also of
PE-HO with a low molecular weight (c - 0360, !!.a = 0.4 rom) and a high molecular
weight (d - 0550, !!.a = 0.55 rom) (A, initial crack length; B, stable crack growth;
C, unstable crack growth)
nity and a lower molecular weight show a larger amount of crazing at the fracture
surface (Figs. B20 a, c). However, a larger amount of shear flow can be detected
for the materials with a higher crystallinity of about 85 % and a higher molecular
weight of approximately 550 kg/mol (Figs. B20 b, d).
6 Conclusions
The toughness behaviour of semicrystalline polymers can be described by

geometry-independent fracture mechanics parameters. Furthermore, using fracture
mechanics investigations, a correlation between the structure and the toughness
properties, as well as the mechanisms of crack initiation and crack propagation, is
possible. The use of side grooves is recommended for the overall characterization
of toughness behaviour for unstable crack growth, for all the high-density poly-
ethylenes investigated.
In conclusion, it can be said in summary that for optimization of the toughness
properties by appropriate choice of processing conditions and material selection, it
is necessary to consider the influence of molecular weight and also the influence of
crystallinity on the toughness behaviour. As a result of the tough material behav-
iour of high-density polyethylenes that results in stable crack growth, the estima-
tion of the resistance against stable crack growth is of great importance in addition
to the consideration of the resistance against unstable crack growth. Furthermore,
the structure of the semicrystalline polymer has a greater influence on the crack
propagation behaviour than on the crack initiation behaviour.
Acknowledgements
The authors gratefully acknowledge the support of the German Research Founda-
tion (DFG) within the context of the graduate course 'Polymer Sciences' (Hetero-
geneous Polymers) at the Martin-Luther-University of Halle-Wittenberg. Our
thanks also go to Dr. F. Ramsteiner, Dr. G. Pfrrrmann and Dr. Haberkorn ofBASF
AG, Ludwigshafen, Germany for providing the materials and for the successful
cooperation.
References
I. Seidler S., Koch T., Grellmann W., Kotter I. (2001): Crack initiation behaviour of semi-
crystalline thermoplastics. Theor. Appl. Fract. Mech., submitted
4. Egan B. J., Delatycki O. (1995): The morphology, chain structure and fracture behaviour of
high-density polyethylene. Part 1: fracture at a constant rate of deflection. J. Mat. Sci. 30:
3307-3318
5. Huang Y.-L., Brown N. (1991): Dependence of slow crack growth in polyethylene on butyl
branch density: morphology and theory. 1. Polym. Sci., Polym. Phys. 29: 129-137
6. Brown N., Zhou Z. (1995): Relationship between the structure and slow crack growth in
blends of high-density polyethylene and model copolymers. Macromolecules 28:
1807-1811
7. Zhou Z., Brown N., Crist B. (1995): Slow crack growth in blends of HDPE and model
copolymers. J. Polym. Sci., Polym. Phys. 33: 1047-1051
8. Barry D. B., Delatycki O. (1992): The effect of molecular-structure and polymer morphology
on the fracture-resistance of high-density polyethylene. Polymer 33: 1261-1265
9. Lu X., McGhie A., Brown, N. (1993): The dependence of slow crack growth in a linear
polyethylene on test temperature and morphology. J. Polym. Sci., Polym. Phys. 31: 767-772
10. Friedrich K. (1983): Crazes and shear bands in semi-crystalline thennoplastics. Adv. Polym.
Sci. 52/53: 226--274
II. Brown N., Lu X., Huang Y.-L., Qiau R. (1991): Slow crack growth in polyethylene - a
review. Macromol. Chern., Macromol. Symp. 41: 55--67
12. Dekker J. C., Bakker A. (1994): Dynamic fracture testing of polyethylene. In: Schwalbe K.-
H., Berger C. (Eds.) Structural Integrity: Experiments - Models - Applications. 10th Bien-
nial European Conference on Fracture, Berlin, Gennany, September 20-23, Engineering
Materials Advisory Services Ltd, Cradley Heath, Warley, West Midlands Vol. 1: 571-580
13. Quirk R. P., Alsamarraie M. A. A. (1989): Physical constants ofpoly(ethylene). In: Bandrup
J., Immergut E. H. (Ed.) Polymer Handbook. 3rd Ed. Wiley Interscience, New York: V/15-
V/26
14. Illers V. K. H., Hendus H. (1968): Melting point and crystal dimension of polyethylene
crystallized from the melt and the solution. Macromol. Macromol. Chern. 113: 1-22
15. Strebel J. J., Moet A. (1995): The effects of annealing on fatigue crack propagation in poly-
ethylene. J. Polym. Sci., Polym. Phys. 33: 1969-1984
16. Brown N., Ward I. M. (1983): The influence of morphology and molecular weight on ductile-
brittle transitions in linear polyethylene. J. Mat. Sci. 18: 1405-1420
17. Yeh J. T., Runt J. (1991): Fatigue crack propagation in high-density polyethylene. J. Polym.
Sci., Polym. Phys. 29: 371-388
18. Standard Draft ESIS TC4 (1995): A Testing Protocol for Conducting J-Crack Growth
19. Beerbaum H., Grellmann W., Seidler S. (1999): Microstructure and toughness behaviour of
high-density polyethylene under impact load. In: Proceedings of Technomer '99, Chemnitz,
Gennany, November 11-13: P42/1-8
20. Beerbaum H., Grellmann W. (1999): The influence of morphology and structure on the
mers, Composites and Adhesives. ESIS Publication 27, Elsevier Science Ltd, Oxford:
163-174
21. Beerbaum H., Grellmann W., Seidler S. (2000): The influence of molecular weight, crystal-
linity and tie molecule density on the toughness behaviour of high-density polyethylene. In:
Proceedings of Defonnation, Yield and Fracture of Polymers, Cambridge, UK, April 10-13:
548-551
B 1.3 Toughness and Relaxation Behaviour
of PMMA, PS and PC
W. Grellmann and R. Lach, Merseburg
1 Introduction and Motivation
Innovative polymer applications which use the specific properties of polymeric

materials for selection necessitate, besides the improvement of strength and stiff-
ness, an optimization of toughness behaviour and a well-grounded knowledge of
the strength- and toughness-determining mechanisms [1].
It is well known that viscoelastic behaviour is a characteristic of polymeric ma-
terials [2]. This causes a strong dependence of the mechanical properties on tem-
perature and loading speed. The reason for this are polymer-specific relaxation
phenomena, which influence the deformation behaviour of polymeric materials
from small strain values to the beginning of plastic flow. Therefore, these relaxa-
tion phenomena are decisive factors for fracture processes, too.
From the two points we can deduce that quantitative and qualitative recording
of relationships between structure, relaxation processes and toughness is of general
interest from the point of view of engineering science, and this is the motivation
for the present study. The use of simple materials systems is advantageous in at-
taining this aim. Therefore, the investigations described here were restricted to
amorphous, homopolymeric materials. Hence, polycarbonate (PC) was the main
material selected for the fracture mechanics investigations, because of its suffi-
ciently separated dispersion ranges concerning the temperature. Poly(methyl meth-
acrylate) (PMMA) and polystyrene CPS) were used for comparative investigations.
2 Basic Concepts of Fracture Mechanics

and Experimental Methods
For the quantitative evaluation of the fracture behaviour of brittle materials, i.e.
materials with elastic or nearly elastic behaviour and predominantly unstable crack
propagation, the concepts of linear elastic fracture mechanics (LEFM) [3] have
been proved to be successful. However, if the size of the plastic zone is no longer
small in comparison with the initial crack length and the specimen geometry, the
concepts of elastic-plastic fracture mechanics (EPFM) must be applied. These
include the crack-tip-opening displacement (CTOD) concept, the deformation-
determined crack field parameter the crack-opening displacement, and the J-
integral. The parameter used in the present work to describe the stress and strain
field close to the crack tip is the energy-determined J-integral. The validity of the
J-integral concept is guaranteed if large flow areas in the crack tip region are

182 W. Grellmann, R. Lach
formed, independent of the specific material behaviour. Therefore, this concept is

well-suited to the fracture mechanics evaluation of polymeric materials.
Starting from its defmition [4], the J-integral can be interpreted energetically:
the J-integral describes the change in the potential energy of the cracked body in
relation to the change in crack extension and stands for the effective specific inter-
face energy of the growing crack. From this energetic definition, different J-
integral evaluation methods have been derived [5,6], e.g. the method of Sumpter
and Turner (ST) [7], which is one of the most widespread procedures for plastics
[8-10] and was also used in this study. In this connection, the analysis of recorded
load-deflection diagrams forms the experimental basis. The general deformation
energy AG up to the beginning of unstable crack propagation, which is equal to the
potential energy, is separated into an elastic and a plastic part, Ael and Api. The
elastic and plastic geometrical functions TIel and Tlph respectively, contain informa-
tion about the specimen as well as the crack geometry. The actual crack propaga-
tion is considered in terms of the effective crack length aeff (pp. 71-86).
Within the scope of the EPFM, the CTOD concept, which starts from the crack
model of Dugdale [11], has been well established. An initially sharp notch blunts
by forming a stretch zone; this is produced in polymeric materials, in most cases,
by shear flow or coalescence of cavities. The profile of the stretch zone can be
measured by means of a microscope after breaking the specimen. In the present
work, the size of the stretch zone in the direction of crack propagation was quanti-
fied as the stretch zone width (SZW), and the size perpendicular to that direction
as the stretch zone height (SZH). Crack-opening displacement (J) values were
calculated using the plastic-hinge model (pp. 71-86, [12-14]), because visual
observation of 8 during crack propagation is not possible for impact-like loading
conditions.
For the determination of fracture mechanics parameters related to resistance
against the beginning of unstable crack propagation, an instrumented Charpy im-
pact tester was used (pp. 71-86, [15]). For the measurements, single-edge-notched
bend (SENB) specimens (80 x lOx 4 mm3) were used. The initial crack consisted
ofa 1 mm deep V-notch and a 1 mm deep razor-blade notch. Because of the small
notch tip radius of 0.2 /lm, a nearly ideally sharp notch could be observed after
relaxation of the notch flanks. The pendulum hammer speed was set at 1.5 mls. If
large amounts of stable crack growth can be found, the determination of fracture
mechanics parameters related to resistance against stable crack initiation and
propagation should be included in the evaluation of toughness behaviour. In the
instrumented impact tester used in the present work, an auxiliary device for limit-
ing deflection values - the so-called stop block - makes it possible to obtain different
stable crack growth sizes by a defined variation of the specimen deflection [13,16-
19]. By this means, an experimental basis for the construction of crack resistance (R)
curves under impact loading was created. The R-curve represents a functional des-
cription of a loading parameter like J or 8 versus the amount of stable crack growth
!:la. Applying specific R-curve procedures, material parameters related to resistance
against stable crack initiation and propagation can be calculated (pp. 141-150).
Toughness and Relaxation Behaviour ofPMMA, PS and PC 183
3 Correlations between Fracture Mechanics

Material Parameters and the Secondary (P) Relaxation
The toughness behaviour at higher temperatures, for example at room temperature,

is strongly influenced by relaxation processes which can be observed at lower
temperatures. Therefore, fracture mechanics investigations were performed over a
wide temperature range from -200 DC to 100 DC. Figure B21 shows toughness
values for resistance against the beginning of unstable crack growth J'd ST versus
temperature. The J'd ST values for PC at T> 50 DC were determined on the basis of
J-TJ stability assessment diagrams, as described in A 3.4 (pp. 141-150).
40
20
10
C?J! ~ PC=
E
E
....... 7'
/"'. ~ 1
6- 4 0 ri' '~T
1--0 v
IV!:~
Cf)-
""")
2 b ln~ r,..
p ~ V V"v
h
PS
,
PMMA
0.4
-200 -100 0 100 200
TeG)
Fig. B21. Critical Jvalues Jld ST as a function of temperature for polycarbonate (conventionally
determined, 0; approximately determined, 0 (pp. 141-150), polystyrene (~) and
poly(methyl methacrylate) (V)
From Fig. B21, one can see that the temperature dependence of ~dST contains
various peaks. In the following, it is emphasized that both of the local maxima, at
-60 °C for PC and at 50°C for PMMA, correlate with the secondary (~) relaxa-
tion. To verify this correlation, an estimation of the frequencies corresponding to
the loading conditions of the instrumented impact test is necessary. On the one
hand, these so-called fracture frequencies can be calculated indirectly by means of
the frequency-dependent shift of the Young's modulus (Fig. B22). On the other
hand, they can also be estimated directly from the time to fracture or the time to
the load maximum; this method was used in this study. To interpret the toughness
increase, the results of the dynamic mechanical analysis were considered in the
form of the mechanical loss factor tan 0, plotted in Fig. B23. Lines for the maxima
of tan t5, i.e. the maxima of the ~ and main (a) relaxation, and the fracture fre-
quency, are plotted together in an Arrhenius-diagram in Fig. B24. Here, the inter-
sections of the ~ relaxation lines and the fracture frequency verify the correlation
between the local toughness increase and the ~ relaxation. For example, the inter-
section for PC was found at a temperature of -60 °C and at a frequency of 300 Hz.
7
6 PMMAv v
ICIT
-IE
5
4
(J
........
'0 3 PS
UJ
2 PC
0-250 -200 -150 -100 -50 0 50 100 150 200

T (0C)
Fig. B22. Dynamic Young's modulus as function of temperature for polycarbonate (D), poly-
styrene (L'1) and poly(methyl methacrylate) ('\7)
5
1 Hz
0.5
-
00
c:
PMMA, ..l3--- __
C'O
,
0.05 ./,'
-,
0.005 '--'-_"----'-_.1..---1-_"---'---"1...--1----'
-250 -200 -150 -100 -50 0 50 100 150 200
T (0C)
Fig. B23. Mechanical loss factor tan 8 versus temperature for polycarbonate, polystyrene and
poly(methyl methacrylate) (a, a relaxation; /3, /3 relaxation)
10 3
10 2 Fracture Frequency
N
;:s 10 1
;or
10°
oPMMA
oPS
a PC
10.1
6 10 14
a 1IT (1/1000 K)
Fig. B24. Arrhenius diagram: lines for a relaxation (a) and ~ relaxation (13), and fracture fre-
quency for polycarbonate (D), polystyrene (~) and poly(methyl methacrylate) ('\7)
The toughness behaviour is connected with microscopic deformation processes

which result in various fracture surface phenomena. These were characterized by
optical and scanning electron microscope investigation methods. Figure B25
shows a generalized schematic illustration of the fracture surface of PC at room
temperature. In the direction of crack propagation, the notch, consisting of a V-
shaped and a razor-blade part, is followed by the fracture mirror length, corre-
sponding to stable crack growth, which is separated from the initiation area of
unstable crack growth and the transitional areas I and 2 by the stretch zone. After
the transitional areas, the actual area of unstable crack growth and, far from the
notch, a compression zone are observed. Thus, the chosen fracture surface phe-
nomena can be quantified and related to the material behaviour. This is illustrated
by an example of the fracture mirror length in PC (Fig. B26). The temperature
dependence of the fracture mirror length is qualitatively comparable to that of the J
values. The local maximum at -40 °C, related to the ~-stimulated toughness in-
crease, is a direct indicator of the fact that molecular relaxation processes play an
essential role in fracture processes because stable crack propagation indicates non-
linear viscoelastic plastic behaviour.
Furthermore, a relationship between the value of the mechanical loss factor, the
toughness-determining deformation mechanisms and the value of the f3-stimulated
toughness increase could be found. Because of the good separation of the a and
f3 dispersion areas and the related dominant occurrence of shear flow, PC shows a
relatively large local toughness increase. In contrast, the f3 relaxation of PS is not
very distinct and is not separated from the a dispersion area. This is the reason
why only crazing and no local toughness increase can be detected. On the evidence
of the coexistence of crazes and shear bands, the behaviour of PMMA is the aver-
age of that ofpC and PS.
Summarizing, the f3 relaxation causes a significant toughness increase by
stimulation of shear deformation mechanisms because shear deformation influ-
ences a much greater volume than does crazing.
1 Cut V-notch
2 Sharp notch by razor blade
3 Stable crack growth
4 Stretch zone
5 Initiation area of unstable
crack growth
7 6 Transitional area 2
6 7 Transitional area 1
,-~-5
8 Actual unstable crack growth
~...,.s;;;imm:m4<4 9 Compression zone
3
2
1
Fig. B25. Schematic illustration of a fracture surface for polycarbonate at room temperature
(top); optical micrographs of fracture surfaces at selected temperatures for polycar-
bonate (bottom)
20~--~--L-~--~~--~--~~~
-250 -200 -150 -100 -50 0 50 100 150 200
TrC)
Fig. B26. Fracture mirror length ass in PC as a function of temperature
4 Correlations between Fracture Mechanics

Material Parameters and the Main (ex.) Relaxation
(Dynamic Glass Transition)
From their defmition, the crack-opening displacement values bid make it possible
to quantify the predominantly plastic deformation close to the notch (pp. 71-86,
[11]).
Because of the dominantly athermal fracture behaviour of PS and PMMA, on
the one hand, only small crack-opening displacement values can be detected for
these two materials. For PC, on the other hand, a very marked maximum of 0 as a
function of temperature was observed. The difference is made clear in Fig. B27.
Here, values of bid at temperatures higher than 50°C were estimated by using
~To stability assessment diagrams (pp. 141-150). This maximum can be ex-
plained by a shear flow process corresponding to the dynamic glass transition,
which is caused by the temperature dependence of the dynamic yield stress and by
the ratio of the nominal stress in the periphery of the specimen to the dynamic
yield stress [15].
Cooperative molecular motion processes like ex relaxation, i.e. the dynamic
glass transition, can be described by means of the Williams-Landel-Ferry (WLF)
equation [20]:
In [~ (T,P)] [To -Too(p)] = In[~ (T,P)] [To - Too (p)] , (Bll)
where T is the temperature, If/' is the frequency and p is the pressure.
300~--------------------------~
DO PC
250 PS
It.
v PMMA
. . . . 200
E
E
b 150
"0
us 100
50
oL-~~~~::~~
-250 -200 -150 -100 -50 0 50 100 150
T (0C)
Fig. B27. Critical crack-opening displacement ~d versus temperature for polycarbonate (con-
ventionally determined: 0 , approximately determined: 0 (pp. 141-150)), polysty-
rene (d) and poly(methyl methacrylate) (V)
By choosing an arbitrary reference point (lJIo, To), the glass transition can be de-
scribed mathematically by a suitable hyperbola. Here, the lower marginal tem-
perature Too is the so-called Vogel temperature. Q is the high-frequency asymptote,
which means the relaxation frequency of local parts of chains (for the glass transi-
tion of polymers, Qr:;:; 1013 Hz [21]). The WLF equation can also be formulated for
thermodynamic quantities other than temperature, e.g. pressure. Analogously to
(B 11), Poo is the lower limit pressure (Vogel pressure). For isochronous situations
(IJI = const.), the validity of the WLF equation in both its pressure and its tem-
perature formulation yields a hyperbolic pressure versus temperature function [2].
This assumption can be verified in the present case if the dynamic yield stress CTd is
taken as a negative pressure (Fig. B28).
10' 0
109
rn 10 8 PS
~ PMMA
(? 107 PC
106
e
105
0
0 PC
-50
rn
a...
6 -100
~
-150 PMMA 0
0
Fig. B28. Temperature dependence of the dynamic modulus GO corresponding to the characteris-
tic frequencies of the Charpy impact test (top) and dynamic yield stress O"d as a function
of temperature for polycarbonate (0), polystyrene (0) and poly(methyl methacrylate)
(0) (bottom); n and Bare the pressure and temperature asymptotes, respectively
Toughness and Relaxation Behaviour of PMMA, PS and PC 189
The curvatures of the resulting hyperbolae are dependent on frequency, which

can be replaced by the inverse fracture time or the local strain rate. The extrapo-
lated temperature values for p ~ 0 obtained from these hyperbolae are nearly
equal to the glass transition temperature.
Up to now, only material parameters related to resistance against the beginning
of unstable crack propagation have been discussed. For an interpretation of the
crack initiation and propagation behaviour, crack resistance curves for PC were
recorded at selected temperatures. The technical crack initiation values JO.2 and ~.2
at an amount of stable crack growth Aa = 0.2 mm (Fig. B29) were ascertained
because a determination of crack resistance values at the point of physical initia-
tion is not possible.
10 200
8 160
E 6 120 E
E
--
~
E C')
b
.,....
........
N
ci 4 80 N
"""J ci
c.o
2 40
0 0
0 20 40 60 80 100
T (0C)
Fig. B29. Technical crack initiation values JO.2 (0) and &.2 (6) (determined at an amount of
stable crack growth 6a = 0.2 rum) as a function of temperature for polycarbonate
The strong increase of JO.2 at 60°C indicates a change in the crack-blunting

process.
Scanning electron microscope investigations of the stretch zone were performed
to clarify the temperature dependence of the JO.2 values. The crack initiation and
propagation mechanisms of amorphous polymeric materials such as PC and some
other semicrystalline polymers can be schematically described as follows [13].
First, the sharp initial notch blunts by bulging of the notch tip and a stretch zone is
formed. After crack initiation by repeated sharpening of the crack tip, the crack
grows by translation of the whole crack propagation front [24].
As a consequence, a stretch zone can be found at the far end of the fracture mir-
ror length from the notch, as can be verified for PC in Fig. B30.
Fig. B30. SEM micrographs of stretch zone height (SZH, left) and stretch zone width (SZW,
right) for polycarbonate at 40°C (the sizes of the SZH and SZW are marked using
arrows)
60 - - - SZW
- 0 - SZH
-E
2: 40
:r:
~
~ 20 ,
I
I
,,
(f)
DI
6 ",'
___ g ____ a----~D
o~~~±=~~~~~~~
-250 -200 -150 -100 -50 0 50 100 150
T (0C)
Fig. B31. Stretch zone height (SZH; d) and stretch zone width (SZW; D) as a function of
temperature for polycarbonate
The temperature dependence of the stretch zone height (Fig. B31) is qualita-
tively comparable to that of the crack-opening displacement because of its geomet-
rical interpretation. The increasing SZHlSZW ratio at higher temperatures reflects

the strong increase of JO.2 at 60°C.
Summarizing this and earlier studies [15,18,19,23] of the application limits of
polymeric materials, it should be noted that material-specific geometry-indepen-
dent results in the experimentally difficult to measure transition area between 'un-
stable' and 'stable' are only possible with knowledge of the relationships between
fracture mechanics parameters and molecular relaxation processes, as well as by
studying selected fracture surface phenomena.
Acknowledgement
Bayer AG, Leverkusen, Germany, - especially Dr. D. J. Dijkstra - is gratefully

acknowledged for providing the PC specimens. Our thanks go to Prof. E. Donth,
Martin-Luther-University of Halle-Wittenberg for critical remarks and discussions
about this problem. We also thank the German Research Foundation (DFG) for
fmancial support within the context of the graduate course 'Polymer Sciences
(Heterogeneous Polymers)'.
References
I. Kausch H.-H. (1987): Polymer Fracture. Springer, Berlin, Heidelberg

2. Donth E. (1992): Relaxation and Thennodynamics in Polymers. Akademie-Verlag, Berlin
3. Irwin G. R. (1957): Analysis of stresses and strains near the end of a crack traversing a
plate. J. Appl. Mech. 24: 361-364
4. Rice J. R. (1968): A path independent integral and the approximate analysis of strain concen-
tration by notches and cracks. J. Appl. Mech. 35: 379-386
ceedings of the lst Conference on Mechanics, Prague, Czech, June 29-July 3, Vol. 5:
129-133
6. Grellmann W., Seidler S. (1999): Possibilities and limits of standards and drafts for J-curve
detennination on polymers. In: Winkler T., Schubert A. (Eds.) Material Mechanics - Frac-
ture Mechanics - Micro Mechanics. DDP Goldenbogen, Dresden: 336-341
8. Bernal C., Rink M. Frontini P. M. (1999): Load separation principle in detennination of J-R
curve for ductile polymers: suitability of different material defonnation functions used in the
nonnalization method. Macromol. Symp. 147: 235-248
10. Cousin-Comet L., Abdalaziz M. N., Boufeau N., Mesmacque G., Cazeneuve C. (1999):
Experimental study and finite element analysis of plasticised silicon acrylic films fracture.
In: Williams G., Pavan A. (Eds.) Fracture of Polymers, Composites and Adhesives. ESIS
Publication 27, Elsevier Science Ltd., Oxford: 201-212
II. Wells A. A. (1961): In: Proceedings of the Crack Propagation Symposium, Cranfield,
Vol. I, Paper 84, Cranfield
12. Elliot D., May M. J. (1969): BlSRA Open Report MG/C/47/69
namic fracture toughness of ABS materials. J. Appl. Polym. Sci. 75:1605-1614
14. Kobayashi T. (1984): On the infonnation about fracture characteristics obtained from in-
strumented impact test of A533 steel for reactor pressure vessel. Engng. Fract. Mech. 19:
67-79
15. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour of poly (methyl methac-
rylate), polystyrene and polycarbonate. Appl. Macromol. Chern. Phys. 253: 27-49
16. Kobayashi T. (1983): Measurement of dynamic fracture toughness J Id by instrumented
charpy test. lnt. J. Fract. 23: R 105-109
17. Grellmann W., Seidler S., Schierjott U., Rutke B. (1988): Anordnung zur Bestimmung des
JTrgesteuerten RiBwachstums bei schlagartiger Beanspruchung. Gennan Patent DD 275 113
18. Grellmann W., Lach R. (1996): Effect of temperature on toughness, fracture surface mor-
phology and molecular relaxation behaviour of polycarbonate. Appl. Macromol. Chern.
Phys.237: 191-208
ness characterization of high impact thennoplastics. Polym. Test. 14: 453-469
20. Williams M. L., Landel R. F., Ferry J. (1955): J. of American Chemical Society 77: 3701ff
21. Lach R., Grellmann W., Schroter K., Donth E. (1999): Temperature dependence of dynamic
yield stress in amorphous polymers as an indicator for the dynamic glass transition at nega-
tive pressure. Polymer 40: 1481-1485
22. Seidler S., Koch T., Kotter I., Grellmann W. (2000): Crack initiation behaviour of PP-
materials. In: Proceedings 13th European Conference on Fracture (ECF 13), San Sebastian,
Spain, September 6-9, CD-ROM Polymer and Composites No. 17: 1-8
23. Grellmann W., Seidler S., Jung K., Kotter I. (2001): Crack-resistance behavior ofpolypro-
pylene copolymers. J. Applied Polym. Sci. 79: 2317-2325
8 1.4 Crazing in Amorphous Polymers
- Formation of Fibrillated Crazes
Near the Glass Transition Temperature
G. H. Michler, Merseburg
1 Introduction
Crazing is a widespread phenomenon in polymeric materials. Crazes are zones of

highly plastically deformed and highly oriented polymer material, consisting
mainly of fibrils along the direction of applied stress interconnected by cross-tie
fibrils. The structure of the crazes has been investigated mostly by transmission
electron microscopy [1-3] and by X-ray [4] and electron [5] diffraction. The
shapes of crazes have been examined by optical methods [6,7]. Various publica-
tions review the structure of craze zones in different polymers and the models of
craze formation [3,8-10].
Although crazes are typical of amorphous, glassy polymers, craze-like features
have also been observed in many semicrystalline polymers (e.g. in polyethylene,
isotactic polypropylene, isotactic polystyrene, polyamide and poly(ethylene ter-
ephthalate)) [11]. Crazing is well known as the energy-absorbing mechanism in
rubber-modified high-impact polymers, e.g. high-impact polystyrene (HIPS),
acrylonitrile--butadiene-styrene polymers (ABS) and high-impact poly(methyl
methacrylate) [3,12]. In the literature, crazes are often referred to microcracks,
fibrillar deformation zones (DZs), yield zones, true crazes or simply crazes.
After a short review of the formation and typical features of crazes in amor-
phous polymers in the following section, changes in the formation and internal
structure of crazes at a higher temperature near (and below) the glass transition
temperature are discussed in more detail in the subsequent sections.
2 The Phenomenon of Crazing
In amorphous polymers, crazes are very long, narrow plastic-deformation zones,

sharply delimited from the surroundings (see reviews in [3,13]). The polymeric
material is plastically deformed to a high degree in crazes (e.g. in polystyrene
(PS), usually up to - 150-250 %). The fibrils in the crazes lie in the direction of
tension, and the long dimensions of the crazes are perpendicular to the tension.
One method of determining the centre-to-centre distance of the fibrils is analysis
of electron micrographs by laser light diffraction or Fourier transformation. By
this method, centre-to-centre distances of 10-50 run have been determined [14].
The structure first visible ahead of the craze tip, i.e. in the transformation or proc-
essing zone of the craze, shows bright spots like 'domains', with diameters be-

194 G. H. Michler
tween 20 and 50 run. All domains are arranged in a narrow band with a thickness
of a few tens of nanometres and a length of, usually, some micrometres. These
zones ahead of the fully developed crazes are called 'pre-crazes' [3]. They appear
to be relatively stable zones and are important to a better understanding of the
formation mechanism of crazes (see Fig~ B32).
Fig. B32. Structure of a craze in PS (deformed semi-thin section in HVEM): (a) pre-craze
domains, (b) fully developed fibrillated craze, (c) laser light diffraction pattern of
fibrillar arrangement
The initiation and formation of crazes can be discussed in connection with the
entanglement model [10]. In amorphous polymers the macromolecules are as-
sumed to form topological connections, the entanglements, responsible for the
strength and deformability of the polymeric material. Entanglements are formed
only if the length or molecular weight of the macromolecules is high enough. The
critical molecular weight for entanglement Me is - 17 500 for PS; if the molecular
weight is lower than double the molecular weight for entanglement, the material is
brittle [IS]. Adjacent entanglements can be connected to form a network (Fig.
B33a). The average diameter of the meshes of this network is - IS run in PS (with
an entanglement distance of - 10 run). The material between the entanglements (i.e.
in the meshes of the network) is assumed to be mechanically weaker. During
stretching, the meshes are preferentially deformed, yielding the domain-like struc-
Crazing in Amorphous Polymers 195
10 run
c
Fig. B33. Schematic illustration of the connection between an entanglement network, a pre-
craze and a fibrillated craze (from [3]): (a) a network is formed by entanglements
and contains weaker meshes; in a stress concentration zone the meshes are locally
plastically stretched; (b) stretched meshes form a pre-craze; (c) a fibrillated craze
appears by void formation and stretching of the network
ture of the pre-craze (Fig. B33 b). There is a good correlation between the diame-
ter of the undefonned meshes (- 15 run) and the minimum diameter of the de-
fonned domains of the pre-craze (- 20 run). After stronger defonnation, the do-
mains rupture, forming micro-voids. The entangled material between these micro-
voids is plastically stretched and transfonned into the fibrils of a developed craze
196 G. H. Michler
(Fig. B33c). During this process, the micro-voids partially coalesce and are trans-
formed from a closed-cell structure into an open-cell structure [3]. Therefore, the
long period of the fibrils correlates with the distance between meshes of the en-
tanglement network (20-30 nm). This crazing mechanism, based on such a 'two-
stage entanglement network model' or 'domain model' [16], probably describes
the details of craze formation better than the model of 'meniscus instability' usu-
ally used. The influence of molecular weight on crazing behaviour can be demon-
strated in a PS blend containing 50 % PS with a molecular weight of - 80 000
(above the critical molecular weight 2Me) and 50 % brominated PS with an Mw of
20000 (below 2Me) [17]. In this blend, the separation of these components into
regions of size 0.1-1 Ilm can be seen in the transmission electron microscope
(TEM) very clearly, since the low-molecular-weight PS appears dark owing to the
bromine content. A deformation zone propagating through this material shows
fibrillated crazes only in the high-molecular-weight part. A crack without any
fibrillation is visible in the low-molecular-weight part because of the missing
intermolecular connections and strength.
A fibrillar structure of crazes is typical of PS. Another glassy polymer is a sty-
rene-acrylonitrile (SAN) copolymer with a PS content of - 74 %. In this material
the dominant deformation structures are homogeneously deformed zones, but in
many places they coexist with fibrillated crazes. Depending on the loading condi-
tions, for example loading temperature, the deformation character can be shifted
from one type to another [2,3].
In [18], the results of microindentation experiments on crazes in the micron
range are reported, yielding additional information about the mechanical proper-
ties of the craze fibrils. The micro-hardness of the craze zones was determined,
and the elastic moduli of the crazed material were calculated from the elastic re-
covery curves of indentations. The specimens were miniaturized compact tension
(CT) specimens (10 mm x 8 mm x 4 mm) with a crack tip craze in front ofa sharp
notch, as shown in Fig. B34. The crazes were created during slow crack propaga-
tion under fatigue loading at room temperature. The micro-hardness was deter-
mined using a dynamic ultramicro-hardness tester (Shimadzu DUH 202) with a
Vickers square-pyramid diamond indenter, which had an angle a between non-
adjacent faces of the pyramid of 136°. The experiments were performed at room
temperature. During microindentation, the load was applied up to a maximum load
of 10 mN at a constant loading rate of 1.4 mN S-I. The penetration depth was less
than 1.5 Ilm in all experiments. In all cases the maximum load was applied for 6 s
(creep), and then the specimen was unloaded at the same rate. Indentations were
performed on a straight line along the direction of crack propagation (Fig. B34).
Figure B35 shows the experimental micro-hardness data from various indenta-
tions on a PS specimen. The horizontal solid line in the upper part represents the
average micro-hardness value of the bulk polymer and the dotted lines represent the
zone of scatter of the data. The experimental results show three distinct regions of
behaviour: crack zone, craze region and bulk. The hardness values measured along
the crack zone are lower than the micro-hardness of the bulk. In this zone the
indentations were made close to the notch tip. The indenter tip was placed into the
crack and the outer part of the indenter pressed into the bulk polymer. Assuming a
Fig. B34. Schematic view of the compact tension specimen illustrating the following posi-
tions: 0, notch region; I, crack region; 2, craze region; indentations (3) were made
along the straight line x (from [18])
200
190
'2'
~ 180
'-'
J170 Crack - --B~lk

160 • Micro-
• Hardness
150 '--r--~~.'--~~~~--\-----r-~--'
-1.2 -0.8 -0.4 0.0 0.4
x (mm)
Bulk
Fig. B35. Micro-hardness values measured along the crack propagation direction in PS; the
cracked, crazed and bulk zones are shown (from [18])
two-phase system consisting of the crack and the bulk material, a lower hardness
value than that of the bulk must be measured. In the craze zone, the micro-
hardness increases rapidly (see Fig. B35). Table B5 shows the bulk micro-
hardness values and the calculated elastic moduli of the craze fibrils of some
polymers. The data suggest that the mechanical properties of the craze fibrils are
higher than those of the bulk polymer. This is consistent with the concept of
highly oriented polymer chains within craze fibrils.
198 G. H. Michler
Table B5. (a) Molecular weight, micro-hardness H and elastic moduli E of various polymers;
(b) micro-hardness and elastic moduli of the craze fibrils and the ratios of these val-
ues to the bulk properties
(a) Polymer Mw MwiMn HMatrix EMatrix

(kg/mol) (MPa) (GPa)
PMMA 2200 2.5 182±2 4.8 ± 0.2

SAN 249 2.3 178 ± 2 4.7 ± 0.2
PS 476 4.2 164±3 4.3± 0.1
PVC 181 1.7 145 ±2 4.2 ± 0.2
(b) Polymer Hcraufibril HfibrillHbulk ECraufibril EfibrillEbulk

(MPa) (GPa)
PMMA 537 ± 16 3.0 13.4±1.1 2.8

SAN 510 ± 20 2.9 12.2 ± 1.4 2.6
PS 414 ± 33 2.4 11.2 ± 1.6 2.6
PVC 698 ± 82 4.8 11.4 ± 1.5 2.7
15 PMMA ~
... 14
SAN
'"
813
]
~ 12
11
10
9+---~---.---.---.---.--~
0.04 0.06 0.08 0.10 0.12 0.14
v, (mole/em')
15
14
10
12 16 20 24 28 32
0", (MPa)
Fig. B36. Elastic modulus Efibril of craze fibrils as a function of (a) entanglement density Ve,
(b) craze initiation stress CTc
The results in [18] emphasize the role of crazes as precursors of cracks, pro-
viding resistance against crack propagation within the polymer. It was shown that
the elastic moduli of craze fibrils (in PMMA, SAN and PS) are proportional to
both the crazing stress and the entanglement density (Fig. B36).
The correlation shown in Fig. B36 agrees with the known relation between
craze initiation stress and entanglement density [19].
According to Donth and Michler [20], the transition from a glassy behaviour of
the undeformed polymer material at temperatures below Tg to a mobile state of the
macromolecules under a mechanical tensile stress can be described in terms of a
universal phenomenon of multiple glass transitions of macromolecules [21]. The
basic idea is that a localized stress concentration increases the local free volume
and reduces locally the glass transition temperature to the level of the test tem-
perature. Therefore, an increase in external stress results in a step-by-step change
of the dynamic state of the polymer from hard glassy behaviour to a more flexible
mode. In particular, the existence of rapid changes of the polymer state has been
demonstrated by electron micrographs of the pre-craze structures and the corre-
sponding real crazes with fibrils [22].
3 Materials and Experimental Methods
3.1 Materials
Experiments were performed with bisphenol-A-polycarbonate (BPA-PC) and a

random poly(styrene-co-acrylonitrile) (SAN) with a ratio SIAN = 76/24. Both
materials were commercial, amorphous polymers. As examples of high-tempera-
ture polymers, chemically modified PC polymers with an increased glass transi-
tion temperature were also used [23]. A PC based on trimethy1cyclohexane (TMC-
PC from Bayer AG), with a high glass transition temperature of Tg = 238°C, is of
particular importance. The high Tg of this polymer is due to a reduced chain mo-
bility. Because of its good compatibility with BPA-PC, blends with defined glass
transition temperatures between 148°C and 238 °C are available (Table B6).
Table B6. Glass transition temperatures and molecular weights of the polycarbonates studied
Material Tg Mw MwlMn
(0C) (g/mol)
PC (Macrolon 2803) 148 28000 1.33

PC-2 180 34000 1.23
PC-3 200 35000 1.26
200 G. H. Michler
Specimens with cross-sections of about 2 mm2 and lengths of 60 mm were pro-

duced from injection-moulded polymer sheets. The thickness of the injection-
moulded sheets was about 1 mm.
3.2 Procedure for Creation of Crazes
In one test, crazes were created by creep experiments. Consequently, the external
stress was constant, which was often coupled with a dramatic change of strain rate
during the experiment (Fig. B37).
Va,
L------- CJ2
~l
- - - - - - - - - - - - - -___
Fig. B37. Schematic representation of strain rate (c ) versus time (t) -at three characteristic
tensile stresses (al < a2 < a3)
The creep experiments on SAN were carried out at four temperatures T (T < T;,),
such that the differences between the glass transition temperature Tg and the test
temperature T were 45 K, 34 K, 25 K and 20 K. To compare the measurements on
SAN and PC, the temperature of the creep tests on PC was selected so that its differ-
ence was always equal to Tg-T= 25 K. The specimens were heated for 10 min at the
test temperature. Then they were deformed at constant stress up to a final tensile
strain of about 3.5 %, which was a constant in each experiment. After reaching the
final tensile strain, the specimens were cooled to room temperature in the loaded
state. In addition to the strain rate, the time necessary for reaching the final extension
varied, from some seconds at high stresses to some days at low stresses.
In another test, specimens of PC were deformed in deformation tests of the
usual kind at temperatures between room temperature and temperatures below the
glass transition temperature.
3.3 Small-Angle X-Ray Scattering
Measurements of the equatorial X-ray scattering of the crazes were performed in

the unloaded state using a Kratky camera system. The monochromatization of the
radiation from the Cu tube used was carried out by reflecting the beam from a
glass sheet near the limiting angle of total reflection of the Cu K.. line. The fibril
separation Do was calculated from
2
DO =-=:--- (B12)
.Jj smax
where Smax is related to the scattering vector of the strong interference maximum.
The numerical factor in (B12) arises from assuming that the arrangement of fibrils
in the craze corresponds to a hexagonal packing of cylinders [24]. The scattering
vector is equal to S = 2A.~: sin () (I!.cu = wavelength, () = scattering angle). The
existence of cross-tie fibrils, observed by electron microscopy by Berger et al. [25]
and Michler [3], and discussed in more detail by Brown [26], was ignored.
3.4 Electron Microscopy
For electron microscope investigations by means of a high-voltage transmission

electron microscope (HVEM), semi-thick sections were produced from selected
specimens. The sections, taken parallel to the surface of the specimen, had a
thickness of approximately 1-2 Jim. The cutting direction was opposite to the
craze propagation direction. Therefore, it could be ensured that the crazes did not
propagate under the pressure of the knife during cutting, and consequently the
original craze microstructure survived. This special method of preparation of
crazes from externally deformed specimens makes it possible to compare results
of electron microscopy with fmdings from other measurements (e.g. SAXS).
Ultra-thin sections and solution-cast films of the PC materials with a thickness
of about 100 nm were mounted onto copper grids and were deformed in a Zwick
tester of the usual kind at different temperatures. Small pieces of deformed speci-
mens were inspected in a 200 kV TEM.
Deformed semi-thin specimens were investigated by scanning electron micros-
copy, too. Before inspection a carbon layer a few nanometres thick was evapo-
rated onto the surface of the specimens.
4 Craze Formation at Higher Temperatures
The results reveal a great variety of possible craze microstructures as a function of

temperature and external stress. This is demonstrated in Fig. B38 for PC.
The craze shown in detail in Fig. B38a has a coarse microstructure with excep-
tionally thick fibrils. Distances between the fibrils are 200 nm and more. Figure
B38b, on the contrary, shows fme fibrils within the craze, with fibril distances of
about 35 nm. Figure B38c represents a 'homogeneous craze' [27] (also called
'homogeneous deformation zone' [9]), which has often been observed, by several
methods of electron microscopy. This kind of craze was produced at compara-
202 G. H. Michler
tively low temperatures and high stresses. By SAXS measurements, however, a

fibril distance of 30 nm was determined, on averaging over the whole specimen.
Although many crazes were macroscopically visible in these specimens, the scat-
tering intensity was low. In combination with necking processes of the specimen,
the discrepancy between the still measurable fibril distance and the homogeneous
microstructure of the crazes may be explained by the existence of a small number
of finely fibrillated crazes or mixed structures of fibrillated and homogeneous
crazes. In contrast to the homogeneous crazes, this small amount of fibrillated
material, distributed over large specimen areas, cannot easily be detected by
means of electron microscopy. The trend towards formation of homogeneous
crazes with increasing stress or decreasing temperature can also be observed for
deformation processes in SAN.
Fig. 838. Internal structure of crazes in PC (HVEM, defonnation direction vertical): (a) coarse
craze fibrils, Tg - T= 16 K, (J= 4.5 MPa; (b) fine craze fibrils, Tg- T= 37 K, (J= 40.3
MPa; (c) homogeneous craze structure, Tg- T= 45 K, (J= 48.2 MPa
The thickness (diameter) D of the crazes and the surface distance Do between
fibrils ('fibril distance') were calculated, and show a decrease with increasing
stress at constant temperature and with decreasing temperature at constant load.
This is illustrated by two examples in Fig. B39 for the fibril distances Do in PC
and SAN.
175 a
• Unannealed PC
o Annealed 70h at 128°C
150
!125
Q
100
75
50~--'---~--~r---r---.---.-~r--.-J
10 12 14 16 18 20 22 24 26
Go (MPa)
b
40
SAN
~o 35
30
28 r-------------~~---------
20 30 40 50 60 70 80
Tg-T(K)
Fig. B39. Decrease of the fibril distance Do with (a) increasing stress in PC (Tcraze = 123 QC),
(b) decreasing temperature in SAN (O"craze'" 25 MPa)
The different internal structures of crazes can be roughly divided into three re-
gimes (see Fig. B40) [22]:
• Regime I. Crazes are characterized by large values of fibril thickness D and
fibril distance Do and appear at low stresses and/or higher tem-
peratures. Fibril formation is due to the melt-like behaviour of the
molecules.
• Regime II. Crazes show a relatively fme fibrillated structure. The mobility of
macromolecules is increasingly restricted by entanglements. The
entanglements fix the centres of the polymer chains and only
shorter units of the macromolecules contribute to craze formation.
• Regime III. Homogeneous crazes dominate or coexist with finely fibrillated
crazes. The high stresses cause entanglements to react in a way
similar to chemical crosslinks in the short time available. Disen-
tanglement, and hence the formation of fibrillated crazes, will al-
ways be improbable.
204 G. H. Michler
Craze Microstructure Regime
II
III
Fig. B40. Schematic representation of characteristic craze microstructures in regimes I, II and

III (see Figs. 838 a, b, c, respectively)
More details of fibril formation in the different regimes are discussed in [22] in
connection with macromolecular flow processes.
In Fig. B39 a the decrease in fibril distance Do with increasing stress is also
shown for physically aged specimens of PC. The unaged and aged specimens
show similar tendencies, but with bigger values for Do (and also for the fibril
thickness values D) in the aged specimens. This behaviour can be correlated with
the reduced free volume due to the physical ageing and, therefore, with an additional
steric hindering of chain mobility. A coarsening of the craze structure in annealed
polymers is a general effect and has been found also in SAN and PMMA [3].
It is generally accepted that crazes with a fibrillated internal structure appear if
the entanglement density is low, i.e. if the distance between entanglements or the
entanglement molecular weight is high (as in polystyrene, with a critical entan-
glement molecular weight of about 17 500). With decreasing entanglement molecu-
lar weight, there is a transition to a fmely fibrillated form of craze and to homogene-
ous deformation bands. Examples are the coexistence of fibrillar and homogenous
crazes in SAN and the existence of homogeneous crazes in PMMA [2,3].
This effect can be demonstrated in high-temperature-resistant amorphous
polymers such as PC, too. The usual bisphenol-A polycarbonate, with Tg = 148°C,
and the chemically modified versions with higher glass transition temperatures
show a characteristic transition in the deformation behaviour. At room tempera-
ture and somewhat higher temperatures, typical shear bands and homogeneous
deformation zones appear. At about 60°C and 80 °C below the glass transition
temperature in the usual PC type and in the materials with the higher glass transi-
tion temperatures, respectively, there is a transition to the formation of fibrillated
crazes. With increasing deformation temperature, the fibrillation of the crazes
becomes coarser (an increase of the long period of the fibrillation). This is summa-
rized in Figs. B41 and B42.
Crazes
Detail: Craze Microstructu re
Homoge neous
Deformation
Shear Bands
25°C T
Fig. B41. Dependence of type and internal structure of defonnation zones in PC on the defor-
mation temperature T (defonned thin specimens, 200 k V TEM)
/?7--.J---··
Crazes
ex = 90°
Homogeneous
Deformation Bands and
Crazes ex = 90° . . . ..... / .•...~.. /,"",'.
Shear Bands and
Homogeneous
Deformation ... f
.'
......
.
.....
~.
,,
,
•- PC2803
·····--PC-2
ex = (45 ...9W ./............ "
x ••••••• PC-3
Shear Bands .x----
.... _.* "
ex = 45°
I I r I I I -II
120 100 80 60 40 20 0
~T(K)
Fig. B42. Schematic drawing of the transition in the defonnation behaviour as a function of the
difference between the test temperature T and the glass transition temperature Tg (~T =
Tg- T; (X = angle between the length of the defonnation bands/crazes and the direction
of loading)
These transitions in the deformation character can be understood in terms of

processes on the macromolecular level that depend on temperature:
206 G. H. Michler
1. RT, T< Tg-AT(AT= 60 °C-80 0C): high entanglement density and high mac-
romolecular mobility result in homogeneous yielding (shear bands and defor-
mation zones DZs).
2. T == Tg - AT: thermally induced disentanglement, reduction of entanglement
density results in a transition from homogeneous yielding to crazing.
3. T< Tg, T ~ Tg: intense thermally induced disentanglement results in coarsen-
ing of craze structure with increase of D and Do.
With increasing temperature, the stress necessary for thermally induced disen-
tanglement drops more rapidly than the stress necessary for molecular mobility
(see Fig. B43). The crossing point marks the transition from homogeneous yield-
ing to crazing. The coarsening of the craze structure of PC (increase of diameter
and distance of fibrils) with increasing temperature is also revealed by X-ray
scattering, as mentioned above.
RT
Fig. B43. Schematic illustration of decrease in the stresses necessary for dist<ntanglement ObE
and molecular yielding O"y; region 1 defmes homogeneous yielding, region 2 is the
transition region and region 3 defines the region of crazing
Crazes are thinner than homogeneous deformation bands. Therefore, the total
amount of polymeric material which is plastically deformed is smaller in the case
of crazes and larger in the case of shear bands or DZs. The result is shown macro-
scopically in the unusual effect of a decrease in toughness and an embrittlement
with increasing temperature.
5 Summary
Deformation tests of amorphous polymers, including in situ tests, have been perfor-
med using various techniques of electron microscopy. The results could be directly
correlated with the results of SAXS methods applied to deformed specimens. Using
these techniques, a better understanding of micro-mechanical properties and a mi-
croscopic modelling of the mechanical behaviour of polymers have been achieved.
In PS, SAN and PC the formation of fibrillated crazes and homogeneous de-
formation bands depends decisively on the size level of entanglements (which is
on the order of 100m). The highly stretched entanglement network inside the
crazes results in a micro-hardness and elastic modulus of the crazed material nota-
bly larger than those of the bulk polymer. The results emphasize the role of crazes
as precursors of cracks, providing resistance against crack propagation within the
polymer.
Using creep experiments, different microstructures of crazes were produced in
SAN and PC and measured by TEM and SAXS. The arrangement of fibrils within
the crazes obeys liquid-like statistics. At the highest stresses and the lowest tem-
perature in the range investigated, the entanglements act as chemical crosslinks.
The loss of entanglement that is geometrically necessary for fibril formation is
suppressed and, therefore, the number of fibrillated crazes is reduced and the for-
mation probability of homogeneous crazes (homogeneous deformation bands) is
increased. With decreasing stress and increasing temperature, the mobility of
polymer chains increases, longer units of macromolecules are mobile and disen-
tanglement becomes more probable. This behaviour results in formation of coarser
fibrils and longer distances between the fibrils, and in a transition from the forma-
tion of homogeneous crazes (or shear deformation processes) to fibrillated crazes.
At the lowest stresses and at the highest temperatures, below the glass transition
temperature Tg, the increased motion of the macromolecules can be characterized
by localized flow processes. A thermally induced disentanglement becomes much
more probable, yielding thick fibrils and large interfibrillar spaces. These changes
in the microstructure of crazes are very clearly pronounced also in high-tempera-
ture-resistant grades of PC.
Acknowledgements
The author thanks Dr. J.-U. Starke, Dipl.-Phys. M. EnBlen and Dipl.-Phys. S.
Henning for performing deformation tests by TEM, HVEM, SEM and SAXS, and
the director of the Max Planck Institute of Microstructure Physics, Halle/Saale,
Germany for providing the opportunity to carry out deformation tests in the 1000
kV HVEM. The financial support of the German Research Foundation (DFG)
{lnnovationskolleg 'New Polymer Materials' and SFB 418 'Nanometerstruk-
turen') is greatly appreciated.
References
1. Lauterwasser B. D., Kramer E. 1. (1979): Microscopic mechanisms and mechanics of crazes

growth and fracture. Philos. Mag. A 39: 469-495
2. Michler G. H. (1979): In-situ deformation of polymers in the HVEM. Krist. Technik 14:
1357-1363; (1990): Correlation between craze formation and mechanical behaviour of
amorphous polymers. J. Mat. Sci. 25: 2321-2334
208 G. H. Michler
3. Michler G. H. (1992): Kunststoff-Mikromechanik: Morphologie, Deformations- und Bruch-

mechanismen. Carl Hanser, Munich, Vienna
4. Brown H. R. (1987): Mater. Sci. Rep. 2: 315-370
5. Yang A. C.-M., Kramer E. I. (1986): Craze microstructure characterization by low-angle
electron diffraction and fourier transforms of craze images. I. Mat. Sci. 21: 3601-3610
6. Doll W., Konczol L. (1990): Micromechanics of fracture under static and fatigue loading:
optical interferometry of crack tip craze zones. Adv. Polym. Sci. 91-92: 137-214
7. Konczol L., Doll W., Bevan L. (1990): Mechanisms and micromechanics of fatigue crack
propagation in glassy thermoplastics. Colloid Polym. Sci. 268: 814-822
8. Kambour R. P. (1973): I. Polym. Sci., Macromol. Rev. 7: 1-154
9. Kramer E. I. (1983): Microscopic and molecular fundamentals of crazing. In Kausch H. H.
(Ed.) Crazing in Polymers. Springer, Berlin, Heidelberg: 1-56 (I. Adv. Polym. Sci. 52-53:
I-56)
10. Kramer E. I., Berger L. L. (1990): Fundamental processes of craze growth and fracture. Adv.
Polym. Sci. 91-92: 1--68
II. Kausch H. H., Gensler R., Grein c., Plummer C. I. G., Scaramuzzino P. (1999): Crazing in
semicrystalline thermoplastics. I. Macromol. Sci. - Phys. B 38: 803-815
12. Bucknall C. B. (1977): Toughened Plastics. Applied Science, London
13. Kausch H. H. (Ed.) (1990): Crazing in Polymers Vol. 2. Springer, Berlin, Heidelberg
14. Michler G. H. (1993): Electron microscopy in polymer science. Appl. Spectrosc. Rev. 28/4:
327-384
15. Aharoni S. M. (1983): On entanglements of flexible and rodlike polymers. Macromolecules
16: 1722-1728
16. Michler G. H. (1991): Crazes und crazeartige Deformationen in Polymeren, 4. Riickschliisse
auf die Struktur der amorphen Polymere. Plaste Kautschuk 38: 268-272
17. Dettenmaier M., EnBlen M., Michler G. H.: unpublished results
18. Michler G. H., EnBien M., Balta Calleja F. I., Konczol L., Doll W. (1999): Mechanical
properties of crazes investigated by ultramicrohardness. Philos. Mag. A 79/1: 167-178
19. Wu S. (1990): Chain structure, phase morphology, and toughness relationships in polymers
and blends. Polym. Engng. Sci. 30: 753-761
20. Donth E., Michler G. H. (1989): Discussion of craze formation and growth in amorphous
polymers in terms of the multiplicity of glass transition at low temperature. Colloid Polym.
Sci. 267: 557-567
21. Donth E. (1981): Glasiibergang. Akademie Verlag, Berlin
22. Starke I.-V., Schulze G., Michler G. H. (1997): Craze formation in amorphous polymers in
relation to the flow and main transition. Acta Polym. 48: 92-99
23. Kampf G., Freitag D., Fengler G. (1992): Neue Polycarbonate erreichen erhOhte Wiirme-
formbestandigkeit. Kunststoffe 82: 385-390
24. Paredes E., Fischer E. W. (1979): Rontgeukleinwiukeluntersuchungen zur Strnktur der
Crazes (FlieBzonen) in Polycarbonat und Polymethylmethacrylat. Makromol. Chern. 180:
2707-2722
25. Berger L. L., Buckley D. I., Kramer E. I. (1987): Low-angle electron diffraction from high
temperature polystyrene crazes. 1. Polym. Sci., Polym. Phys. 25: 1679-1697
26. Brown H. R. (1991): A molecular interpretation of the toughness of glassy polymers. Mac-
romolecules 24: 2752-2756
27. Michler G. H. (1984): New possibilities of the electron microscopical determination of the
morphology of polymers. Ultramicroscopy 15: 81-100
B 1.5 Influence of Temperature and Moisture
B. Langer, S. Seidler and W. Grellmann, Merseburg
1 Introduction
The thennoplastic polymer polyamide (PA) has a large field of application be-
cause of its especially well balanced relation between toughness and stiffness, its
good chemical resistance and its uncomplicated processing. However, an impor-
tant factor in any decision to use polyamide materials is the water absorption,
which is influenced by the ratio of methylene to amide groups. The achievable
mechanical properties of all polyamide types are strongly influenced by this prop-
erty.
The mechanical property spectrum, including the toughness, of several types of
PA results from differences in the water absorption behaviour and from its time
and environmental dependence. Considering the specific material behaviour ofPA,
it is of interest to quantify the moisture-influenced toughness behaviour and the
effect of water absorption on the resistance against crack initiation and propaga-
tion. Furthennore, a toughness characterization requires a consideration of the
moisture and time dependence of the toughness parameters. The aim of the work
described in this chapter was to characterize the influence of temperature and
moisture on the toughness behaviour of PA on the basis of fracture mechanics
concepts.
2 Structure, Application and Fracture Mechanics

Properties of Polyamide Materials
Owing to the hygroscopic properties of PA materials, the toughness level can be

adapted to the required conditions of practical use by variation of the moisture
content.
A basic knowledge of the interaction between the moisture content of PA and
the ambient temperature leads to the conclusion that the field of application of this
material group can be expanded. PA 6, with a ratio of methylene to amide groups
of '5', absorbs 10 % water in an atmosphere of 100 % relative humidity. For PA
12, with a ratio of methylene to amide groups of' 11 " the water absorption is re-
duced to 2 % [1]. An increasing moisture content in PA shifts the glass transition
temperature to lower values. By way of example, the glass transition temperature
of PA 6 shifts from 60°C for the moisture-free material to 5 °C at 3.5 % moisture
content and to -15°C at 10 % moisture content [1,2]. Furthennore, the relaxation

210 B. Langer, S. Seidler, W. Grellmann
processes of the strongly hygroscopic types PA 6 and PA 66 change by a particu-

larly large amount owing to the physical incorporation of water. The ~ relaxation
area occurs between -50°C and -70 °C and is characterized by motion of seg-
ments without hydrogen bonds and of molecules coupled to the amide group, for
example water molecules. It only occurs if the content of moisture is low [2].
More recent investigations assume a separation of the ~ relaxation area into two
partial processes ~1 and ~2 [3]. The mobility of the chain segments increases in the
presence of moisture, and the ~ 1 relaxation occurs. Its intensity and position with
respect to temperature depend on the moisture content. The ~1 relaxation starts at a
higher activation energy than the ~2 relaxation, which happens independently of
the moisture content.
If PA is in contact with water molecules, regardless of whether the H 20 is in
gaseous fonn in the atmosphere or in the liquid state in direct contact, the water
molecules are incorporated into the amorphous phase and linked with the amide
groups by hydrogen bonds. The incorporation speed and quantity depend on
- the proportion of CH2 to CONH
- the degree of crystallinity
- the specific contact surface between water molecules and PA
- the relative humidity and temperature of the environment.
The structure of the crystalline phase is completely maintained when PA takes
up moisture. In the amorphous phase, the incorporated water molecules increase
the distance between the chains and reduce the intennolecular forces so that the
chains become more flexible. Therefore, the energy absorption and dissipation
capacity of the amorphous phase increase with increasing moisture content and
increasing chain mobility. Impact and shock loads, which in moisture-free PA lead
to stress concentrations, breakage of hydrogen bonds and, finally, to macroscopic
cracks, can be better compensated in PA stored in contact with water or in stan-
dard climate-conditioned PA [2,4,5].
Table B7 shows a summary of fracture mechanics parameters of some well-
studied polyamide-elastomer blends with semicrystalline (Zytel ST801) and
amorphous (Zytel ST90 I) structures. These blends are characterized here by frac-
ture mechanics parameters from elastic-plastic fracture mechanics (EPFM), e.g. by
J,c or J'd at crack initiation. These blends have also been used as model materials to
verify the influence of test conditions on the measurement of fracture mechanics
values. From the material characterization, it can be seen that the semicrystalline
PA blends have a more favourable toughness than the amorphous blends. The
reason for this is the defonnation behaviour of the crystalline areas. The results of
examination of the influence of temperature on the toughness properties of these
PA blends support this statement [6]. One can assume that, owing to the defonna-
tion behaviour of the crystalline phase, the crack initiation values of only the semi-
crystalline PA blend are reduced with decreasing temperature.
Influence of Temperature and Moisture 211
Table B7. Fracture toughness data of semi crystalline and amorphous elastomer-modified PA
materials
Data
Test conditions P A-elastomer blend Remarks Refe-
Semicrystalline Amorphous rences
SENB a lIe (N/mm)b Values indepen- [7]

Razor blade notch dent of specimen
B=3.2mm 33.7 16.4 thickness
B=6.4mm 26.5 15.8
B=1.7mm 29.2 16.3
SENB a lId (N/mm) Values indepen- [8]

Razor blade notch dent of impact
B= 12.7mm speed
VH = 0.6 mls
to '" 20 ",20
VH = 3.3 mls
B= 1.8 mm le (N/mm) Values depend on [9]

VT = 10 mrnImin specimen geome-
SENB 11.9 30.9 try
DENT 10.7
DENB a lIe (N/mm) (MSMt Values indepen- [10]

RK = 12 IJlll 33.6 17.2 dent ofR-curve [11]
technique
vT = 2 mrnImin lIe (N/mm) (SSM)d
B= 12.6mm 34.6 18.5
SENB lIe (N/mm)b [12]

VT= 25 mrnImin
B= 12.7 mm ",10
SENB lIe (N/mm) [13]

RK = 13 IJlll
•
F = 8.9N/s
B= 12.7mm 5
SENB a le (N/mm) Values depend on [14]

vT = 1500 mrnImin specimen thick- [15]
B= 3.2 mm to 8.1-25.1e 2.5-12.0e ness and evalu-
B= 12.7mm 26.5-32.7 b 14.0-16.4b ation method
SENB a lIe (N/mm) Only the values [6]

RK = 5-12 IJlll for the semicrys-
VT = 1560 mrnImin talline blend de-
B= 12mm pend on tempera-
T= 23°C 27.5-32.5 16.0-19.0 ture
T=-10 °C 29.5-31.3 14.8-15.0
T=-40 °C 21.0-23.8 15.0
Continuation of Table B7
Data
Test conditions PA-elastomer blend Remarks Refe-
Semicrystalline Amorphous rences
SENBa Jle(N/mm) Only the values for [6]

RK, = 5-12 IUD the semicrystalline [16]
B= 12mm blend depend on
vT(mmlmin) crosshead speed
15.6 27-32 13-19
156 33-41 19
1560 39-44 18
DDENT A. (mJ/m2) Values depend [17]

B= 1.68mm 19 negligibly on
B= 3.17 mm 21 specimen thickness
SENB, CT, SENT .he (N/mm) Values independent [18]

B=20mm 22 ofldnd ofloading
RK, = 13 IUD dJ/d(Aa) (N/mm2) and specimen ge-
10.3 ometry
a Moisture-free specimen. b ASTM E813-87. e Multiple-specimen R-curve method.

d Single-specimen R-curve method. e ASTM E813-81.
3 Experimental
The instrumented Charpy impact test (lCIT) and the quasi-static fracture mechan-
ics test represent the experimental basis for the fracture mechanics toughness char-
acterization of the PA 6 and cast PA 6 materials tested here. These methods guar-
antee the separate evaluation of the strength and the deformation parts of the total
deformation process. This is a requirement for the evaluation of fracture mechan-
ics parameters related to resistance against unstable and stable crack growth under
static and dynamic load. The estimated fracture mechanics parameters (Table B8)
describe the resistance against unstable and stable crack growth. Here, load-de-
flection diagrams recorded using the ICIT and the quasi-static fracture mechanics
test are the basic elements, besides the utilization of the recommendations de-
scribed in A 2.2 (pp. 71-86) and [19-21].
For the determination of the resistance against unstable crack growth, the J-
integral concept of elastic-plastic fracture mechanics was used because the tested
PA materials showed elastic-plastic material behaviour with formation of an en-
ergy dissipation zone which was large in comparison with the initial crack length.
For describing the resistance against stable crack growth, the crack resistance (R-)
curve concept was used. An R-curve completely reproduces the process of stable
crack growth. Because of the functional connection between the loading parameter
J and the amount of stable crack growth !la, the phases of the stable cracking pro-
Table B8. Fracture mechanics parameters
Fracture mechanics test Parameters related to Parameters related

EPFM to R-curve
Quasi-static
fracture mechanics test
Instrumented Charpy
impact test
cess of blunting, crack initiation and stable crack propagation can be represented.
Fracture mechanics parameters of R-curves were estimated according to the pro-
cedures of (pp. 71-86, [20,21]. Furthermore, energy dissipation processes were
included in the examination by application of the JTJ concept [22]. It was possible
to record an R-curve by two different techniques:
- the single-specimen R-curve method (SSM)
- the multiple-specimen R-curve method (MSM).
In the application of the SSM, one specimen is loaded several times, and in the
application of the MSM, several specimens are loaded once with different ener-
gies. For estimation of the resistance against unstable and stable crack growth
using the quasi-static fracture mechanics test, compact tension (CT) specimens
were utilized. By means of a Zwick 1386 universal testing machine quasi-static
loading was performed. The deformation of the specimen was recorded by an
inductive displacement transducer as load-line-displacement values. The internal
recording unit of the universal testing machine documented the load received from
the specimen.
The instrumented Charpy impact test was carried out according to the procedure
described in A 2.2 (pp. 71-86). For the determination of R-curves it is necessary
that several specimens contain a defined amount of energy. For that purpose, a
number of experimental R-curve methods exist [23,24]:
- energy method: variation of the pendulum hammer weight or of the hammer
speed ('low blow')
- stop block method: variation of the specimen deflection
- slL method: variation of the ratio of support span to specimen length
- slWmethod: variation of the ratio of support span to specimen width.
In the present work the stop block method and the multiple-specimen R-curve
method were predominantly used. The energy which a specimen contains can be
varied by restricting the deflection by steel barriers with a defined width gradation
or by using a device for catching the pendulum hammer with a defmed, adjustable
deflection (pp. 71-86). In the application of the slL method here, the specimen
length L was reduced. The specimen had a length equal to the support span s plus a
variable excess (Ln = s + n x 0.1 mm). A number of specimens were pulled

through the support, and so different amounts of stable crack propagation were
created.
The specimens used were conditioned before the experiments started. The
'moisture-free', 'normal-moisture' and 'high-moisture' had moisture contents as
shown in Fig. B44, in the case of PA 6. These different moisture contents of the
PA materials were achieved by exposure the specimens to water or other media as
summarized in Table B9.
12 PA 6 1 - - - - - - - - - ,
8 Exposure to Water
~
9.5 wt.% Water
~
~
a
..c
u 4
'"'" Rapid Conditioning
os 2.7 wt.% Water
~
~ _ _ _ _ _ Exposure to Vacuum
Owt.% Water
-4~--------~--------~----------------~
o 20 40 60 80
Time (d)
Fig. B44. Conditioning ofPA 6
Table B9. Conditions of exposure
Conditioning state Procedure
Moisture-free 1. Exposure to vacuum at room temperature directly after the in-

jection-moulding process or the casting process
2. After storing in other climatic conditions, removal of water by
vacuum drying at 80°C [25] (Fig. B44)
Normal-moisture 3. Exposure to standard climate in accordance with DIN EN ISO

1110 [26]; a constant weight occurs after about 30-40 days [2]
4. Exposure over sodium nitrite at 80°C (rapid conditioning ac-
cording to [25]; constant weight occurs after about 10 days)
(Fig. B44)
High-moisture 5. Exposure to water at room temperature (Fig. B44)

4 Results
4.1 Influence of Test Conditions on the Investigation

of Fracture Mechanics Parameters
4.1.1 Comparison of Different R-curve Procedures
For the determination of R-curves in the ICIT, no standard or recommendation

exists which advises the application of a particular technique. For this reason in-
vestigations were carried out to compare different R-curve techniques and R-curve
methods. Any possible influence of these different techniques and methods on the
values of parameters so determined should be studied, so that one is able to assess
the results obtained from the experiments [27]. Different results are obtained by
application of the stop block method, the low-blow method and the slL method
(Fig. B45). The amount of stable crack growth is higher, and stronger energy dis-
sipation during stable crack growth appears, when the slL method is used, in com-
parison with the stop block method (Fig. B45a).
The higher compliance of the specimens used is the cause of the stronger energy
dissipation in the slL method. A comparison between the two methods is only
possible under certain conditions; in the slL method, every point of the graph is
determined by variation of the specimen compliance, while in the stop block
method, the specimen compliance is kept constant. Furthermore, it is a problematic
fact that, by application of both methods, no valid R-curve was determined, and the
5 10
0.98
4 ~max=0.6mm J=48.1~
0 75
t;.
a 3 J= 13~
0.96
t;.
~-. &min
<0
50
~min Mmax
2 0 t;.
0.39
J = 2.9.&i 25
oslL Method LowBlow

t;..
• Stop Block '" Stop Block
0 0.1 0.2 OJ 0.4 0 0.5 1.5 2

.&i(mm)
Fig. 845. R-curves of 'normal-moisture' PA 6 and PA 66-HI (high impact) determined by

application of (a) stop block and slL methods and (b) stop block and low-blow
methods
estimation of the technical crack initiation value J do. 2 (cf. pp. 71-86) was not
meaningful. The comparison of the results of the stop block method and the low-
blow method shows that in this case the points of the graph determined have the
same fit function, independent of the method used. Differences occur in the adjust-
able amount of maximum stable crack growth fla and the distribution of the data
points. An important problem in the application the low-blow method is the influ-
ence of the loading rate on the mechanical behaviour. Because of their viscoelastic
behaviour, polymers react sensitively to changes in loading rate. The variable
influence of the loading rate on the elastic and the elastic-plastic toughness be-
haviour can lead to considerable differences [23,24].
Regarding the influence of the single- and multiple-specimen methods on the
results of the instrumented Charpy impact test (Fig. B46a) and the quasi-static
fracture mechanics test (Fig. B46b), the results show that, independent of the
method used, in the regime of small amounts of stable crack growth, almost identi-
cal J values are determined. With increasing crack growth, the differences become
greater. The external conditions (support span, loading rate, initial crack length)
and internal conditions (notch sharpness) are, in the first measurement, identical
for both of the methods. The same supplied energy leads to the same J values.
When the MSM is used, each subsequent crack growth occurs under the same
conditions, while when the SSM is used, these conditions change, particularly the
notch sharpness and the initial crack length. The reason for the different toughness
behaviour seems to be the varying notch sharpness, because the varying initial
crack length is taken into account in the J-integral calculation (equation (A65) of
A 2.2 (pp. 71-86)).
10raar------------L:~~~------------~b
VH= 1 mls Vr = 2 rmnImin
,.
7.5 &max=3.2mm
75
•
&max
5 50
2.5 25
SSM· SSM·
MSM6 MSMO
0~--~0.~2--~0.4~~0~.6~~0.8 0.5 1.5

~a(nm)
Fig. B46. R-curves of 'normal-moisture' PA 6, determined by application of MSM and SSM;

(a) ICIT and (b) quasi-static fracture mechanics test
4.1.2 Influence of Specimen Thickness
Dependent on the geometry, different multi-axial states of strain are formed in

front of a crack tip in components and in specimens. In the case of a state of plane
strain, the toughness values determined are independent of the geometry. Then
they characterize influences such as the temperature, the loading rate and the
toughness behaviour resulting from the material behaviour [28,29]. Under equal
loading conditions, it depends on the specimen geometry, particularly on the
specimen thickness, whether a state of plane strain or plane stress is present.
The geometrical criterion (equation (A60) of A 2.2 (pp. 71-86)) for the limit of
a plane strain state to be valid was used to estimate the necessary geometrical
quantities of the specimen thickness E, the initial crack length a, and the difference
between the specimen width and initial crack length (W - a). The constant 6 de-
pends on the material behaviour and can take on values between 25 and 200 for
metals [30,31].
1O,--------i
7.5
o
I
.... 5
a"'" 0 0 0 J O.2
o 5 10 15
B{mm)
• 2mm • 6mm
• 3mm + 8mm
.. 4mm x lOmm
0.4 0.6 0.8

~(mm)
Fig. B47. Influence of specimen thickness on (a) R-curves and (b) resistance against unstable
and stable crack growth of 'normal-moisture' PA 6
For determining the resistance against unstable crack growth for polymers, 6
values between 10 and 90 are given in [32,33]. Investigations of the 'normal-
moisture' PA (Fig. B47) using the ICIT support this statement. The specimen
thickness has no influence on the resistance against unstable and stable crack
growth under dynamic loading (Fig. B47), and by using the standard specimen of
80 x 10 x 4 mm3 [34], geometry-independent fracture mechanics parameters of the
material were determined.
4.2 Dependence of Fracture Mechanics Parameters

on Moisture Content and Test Temperature
4.2.1 Influence of Moisture Content
All mechanical properties of PA materials are considerably affected by the quan-

tity of moisture contained; in particular, PA 6 materials are strongly influenced
because of their high water absorption ability [1-5,25,35], as described in Section 2.
Under quasi-static loading, the mechanism of crack growth changes with in-
creasing moisture content. The 'normal-moisture' and the 'high-moisture' cast
polyamides fail by dominantly stable crack growth, whereas in the 'moisture-free'
cast polyamide the failure process is characterized by unstable as well as stable
crack growth. R-curves were correspondingly determined to assess the effect of
different conditioning states on the toughness behaviour of cast PA 6 [27]. Figure
B48 shows R-curves of cast PA in the 'moisture-free', 'normal-moisture' and
'high-moisture' conditioning states.
The conditioning was performed under the conditions given in Section 3. The
'normal-moisture' and 'high-moisture' conditioning states are characterized by a
higher amount of stable crack growth than the 'moisture-free' state because of the
change of the crack growth behaviour. The highest J values were determined for
'high-moisture' cast PA 6. Table BIO describes the changes of yield stress O'zS,
modulus of elasticity and fracture mechanics values, as a function of the condi-
tioning state of cast PA 6.
10(h----------i
CT Specimen
VT=2mmlmin
High-Moisture
6.amin
•
E Normal-Moisture
~ 5
Conditioning State
Moisture-Free High-Moisture 35.8

Normal-Moisture 20.8
4 • Moisture-Free 9.1
0.5 1.5
Lla(mm)
Fig. B48. R-curves of cast PA 6, detennined under quasi-static loading, as a function of the
moisture content
Table BIO. Properties of cast PA 6 under quasi-static loading as a function of the

conditioning state
Conditioning state Et J 0.2 TI2 JTJ

O"zS Ie
(MPa) (MPa) (N/mm) (N/mm)
Moisture-free 3800 80 9.1 30 106

Normal-moisture 2750 64 20.8 34 603
High-moisture 1400 32 35.8 140 5840
The yield stress CTzS and the modulus of elasticity decrease continuously with in-
creasing moisture content, in agreement with statements in the literature [1,2,25].
These changes affect the resistance against stable crack initiation and propagation,
too. The relatively high strength and elasticity of the 'moisture-free' conditioning
state lead to a small amount of stable crack growth and to low fracture mechanics
values, as already described. 'High-moisture' cast PA exhibits, in comparison with
the 'normal-moisture' cast PA, a higher resistance against crack initiation and
propagation as well as a stronger energy dissipation capability.
The results of the investigations regarding the influence of moisture content on
resistance against stable crack propagation under quasi-static loading lead to the
conclusion that stable crack processes in PA 6 are influenced by the moisture con-
tained. A PA 6 material with a high moisture content (corresponding to the 'high-
moisture' conditioning state) shows, under dynamic loading, an improvement of
the resistance against stable crack initiation and propagation, too (Fig. B49). The
resistance against stable crack initiation and propagation of cast PA increases with
increasing moisture content. This is analogous to the results of the quasi-static
fracture mechanics test. The extremely low amounts of stable crack growth in the
'moisture-free' and 'normal-moisture' conditioning states are obvious, and make
the fracture mechanics assessment difficult in comparison with the 'high-moisture'
conditioning state. Additionally, the J values determined have a high scatter be-
cause of the different behaviour of the crystalline and amorphous phases. The
estimation of the technical crack initiation value (pp. 71-86) is only possible by
approximation of the resulting fit to I'!a = 0.2 mm, which strongly restricts the
validity of this value regarding the crack initiation behaviour of the material ex-
amined. The energy dissipation during the stable crack growth process is charac-
terized by the product JTj , and this makes possible comparison of the R-curves
concerning the resistance against stable crack propagation as a function of mois-
ture content. A check of validity was carried out with the help of the OJ values
(equation (A77) in A 2.2 (pp. 71-86)).
In addition, the tearing modulus Tj was estimated in order to characterize the
resistance against stable crack propagation. For a comparison, all Tj values were
estimated at the point I'!a = 0.1 mm using a root function for JTj (Fig. B49a; Ta-
ble B II). Additionally, constant Tj values were determined using a linear function
(Fig. B49b; Table B 11) because the distribution of the present J-I'!a data allows a
linear fit. Table B 11 also contains the dynamic flexural modulus Ed and the yield
stress ad detennined by the Charpy impact test under dynamic loading, as well as
the fracture mechanics values detennined using the R-curves. The influence of
moisture content on the mechanical parameters and stable crack growth behaviour,
as detennined by the Charpy impact test, essentially agrees with the influence of
moisture content on the parameters of cast PA 6, as ascertained using the quasi-
static fracture mechanics test method. The dynamic flexural modulus Ed decreases
continuously with increasing moisture content. The yield stresses ad of 'moisture-
free' and 'nonnal-moisture' PA under impact loading are the same. However, the
corresponding value for 'high-moisture' PA, detennined by the Charpy impact test
as well as by the quasi-static fracture mechanics test, is strongly reduced. The
parameters of the R-curves TIland JTJ increase with increasing moisture content.
10
a b
7.5
Ie 5
-.
2.5
O.S J = 0.6+6.1 L\a
• J = (O.l+13L\a)
0.5 J = 0.5+ 1.2 L\a
1oI!III'<~-+,A J = (OJ+1.3L\a)
0.1 0.2 OJ 0.1 0.2 OJ
L\a(mm)
Fig. B49. R-curves of PA 6 with different moisture contents, determined by ICIT, with two
fits shown: (a) root function and (b) linear function
The dependence of the mechanical and fracture mechanics parameters on the

moisture content detennined above allows one to conclude that the improvement
of the defonnation properties because of incorporated water greatly influences the
resistance against stable crack propagation and the energy dissipation. The TJ
values calculated from the linear function show the same trend as the TIl values
calculated from the root function at the point Ila = 0.1. The detennination of a
constant TJ value from a linear fit to an R-curve represents an uncomplicated pos-
sibility for quantifying the resistance against stable crack propagation if the rela-
tion between J and Ila is nearly linear.
The results of the investigations of the toughness behaviour of PA materials

containing different amounts of moisture under dynamic load cannot be discussed
without taking the relaxation process into account.
Because of the different contents of incorporated water, the locations of the re-
laxation area are differently influenced [1,2]. If the assessment of the toughness
behaviour of the three conditioning states of PAis performed only at room tem-
perature, the influence of the location of the relaxation process on the fracture
mechanics parameter need not be considered. For that reason, the fracture me-
chanics values of PA 6 in the 'moisture-free', 'normal-moisture' and 'high-
moisture' conditioning states are discussed as a function of the temperature in the
following section.
Table Bll. Properties of PA 6 as a function of the moisture content, determined under im-
pact load
Conditioning state Ed ad T/.l TJ JTJ

(MPa) (MPa) (root function) (linear function) (N/mm)
Moisture-free 1900 35 1.6 1.9 1.0

Normal-moisture 1400 38 5.3 5.9 6.4
High-moisture 420 15 54 70 264
4.2.2 Influence of Test Temperature
Investigations of the influence of the test temperature on the toughness behaviour

of semicrystalline polymers lead to the main conclusion that the load- and energy-
determined fracture mechanics parameters pass through a maximum and then de-
crease as the test temperature approaches the glass transition temperature (pp.
51-70)
The influence of the test temperature on the toughness behaviour of PA materi-
als has great significance because of the simultaneous influence of the conditioning
states [6,36-40]. Figure B50 shows the dependence on the test temperature of the
.Ad values of PA 6 in the 'moisture-free', 'normal-moisture' and 'high-moisture'
conditioning states, as well as that of an impact-resistant material PA 6-HI.
The resistance against unstable crack growth is increased with increasing test
temperature. The toughness of 'normal-moisture' and 'high-moisture' PA and of
the PA-HI increases continuously, whereas 'moisture-free' PA 6 shows a maxi-
mum at T = -20°C. This material still exhibits dominantly unstable crack growth
behaviour at a temperature of T = 60°C. The crack growth behaviour of the PA
materials in the 'normal-moisture' and 'high-moisture' conditioning states and of
the PA-HI changes at the critical temperatures from dominantly unstable to domi-
nantly stable. A complete fracture mechanics assessment as a function of test tem-
perature is not given here because of the transition of the crack propagation be-
haviour from dominantly unstable to dominantly stable.
2~--------------~
.. Mlisture-Free
15 • Nonna1-Moisture
° High-Moisture
• High-Impact;
Nonna1-Moisture
-150 100 150 200
Fig. B50. Influence of temperature on the resistance against unstable crack growth of PA 6 in
the 'moisture-free', 'normal-moisture' and 'high-moisture' conditioning states, as
well as that of a 'normal-moisture' impact-resistant material, PA 6-HI
The PA 6-HI is the material with the best toughness properties in the tempera-
ture range examined. This behaviour is caused by energy dissipation processes in
the matrix, initiated by the rubber modification (pp. 243-256). The transition from
unstable to stable crack propagation is material-dependent. For uniform assess-
ment of the tested materials regarding the resistance against unstable crack propa-
gation, a transition temperature was identified. A conventional concept of a transi-
tion temperature, developed for steel with a space-centred cubic lattice, was
transferred to the dependence of the J values on temperature (Fig. B51). From
load-deflection diagrams showing linear elastic material behaviour, averaged JId
values were determined which represent the lower shelf of J Id (marked in Fig. B51
as (1», comparable with the low-temperature toughness. In this temperature range
polymers are characterized by a nearly temperature-independent (athermal) frac-
ture behaviour. The JId value calculated from a load-deflection diagram showing
unstable crack growth at the end represents the upper shelf of J Id for the tempera-
ture range examined (marked in Fig. B51 as (2», comparable with the high-
temperature toughness.
A brittle-to-tough transition temperature was determined by finding the average
of the JId values for the lower and upper shelf. The brittle-to-tough transition tem-
perature TBn expresses the safety against crack growth and is independent of the
specimen geometry, if the geometrical criterion is fulfilled. In this example, the
brittle-to-tough transition temperature TBn is 35 °c (Fig. B51). As expected, the
brittle-to-tough transition temperature shifts to higher temperature with decreasing
moisture content (Table BI2). In application, the impact-resistant PA 6-HI can
withstand stronger loads without any critical unstable crack growth than can the
'normal-moisture' and 'high-moisture' PA. The seeming toughness differences

between 'normal-moisture' and 'high-moisture' PA 6 at room temperature are
eliminated if the toughness parameters of the materials with the two moisture con-
tents are related to the transition temperatures (see Fig. BSO).
10.------------1
(2)
• Linear Elastic
8 0 Elastic-Plastic
( I) lower shelf of lJd values

(2) upper shelf of lId values
6
2 (I)
..............................I=*---~h~~:p;+
O~-----+------~------~-----+----~~----~
-200 -ISO -100 -50 0 TBIT50 100
T (0C)
Fig. B51. Identification of the brittle-to-tough transition temperature, using 'normal-moisture'

PA 6 as an example
Table BU. Toughness parameters ofPA 6 materials at TBrr
Conditioning state Transition temperature

(0C) JIdlTBTT
(N/mm)
Normal-moisture; high-impact to 10.6

High-moisture -22 4.2
Normal-moisture 35 4.5
Moisture-free > 60 >2.4
For a description of the resistance against crack propagation at temperatures

above the upper shelf of .h.d, the application of the crack resistance (R-) curve con-
cept is necessary because of the dominantly stable crack growth. The influence of
temperature on the stable crack propagation will be presented here using 'mois-
ture-free' PA 6 as an example. R-curves were determined at the lower and upper
shelf of J 1d • The experiments were very difficult, particularly in the temperature

range of the lower shelf of J1d because of the nearly linear elastic material behav-
iour with small-scale yielding. Furthermore, very small amounts of stable crack
growth appear which are verifiable at the crack surface in the form of the fracture
mirror length aBS. Thereby, extremely small amounts of stable crack growth arise
at the temperature of the lower shelf of J1d , as shown in Fig. B52a. With increasing
test temperature, the experimentally determined values of damax increase. How-
ever, they are so small that a determination of the technical crack initiation value at
the point da = 0.2 mm is not possible. Figure B52b shows the dependence of the
JTj and Tj values on the test temperature for the 'moisture-free' PA 6.
0.2 PA 6 Moisture-Free
a b 60
0.2
3
40
0.15 E
.§
E b 20 ~
E- 0.1 ~
-.
~ • )
20
oS
<I • •
10
0.0
•
0
-140 -100 -60 -20 20 60 100 -140 -100 -60 -20 20 60 100
TCOc)
Fig. B52. 'Moisture-free' PA 6: (a) dependence of the amount of stable crack growth ~amax on
the temperature, (b) influence of temperature on JTJ and TJ
Because of the small amount of stable crack growth, the quantification of the
crack resistance behaviour was carried out using the product JTj • The parameter
Tj , determined from a linear fit, describes the resistance against stable crack
propagation. The highest JTj values and the highest resistance against stable crack
propagation Tj were found at a temperature of -20°C. At this temperature, an in-
creased J1d value was also found (see Fig. B50). Above and below this temperature,
no fundamental changes in either of the two fracture mechanics parameters were
noticed. This phenomenon can be associated with the secondary relaxation of PA 6
[37,38] and the ~2 relaxation mechanism of 'moisture-free' PA [3], because damax
(see Fig. B52a), which characterizes the amount of stable crack propagation,
strongly increases at a test temperature of -20°C. The relaxation processes occur
at the temperature of -50°C with a characteristic load of approx. I-10Hz. Assuming
a characteristic frequency of 1 kHz in the ICIT, a shift of the secondary relaxation
process by 30 K to -20°C can take place. Thus, the high JTJ and TJ values result
from the particularly high mobility of the molecule's chain segments. The two
fracture surfaces in Fig. B53 document the different types of deformation behav-
iour.
Fig. B53. Fracture surfaces ofPA 6: (a) brittle and (b) ductile material behaviour
5 Summary
When fracture mechanics experiments of the type described above are performed,
the determination of reproducible and geometry-independent toughness parameters
can be guaranteed. Therefore, the investigations presented allow statements as to
the influence of temperature and moisture content on the toughness behaviour of
polyamide. Comparing investigations in which different procedures were used to
determine R-curves from the ICIT and the quasi-static fracture mechanics test
leads to the conclusion that the stop block method in combination with the multi-
ple-specimen method guarantees a reproducible determination of values.
The investigations of the influence of specimen thickness, using cast PA 6 as an
example, show that the standard specimen size of 80 x lOx 4 mm3 is suitable for
determining geometry-independent fracture mechanics parameters related to resis-
tance against unstable and stable crack growth under impact loading, using the J-
integral concept.
The influence of moisture content on the toughness behaviour of 'moisture-
free', 'normal-moisture' and 'high-moisture' PA 6 and cast PA 6 was characterized
by using the resistance against stable crack propagation. The most favourable
toughness properties were found for 'high-moisture' cast PA 6 using the quasi-
static fracture mechanics test and for 'high-moisture' PA using the ICIT.
Additionally, using PA 6 with various moisture contents as an example, evi-
dence was produced that the determination of a constant TJ value is a suitable
possibility for describing the resistance against stable crack propagation if the
relation between J and /).a is almost linear.
Investigations of the influence of the test temperature on the resistance against
unstable and stable crack growth of PA 6 were performed by using the quasi-static
fracture mechanics test and the ICIT at the 'moisture-free', 'normal-moisture' and
'high-moisture' moisture contents and on a PA 6-HI material.
With the help of the dependence of the -".d values on the temperature, material-
specific transition temperatures TBrr were determined which express the safety
against unstable crack propagation. With increasing moisture content, the transi-
tion temperature shifts to lower temperatures. The highest JTJ and TJ values were
found for 'moisture-free' PA 6 at a test temperature of T= -20°C. The maxima of
the toughness parameters that appeared are associated with the secondary relaxa-
tion process of PA 6. The results present provide evidence that for a characteriza-
tion of the influence of temperature and moisture content, the complex effect of
both of these factors has to taken into account.
References
I. Domininghaus H. (1998): Die Kunststoffe und ihre Eigenschaften. Springer, Berlin, Heidel-
berg: 601-716
2. Bottenbruch L., Binsack R. (1998): Polyamide. In: Becker G. W., Braun D. (Eds.) Kunst-
stoffhandbuch 3/4 Technische Thermoplaste. Carl Hanser, Munich, Vienna
3. Ning X., Ishida H. (1991): Dynamic mechanical analysis of RIM nylon-6. J. Polym. Sci.,
Polym. Phys. 29: 1479-1492
4. Speroni F., Castoldi E., Fabbri P., Casiraghi T. (1989): Mechanisms of energy dissipation
during impact in toughened polyamides: a SEM analysis. J. Mat. Sci. 24: 2165-2176
5. Hahn M. T., Hertzberg R. W., Manson J. A. (1986): Effects of rubbery phases and absorbed
water on impact-modified nylon 66. J. Mat. Sci. 21: 39-45
6. Huang D. D. (1989): Fracture toughness testing of tough materials. In: Riew C. K. (Ed.)
Rubber Plastics Toughened Advances in Chemistry. Series 222, Am. Chern. Soc., Washing-
ton: 119-133
7. Zhou Z., Landes J. D., Huang D. D. (1994): J-R-curve calculation with the normalization
method for toughened polymers. Polym. Engng. Sci. 34,2: 128-134
9. Chan W. Y. F., Williams J. G. (1991): Deformation of fracture toughness by the essential
work method and the J-integral method. In: Proceedings of the 8th International Conference
on Deformation, Yield and Fracture, Cambridge, UK: 74/1-74/4
10. Hashemi S., Williams 1. G. (1991): Single and multi-specimen R-curve methods for deter-
mination of toughened nylons. 1. Mat. Sci. 26: 621-630
II. Williams 1. G. (1990): Single specimen J testing of tough polymers. In: Proceedings of
Toughness Fracture and Fatigue of Polymers and Composites, Yamagata, Japan, October 8-
11: 146-153
12. Huang D. D., Williams 1. G. (1990): Comments on 'fracture toughness of impact modified
polymers based on the J-integral'. Polym. Engng. Sci. 30, 21: 1341-1344
13. Narisawa I., Takemori M. T. (1989): Fracture toughness of impact modified polymers based
on the J-integral. Polym. Engng. Sci. 29,10: 671-678
toughened polymers. In: Proceedings of the 7th International Conference on Fracture (ICF
7), Houston, Texas, USA, March 20-24: 2725-2732
15. Huang D. D. (1988): Experimental investigations of the J-integral. Polym. Preprints 29: 159
-160
16. Huang D. D., Williams J. G. (1987): J testing of toughened nylons. J. Mat. Sci. 22:
2503-2508
17. Saleemi A. S., Nairin J. A. (1990): The essential work of fracture technique as applied to
toughened nylon. Polym. Engng. Sci. 30: 211-213
18. Hashemi S., Williams J. G. (1986): The effects of specimen configuration and notch tip
radius on the fracture toughness of polymers using Jc• Plast. Rubb. Process. Appl. 6:
363-375
19. ISOIDIS 13586.2 (2000): Plastics - Determination of Fracture Toughness (Gc and Kc) -
20. ESIS P2-92 (1992): Procedure for Determining the Fracture Behaviour of Materials
sistance Curve Test on Plastics
22. Will P. (1994): R-curves of energy dissipative materials. J. Mat. Sci. 29: 2335-2340
ness characterization of high impact thermoplastics. In: Williams 1. G., Pavan A. (Eds.) Im-
pact and Dynamic Fracture of Polymers and Composites. ESIS Publication 19: 171-179
25. Doant G. (1988): Zorn EinfluB einer komplexen Beanspruchung auf die Kurz- und Lang-
zeiteigenschaften von Polyamid-6-Verbundwerkstoffen. In: Proceedings of the 12th Tagung
'Verstiirkte Plaste', Berlin, Germany, September 21-22: 7.6/1-7.6/6
26. DIN EN ISO 111 0 (1998): Kunststoffe - Polyamide - Beschleunigte Konditionierung von
Probekorpern
27. Langer B. (1998): Bruchrnechanische Bewertung von Polyamid-Werkstoffen. Logos, Berlin
28. Grellmann W., Che M. (1997): Assessment of temperature-dependent fracture behaviour
with different fracture mechanics concepts on example of unoriented and cold-rolled poly-
propylene. J. Appl. Polym. Sci. 66: 1237-1249
29. Che M., Grellmann W., Seidler S. (1997): Crack resistance behaviour of polyvinyl chloride.
J. Appl. Polym. Sci. 64: 1079-1090
30. Begley J. A., Landes J. O. (1973): A comparison of the J-integral fracture criterion with the
equivalent energy concept. In: ASTM STP 536: 246-263
31. Anderson T. L. (1995): Fracture Mechanics: Fundamentals and Application. CRC Press,
Boca Raton
32. Grellmann W., Seidler S. (1988): Determination of geometry independent fracture mechani-
cal parameters of polymers. In: Proceedings of the European Conference on Fracture, Buda-
pest, Hungary, September 19-24: Vol. 1: 517-519
ics values of polymers. Int. 1. Fract. 68: RI9-R22
34. ISO 179 (1993): Plastics - Determination of Charpy Impact Properties: Part 1: Impact
Strength
35. Mandell 1. F., Darwish A. Y., McGarry F. J. (1981): In: ASTM STP 734: 73ff
36. Karger-Kocsis J., Czigany T. (1993): Fracture behaviour of glass-fibre mat-reinforced

structural nylon RIM composites studied by microscopic and acoustic emission techniques.
J. Mat. Sci. 28: 2438-2448
37. Wu J., Mai Y., Cotterell B. (1993).: Fracture toughness and fracture mechanisms of
PBTIPCIIM blend. J. Mat. Sci. 28: 3373-3384
38. Lu M., Keskkula R., Paul D. R. (1994): Acrylic acid containing copolymers as reactive
compatibilizers for toughening nylon 6. Polym. Engng. Sci. 34, 1: 33-41
39. Carling M. J., Williams J. G. (1987):· On the applicability of fracture mechanics to short
fibre composites. In: Proceedings of the ICCM-VI and ECCM-2, London, UK, July 20-24:
3317-3326
40. Karger-Kocsis J. (1990): Dynamic fracture mechanics of a RIM polyamide block polymer
reinforced with glass strand. Mat. Polym. Bull. 24: 341-348
B2 Blends
B 2.1 Relationship Between Fracture Behaviour

and Morphology in PE/PP Blends
U. Niebergall, J. Bohse and H. Sturm, Berlin,
S. Seidler, Vienna, and W. Grellmann, Merseburg
1 Introduction
Inexpensive combined thermoplastics are of great interest for technological appli-

cation in the marketplace; in particular, blended materials of increased fracture
toughness combined with balanced stiffness and strength, as well as appropriate
processing and production properties, are highly desirable. At present this goal is
reached by methods such as copolymerization or modification with a suitable
elastomer.
Another possibility for improving the performance of thermoplastics is to mix
two components, with or without a compatibilizer (CP) [1]. However, to our know-
ledge, the mechanical properties of polyolefm blends found in the literature, espe-
cially the modulus, mostly follow the rule-of-mixture Voigt model and the mor-
phologies show coarse dispersion, but there are also antagonistic and synergistic
effects reported.
The production of blended materials with toughness and flow properties beyond
the mixing-rule level is highly desirable. To optimize the toughness properties, it is
necessary to know the relationship between fracture behaviour and morphology [2].
2 Experimental
Isotactic polypropylene (PP) and high-density polyethylene (PE-HD) were pro-

duced by Hoechst AG, Frankfurt, Germany. The PE-HD and PP were mixed, with
addition of a stabilizer and with addition of a stabilizer and a compatibilizer, in
different mixing ratios ranging from 5 wt.% PE-HD/95 wt.% PP to 95 wt.% PE-
HD/5 wt.% PP [3]. The blend samples were prepared by means of a single screw
extruder. Injection moulding of test specimens was performed according to
ISOIDIS 3167 Type A. Processing was carried out under typical conditions used
for PP. Ultra-thin sections of the injection-moulded test specimens of the pure
materials and of the blends were stained with RU04 and imaged by transmission
electron microscopy (TEM). The fracture surfaces were imaged by scanning elec-

230 U. Niebergall, J. 8ohse, H. Sturm, S. Seidler, W. Grellmann
tron microscopy (SEM). The morphology imaged by TEM was analysed by deter-
mination of the particle size and distance of the embedded phase, and determina-
tion of the specific interfacial area. The notched impact test was performed ac-
cording to ISO 179, the tensile test was performed according to DIN 54455 and
the Young's modulus was determined according to DIN 53457. For the fracture
mechanics measurements, an instrumented Charpy impact tester with 4 J work
capacity was used, and load-deflection (F-j) diagrams were recorded (pp. 71-86,
[4,5]). Single-edge-notched bend (SENB) specimens with dimensions 80 x 10 x 4
mm3 were used for this study.
These test conditions made it possible to determine geometry-independent
fracture mechanics parameters related to unstable crack propagation. The J values
were determined according to the evaluation method of Sumpter and Turner [6],
and the crack-opening displacement (COD) values were calculated on the basis of
the plastic-hinge model [7]. During the tensile test, the deformation behaviour was
investigated by acoustic-emission analysis. Combined with data extracted from
SEM images of the stretched test specimen, this gave important information about
mechanisms of toughness. As characteristic properties derived from the acoustic-
emission (AE) analysis, the rate and the energy of the AE signals were chosen.
3 Results
3.1 Electron Microscopy
The SEM images display a very good distribution ofthe two polymers in the blend
(Fig. B54). During crystallization, PP forms spherulites in the melt of PE-HD.
Growing spherulites enclose regions of the PE-HD melt or move them to the PP
spherulite boundaries. Finally, the PE-HD component crystallizes within these
regions. Measuring the size of the spherulites in the blend and in the pure materials
as reference states, we conclude that the PP spherulites are decreased in size by a
factor of 1.6 and the PE-HD spherulites by a factor of3 .
Fig. 854. SEM images of free surfaces of PP, not etched (a) and etched (b)
Relationship Between Fracture Behaviour and Morphology 231
We focused on the TEM images (Fig. B55) of the blend material 50PE-
HD/50PP, with and without compatibilizer, to obtain information about the mor-
phology since this specific content shows a significantly higher impact strength. PP
is the matrix, enclosing a dispersed phase of PE-HD. The blend system can be
described as a bicontinuous network. The addition of compatibilizer reduces the par-
Continuation of Fig. B54
01~ml~0kU 925E2 0~94/95 BAN6 22
Fig. B54. PE-HDIPP (c, d) and PE-HD (e, f). SEM images of free surfaces. not etched (left)
and etched (right)
Fig. B55. TEM images of ultra-thin sectioned blend specimens: 50PE-HD/50PP without
compatibilizer (a, b)
232 U. Niebergall, J. Bohse, H. Sturm, S. Seidler, W. Grellmann
Continuation of Fig. B55
f~~E
Fig. B55. TEM images of ultra-thin sectioned blend specimens: SOPE-HD/SOPP with com-
patibilizer (c, d)
ticle size in the dispersed phase by a factor of 1.7. The images taken at the highest
magnification show lamellae of the dispersed phase. The PE-HD phase consists of
thick lamellae; the PP lamellae are very short and thin. A very detailed report on
this topic can be found in [2,8,9].
3.2 Mechanical Properties
In the following, the effect of the morphology of the blend on the mechanical
properties is described [2].
1.4 • withoutCP
• with CP
MR
1.2
0.8
0.6 -I--.--.....-....,....-,..-----.----r--...,.....___.--~-_;
o 20 40 60 80 100
PE-HD Content (%)
Fig. B56. Dependence of Young's modulus on the PE-HD content for the blends with and
without compatibilizer (CP); the straight line displays the calculated curve accord-
ing to the rule of mixtures (MR)
The Young's modulus (Fig. B56) for a PE-HD content of up to 70 % is higher

than that calculated by the rule of mixtures (B13):
(BI3)
where E is the Young's modulus and tPthe content of each component. At high PP
contents a pronounced stiffening effect is found which is caused by interactions of
the structures ofPE-HD and PP.
The yield stress, as a function of the content of each component, varies within
the limits given by the pure materials. The influence of the compatibilizer is small.
The elongation at the yield point exhibits a small maximum. A similar effect, but
much more significant, is shown by the values of the impact strength (Fig. B57).
100
• withoutCP
• with CP
80
....E 60
~
15
111 40
20
0
0 20 40 60 80 100
PE-HD Content (%)
Fig. B57. Impact strength as a function of the PE-HD content for the blends with and without
compatibilizer (CP)
The maximum of the impact strength is found at a composition of 50PE-

HDI50PP plus compatibilizer. The impact strength is five times higher than that of
PE-HD and eight times higher than that ofPP. In the blend without compatibilizer,
the maximum is found for 60PE-HD/40PP. Both maxima correlate with the de-
pendence of the elongation at the yield point [2].
Additionally, fracture mechanics investigations were performed with the in-
strumented Charpy impact test to determine the unstable-crack-propagation be-
haviour (Fig. B58). PP behaves almost linearly; the crack propagation is unstable.
The fracture of PE-HD can be characterized as elastic-plastic behaviour possess-
ing a high percentage of stable crack propagation. The samples containing 40 % or
more PE-HD indicate elastic-plastic behaviour, and the crack propagation changes
at 50 % PE-HD content from dominantly unstable to dominantly stable.
234 U. Niebergall, J. Dohse, H. Sturm, S. Seidler, W. Grellmann
30PEnopp 50PEl50PP
60PE/40PP 80PEl20PP PE
Deflection I 1 mm
~I
5PEl95PP CP 30PEI70PP CP
70PEl30PP CP 80PE/20PP CP 95PEl5PP CP
Deflection 1 mm
H
Fig. D58. Load-deflection curves from instrumented Charpy impact test for the blended mate-
rials without (top) and with (bottom) compatibilizer
The compatibilizer causes an elastic-plastic fracture of the blend over the whole
content range; the crack propagation is dominantly stable.
The toughness behaviour of the blends is well described by the J 1d values
(Fig. B59) and the values of the bid, the crack-opening displacement (Fig. B60).
10
•
8
E 6
.E
z
I-
If) 4
'""" •
2 • without CP
• with CP
0
0 20 40 60 80 100
PE-HD Content (%)
Fig. B59. Dependence of JId on PE-HD content for the blends with and without compatibilizer
(CP)
The influence of the compatibilizer on the fracture toughness with respect to

initiation of unstable crack propagation (Figs. B59 and B60) is clearly much less
than that with respect to the impact strength.
Nevertheless, there is some effect caused by the compatibilizer; the material stores
energy, which can be detected from the comparatively large crack propagation en-
ergy (Fig. B61).
300
250 •
E • • • •
E
..,~ 200 • • ••
~
• • •
•
"C
00-
150
• • without CP
• with CP
100
0 20 40 60 80 100
PE-HD Content (%)
Fig. B60. Dependence of bid on PE-HD content for the blends with and without compatibilizer
(CP)
236 U. Niebergall, J. Dohse, H. Sturm, S. Seidler, W. Grellmann
1000 • without CP
• with CP
BOO
E 600
E
z
.......
400 •
<t
200
0 20 40 60 80 100
PE-HD Content (%)
Fig. D61. Crack propagation energy after initiation of unstable crack growth as a function of
the PE-HD content for blends with and without compatibilizer (CP)
3.3 Acoustic-Emission Analysis
The results presented above allow us to conclude that the most important property
of the blends investigated is the high toughness, especially regarding technological
application.
To identify the mechanisms causing the increased toughness, acoustic-emission
analysis [10] was performed. The questions mainly considered were, 'at which
stage of deformation does acoustic emission happen in the test specimen?', 'which
phase in the polymer blend is responsible?' and 'how can this mechanism be de-
scribed?'
For this purpose, tensile tests (100 mmlmin crosshead speed) combined with
acoustic-emission analysis were performed. Under these conditions, PE-HD and
PP emitted only a few signals (Fig. B62).
Pure PE-HD gave some signals beyond the yield point, e.g. in the oriented re-
gion (Fig. B62). It is known that in semicrystalline materials processes such as
orientation and fibrillation, splitting of fibrils and rupture of fibrils can occur.
Since acoustic emission is due to elastic recovery during the fracture process, the
measured signals are caused by fibrillation and fracture of fibrils. The primary
process detected by AE is splitting of fibrils.
Compared with the pure materials, the blends with and without compatibilizer
(Figs. B63 and B64) exhibit much more pronounced AE signals. The highest
acoustic-emission energy was detected in the blend with compatibilizer [2].
40 700 50
30
! a PP PE-HD
600
40
500 ~
f'
g
m 400~
~ 30 e;
Ql
0.. c:
620 ~
w
w
>
Z! 300~
<I:
~ w
20 j
<I: .!l!
:::J
C E
200 :::J
10
~
8
c 10 <I:
100
0
b
1,1 ~b 0 0
0 5 10 15 20 0 5 10 15 20 25
Time (5)
Fig. B62. Acoustic emission of PP and PE-HD (a, stress-time behaviour; b, AE activity; c,
accumulated AE energy)
40~--------------------------------------~
5OPE-HD/50PP
10
O+---~--~~~~~~~~L+LU~LU~hL~--~
o 5 10 15 20
TIme (5)
Fig. B63. Acoustic emission of the 50PE-HD/50PP blend without compatibilizer (a, stress-
time behaviour; b, AE activity; c, accumulated AE energy)
238 U. NiebergaIl, J. Bohse, H. Sturm, S. Seidler, W. GreIlmann
40- c 700 50
50PE-HD/50PP CP
b 600 (i)
30 a 40 l
500 .!!!
>-
m
a..
~ e>
e 400
~
30 ~
w
w
(/) 20
(/)
~
~ 300
(/)
w 20 ~
til
~ "3
200 E
10 :J
10 l::l
+~"
~
100
0
U L
I 0 0
0 5 10 15 20 25
Time (5)
Fig. B64. Acoustic emission of the 50PE-HD/50PP blend with compatibilizer (a, stress-time
behaviour; b, AE activity; c,accumulated AE energy)
Fig. B65. SEM images of surfaces of stretched tensile-test specimens of the pure materials PP
(a) and PE-HD (b), and the 50PE-HD/50PP blend with compatibilizer (c)
SEM images taken from stretched tensile-test specimens verify the results de-
scribed above (Fig. B65). The pure PP displays only very low elongation; no fib-
rillation is evident. The fibrillation in the pure PE-HD is slightly higher: single
fibrillation and a few fractured fibrils are recognized. In the 50PE-HD/50PP blend
with compatibilizer, strong elongation and fibrillation are observed by SEM. The
number of fractures of fibrils is strongly increased compared with the pure materi-
als.
Similar mechanisms were found for the dynamic fracture processes (Fig. B66).
Fracture surfaces of unnotched test specimens imaged by SEM demonstrated the
effect of the compatibilizer even under these extreme relaxation conditions (T = -
50 QC). Adding the compatibilizer improves the plastic-deformation performance
for both components in the blend. This process is considered as the most important
mechanism for the increase of the toughness [2].
Fig. B66. SEM images of fracture surfaces of PE-HDIPP blends: 60PE-HD/40PP (a) and
60PE-HD/40PP with compatibilizer (b)
3.4 Modelling of Toughness
The highest acoustic-emission energies were detected for the blend with compati-
bilizer. This was correlated with the observation of very pronounced fibrillation in
the material by SEM. In the model described below, a correlation between the
particle size and the impact strength was obtained.
To interpret the toughness quantitatively, the mechanisms described above were
modelled according to (B14) [11]:
acNBlend = AspeClnt 2rp G lclnt +tPM acNM* +tPl acN r* +tP; acNw' (B14)
'----v---' ~ '----v---"
interface matrix inclusion
where Aspec is the specific interfacial area, rp is the radius of the fracture process
zone, G(c is the energy release rate, <l\i is the matrix volume, ~ is the inclusion
volume, ~' is the inclusion volume which causes additional plastic deformation,
acN is the impact strength and * denotes the modified phase.
The dimension of the fracture process zone rp was calculated from the critical
crack-opening displacement bid, and the strain at rupture of the fibrils was ap-
proximated from the acoustic-emission results.
Table B13. Characteristic values of parameters used for modelling of impact toughness of
SOPE-HD/SOPP with and without compatibilizer
Parameter SOPE/SOPP without CP SOPE/SOPP with CP
tPpp
0.5 O.S
tPPE O.S O.S
,
tPPE O.S 0.5
Aspeeln! (mrnlmm 2)
2466 3682
2rp (mm) 0.026 0.034
GIClnt (N/mm)
0.047 O.4SS
aeN pp ' (kJ/m2)
11 26
aeN • (kJ/m 2)
PE 18 23
affiodel
eN blend (kJ/m 2) 23 94
a experiment
eN blend
(kJ/m 2) 2S 98
The specific interfacial area Aspee was determined by computerized image analy-
sis of the TEM images (Fig. B55). In principle, this value changes as a function of
the composition of the blend. G1c was determined by the double-cantilever beam
(DeB) test. By this experiment, the order of magnitude of the adhesion in the
blend, with and without compatibilizer, was estimated. The energy release rate at
the interphase increases by a factor of 10 in the presence of the compatibilizer. As
shown in Table B 13, ~. is equal to ~. According to the model, the energy of the
observed maximum of the impact toughness consists of three contributions: 60.3 %
due to the interphase, 13.8 % due to the matrix and 25.9 % due to the effects of the
inclusions [2].
Figure B67 presents the calculated and experimental values. According to the
model, the maximum of toughness is caused mainly by two effects: interphase and
embedding effects. The matrix component has a minimum in this range of compo-
sition. Although (BI4) is just a model, the agreement between experiment and
calculation is quite good.
• aeN without CP
100 • • a eN with CP
--- without CP/Model
- with CP/Model
80
•
r;--
~
60
cz
""')
III" 40 •
20 ... _._--.---.
-- ---
. ----------
• •
.- ~-II-.-
O+---~-.--~--.---~-.--~--.---~~
o 20 40 60 80 100
PE-HD Content (%)
Fig. B67. Comparison of modelled and experimental impact toughness
4 Conclusion
The investigations described above show that the compatibilizer strongly affects
the toughness of the polyolefm blends analysed. There are two main reasons for
the high impact toughness:
1. The high phase dispersion and the compatibilizer cause strong interfacial
interaction. The results are an increased impact toughness of both phases
and a large amount of debonding energy.
2. The embedding of small inclusions initiates an additional fibrillation pro-
cess in the surrounding matrix, which dissipates impact energy.
Acknowledgement
We are grateful to Dr. Frechen and Dr. Heckmann from BASF AG, Ludwigshafen,
Germany for providing the TEM images.
References
1. Teh J. W., Rudin A., Keung J. C. (1994): A review of polyethylene-polypropylene blends

and their compatibilization. Adv. Polym. Techno!. 13: 1-23
2. Niebergall U. (1998): Morphology and Fracture Behavior of PolyethyleneIPolypropylene
Blends. PhD thesis
3. Niebergall U., Bohse J., Grellmann W., Schiirmann B. L. (1997): Very Tough Polyolefin
Blends. German Patent DE 19619625, European Patent EP 0807663A2, South Africa Patent
974152, Japan Patent 125526/97, Brazil Patent PI 9703160-7, US Patent 6.028.145
(22.02.2000)
4. Kobayashi T. (1990): Recent Advances in Instrumented Charpy Impact Test. Laboratory for
Evaluation and Failure Prevention of Materials, Department of Production Systems Engi-
neering, Toyohashi, Japan
Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Series No.3. Insitut fUr
Mechanik, Berlin, Chemnitz: 102-111
6. Sumpter J. G. D., Turner C. E. (1976): Cracks and fracture. In: ASTM STP 601: 3-18
Charpy impact test. Fracture Mechanics, Micromechanics, Coupled Fields (FMC) Series 37.
Insitut fUr Mechanik, Berlin, Chemnitz: 186--192
8. Niebergall U., Schiirmann B. L. (1996): Morphology and fracture toughness of PEIPP
blends. In: PPS-12, Sorrento, Italy, May 27-31: 255-257
9. Schiirmann B. L., Niebergall U., Severin N., Burger C., Stocker W., Rabe 1. P. (1998):
Polyethylene (HDPE)/polypropylen (iPP) blends: mechanical properties, structure and mor-
phology. Polymer 39: 5283-5291
10. Bohse J., Kroh G. (1992): Micromechanics and acoustic emission analysis of the failure
process of thermoplastic composites. J. Mat. Sci. 27: 298-306
11. Bohse J., Grellmann W., Seidler S. (1991): Micromechanical interpretation of fracture
toughness of particulate-filled thermoplastics. J. Mat. Sci. 26: 6715-6721
B 2.2 Influence of Modifier Content
and Temperature
I. Bethge, K. Reincke, S. Seidler
and W. Grellmann, Merseburg
1 Introduction
The modification of polymers for improvement of properties is becoming in-

creasingly important. In particular, polyamide has a wide application field as a
construction polymer material. Because of the great variety of possibilities for
modification for improvement of the mechanical properties, polyamide is a suit-
able material for made-to-measure solutions to polymers.
The investigations of materials presented below were performed to optimize the
toughness behaviour by varying the kind of modifier and the modifier content. For
this purpose, dynamic test methods were used. The main focus of the investiga-
tions was the determination of fracture mechanics parameters related to resistance
against unstable crack propagation in order to determine the brittle-to-tough tran-
sition (BIT) temperature, and the determination of the resistance against stable
crack propagation in order to characterize the energy dissipation capacity of the
blends as a function of temperature, the impact modifier used and its content in the
matrix material, polyamide 6 (PA 6).
2 Experimental
2.1 Materials
The investigations were performed on PA 6 and impact-modified PA 6; the fol-

lowing impact modifiers were used:
• ethylene-vinyl acetate copolymer (EVAC)
• ethylene-propylene copolymer (EPM)
• butyl-acrylate copolymer (BA)
Modifier contents of 5, 10,20,25 and 30 wt.% were used. Injection-moulded
standard test specimens with the dimensions length L = 80 mm, width W = 10 mm
and thickness B = 4 mm (according to ISO 179, type 1) in the 'normal-moisture'
conditioning state were used (see B 1.5, pp. 209-228).

244 I. Bethge, K. Reincke, S. Seidler, W. Grellmann
2.2 Investigation of the Material Resistance

Against Unstable Crack Propagation
The instrumented Charpy impact test (ICIT) was carried out using a Fractovis
falling-weight tester made by CEAST, Italy (Fig. B68) .
.t. F Acceleration Un~
1:
, Falling
i I Weight
Strain G.!':'.~e
Spedmenfor
:=;t;:;:;===€p~~ --- Charpy Test =
Support
Instrumented Falling-
Weigh! Impact Tester
"-.
~!5I_I_I~
-----
Cross Head
Temperature Chamber
(-70 ' C to 150 'C)
Fig. B68. Set-up of the Fractovis instrumented falling-weight tester for perfonning instru-
mented Charpy impact tests
Instrumented Charpy impact tests can be performed using either a pendulum

impact tester or a falling-weight tester because the procedure is in principle the
same for both experimental methods; both methods use a Charpy falling weight
and a support. The difference is in the different paths of the impact hammer. Be-
sides the load, another directly measured value is the time to fracture. For the
examinations described here, the following test conditions were used:
• initial crack length a = 2 mm, i.e. a/W= 0.2; razor blade notch
• support span s = 40 mm, i.e. s/W = 4
• test speed VH = 1 mls.
The assessment of the toughness behaviour was performed on the basis of the
J-integral concept, and the evaluation method of Sumpter and Turner was used to
determine J values (pp. 71-86).
Influence of Modifier Content and Temperature 245
2.3 Investigation of the Material Resistance

Against Stable Crack Propagation
For the determination of R-curves under impact loading, an instrumented Charpy

impact tester with 4 J work capacity at the maximum drop height was used. The
R-curves were determined at room temperature and at 80°C by applying the stop
block method (pp. 71-86, [I)). A test temperature of 80°C was selected because
in this range all materials are above the BIT temperature. A possible method of
including stable crack propagation in the analysis of the R-curve and of quantify-
ing energy dissipation processes is provided by the JTJ concept of Will et al. [2],
which was used here for evaluation of the R-curves. Additionally, JTJ values were
determined according to the equations described in A 2.2 (pp. 71-86). The fol-
lowing test conditions were used:
• initial crack length a = 4.5 mm, i.e. alW= 0.45
• support span s = 40 mm, i.e. slW = 4 mm
• test speed VH = 0.76 mls.
3 Results
3.1 Temperature Dependence of Toughness Behaviour
The fracture mechanics evaluation of the temperature dependence of toughness is

of a special significance when applications are considered [3,4]. The BIT tem-
perature is a decisive target parameter in material development.
In [5] the basic process for evaluation of the temperature dependence of the
toughness behaviour is explained in detail, using PA 6 as an example. In that
work, the material was examined in a temperature range between -30°C and
+80 °C. As a result of these investigations, J values were determined which are
nearly constant at 2 N/mm between -30°C and +20 0c. From +40°C there is a
noticeable increase in the J values, up to 14 N/mm at +80 °C. The changes in the
temperature dependence of the J-integral correlate with the relaxation behaviour
ofPA6.
The increase in the J values at high temperatures is associated with reaching
and exceeding the glass transition temperature, for which a value of +60 °C is
quoted in the literature [6]. The load--deflection diagrams described in [5] show
linear elastic material behaviour; this means that PA 6 is brittle at temperatures
lower than 20°C. With increasing temperature, the material behaviour changes to
elastic-plastic. However, it must be taken into account that the shape of a load-
deflection diagram depends on the specimen geometry and arrangement. Further-
more, [5] describes the transfer of a conventional transition temperature concept,
known from metallic materials, to the temperature dependence of the toughness.
For the lower shelf of toughness, the mean of those J values which were evaluated
from linear elastic load-deflection diagrams was determined. On the other hand,
for the determination of the upper shelf, that value at which the last unstable crack
propagation occurred at that particular temperature was determined. The BTT
temperature TBTT was determined from half of the difference between the lower
and the upper shelf; this describes a mean temperature for the transition from
unstable to stable crack propagation. So, for all materials tested a uniform evalua-
tion criterion for calculation of the BIT temperature was used.
Figure B69 shows the influence of temperature and modifier content on the
toughness of PA 6IBA blends in comparison with unmodified PA 6. With in-
creasing modifier content, the JdST values for the lower shelf of toughness are
insignificantly increased, but they are considerably enhanced if the material be-
haviour becomes elastic-plastic. Besides the increase of the J values, the brittle-
to-tough transition is displaced to lower temperatures with increasing modifier
content (Fig. B70).
30
25 wt.%
20
-
E
E
~
....,
~"'C
15
10
0
-60 -40 -20 o 20 40 60 80
Temperature (OC)
Fig. B69. Temperature dependence of toughness of the PA 6IBA blends
For PA 6IEPM blends, the JiT values are increased with increasing modifier
content of up to 25 wt.% in the whole temperature range investigated. The BIT
temperature (Fig. B70) shifts to lower temperatures with up to 20 wt. % EPM. It
was impossible to determine BIT temperatures for the blends with 25 wt.% and
30 wt.% EPM because there was no lower or upper shelf of toughness.
Summarizing, it may be pointed out that the PA 6IEPM and PA 6/BA blends
with modifier contents higher than 10 wt. % are qualified for low-temperature
applications because their BIT temperatures are displaced to values lower than
o0c.
80
60
PA6/EVAC
6' 40
L- /
~~ 20 PA6/EPM
/
0
-20 PA6/SA
0 5 10 15 20 25 30
Modifier Content (wt.%)
Fig. B70. Brittle-to-tough transition temperatures TOTI of the PA 6 blends investigated
3.2 Influence of Load and Deflection Behaviour

on the Temperature Dependence of the J Values
In Fig. B71a the temperature dependence of Jl T is shown for the example of the
PA 6IEPM blend with 5 wt. % EPM, where a continuous increase of the values
with increasing temperature is visible. The maximum load and deflection
(Fig. B71 b) show different characteristic behaviours as a function of temperature.
After passing through a maximum, in the temperature range from 10 to 70°C, the
maximum load decreases by 28 %, and the deflection increases by a factor of 3.5.
This means that the deformation mechanisms are both load- and deflection-
controlled for T'5, 10°C. Above T = 10°C the increase in the J values is exclusively
deflection-controlled. Obviously, a description of the temperature dependence of
toughness with the LEFM concept does not lead to material-equivalent statements
[3].
3.3 Morphology
Morphological parameters such as particle size and particle distance were meas-
ured by examining cryo-fracture surfaces of undeformed injection-moulded test
specimens. These surfaces were coated with gold and scanning electron micro-
scope (SEM) micrographs were taken.
Because of the poor particle-matrix adhesion in PA 6IEVAC blends, the aver-
age particle size and particle distance could be directly obtained from SEM micro-
graphs by an image analysis system, averaging over more than 1000 particles.
The EPM was grafted with 0.7 wt.% maleic anhydride (EPM-g-MAH), and the
amine end groups of the PA 6 phase are able to react with the maleic anhydride
groups in the rubber phase. The number of reacting amine end groups influences
25 a
20
15
E
~ 10 •
..,
~
Iii"
0
-20 -10 0 10 20 30 40 50 60 70
Temperature (OC)
4
b
300
280 3
260
z 240 •
~
1
• 21
11.. _I
220
200
180 0
-20 -10 0 10 20 30 40 50 60 70
Temperature (OC)
Fig.B71. JlT values (a), and maximum load and maximum deflection (b) of the PA 6IEPM
blend with 5 wt. % modifier as a function of temperature
the size, shape and size distribution of the rubber phase formed during blending
[7]. Because of the very good particle-matrix adhesion no particles were visible
on the cryo-fracture surfaces; because of this, the surfaces were etched in boiling
toluene for I h to dissolve the EPM particles.
Butyl-acrylate is a core-shell elastomer with a basis of acrylate. The particle-
matrix adhesion is very good too, and therefore the cryo-fracture surfaces of the
PA 6IBA blends were etched in acetone for 112 h to dissolve the rubber particles.
The SEM micrographs of these specimens were analysed by image analysis too.
The average and maximum particle sizes and the particle distances of all materials
investigated are summarized in Table B14.
Figure B72 illustrates the morphological parameters as a function of modifier

content. Because of agglomeration of EV AC particles, both particle size and
distance increase in the range from 5 to 25 wt.% EVAC.
For the PA 6/EPM and the PA 6IBA blends it can be said that the elastomer
particles are well dispersed. The particle size (Fig. B72a) is independent of the BA
content, with a value of about 0.2 11m, whereas for the PA 6IEPM blends d gradu-
ally decreases for BA contents between 5 and 30 wt. %. It can be observed that the
particle distance (Fig. B72b) for the PA 6IEPM and the P A 6/BA blends rapidly
decreases in the range from 5 to 20 wt. % modifier, whereas Ad only gradually de-
creases between 20 and 30 wt.% modifier.
2.5
a
2.0 ~y--------.
.---.
1.5 .________ PA 6/EVAC
E
2::
.-r.---.
"0 1.0 / PA6/EPM
0.5 - / PA6/SA
... ... ... ... ...

0.0
5 10 15 20 25 30
5
b
~
.---y----.
4
E 3
~
• .___________
'" PA 6/EVAC
2::
~ ~.PA~EPM
------- ------.
4:." 2
/ /
... ...•
PA6/SA
... ...
0
5 10 15 20 25 30
Fig. B72. Average particle size (a) and average interparticle centre distance (b) as a function
of modifier content for blends investigated
Table B14. Morphological parameters of blends investigated
Modifier content Average particle size Maximum particle Average interparti-

(wt.%) d size dmax cle centre distance
(/lm) (/lm) Ad (/lm)
PA6IEVAC
5 1.02 3.51 2.63
lO 1.43 4.03 2.76
20 1.65 5.27 3.45
25 2.17 10.70 3.67
30 1.99 11.50 3.31
PA6IEPM
5 0.72 1.68 3.90
10 0.58 1.90 2.51
20 0.47 2.57 1.34
25 0.51 2.68 0.98
30 0.39 2.24 0.88
PA6IBA
5 0.l6 0.26 2.l9
lO 0.20 0.28 1.04
20 0.20 0.30 0.59
25 0.20 0.31 0.59
30 0.20 0.29 0.50
3.4 Determination of R-curves and Correlations

Between Morphological Parameters and Toughness Values
The influence of the modifier content on the load-deflection behaviour at room

temperature is shown in Fig. B73 for the example of the PA 6IBA blends. With
increasing modifier content the crack propagation behaviour changes from unsta-
ble to stable; this is connected with an increase in the plastic part of the general
deformation energy. A modifier content of 20 wt.% leads to the occurrence of a
crack propagation energy that characterizes a decrease in crack propagation speed
[8,9]. For this reason, the determination of R-curves is necessary for assessing
these blends, which show elastic-plastic material behaviour and dominantly stable
crack propagation.
The investigation of R-curves under dynamic loading for the blends with 5 and
10 wt. % modifier at room temperature was carried out with the aim of assessing
the toughness reserve (the stable part of the general deformation energy) because a
small modifier content is of economic significance. At a temperature of 80°C all
blends are above the transition temperature and show dominantly stable crack
propagation. Therefore, at this test temperature the recording ofR-curves is neces-
sary for a fracture mechanics evaluation. The R-curves of the PA 6IEVAC,
PA 6IEPM and PA 6IBA blends show generally the same behaviour as a function
of modifier content, as can be seen in Fig. B74 for the example of the PA 6IEPM
blends at room temperature. There is a significant increase of the slopes ofthe R-
curves up to a modifier content of 20 wt. %.
PA6 5wt.%BA
slI :/1 Ael ~I

0.5m!'VV ,~ 0.5mm
t------! f-----i
10 wt.% BA 20wt.%BA
~
LO
I
0.5 mm 0.5 mm
t--------l 1--.'.--1
Fig. 873. Dependence of the load-deflection behaviour of PA 6IBA blends on modifier contel
20.---------------------------------~
20wt.%
•
15
O~~~----._------_.--------,_------~
0.0 0.1 0.2 0.3 0.4
t.a (mm)
Fig. 874. Comparison of R-curves of PA 6IEPM blends with different modifier contents at
room temperature
A quantitative comparison of R-curves for these materials and a distinction

between the effects of the various modifiers regarding the energy dissipation ca-
pacity is only practicable on the basis of the JTJ concept [2]. The energy dissipa-
tion capacity of the materials examined is significantly improved with increasing

modifier content (Table B 15).
For all blends, the energy dissipation capacity is optimal for a modifier content
of20 wt.%, independent of the kind of modifier. This correlates with statements in
[10], where an optimal impact-resistance modifier content of 20 wt.% is men-
tioned. In relation to TBn. only the PA 6IEPM and PA 6/BA blends with this
modifier content are comparable, because they are above TBTT at room tempera-
ture.
All blends with 5 wt. % modifier are comparable too, because they are below
TBTT . In comparison with the unmodified PA 6, the JTJ values of the PA 6IBA
blend are doubled, those of the PA 6IEVAC blend are increased by fivefold and
those of the PA 6IEPM blend are increased by sevenfold.
Table B15. JTJvalues of the blends investigated at room temperature and 80 °C
Modifier content (wt.%) JTJ (N/mm)

T=23°C T= 80 °C
PA6 0 26 268
PA6IEVAC 5 119 399
10 261 520
20 447 793
25 203 546
PA6IEPM 5 182 556
10 424 806
20 720 756
25 310 609
PA6IBA 5 59 532
10 480 976
20 558 856
25 535 780
In contrast to the PA 6/EVAC blends - at room temperature these are below

TBTT - the increase of the JTJ values for the PA 6/EPM and PA 6IBA blends is
attributed to differences in crack propagation behaviour as a result of testing be-
low and above TBTT . As the particle size of the PA 6/EPM blends is nearly con-
stant at approximately 0.5 )lm, it is assumed that the decrease of the JTJ values at
25 wt. % EPM can be attributed to the existence of a critical particle distance. The
JTJ values of the blends with 20 and 25 wt. % BA are nearly equal, which corre-
lates with a constant particle size and particle distance (Table B14, Fig. B72a, b).
According to [7,11], an effective toughness increase of impact-modified polyam-
ide is only possible with a particle size between 0.2 11m and 111m. Therefore, an
optimization of the morphological parameters of the PA 6/EVAC blends (constant
particle size below 111m, improvement of particle-matrix adhesion) could result
in higher toughness.
Fig. 87S. Stable crack growth and damage area at RT for PA 6 (top left), PA 6IEVAC blend
(bottom left), PA 6IEPM blend (top right) and PA 6IBA blend (bottom right); all
specimens with 5 wt.% modifier content
At 80°C all materials are above the BTT temperature and above the glass tran-
sition temperature of the matrix material, which correlates with a significant in-
crease in the energy dissipation capacity. A maximum of the energy dissipation
capacity for the PA 6IEVAC blends is found at 20 wt.% modifier.
For the PA 6IEPM and PA 6IBA blends, the highest JTJ values were found at a
modifier content of 10 wt. %, where the energy dissipation capacity of the
PA 6IBA blend is distinctly higher than that of the PA 6IEPM blend.
A micro-fractographic fracture surface analysis was carried out to investigate
the active toughness mechanisms. Regarding toughness-increasing mechanisms,
the particle distribution, the particle size and distance, the particle-matrix adhe-
sion and the Young's modulus of the impact modifier are of decisive significance.
The toughness mechanism in modified polyamides is essentially based on cavita-
tion in or around the modifier particles [12]. Thereby, triaxial stress states are
relieved and an intense shear deformation of the matrix bridges between the cavi-
ties is possible.
Scanning electron microscope investigations of the PA 6IEVAC and PA 6IEPM
blends show cavities and stretching of matrix bridges in the crack propagation
direction (Fig. B75). The PA 6IBA blend also shows these toughness mechanisms,
but here the structure is finer. Additionally, at 80°C stronger cross structures
occur in the area of stable crack propagation. At this temperature, in all blends, the
toughness mechanisms mentioned above appear and are initiated by the incorpo-
rated modifier particles. There are also shear deformation mechanisms in the ma-
trix which significantly contribute to the increase in the energy dissipation capac-
ity. The differences in the energy dissipation capacity of the blends with various
impact modifiers cause differences in the matrix stretching and crack-tip forma-
tion in the area of stable crack propagation.
3.5 Calculation of R-curves Using a Hinge Model Test Specimen
An R-curve procedure (pp. 133-140, [13,14]) developed at the Institute of Materi-

als Science, Merseburg, allows the investigation of important material parameters
related to the rigidity, strength and toughness behaviour using a single-specimen
method. In this procedure, one load-deflection diagram, the specimen dimensions
and the initial crack length serve as the starting point.
50
PA6/BA
E
E
..... 40
~
8
c: 30
ca
U)
'iii
Q)
c::: 20
15
!!!
()
10 PA6/EVAC
0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Crack Growth (mm)
Fig. B76. R-curves of different modified PA 6 materials with 25 wt.% modifier content in all
cases, calculated with the 'automatic R-curve procedure' from load-time curves
Comparisons of this method using experimental results and results from a fi-
nite-element simulation method confirm the suitability of this procedure for the
determination of material parameters [13]. Comparisons with R-curves determined
using the multiple-specimen method show quantitative differences, but confirm
the validity of R-curves obtained in such a manner qualitatively. The present sin-
gle-specimen method allows material optimization. Figure B76 shows the R-
curves of dynamically loaded specimens (VH= 0.76 mis, RT) of different modified
PA 6 materials. The R-curves, calculated from one load-time curve in all cases,
allow the estimation of toughness values related to impact loading. These values
correspond qualitatively with the results obtained using the mUltiple-specimen
method.
4 Summary
Depending on the kind and content of the impact modifier, the toughness modifi-
cation of PA 6 materials leads to an improvement in the crack toughness related to
resistance against unstable crack propagation and to an increase in the energy
dissipation capacity in the temperature range investigated. A description of the
temperature dependence with the LEFM concept results in an underevaluation of
the toughness behaviour.
The BIT temperatures determined by using fracture mechanics methods shift
to lower temperatures for all modifiers, but the shift is more pronounced for the
PA 6IBA and PA 6IEPM blends. Materials with modifier contents from 10 wt.%
upwards are especially well qualified for low-temperature applications.
The crack toughness related to stable crack propagation show no changes in the
deformation mechanism with temperature. An optimal energy dissipation capacity
is found for an impact modifier content of 20 wt.%. For the PA 6IEPM and
PA 6/BA blends the reason for this is, apart from good particle-matrix adhesion, a
nearly constant particle size of less than 1 Ilm, and the fact that the testing tem-
perature (room temperature) is above the BIT temperature. Taking account of
economic considerations, a modifier content of 10 wt. % seems to be sufficient for
toughness improvement for a large number oftechnological applications.
Acknowledgements
The authors would like to thank LEUNA-Miramid GmbH, Leuna, Germany, espe-
cially Mr. H. Schumacher and Mr. G. Doant, for providing the materials, and the
Ministry of Culture and Education of Sachs en-Anhalt, Germany for fmancial
support within the scope of the project 'High dynamic material behaviour of poly-
amide - fracture behaviour and morphology for stable crack propagation'.
References
1. Seidler S., Grellmann W. (1995): Application of the instrumented charpy impact test to the
Fract. Mech. 37, 2: 275-281
3. Grellmann W., Che M. (1997): Assessment oftemperature-dependent fracture behavior with
4. Grellmann W., Lach R. (1996): Influence of temperature and toughness, fracture surface
morphology and molecular relaxation behaviour of polycarbonate. Appl. Macromol. Chem.
Phys.237: 191-208
5. Zwanzig I., Seidler S., Grellmann W. (1996): Zllhigkeitseigenschaften von modifizierten

Polyamiden. In: Proceedings ofWerkstoffwoche 1996, Symposium 6: Werkstoff- und Ver-
fahrenstechnik, Stuttgart, Germany, May 28-31: 257-262
6. Fedors R. F. (1979): Glass transition temperature and critical properties. Polymer 20:
1255-1258
7. Oshinski A. J., Kekkula H., Paul D. R. (1996): The role of matrix molecular weight in rubber
toughened nylon 6 blends: 1. Morphology. Polymer 37, 22: 4891-4907
8. Grellmann W., Lauke B., Seidler S. (1991): Application of J-integral concept for the de-
scription of toughness properties of fibre reinforced polyethylene thermoplastics. Polym.
Compos. 12,5:320-326
9. Grellmann W., Seidler S., Bohse 1. (1991): Toughness and morphology of thermoplas-
tic/particle composites. Kunstst. - Germ. Plast. 81, 2: 29-32
10. Sunderland P., Kausch H. H., Schmid E., Arber W. (1988): The application of fracture
mechanics to the impact behaviour of rubber-toughened polyamides. Macromol. Chern.
Macromol. Symp. 16: 365-378
II. Borggreve R. J. M., Gaymans R. J., Schuijer J. (1989): Impact behaviour of nylon-rubber
blends: 5. Influence of the mechanical properties of the elastomer. Polymer 30: 71-76
12. Michler G. H. (1992): Kunststoff-Mikromechanik - Morphologie, Deformations- und Bruch-
der Basis des Gelenkpriltkorpers. Fortschritt-Berichte VOl-Series 18: MechaniklBruchme-
chanik No. 208, VOl, Dusseldorf
14. Steiner R. (1997): Verfahren und Anordnung zur automatischen Ermittlung von J-R-Kurven
sowie Zllhigkeits-, Festigkeits- und Steifigkeitskennwerten unter dynamischer Bean-
spruchung von SENB-Priltktlrpern. German Patent DE 19635968
B 2.3 Morphology and Toughness
of PP/EPR Blends
T. Koch, S. Seidler, K. Jung and W. Grellmann, Merseburg
1 Introduction
Polypropylene (PP) is a semicrystalline polymer with a wide range of applications,

mainly standard applications but also in the field of construction. The glass transi-
tion temperature of about 4°C limits the use of PP. Compounding with a soft
elastomeric phase of a much lower glass transition temperature provides a possi-
bility of expanding the application range of PP to temperatures below 0 °C. The
mechanical properties of such blends depend on the structure of the matrix mate-
rial, on the type, diameter and volume content of the dispersed phase, and on the
interaction between the phases. Toughness optimization can only be performed
with knowledge of the quantitative correlations between morphology and tough-
ness.
The toughness behaviour was characterized by means of the instrumented
Charpy impact test and fracture mechanics methods. Owing to the wide range of
toughness behaviour of the materials investigated the J-integral concept was used.
The investigations covered the evaluation of J values related to resistance against
unstable crack growth and J values related to resistance against stable crack
growth.
2 Experimental
The investigations were performed on blends of PP and ethylene-propylene rub-

ber (EPR) (0, 1, 5, 15,20, 30, 45 and 50 wt.% EPR) provided by Borealis AG,
Linz, Austria. The blends were prepared by mixing PP homopolymer KS 10 and
EPR Dutral CO 038 in a Collin extruder.
The dynamic mechanical behaviour was measured with a Rheometric Scientific
DMfA MKIII instrument using a torsional device, a frequency of 1 Hz and a heat-
ing rate of 2 Klmin.
To characterize the spherulitic structure, thin sections (thickness 10 J.1m) were
cut with a microtome and subsequently observed in a polarizing microscope.
Wide-angle X-ray-scattering (WAXS) experiments were performed with a
system made by Rich. Seifert & Co. using CuK.. radiation and an accelerating
voltage of 40 kV.
In addition to WAXS, the degree of crystallinity was determined by differential
scanning calorimetry (DSC) using a Perking Elmer DSC-2C instrument at 10
Klmin.

258 T. Koch, S. Seidler, K. Jung, W. Grellmann
Specimens were brittle-fractured to observe particle morphology. The diameter,

shape and distribution were measured with a Quantimet 570 image analysis sys-
tem.
An instrumented Charpy impact test, with separate electronic registration of
load and deflection during the impact process, was used to characterize the tough-
ness behaviour. The measurements were performed using a 4 J pendulum. The
arrangement of the device is described in [1-3]. The procedure described in A 2.2
(pp. 71-86) was the basis of the fracture mechanics approach used. A test tem-
perature of -20°C was obtained with a special device [4].
The materials showing dominantly stable crack growth were characterized by
the R-curve method. Stable cracks were generated using the stop block method,
not with the help of steel barriers [5,6], but using a special pendulum-catching
device [7]. The procedure described in A 2.2 (pp. 71-86) was also used for deter-
mining fracture mechanics parameters.
3 Results
3.1 Morphology
The dependence of the mechanical loss factor tan 0 on temperature is illustrated in

Fig. B77. For all compositions, the maxima corresponding to the glass transition
temperatures are identical to those of the pure components, indicating a two-phase
system. The locations of the glass transition temperatures of PP and EPR are al-
most unaffected by the blend composition.
Generally, the toughness behaviour of PP blends with low elastomer content is
expected to be dominated by changes in matrix structure caused by the addition of
EPR. That is why the spherulitic structure and the degree of crystallinity of the
matrix were investigated here.
The injection-moulded specimens show a typical skin-core morphology with
an increasing size of spherulites from skin to core. Owing to the processing, there
is a skin approximately 10 J.1m thick. The addition of EPR acts as a nucleating
agent. Increasing the rubber content leads to a decrease in spherulite size and an
increase in spherulite number per unit volume.
The W AXS and DSC measurements were intended to clarify whether the ma-
trix crystallinity was influenced by the elastomeric phase. In Fig. B78 one can see
the wide-angle X-ray diffraction spectra of the pure components and of some
blends. The homopolymer KS 10 shows maxima typical of the a form of PP; the
EPR shows a wide amorphous halo. The amorphous halo increases in height and
the heights of the crystalline peaks decrease with increasing rubber content. The
specimens are oriented, as can be seen in the different ratios of the (040) and (110)
peak heights. The half widths of the reflection peaks are not clearly different,
indicating that the crystallite size is not influenced by EPR content. The degree of
crystallinity does not show any dependence on rubber content. It was not clarified
Morphology and Toughness ofPPIEPR Blends 259
whether the small apparent differences come from the material or from an inaccu-
racy of the evaluation method, which is based on [8]. It is not possible to sayany-
thing about the influence on the orientation.
0.10
0.05
- - 100/0
................. 9515
- .. - ..- ..- 85115
-.-.-. 70/30
----. 50/50
0.01 .....--.---,--...--,---..--,...---.---r-----r-.....
-100 -50 o 50 100
Temperature (OC)
Fig. B77. Mechanical loss factor tan oversus temperature for some PPIEPR blends
100/0
95/5
~
·iii
c 85/15
.!!
.=
70/30
50/50
0/100
5 6 7 8 9 10 11 12 13 14 15
e (0)
Fig. B78. X-ray diffraction patterns of some selected PPIEPR blends
Because of the difficulties in defining the amorphous halo exactly and the diffi-
culties with the measured orientations, the results of WAX.s were not very reli-
able. For this reason, DSC measurements were performed for comparison. In Fig.
B79, heating runs for some blends and their components can be seen. The height
of the melting peaks at about 165°C decreases with increasing rubber content.
The enthalpies of melting (Ml) are also shown in Fig. B79. If the enthalpies of
melting were related to the amount of PP in the blends, there would be a constant
value of MI. This is in agreement with the conclusion from the WAXS results that
addition of EPR does not influence the degree of crystallinity of the PP matrix. A
constant peak width indicates a constant thickness of lamellae.
120,--------,
100
endo :§ 80
....,
~ 60
J:
<l 40 0 Blend Mass
20 • PP Content
o 10 20 30 40 50
PP/EPR
dO EPR Content (wt.%)
ill 100/0
~)/l--95/5
j ---~=====::::::::::::::::::::=?It= ~~;:
_______
exo ~_________~_. ~ 50/~
0/100
o 50 100 150 200 250

T (0C)
Fig. B79. Results of the DSC measurements
The brittle fracture surfaces of the blends show nearly circular particles at rub-
ber contents up to 30 wt. %. The distributions of diameters are relatively wide, but
the arithmetic mean value of the particle diameters is approximately constant
(about 2 /lm). This means that the average interparticle distance decreases with
increasing rubber content (Fig. BSO).
Because the specimens were produced by injection moulding, particle defor-
mation has occurred. During processing, shear loads elongate the particles in the
injection direction. In the core region the particles show an aspect ratio of about
two; immediately under the skin this ratio is much larger (Fig. BS1). However, the
consequences of this particle deformation are not the subject of this work.
The appearance of the EPR phase in the case of high amounts of rubber (45 and
50 % EPR) is very different. The shape is very uneven and branches are visible for
Morphology and Toughness ofPPIEPR Blends 261
the first time. This indicates the beginning of the transition to an interpenetrating
network. As PP is the continuous phase even in the 50150 blend, the interpenetra-
ting network is not formed until a content greater than 50 wt. % EPR is reached.
6 .---------------------------------~
o Average Interparticle Distance A
··
(jJ
• Average Particle Diameter d
•
·••
4
·,
\
\
\
,
.
\
• •
\
• ,, •
"' - - - ~ -- El
o 5 10 15 20 25 30
EPR Content (wt.%)
Fig. B80. Average particle diameter and interparticle distance of the blends with up to 30 wt.%
EPR
Fig. B81. Particle morpho logy in the injection-moulded specimens with up to 30 wt.% EPR
(ID - injection direction)
3.2 Toughness Behaviour under Impact Loading
An initial survey of the toughness spectrum of the blends is given by the load-
deflection diagrams in Fig. B82. It can be seen that the toughness behaviour
ranges from linear elastic unstable (0, 1, 5 wt. % EPR) through elastic-plastic
unstable (15, 20, 30 wt.%) to elastic-plastic stable (45, 50 wt.%). Owing to these
very different behaviours, a quantitative comparison between the materials is
impossible with one fracture mechanics parameter.
Up to an EPR content of 30 wt. %, the resistance against unstable crack growth
was characterized by critical J values JIlT evaluated according to Sumpter and
Turner's method [9]. In the case of the blends with high rubber contents (45 and
50 wt.%), crack resistance curves under dynamic loading were measured. Suffi-
ciently large plastic-deformation parts for the blends with 20 and 30 wt.% EPR
made investigation of resistance against stable crack growth also possible.
PP Homopolymer PP/EPR 95/5
I-----!
0.5mm
iO.5mmi
PP/EPR 70/30 PP/EPR 50/50
I--t
0.5mm
Fig. B82. Load-deflection curves of some PPIEPR blends
3.2.1 Resistance Against Stable Crack Propagation
Figure B83 shows J-fla curves for the blends. The materials with high rubber
content show a comparable, high toughness level. They have relatively high val-
ues of the technical crack initiation parameter JO.2 and the tearing modulus TJ (Ta-
ble B 16). The J-R curves of the blends containing 20 and 30 wt. % EPR are much
lower. On fracture surfaces, differences between blends with middle and high
rubber contents can also be seen (SEM images in Fig. B83).
Morphology and Toughness of PPIEPR Blends 263
14
12
10
E 8
--
~
E
"""') 6
0
0 0.2 0.4 0.6 0.8 1.0
M(mm)
Fig. B83. J-R curves measured under impact loading of PPIEPR blends; details of the end of
stable crack growth (a, b)
In the regime of stable crack growth there is a failure of the matrix-elastomer

interface. The matrix ligaments are plastically stretched, but the matrix ligaments
of blends with 45 and 50 wt.% EPR are much more deformed.
The tearing modulus T}-2 determined from J-!J.a curves at !:J.a =0.2 mm, which
characterizes the resistance against stable crack growth, depends more strongly on
EPR content than does the technical crack initiation value JO.2 (Table B 16).
Table B16. Characterization of the resistance against stable crack growth; fracture mechanics
values
Material T=23°C T=-20 °C

Jo.l(N/mm) TIl Jo.l(N/mm) TIl
80120 0.6 0.5 Linear elastic

70/30 1.1 5.0 Linear elastic
55/45 4.5 54.6 1.2 4.0
50/50 5.2 71.2 2.0 7.6
264 T. Koch, S. Seidler, K. Jung, W. GreUmann
Toughness improvement at lower temperatures is one main goal of polymer

modification. For this reason, specimens were measured at -20 °c. During the
first tests it was found that only blends with 45 and 50 wt.% EPR deform with
significant plastic parts, i.e. significant dissipated energy. Only for these two ma-
terials could the resistance against stable crack growth at -20 °C be characterized
(Fig. B84). The toughness values are much lower at - 20 °C (Table BI6), but suf-
ficient.
14
T = -20 °C
12
• SO/50
o 55/45
I·
10
E 8
--z...,
E
6
•
4
0
0 0.2 0.4 0.6 0.8 1.0
t:P (mm)
Fig. B84. J- R curves measured under dynamic loading of PPIEPR blends at -20°C; SEM
image, detail of the end of stable crack growth
3.2.2 Resistance Against Unstable Crack Propagation
The critical J values obtained according to the method of Sumpter and Twner [9]
are plotted versus EPR content and average interparticle distance in Fig. B85.
Because the average particle diameter is almost constant and the interaction be-
tween matrix and rubber does not change, it is assumed that the interparticle dis-
tance is the critical feature for the different toughness behaviour. This agrees with
results on other toughness-modified semicrystalline thermoplastic polymers [10-
13].
An addition of 5 wt.% EPR does not increase toughness. There are no notice-
able interactions between the particles, because of the large interparticle distances.
The matrix ligaments are in a plane strain state. Increasing the rubber content from
5 to 15 wt. % reduces the interparticle distance to half of its previous value. Weak
dissipation processes start. An interphase failure occurs with only a small amount
of stretching of matrix ligaments, which are now in a plane stress state. So it is
Morphology and Toughness ofPP/EPR Blends 265
assumed that the threshold for the brittle-to-tough transition, the critical interparti-
cle distance, is reached in this polymer system.
If more rubber is added, the number of matrix ligaments which are able to yield
is increased and the toughness is increased. This involves a transition from mainly
unstable to mainly stable crack growth.
6~----------~----------1
4 ..,
E
.e3
~
o~--~----~--~
o 10 20
EPR Content (wt.%)
30
1-", ..,
";2
O+-~--r-~~--~-r~--~~~r-~~
o 2 3 4 5 6
A (IJm)
Fig. B8S. J values JId related to resistance against unstable crack growth as a function of average
ihterparticle distance A and EPR content
4 Summary
In this work, PPIEPR blends with contents up to 50 wt.% EPR were investigated.
In the extruded polymer mixtures, the adhesion between the phases is low. The
added rubber acts as a nucleation agent, but the degree of crystallinity of the ma-
trix, measured by W AXS and DSC, does not show a dependence on EPR content.
During impact loading in the instrumented Charpy impact test, the blends show
two characteristic transitions in toughness behaviour. In the regime of unstable
fracture there is a transition from a low level to a high level of J values. This tran-
sition correlates with decreasing interparticle distance because of the almost con-
stant average particle diameter and comparable diameter distributions. Above
30 wt. % EPR a change to elastic-plastic stable material behaviour occurs. At
45 wt.% EPR and under the testing conditions described the specimens do not
break.
References
In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC)-Series, No.3. Institut
flir Mechanik, Berlin, Chemnitz: 102-111
mechanics and Coupled Fields (FMC) Series, No.3. lnstitut fiir Mechanik, Berlin, Chem-
nitz: 142-151
4. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour of poly (methylmeth-
acrylate), polystyrene and polycarbonate. Appl. Macromol. Chem. Phys. 253: 27-49
5. Savadori A., Bramuzzo M., Marega C. (1984): J-integral analysis of ductile fracture of
PPIEP rubber blends. Polym. Test. 4: 73-89
6. Kobayashi T. (1983): Measurement of dynamic fracture toughness Jld by instrumented
Charpy test. Int. J. Fract. 23: RI05-RI09
7. Grellmann W., Seidler S., Schierjott U., Rufke B. (1985): Anordnung zur Bestimmung des
JT.rgesteuerten RiBwachstums bei schlagartiger Beanspruchung. German Patent DP 275113
8. Hlavata D., Horak Z. (1994): WAXS and SAXS investigation of polypropylene crystalline
phase in blends with high-impact polystyrene and compatibilizers. Eur. Polym. J. 30: 597-
600
9. Sumpter J. D. G., Turner C. E. (1976): Cracks and fracture. ASTM STP 601: 3-18
10. Wu S. (1985): Phase structure and adhesion in polymer blends: A criterion for rubber tough-
ening. Polymer 26: 1855-1863
11. Wu x., Zhu X., Qi Z. (1993): Percolation model of brittle-ductile transition in PPIEPDM
blends. J. Polym. Engng. 12: 229-238
12. Grellmann W., Seidler S., Kotter 1. (2000): Brittle-to-tough transition in toughened polypro-
pylene copolymers. In: Proceedings of the 13th European Conference on Fracture (ECF 13),
San Sebastian, Spain, September 6--9, CD-ROM Polymer and Composites No. 17: 1-8
13. Kotter 1., Grellmann W., Seidler S. (2000): Crack toughness behaviour and morphology of
PP blends. In: Proceedings of Polymeric Materials (p 2000), Halle, Germany, September 25-
27: 368-371
B 2.4 Morphology and Micro-Mechanics
of Phase-Separated Polyethylene Blends
R. Godehardt, W. Lebek and G. H. Michler, Merseburg
1 Introduction
A better understanding of the mechanical properties of polymeric materials is a

task of particular scientific and economic importance. Because of the large variety
of macromolecular and supermolecular structures (the morphology), polymers are
inherently capable of large property modifications. The structure or morphology
and the mechanical properties are linked by the micro-mechanical processes of
deformation and fracture, i.e. by the micro-mechanics [1]. Improved knowledge of
the micro-mechanical processes provides a very direct way of improving the me-
chanical properties of polymers, as shown in Fig. B86.
The term 'micro-mechanics' covers all processes on mesoscopic, microscopic
and nanometre scales that occur inside a material in response to external mechani-
cal loading [I]. These processes range from reactions of macromolecules, such as
reptation (snake-like) motion, stretching and scission of macromolecules, up to
crack propagation and fracture processes. On the one hand, these processes are
influenced by the diverse molecular structures and the morphology of the poly-
meric materials and, on the other hand, they are also influenced by loading condi-
tions such as temperature, loading velocity or stress state.
Progress in techniques, mainly in electron microscopy and scanning force mi-
croscopy, enables a wide range of experiments that were previously impossible,
including the direct study of the micro-mechanical behaviour of a sample in situ
[2-5]. Using a transmission electron microscope (TEM) and, above all, a scanning
force microscope (SFM), these direct microscopic techniques were applied to
study the micro-mechanical properties of phase-separated binary blends of ethyl-
ene/l-hexene copolymers. The aim of producing segregated blends of two ethyl-
ene/l-hexene copolymers of markedly different densities was to develop a mate-
rial of high toughness, combining two mechanical properties, viz. a high strength
and a large plastic elongation at break. Here, toughness, or fracture toughness,
denotes a property defined by physical energy absorption during deformation up
to fracture. This is only possible with a structural modification of the polymeric
material with the aim of initiating many local energy-absorbing yield events. Im-
portant prerequisites for an intense initiation of these local yield events are local
cavitation and stress concentration [6].

268 R. Godehardt, W. Lebek, G. H. Michler
Processes of Deformation
and Fracture
Process
-Stretching of Molecular Segments
run -Chain Scission
-Stress Concentration
-Micro-Yielding
-Micro-Cavitation
-Crazing
-Shear Band Formation
-Crack Initiation
-Intense Plastic Deformation
mm
-Crack Propagation
-Fracture
Mechanical
Properties
Fig. B86. Correlation between structure, morphology, micro-mechanical processes and prop-
erties of polymers
2 Micro-Mechanical Methods
For investigating the morphology of polymers, there are in general two direct
methods available [1]:
1. The study of special surfaces (brittle-fracture surfaces, smooth and selectively
etched surfaces) which give information on the internal structure of the material
and are investigated by means of replicas in the TEM or directly in the scan-
ning electron microscope (SEM) or SFM.
Morphology and Micro-Mechanics 269
2. The preparation of (ultra-) thin sections by (cryo-) ultramicrotomy, generally

after special fixation and staining procedures, and investigation by conventional
TEM or high-voltage electron microscopy (HVEM).
A survey of techniques successfully applied to investigate micro-mechanical
processes on the basis of characterization of the morphology is given in Fig. B87,
demonstrating three main techniques [1-3]:
1. Fracture surfaces (Fig. B87a) are mainly investigated with the SEM (micro-
fractography), demonstrating most significantly the final processes of deforma-
tion and fracture. In particular, structural heterogeneity and defects which initi-
ate fracture and influence crack propagation can be detected.
2. Deformation of bulk material is followed by investigating the changes at the
surfaces, using SEM and SFM or, after replication, TEM. Changes inside the
bulk material are studied by preparing ultra-thin sections using an ultramicro-
tome (occasionally after chemical staining) and then using TEM (Fig. B87b).
3. Samples of different thicknesses can be deformed in a tensile device, and the
deformed samples can be investigated directly by SEM, SFM, HVEM or TEM
(Fig. B87c).
Bulk Polymers
a) cr b) c)
i cr
i
~~
TEM
-
- ul1rathin section SEM TEM
-replica
SFM -
SEM
SFM
!
0-
t
cr
cr
Fig. B87. Survey of (electron and scanning force) microscope methods of investigating micro-
mechanical processes in polymers
The advantage of techniques I and 2 is the possibility of studying bulk mate-

rial, deformed under defined stress conditions and loading temperatures. Disad-
vantages lie in the difficulty of detecting the initial stages of deformation in tech-
nique 1, in the danger of damaging the deformed structures in technique 2 and in
the impossibility of carrying out in-situ experiments in either of the techniques.
Such in-situ tests can only be performed by techniques 3, which has become pos-
sible through new developments in micro-tensile devices and in the investigation
techniques of electron microscopy and scanning force microscopy.
In-situ tests in different types of electron microscope and in the SFM are possi-
ble, e.g. by the application of the tensile devices shown in Fig. B88. Using these
tensile stages, samples of different thicknesses can be studied. Thicker samples
(thick sections up to macroscopic specimens) can be investigated with an SEM or
SFM using the tensile devices in Figs. B88a, b, revealing deformation processes
near the surface.
Fig. B88. Several tensile stages usable inside an SFM and various electron microscopes:
tensile device (Oxford Instruments) to deform thicker samples in an SEM (a); tensile
stage of capacity 1000 N (Kanunrath & Weiss) to deform thick or thin samples in an
SFM or in an environmental SEM (ESEM) (b); tensile stage (JEOL) to deform
semi-thin samples in a 1000 kV HVEM (c); straining holder (Gatan) to deform ultra-
thin sections in a TEM (d)
Semi-thin sections, up to several micrometres thick, i.e. with a sufficient thick-

ness to represent the typical behaviour of the polymeric material, can be studied
with the stage in Fig. B88c in a 1000 kV HVEM, whereas ultra-thin films or sec-
tions are necessary when the straining holder (Fig. B88d) is used in a conventional
200 kV TEM at a higher resolution.
3 Micro-Mechanical Mechanisms in Phase-Separated

Blends of Ethylene/1-Hexene Copolymers
3.1 Materials
Using, on the one hand, methylaluminoxane (MAO)-activated bis(pentamethyl-

cyclopentadienyl)zirconium dichloride (Me5Cp)2ZrClz and, on the other hand,
ethylene bis(indenyl)zirconium dichloride Et(Ind)2ZrClz, ethylenell-hexene co-
polymers of markedly different densities were produced under the same conditions
[5]. The I-hexene/ethylene ratio in the monomer feed was 4/l. A binary blend was
formed by melt blending of the high-density ethylenell-hexene copolymer with
the elastomeric ethylenell-hexene copolymer in a wt.% ratio of 80/20. The blend
was mixed in a Brabender Plasticorder PL 2000 at 180°C and 40 rpm for 8 min
before being pressed to form plates (at a temperature and pressure of 180 °C and 3
bar for 330 s and then 180°C and 25 bar for 180 s). The chiral ansa metallocene
catalyst system Et(Ind)2ZrClzIMAO shows significantly better incorporation of 1-
hexene for the same comonomer concentration in the feed than the
(Me5Cp )zZrClzIMAO catalyst does. The different amounts of comonomer incorpo-
rated have a significant influence, e.g. on the density and the melting temperature
of the copolymer. Characteristic data on the copolymers used as blend components
are given in Table B17. P(E-co-H) denotes the copolymer produced with the
catalyst Et(Ind)2ZrClzIMAO, while an asterisk * after the P denotes the copolymer
formed with catalyst (Me5Cp)2ZrClz/MAO.
Table B17. Comparison of the blend components
Copolymer l-Hexene content of Density Melting

Mn Mw
the copolymer peak Tm
(mo\%) (g/cm3) (0C) (kg/mol) Mn
P*(E-co-H) < 0.4 0.930 127 102.0 3.1

P(E-co-H) 17.2 0.867 29.7 47.5 2.4
3.2 Morphology
Using a Jeol200 kV TEM (JEM 2010), morphological studies of the blend were
performed in the usual way. Ultra-thin sections suitable for the TEM study were
cut from a sample previously stained with ruthenium tetroxide vapour at 60°C
using a Leica Ultracut E ultramicrotome at room temperature. The TEM micro-
graph in Fig. B89 shows the morphology of the blend. The dark image structures
in the electron micrograph are caused by selectively stained areas of the specimen.
On the one hand, they show the amorphous areas between the bright crystalline
lamellae, where an additional contrast enhancement in the boundary layers along-
side the crystalline lamellae is a known staining effect, described, for example, in
[7]. On the other hand, the dark areas of various sizes result from strongly stained
segregated particles consisting of the elastomeric blend component. TEM investi-
gations, described in more detail in [5], showed the total segregation of the blend,
which was demonstrated by a quantitative comparison of the total area of the dark-
imaged elastomeric particles with the residual image area showing the matrix
component, both weighted by their densities given in Table B17. Within the accu-
racy limits of the measurement, the result matches quite well the wt. % ratio of
80/20. It is also typical that a few of especially thick crystalline lamellae of the
matrix penetrate into the segregated elastomeric particles.
Fig. B89. TEM micrograph of a stained ultra-thin section of the blend
In addition, SFM investigations were performed using a microscope, model

Dimension 3000, from Digital Instruments. Special SFM modes allow not only the
imaging and evaluation of the surface topography but also the direct observation
of the specimen morphology in the surface region, provided the surface is suffi-
ciently flat. In general, the mapping of the phase of the cantilever oscillation in the
tapping mode is the preferred technique to detect local variations in the composi-
tion, adhesion and viscoelasticity.
In the tapping mode, the cantilever oscillates near its resonant frequency. The
amplitude is reduced compared with the amplitude of the free oscillation if the
distance between the cantilever and the specimen surface is decreased and the
cantilever begins to tap the sample during each vibration. During the surface scan
the reduced amplitude is kept constant, according to the setpoint chosen, by a

feedback loop which drives the z piezo and controls the distance between the
specimen surface and the cantilever. Thus, the feedback signal yields height val-
ues of the surface topography directly. During the scan, an additional measure-
ment of the phase difference between the input and output signals of the cantilever
oscillation mainly records local changes in the stiffness of the sample surface.
However, it has to be taken into account that the contrast of the phase images is
not distinct, and other imaging conditions can result in a vanishing or even inver-
sion of the contrast [8-10].
On the other hand, the contrast formed by force modulation in the negative-lift
mode is unambiguous, with the dark areas indicating stiffer materials and the
brighter areas indicating softer ones [11,12]. With this technique, each scan line is
scanned twice. In the first scan, the surface topography along the line is deter-
mined from the tapping-mode height signal, with the height values stored for all
image pixels of the line. Then the line is rescanned in the force modulation mode.
According to the negative lift height chosen, the cantilever is lowered relative to
the height values previously determined for each pixel, and the cantilever tip in-
dents the specimen. In addition, the cantilever is oscillated at a low driving am-
plitude near the resonant frequency of the bimorph piezo on which the cantilever
is mounted. A higher oscillation amplitude is measured in softer sample areas by
this technique.
Provided the surface is sufficiently flat and the surface structures reflect the
bulk morphology, the SFM techniques described above can be used to study blend
morphologies directly without staining or etching the sample, as shown in Figs.
B90 and B91. The force modulation image in Fig. B90 shows a blend morphology
comparable to the results of the TEM investigation of a stained sample shown in
Fig. B89. Because of their elastomeric properties, the segregated particles appear
bright and can be discriminated clearly from their surroundings. In contrast, the
crystalline lamellae occur as dark structures owing to the higher stiffness than that
of the adjacent amorphous areas. Although the contrast is not so good, details are
better resolved in the tapping-mode phase image in Fig. B91. Furthermore, the use
of the tapping mode is favourable as it is the easiest SFM method of investigation.
3.3 Micro-Mechanical In-Situ Investigations with the SFM
SFM in-situ deformation experiments were performed in a Dimension 3000 mi-

croscope equipped with a deformation unit (a bending module which was modi-
fied for tensile tests) from Kammrath & Weiss GmbH. Suitable samples - 40 ~m
thick, 8 mm long and 4 mm wide - were prepared at room temperature with a
Leica RM2l65 microtome, and were annealed near the melting temperature be-
tween freshly cleaved mica plates to decrease the surface roughness. The cooling
rate after annealing was about 10 °C/min. A periodic pattern with a periodicity of
35 ~m was formed on the back surface of the sample by evaporating silver
through a grid. An area of interest, about 100 ~m long and 350~m wide, where
the deformation would mainly take place, was chosen and was marked by notch-
274 R. Godehardt, W. Lebek, G. B. Michler
ing the sample with a razor blade. Finally, the specimen was glued onto two slid-
ing parts of a special mount, which itself was clamped in the deformation unit so
as to load the sample for performing an SFM in-situ tensile test. During the test,
the deformation and structural changes in this specimen area were observed in
successive steps, with the elongation being stopped to perform the SFM investiga-
tion. The pattern evaporated on the back surface of the transparent sample served
as adjusting marks during the positioning of the specimen via an integrated optical
microscope with a zoom lens.
Fig. B90. Blend morphology imaged by force modulation mode of the SFM
The micrographs in Figs. B90 and B91 show the morphology of a sample spe-
cially prepared for SFM in-situ tensile tests as described above. The tapping-mode
phase imaging revealed the micro-mechanical behaviour of the centre of the sam-
ple during a tensile test. Figure B92 shows a phase image at higher magnification,
with the selected area of about 4 Ilm2 framed in black. Obviously, a segmental
appearance, like beads on a string, is typical of the dark crystalline lamellae; this
has also been observed in other copolymers of ethylene with a. olefins [13,14].
Fig. B91. Blend morphology imaged by the tapping-mode phase signal of the SFM
Fig. B92. SFM tapping-mode phase image of the central part of Fig. B91, with the area chosen
for studying the micro-mechanical behaviour of the blend marked
Figure B93 shows the image in the original state (1) and images of three steps
of the tensile test (2-4) corresponding to elongation by 50 f..lm, 100 f..lm and
200 f..lm of the tensile module. In all four micrographs exactly the same specimen
site is imaged. Thus, on the one hand, the micrographs show a homogeneous strain
in the direction of the external stress, and on the other hand, from the unchanged
width of the images, there is no significant constriction. Relative to the original
length 10 = 2.19 f..lm, the changes of I result in average local strains &t of 8 %, 16 %
and 56 % of the imaged area at the three steps of the tensile test. To illustrate the
strain distribution within the imaged area, the changes of four line segments are
Fig. 893. Micro-mechanical behaviour in the marked area of Fig. B92 shown by SFM tap-
ping-mode images showing the specimen area before defonnation (1) and in three
stages of an in-situ tensile test (2-4)
compared. The line segments, defined by the locations of the image structures at
their endpoints, are marked by arrows and the letters a, b, c and d. The length of
all the segments is 445 om in the original image. While the length of segment a, in
a semicrystalline area, does not change in any of the micrographs, the length of the
adjacent segment b, which is located within an elastomeric particle, changes sig-
nificantly. The corresponding local strain &h of the latter was determined to be
16 %,30 % and 146 % at the three deformation steps. Segments c and d are ar-
ranged parallel to segments a and b. Though, like segment a, they lie in an area of
semicrystalline material, they are both deformed in the same way, and the three
micrographs show local strains of Ge = Gd of 8 %, 16 % and 69 %.
A B A B
A B A B
Fig. B94. SFM tapping-mode height images in a 3d presentation (a, c), and force modulation
images (b, d) showing the same specimen area before deformation (a, b) and after
deformation (c, d)
Another result from the SFM in-situ tensile tests is shown in Fig. B94. The im-
ages show a sample area before (first row) and after (second row) a mean strain of
60 % in the direction of the stress, given below Fig. B94d. On the right, images of
the sample morphology obtained by the force modulation technique are shown.
Two elastomeric particles are marked by arrows and the letters A and B, to allow
one to correlate the images more easily. On the left, the corresponding tapping-
mode height images in a three-dimensional (3d) presentation illustrate the surface
relief. The original surface is very flat. The differences in height of the area im-
aged are smaller than 50 nm. However, in Fig. B94 c, there is a strong increase in
the surface roughness after deformation, mainly caused by the specific drift of the
elastomeric material into the sample interior.
3.4 Micro-Mechanical In-Situ Investigations with the TEM
Using the single-tilt cooling-straining holder shown in Fig. B88d, in-situ defor-
mation tests of thin sections were performed in the TEM. The sections, about
100 nm thick, 3 mm long and 0.5 mm wide, were cut with a cryo-ultramicrotome
(Leica Ultracut E with a freeze chamber FC4E attachment). Each section was
fixed between two specially shaped adhesive tapes and bridges alongside a central
hole of 1 mm in diameter within the tapes. After loading the special mount with
the sample into the straining holder, the tapes were separated in the transverse
direction in the areas adjacent to the central hole. Then the self-supporting thin
section was deformed within the TEM at an elongation rate of 0.5 J.lmls.
Fig. B95. TEM micrograph of a thin section of the blend strained in situ with the stage shown
in Fig. B88d
One example of the results obtained from the in-situ TEM deformation of a thin
section from the blend is shown in Fig. B95. In general, a TEM micrograph of an
Wlstained thin section of the blend shows hardly any image structures with mor-
phological features as there is neither a material contrast nor a contrast caused by
differences in the local thickness. However, it is well known that during loading,
features become visible owing to the so-called straining-induced contrast en-
hancement: during deformation, structural elements which differ in their YOWlg' s
modulus or in their extensibility from the surroWldings (even for small differ-
ences) appear in better contrast [7]. The mainly bright structures in Fig. B95 are
caused by the decrease in thickness of the strongly stretched elastomeric particles.
A few of these particles already show voids. Obviously, strongly increased strain
occurs in beaded strings of neighbouring elastomeric particles that lie roughly
perpendicular to the direction of the external stress. The semicrystalline matrix
between adjacent elastomeric particles, too, is highly strained within the beaded
strings, whereas this does not happen for the dark areas adjacent to strings that lie in
the stress direction. This is in reasonable agreement with the results of SFM studies.
4 Conclusions
Taking advantage of a metallocene-based ethylene/l-hexene copolymerization, a

binary blend was formed by melt blending of a high-density ethylene/l-hexene
copolymer with an elastomeric ethylenell-hexene copolymer in a wt. % ratio of
80/20. TEM investigations of the morphology revealed the total segregation of the
blend components, but it was also typically observed that a few of the thicker
crystalline lamellae of the matrix penetrated into the segregated elastomeric parti-
cles. Furthermore, SFM investigations showed that a segmental appearance, like a
string of beads, is typical of the crystalline lamellae.
Use of the SFM tapping mode and a special force modulation technique en-
abled local deformations of the sample and changes in the morphological structure
in a particular area of the specimen site to be directly recorded while the external
stress was successively increased. The results of these in-situ experiments show
that the deformations, which are homogeneous down to the micron range, are
strongly inhomogeneous in the submicron range. In particular, the strong defor-
mation of the elastomeric particles is accompanied by a very heterogeneous de-
formation of the surroWlding semicrystalline matrix.
These results are in reasonable agreement with the results of in-situ deforma-
tion of thin sections carried out within the TEM. The TEM results show that
strongly increased strain occurs in beaded strings of neighbouring elastomeric
particles that lie roughly perpendicular to the direction of the external stress. The
semicrystalline matrix between the adjacent elastomeric particles is also highly
strained within the beaded strings, whereas this is not observed in areas adjacent to
strings lying in the stress direction.
The micro-mechanical investigations show that beaded strings of strongly de-
formed elastomeric particles are the weak points of the blend and that an improved
toughness should be obtainable by means of an enhancement of the strength of the
segregated elastomeric particles, e.g. by crosslinking.
Acknowledgements
This work was carried out within the framework of the DFG program named Itmo-
vationskolleg 'New polymer materials by purposeful modification of interface struc-
tures/properties in heterogeneous systems' and was supported by the German Re-
search Foundation. The authors thank Dr. S. Rudolph and Dr. S. Giesemann for the
production of the blend material, and Mrs. S. Goerlitz and Mrs. E. Allert for perfor~
ming the morphological investigation and the in-situ test, respectively, in the TEM.
References
1. Michler G. H. (1998): Polymers for biosystems - microstructural construction of polymers

with improved mechanical properties. Polym. Adv. Technol. 9: 812-822
2. Michler G. H. (1992): KunstofI-Mikromechanik - Morphologie, Deformations- und Bruch-
3. Michler G. H. (1995): Electron microscopic techniques for direct investigation of microme-
chanical mechanisms in polymers. Trends Polym. Sci. 3: 124-131
4. Starke J. U., Godehardt R, Michler G. H., Bucknall C. B. (1997): Mechanisms of cavitation
over a range of temperatures in rubber-toughened PSAN modified with three-stage core-
shell particles. J. Mat. Sci. 32: 1855-1860
5. Godehardt R, Rudolph S., Lebek W., Goerlitz S., Adhikari R, Allert E., Giesemann J.,
Michler G. H. (1999): Morphology and micromechanical behavior of blends of ethene/l-
hexene copolymers. J. Macromol. Sci. Phys. B38(5&6): 817-835
6. Michler G. H., Starke J. U. (1996): Investigations of micromechanical and failure mecha-
nisms of toughened thermoplastics by electron microscopy. In: Riew C. K., Kinloch A. J.
(Eds.) Toughened Plastics n. American Chemical Society, Washington, DC, Chap. 17:
251-277
7. Michler G. H. (1993): Electron microscopy in polymer science. Appl. Spectrosc. Rev. 28:
327-384
8. Magenov S. N., Elings V., Whangbo M.-H. (1997): Phase imaging and stifihess in tapping-
mode atomic force microscopy. Surf. Sci. 375: L385-L391
9. Bar G., Thomann Y., Brandsch R, Cantow H.-J., Whangbo M.-H. (1997): Factors affecting
the height and phase images in tapping mode atomic force microscopy. Study of phase-
separated polymer blends of poly(ethene-co-styrene) and poly(2,6-dimethyl-l,4-phenylene
oxide). Langmuir 13: 3807-3812
10. Whangbo M. H., Bar G., Brandsch R (1998): Qualitative relationships describing height and
phase images of tapping mode atomic force microscopy. An application to micro-contact-
printed pattemed self-assembled monolayers. J. Appl. Phys. A 66: 1267-1270
II. Chen J. T., Thomas E. L. (1996): The use of force modulation microscopy to investigate
block copolymer morphology. J. Mat. Sci. 31: 2531-2538
12. Galuska A. A., Poulter R R, McEarth K. O. (1997): Force modulation AFM of elastomer
blends: morphology, fillers and cross-linking. Surf. Interface Anal. 25: 418-429
13. Minick J., Moet A., Hiltner A., Baer E., Chum S. P. (1995): Crystallization of very low
density copolymers of ethylene with a-olefins. J. Appl. Polym. Sci. 58: 1371-1384
14. Beansason S., Minick J., Moet A., Chum S., Hiltner A., Baer E. (1996): Classification of
homogeneous ethylene-octene copolymers based on comonomer content. J. Polym. Sci.,
Polym. Phys. 34: 1301-1315
B 3 Copolymers
B 3.1 Toughness Optimization

of Multi-Phase Polymer Materials
Based on a PP Matrix
Using Fracture Mechanics Parameters
S. Seidler and W. GreUmann, Merseburg
1 Introduction
Besides the development of new materials, the emphasis of international polymer

science is on the extension of the range of application of the currently available
polymers. Possible ways of achieving this purpose are filling and reinforcing the
polymer to increase the strength and stiffness, and the production of blends and
copolymers to increase the stiffness, toughness, heat deflection temperature,
weathering resistance and environmental-stress-cracking resistance, and to im-
prove the processability. These blends and copolymers consist of a continuous
polymer phase and one or several dispersed phases. The dispersed phases may be
either polymers or other components. Such materials are summarized in the term
'multi-phase polymer materials' [1].
Besides the properties of the continuous and dispersed phases resulting from the
structure, the mechanical properties are influenced by many circumstances which
result from the shape, volume content, distribution, average free path length, phase
bonding, phase interactions and processing conditions of the dispersed phase.
An important aim of blend and copolymer development regarding mechanical
properties is an increase of toughness while simultaneously retaining sufficient
stiffuess. Usually this is realized by including a soft dispersed polymer phase with
a glass transition temperature Tg below the desired operational temperature in a
relatively hard thermoplastic matrix material. In the following, the dispersed phase
is referred to 'particles'. During loading, the particles initiate various mechanisms.
As long as the particles are intact, they decrease yield strength, they induce defor-
mation mechanisms in the matrix material, i.e. crazing and/or yielding, or they
form cavities which are able to stabilize or propagate the plastic zone in front of
the crack tip [2]. Both crazing and shear yielding can act as toughness-increasing
mechanisms. Among other things, their occurrence depends on the temperature,
the testing velocity, the structure of the matrix material and of the particles, the
particle diameter, the interparticle distance and phase interactions [2-13].

282 S. Seidler, W. GreUmann
One group of materials of increasing importance is blends and copolymers con-

sisting of a semi crystalline matrix material and an included rubber phase, for ex-
ample copolymers and blends with a polypropylene (PP) matrix material [14-16]
and impact-modified polyamides [17,18].
In the following, the influence of selected structural parameters on crack tough-
ness behaviour is illustrated for the example of ethylene-propylene copolymers.
The reasons for production of PP blends and PP copolymers are discussed in
A 1.1 (pp. 3-26) and B 2.2 (pp. 257-266). Because of the different rheological
behaviour of PP and of the rubbers used for compounding, such as EPDM, a fine
rubber particle distribution is only partly achievable, although dispersing tech-
nologies are very extensively used. The consequences are limits on toughness
improvement that copolymers with included rubber particles such as heterophasic
reactor materials do not show [15].
2 Influence of the Structure

on Crack Toughness Behaviour of PP Materials
2.1 Influence of Molecular Weight and Crystallinity
As mentioned before, the properties of polymer blends and copolymers are influ-
enced by a multitude of factors, including the matrix material properties. Among
properties related to the chemical structure, the average molecular weight, mo-
lecular-weight distribution and crystallinity determine the mechanical properties of
PP. Figure B96 shows the influence of the average molecular weight on the load-
deflection (F-f) behaviour of PP and on the fracture mechanics values determined
according to A 2.2 (pp. 71-86). With increasing average molecular weight, the F-f
behaviour changes from linear elastic to elastic-plastic and the fracture mechanics
values related to resistance against unstable crack propagation increase. In the
homopolymers the stable-crack-growth component is very low.
With increasing crystallinity, the material resistance against unstable crack
propagation decreases, and also the physical crack initiation values (Fig. B97). A
comparable result of the influence of crystallinity on crack initiation behaviour is
shown in [19].
Whereas in the homopolymers dominantly unstable crack growth occurs, in PP
blends and PP copolymers the stable-crack-growth component with increasing
rubber content increases [20-24]. Figure B98 shows the influence of the molecular
weight on the crack initiation behaviour of so-called standard copolymers pro-
duced in a reactor process. These materials are heterophasic copolymers with a PP
matrix material and inclusions consisting of ethylene-propylene rubber (EPR) and
polyethylene (PE). The particle diameter is about 1.8 IJ.m; the particle structure is
shown schematically in A 1.1 (pp. 3-26) and examples are shown in [23,24]. With
increasing molecular weight, the technical crack initiation parameters JO.2 and ~.2
increase. In this connection, it is noteworthy that in the copolymers a marked im-
Toughness Optimization of Multi-Phase Polymer Materials 283
provement of deformation capability at the crack tip occurs at a relatively high

molecular weight and this deformation capability seems to be constant at higher
molecular weights. One would expect, in principle, that this marked improvement
of deformation capability at the crack tip would cause a change in crack propaga-
tion behaviour. Whereas at average molecular weights of 316 kg/mol and 346
kg/mol material failure occurs by unstable crack propagation, an increase of the
average molecular weight to 675 kg/mol causes a change of crack propagation
behaviour from unstable to stable. Above this transition, the toughness properties
are nearly constant.
8 F_~ F_--;r1 F-71

J .~-
~ flTB( \
~_
fma<
LL
f~ /
160 \ /'" 8
/ ./..-
E
E
1:,
......
...
r.O
120
80 - M-•••••••.•••••••.•••••• -.~••~.........
.../
/
.!.........................._...- ............../-'-/-_.
6
40 2
o 200 400 600 800

Mw(kglmol)
Fig. B96. ~d and JIdST versus average molecular weight; typicalload-deflection diagrams
'k
0.4 3
0.3
-
• 2
E
E .E
.E 0.2 ~
~
...:; t;:!!
....
0.1
"
.. ..•.,'-"
'-
'-.
40 50 60 70 80
Crystallinity (%)
Fig. B97. Influence of crystallinity on the physical crack initiation parameters J. and on J1r
<>
• 316 kg/mol
• 346 kg/mol r-------, 120
<> <>
• 675 kg/mol
E
• 821 kg/mol E
4
90 1,
.....
0.3 0.6 0.9 / ...........~.................. N
0
08
E Aa (mm) ............ ......•..•.......•
E
-.
! /// I
k /...•· · / / .../ 200
~ <>
<>
...,
N
0 2
100
<> <>
t···· ~
0.6 0.9
Aa (mm)
o 200 400 600 800 1000
Mw (kg/mol)
Fig. B98. Influence of average molecular weight on crack resistance behaviour of standard
copolymers
The same effect can be demonstrated for high-impact modified copolymers with
different average molecular weights [25]. An increase in molecular weight which
causes a transition from unstable to stable crack propagation behaviour in the
standard copolymers does not affect the crack propagation process in the high-
impact modified copolymers. For these materials no unstable crack propagation
can be proved under the experimental conditions chosen.
From these results, two statements can be derived:
1. The failure behaviour is deformation-determined.
2. After a 'critical' molecular weight is exceeded, the deformation behaviour is no
longer predominantly influenced by matrix material properties, i.e. the tough-
ness behaviour is determined by processes within the particles or in the matrix-
particle interface.
2.2 Influence of Ethylene Content
Besides the influence of molecular weight, the influence of ethylene content on the
crack resistance behaviour of the standard copolymers was investigated. An in-
crease of the total ethylene content can result in an increase of the EPR content
and/or an increase ofthe content of crystalline polyethylene.
In Fig. B99 the results for the influence of ethylene content on crack resistance
behaviour are summarized. The quantification of toughness behaviour followed the
ESIS testing protocol [26]. In addition, the energy dissipation processes which
determined the fracture behaviour of the copolymers could be quantified using the
JTJ concept (pp. 71-86, [27]). The influence of the ethylene content on the crack
initiation behaviour is low in comparison with the influence on energy dissipation
processes.
TEM investigations show an increasing particle number and an increasing
thickness of the EPR shell around the PE inclusions with increasing ethylene con-
tent [28].
A qualitative estimate of the EPR content can be obtained from the area of the
peak of the main relaxation mechanism of the EPR determined by dynamic me-
chanical analysis. The results given in [25] indicate an increasing EPR content
with increasing ethylene content. Owing to a nearly constant particle diameter of
about 1.8 /lm, an increasing EPR content is equivalent to a decreasing interparticle
distance. In these materials the interparticle distance has an important influence on
toughness behaviour too.
20
cr-12f========;=l==~-120
" 16.0 mol.% Ethylene
o 17.6 mol.% Ethylene
E
150 E D 24.4 mol.% Ethylene 15
...,~
E E
E
~ 100 0 10 ~N
f.:'
-:J
d
-:J
JO.2
50 5
10 20 30
Ethylene Content (mol.%)
Fig. B99. Influence of ethylene content on crack initiation behaviour and energy dissipation
capacity ofPP standard copolymers
Investigations of the influence of crack-opening-displacement rate during stable

crack propagation show nearly constant values only in the material with 24.4 mol. %
ethylene, after a certain value of stable crack growth has been reached (Fig.
BIOOa). Simultaneously, the energy necessary for stable crack propagation in-
creases strongly (Fig. B1 OOb).
The determination of the crack-opening-displacement rate followed (B15):
b-dk=8dk , (B15)
tB
where
-
t B- fmax (B16)
,
vH -113 L1v
(B17)
0.15.---------------;---,
a
t:. 16.0 mol.% Ethylene
o o 17.6 mol.% Ethylene

o 24.4 mol. % Ethylene
o
o 0.2 0.4 0.6 o.
M(mm)
12·.-------------~-,
b
t:. 16.0 mol.% Ethylene
o 17.6 mol.% Ethylene ?o
8 o 24.4 mol.% Ethylene
o
o 0.05 0.1 0.1
Odk (m/s)
Fig. BIOO. Relation between crack-opening-displacement rate, amount of stable crack growth
andJ
The relation shown in Fig. B 100 permits the supposition that owing to a de-
creasing interparticle distance and a simultaneously increasing thickness of the
EPR shell, an increasing part of the energy is consumed by defonnation mecha-
nisms in the EPR and in the EPRIPE interface. For this to happen, the stress fields
that fonn during loading have to overlap to a certain extent for an optimum effect.
This suggestion is supported by an example of investigations on PPIEPRIPE
model materials with defined interparticle distances (pp. 3-26, [25]). In these
materials the energy-determined and defonnation-determined characteristic values
are influenced in different ways. This effect is related to the fact that, in the case of
a fine particle morphology, the stress component of the whole deformation process
reacts more sensitively to changes in particle distance than does the deformation
component.
In these materials the dominant failure mechanism is crazing [28] and the fail-
ure process starts at the EPRIPE interface. During this process the particles are not
directly removed from the matrix material. The particles remain fixed to the matrix
even if they are dispersed into fibrils. This process enables a high energy dissipa-
tion in the material, which causes, in consequence, a high material resistance
against crack propagation. The intensity of this process depends on the thickness
of the EPR shell and on the size of the PE sequences included in the EPR [23-
25,29].
2.3 Influence of Interparticle Distance
The deformation behaviour of polymer multi-phase materials is influenced impor-

tantly by the geometric relations which follow from the shape, size, volume con-
tent and distance of the included phase. If this included phase consists of spherical
particles, matrix-specific particle diameters and distances suitable for initiation and
stabilization of energy dissipation mechanisms can be derived [30,31]. Only such
conditions enable high toughness properties in polymer materials.
In the following, the influence of particle diameter and distance on the crack
toughness behaviour of PP copolymers is discussed in more detail. The matrix
material considered here is a PP homopolymer with an average molecular weight,
and the particles are nearly core-shell particles with an EPR shell and a PE core.
The average particle diameter is about 1.5 11m, which it is nearly the same as in the
standard copolymers.
Because of the matrix dependence of the critical particle diameters and dis-
tances, the influence of the test temperature has to be considered, especially for
PP, owing to its unfavourable glass transition temperature. Figure BIOI shows the
influence of test temperature and interparticle distance on the -hlT values deter-
mined according to A 2.2 (pp. 71-86). At T= -20°C, 0 °C and 10 °C all materials
fail in an unstable manner, but with increasing test temperature the plastic compo-
nent of the general deformation energy increases. This increase depends on the
interparticle distance and is especially clear at A = 1.38 11m. This leads to domi-
nantly stable crack propagation at room temperature.
With decreasing interparticle distance, the influence of test temperature be-
comes more clear, which indicates an increasing effectiveness of energy dissipa-
tion mechanisms. These mechanisms ultimately result in the change of deformation
behaviour from linear elastic unstable via elastic-plastic unstable to elastic-plastic
stable described above.
With this material system a decrease in particle diameter to about 0.2 11m does
not affect the toughness behaviour, neither in its dependence on copolymer content
nor in its dependence on temperature. This is especially valid at temperatures be-
low the glass transition temperature of PP (Fig. B 102).
2Gr--------------------------------------.
--t:r- PP ......... A= 1.92 ~m
~ A=3.76~m -+- A=1.79~m
-0- A = 2.23 ~m _____ A = 1.54 ~m

~ A=2.12~m -+- A=1.38~m
E 12 d=1.5pm
.E
~
.,...
Ii;
8
4
A~----------__t:r_----~-----
o'~----+-----+-----+-----+-----~--~
-30 -20 -10 0 10 20 30
T (·C)
Fig. BlOt. Influence oftest temperature and interparticle distance on crack toughness related to
material resistance against unstable crack propagation
d = 0.2 pm
; /•
4
E •
~:
.E
.~. ...• :
~
., .....•
....en
'D •
2 ... ... t
... T=-20·C • T = 10·C
• T= O·C • T= 23·C
0
0 20 40 60 80 100
Copolymer Content (wt.%)
Fig. BI02. Dependence of JIlT on copolymer content and test temperature at an average parti-
cle diameter of 0.2 11m
The reasons for the occurrence of critical interparticle distances are discussed in
the literature with different models, which are based essentially on the assumption
of overlapping stress fields formed around the particles or changes in the stress
state of the matrix between the particles [18,28,30,31].
The shortest interparticle distance in these materials is A = 0.55 J.l.m. The ratio
of interparticle distance to particle diameter is Aid = 2.7. This value is nearly the
same as ratio Aid = 2.5 for the copolymer with A = 3.76 J.l.m for which results are
shown in Fig. BIOI. For this copolymer also, only a small increase in toughness
properties is detectable with increasing test temperature. This correspondence
indicates the narrowness of the range in which interactions between particle di-
ameter and interparticle distance occur.
Besides the characterization of material resistance against unstable crack propa-
gation, the quantitative recording of stable crack growth was of special interest.
Figure B I 03 shows the influence of interparticle distance on the crack resistance
behaviour of PP copolymers with an average particle diameter of 0.5 J.l.m at room
temperature. With decreasing interparticle distance the material resistance against
crack initiation and crack propagation increases. The J-l1a and Odk-l1a curves
show a similar tendency.
15.------------------------------------,
[J A=2.23I1m • A=1.79 I1m 200,-----''---------,
o A=2.12I1m _ A=1.54l1m
12 .t. A = 1.92 11m • A = 1.38 11m ~ 150
~ 100
E 9 J
.€
~ 0~0~-r-0~.4--~0.-8-~1.2
"') 6
.1a(mm)
0.2 0.4 0.6 0.8

Ail (mm)
Fig. BI03. Crack resistance curves measured under impact loading of PP copolymers with an
average particle diameter of d = I.S /.1m at T = 23°C
In Fig. B I 04 the characteristic values JO.2 and TJ0.2, determined from the crack
resistance curves, are shown as a fimction of test temperature. These dependences
also reflect the influence of interparticle distance. With decreasing temperature,
the interparticle distance necessary for a high material resistance against crack
initiation at a given temperature decreases. The tearing modulus TJ at l1a = 0.2
mm, which gives a quantitative description of the material resistance against stable
crack propagation, decreases continuously with increasing interparticle distance.
24
~
5 18
N
~ 12
4
6
3
E 1.5 2.0 2.5
A (11m)
~ 2
...,a
0
&
•
T'O'~
T= 10 'C
• T=23'C
1.0 1.5 2.0 2.5

A (lJm)
Fig. BI04. Dependence of technical crack initiation parameter JO.2 and tearing modulus TJ on
interparticle distance and test temperature
From the dependences of the fracture mechanics parameters related to resis-

tance against unstable crack propagation on the test temperature and interparticle
distance, brittle-to-tough transition temperatures (pp. 209-228 and pp. 243-256)
and critical interparticle distances for the brittle-to-tough transition [18,28,32] can
be determined, following conventional concepts of the transition. The principle of
the determination of the critical interparticle distance Acril on the basis of fracture
mechanics values is explained in [29].
The critical interparticle distance decreases with decreasing test temperature
(Fig. B 105a). The results given in Fig. B 105a were determined from materials with
increased matrix material toughness. The matrix materials here are propylene-
ethylene random copolymers. The random copolymer 1 contains 8 mol. % ethylene
and the random copolymer 2 contains 4 mol. % ethylene. A random copolymeriza-
tion of PP with ethylene is expected, in principle, to lead to a toughness increase.
This means that heterophasic random copolymers are materials systems with
increased matrix material toughness in comparison with heterophasic copolymers.
Furthermore, these materials show increasing interactions as a function of ethylene
content between the copolymer matrix and the EPR copolymer. The average
particle diameters are about 0.5 ~m. To keep the particle diameter nearly constant,
both of the random heterophasic copolymers were diluted with their matrix
materials. The specimen preparation procedure is described in detail in [23,29].
The random heterophasic copolymer 1 contains EPRIPE particles with up to five
PE inclusions (Fig. B 106); the EPRIPE particles in the random heterophasic
copolymer 2 correspond approximately to core-shell particles.
F or both materials, the dependence of the critical particle distance on test tem-
perature results in a linear relation which is not influenced by the glass transition
temperature of the matrix material. These experimentally determined linear rela-

tions can be described by a simple linear fitting procedure, which leads to the
empirical equations given in Fig. BI05a. Comparable results are shown in [32].
The constants of the linear fit depend on molecular parameters of the matrix mate-
rial and the included phase, i.e. the rubber. For this reason, a generalization of such
relations is only qualitatively possible. On the other hand, a knowledge of such
connections is of great importance for effective materials design.
Using the example of the heterophasic copolymers, we have shown that the ex-
clusive consideration of the critical particle distance is not sufficient for the inter-
pretation of the deformation behaviour in such multi-phase materials. Rather, the
A/d ratio has to be considered. Owing to the assumed nearly constant particle di-
ameter in the systems investigated the Acritld ratio versus test temperature shows
qualitatively the same behaviour as Acrit versus test temperature (Fig. B 105b).
a
4
PP Random Heterophasic Copolymer 2
A,.. (T) = 1.5 + 0.05 T
3
Brittle
E
2: 2
~
Tough
o A,.. (T) a 0.8 + 0.02 T
·20 ·10 o 10 20 30 40
T (OC)
b
6
5

o
·20 ·10 o 10 20 30 40
T (OC)
Fig. BI05. Critical particle distance Acrit for the brittle-to-tough transition of random heteropha-
sic copolymers (a) and ratio Acm/d for the same materials (b)
•
~andom Copolymer Matrix
O. 5~m
Fig. 8106. TEM micrograph of the internal structure of random heterophasic copolymer I
As described before, the use of a random copolymer matrix leads to signifi-

cantly higher toughness values in PPIEPRlPE copolymers than does a homo-
polymer matrix. This is clear in the crack resistance behaviour too, as is shown by
the example of the J-/la curves of the random heterophasic copolymer 1 (Fig.
B 107a). The experimentally determined crack resistance curves reflect also the
effect of an increasing material resistance against crack initiation and crack propa-
gation with decreasing interparticle distance. The random heterophasic copolymer
2 shows a comparable behaviour at a lower toughness level [23].
For the quantitative description of toughness behaviour, the technical crack ini-
tiation parameter JO.2 was used, determined according to A 2 .2 (pp. 71-86). For
both materials the JO.2 versus A curves show dependences which correspond quali-
tatively to the dependences J.P = f(A) used for the determination of the critical
particle distance [23,29] (Fig. B 107b). The critical particle distance defines the
brittle-to-tough transition in the material and often this transition is connected with
a basic change in the deformation mechanism. Above the brittle-to-tough transi-
tion, ductile failure behaviour generally occurs in the materials. The occurrence of
a second transition in the ductile failure area depends on the internal state of the
material, i.e. the particle diameter and distance as well as phase interactions be-
tween particles and matrix, and on the test temperature, as the results given in Figs.
B 104 and B 107b show. Owing to the interactions between particle diameter and
distance, the ratio Aid has to be considered for an exact description of the material
behaviour. While the experimentally determined particle diameter in the random
heterophasic copolymer 1 is constant, it decreases with increasing particle distance
in the random heterophasic copolymer 2 [23]. As a consequence there is a clear
but weak transition area in the JO.2 versus Aid curve for the random heterophasic
copolymer 2 (Fig. BI07c).
For the materials investigated two transition areas can be identified under certain
test conditions and for certain material states: the generally known brittle-to-tough
transition and a second, tough-to-high-impact transition in the ductile failure area.
30,.----------------------------------------,-.
a
Random Hetarophasic Copolymer 1
25
• A=O.36l1m
• A =0.39 11m
20 ... A=0.40 11m
" A=O.54l1m
E 15 •
+
A=0.57I1m
A= 1.03 11m
~..., 10
o
0.0 0.1 0.2 0.3 0.4 0.5 0.6
da(mm)
10
c
10..---- 8 Random Heterophl." Copolymer 1

b
E 6
8 z.f!
...
'":1 4 .. C P h••1c Copo'~.r 2
2
'(AId)..
6
E (AId)..
00
~a 4
2 3 <4 5
AJd
...,
2
A: 2,'"
A 2 ,crit
~.o 0.5 1.0 1.5 2.0 2.5

A (11m)
Fig. BI07. Influence of particle distance on the crack resistance behaviour of the random het-
erophasic copolymer 1 (a), and JO,2 versus A (b) and AId (c) for the heterophasic
random copolymers investigated
The nature of the brittle-to-tough transition is a subject of controversy in the lit-

erature. In addition to Wu's well-known percolation theory [31], an increasing
number of interpretations exist which assume that overlapping of the stress fields
starts if the interparticle distance is smaller than Acril [17,18,33,34], or that a
change from plane strain to plane stress conditions in thinner ligaments, i.e. with
decreasing interparticle distance, reduces the critical stress for matrix yielding
[32,35].
294 S. Seidler, W.GreUmann
Fig. BIOS. Craze fonnation in front of the crack tip in a PP homopolymer (a, b) and in a PP
random copolymer (c)
In the materials investigated here the brittle-to-tough transition can be corre-

lated with a change in defonnation mechanism. Both types of matrix materials, the
PP homopolymers (Fig. B I08a, b) as well as the PP random copolymers (Fig.
B 108c) show crazing as the main failure mechanism at room temperature. The size
and shape of the crazed zone and the tendency towards single or multiple crazing
depend on the loading conditions, test temperature and material. In the ductile
failure regime all materials show so-called 'croiding' as the main failure mecha-
nism (Fig. B 109).
Figure B 11 0 a shows an overview of a fracture surface of the PP random hetero-
phasic copolymer 2 with a particle distance A = 1.3 ~m. In this material, fracture
occurs by stable crack propagation. The fracture surface was produced by breaking
the specimen at liquid-nitrogen temperature. Such a fracture surface is typical of
these materials. The amount of stable crack growth l1a and an enlarged damage
zone are clearly visible. The macroscopic toughness behaviour depends very much
on the kind, intensity and stability of the microscopic processes in this zone. Crack
propagation occurs if critical stresses, strains or energies are exceeded, and the
fracture mechanics parameters quantify these critical values. So, it is necessary to
direct one's attention to processes in the regime of stable crack propagation during
ductile failure and to processes in the damage zone during more brittle failure.
Fig. BI09. The 'croiding' defonnation mechanism in front of the crack tip in the ductile failure
regime ofheterophasic PP copolymers and PP random copolymers
In the ductile failure regime, for particle distances below the tough-to-high im-
pact transition, strong cavity formation and matrix fibrillation can be demonstrated
(Fig. B 11 Ob). The cavities form because of the failure of the particles within the
EPR shell, caused by the three-dimensional stress field in front of the crack tip.
According to this description, the failure behaviour of the materials is initiated at
the EPRIPE interface, i.e. in the initial stage of failure the properties of the EPR,
the number and size of the PE inclusions, and the properties of the EPRIPE inter-
face determine the energy dissipation of the process. This initial stage is charac-
teristic of all materials investigated, independent of the kind of matrix material and
particle distance (Fig. BIll).
The subsequent processes are strongly influenced by the interparticle distance
or the ratio Aid, as well as by the matrix properties and the matrixlEPR interface.
Owing to the influences of the matrix and interface, a direct comparison of copoly-
Fig. D110. SEM micrographs of fracture surface morphology of the random heterophasic copoly-
mer 2 with A = 1.3 ~ at room temperature <a) and matrix fibrillation between parti-
cles or cavities in the regime of stable crack propagation
mers with different matrix materials is impossible. The discussion of the influence
of the morphological parameters of particle diameter and distance must be con-
fmed on one system, but in principle the mechanisms in all materials are compara-
ble. Around the cavities, stress fields form, which lead to plastic deformation of
the matrix material, as shown in Fig. BIll for the example of the random hetero-
phasic copolymer 2. With decreasing interparticle distance, the number of particles
increases and therefore the number of initiation processes and the deformed vol-
ume in front of the crack tip increase owing to the increasing number of cavities.
Simultaneously, an increased plastic defonnation of the matrix is initiated, which
depends strongly on particle distance if the critical particle diameter is reached. Fig-
ure BIlla shows that this process is restricted to individual local areas near the
brittle-to-tough transition (A ~ Acril). Above the tough-to-high-impact transition (A
< A 2•crit), all matrix ligaments between the particles or cavities are included in the
deformation process (Fig. Billb) and 'croids' form.
Fig. B111. SEM micrographs of fracture surfaces of the random heterophasic copolymer 2 with
a particle distance of A = 2.9 11m (damage zone formed during unstable crack
propagation) (a) and a particle distance of A = 1.4 11m (regime of stable crack
propagation) (b)
This means that the first transition, from brittle to tough, is connected with a ba-
sic change in defonnation mechanism, from crazing to cavity fonnation with local
plastic defonnation of the matrix ligaments. The second transition, from tough to
high-impact, appears if the whole matrix material is included in the defonnation
process, i.e. most of the matrix ligaments are oriented in the stress direction and
are able to receive load, and 'croidbands' are fonned. In consequence, this process
leads to a strong energy dissipation capacity of the material. According to this
picture, the second transition can be described as a transition to a higher quality of
the general defonnation process above the brittle-to-tough transition.
The aforementioned theories about the nature of the brittle-to-tough transition
are based on conventional notched impact strengths. In conventional impact tests
specimen fracture must occur, otherwise no values can be determined. This means
that only one part of stable crack growth is included in the value for 'notched im-
pact strength', and on this basis it is impossible to separate the stable and unstable
parts of the crack propagation process. There is a possibility that controversy is
due to the fact that the crack propagation behaviour is unknown. The results of the
fracture mechanics investigations presented above, including the discussion of the
relation between morphology and toughness, indicate the validity of Wu's perco-
lation theory [31] for the brittle-to-tough transition. If the region of predominantly
stable crack propagation is reached, the micro-mechanical defonnation mechanism
follows the assumptions of Margolina [32,35], i.e. the critical stress for matrix
yielding is reduced owing to a change from plane strain to plane stress conditions
in the matrix ligaments between the particles with decreasing interparticle distance.
By this mechanism, the stress on the fibrils oriented in the loading direction is
relieved and the energy dissipation capacity of the material increases.
298 S. Seidler, W. GreIImann
3 Summary
The work above demonstrates a correlation between the toughness behaviour and
the morphology of polymer multi-phase materials with a PP matrix. The experi-
mental basis was the instrumented Charpy impact test.
The results show that there is a strong interaction between the matrix properties
and the size, distance and structure of the included particles because the particle
diameter and distance determine the extent and the intensity of the stress fields
formed, which in turn determine, depending on the phase interactions and defor-
mation behaviour of the matrix material, the kind and location of failure initiation.
In the matrix materials, the crack resistance against unstable crack propagation
increases with increasing average molecular weight and decreasing crystallinity.
Simultaneously, the physical crack initiation values increase. The investigations of
the influence of molecular weight on crack toughness behaviour show a deforma-
tion-determined failure behaviour. After a so-called critical molecular weight is
reached, the deformation behaviour is not dominantly influenced by the matrix
properties, but by processes inside the particles or in the matrix/particle interface.
Differences in the failure behaviour are reflected in the dependences of the
crack-opening-displacement rate on the amount of stable crack growth or on J.
In principle, the failure behaviour of the materials investigated is expected to be
determined by the particle distance as well as by the ratio of particle distance to
particle diameter. On the basis of geometry-independent fracture mechanics pa-
rameters, material-specific critical particle distances which enable an optimal ma-
terial design can be determined as a function of test temperature.
The results presented here show that fracture mechanics material testing in
combination with methods of structure analysis and methods of deformation
mechanism analysis can accomplish an important contribution in the field of poly-
mer development.
References
1. Osswald T. A., Menges G. (1995): Materials Science of Polymers for Engineers. Carl Han-
ser, Munich, Vienna
2. Bucknall C. B. (1994): Role of the rubber particle in toughened plastics. In: Proceedings of
Polymat 94, London, U.K.: 567-570
3. Agarwal C. M., Hunter K., Pearsall G. W., Henkens R W. (1992): Application of scanning
tunneling microscopy to the study of fracture morphology of polycarbonate. J. Mat. Sci. 27:
2606--2612
4. BubeckR A., Buckley Jr. D. J., KramerE. J., Brown H. R (1991): Modes of deformation in
rubber-modified thermoplastics during tensile impact. J. Mat. Sci. 26: 6249-6259
5. Michler G. H. (1993): The role of interparticle distance in maximizing the toughness of
high-impact thermoplastics. Acta Polym. 44: 113-124
6. Donald A. M., Kramer E. J. (1982): Craze initiation and growth in high impact polystyrene.
J. Appl. Polym. Sci. 27: 3729-3741
7. Argon A. S. (1988): Sources of toughness in polymers. In: Proceedings of the 7th European
Conference on Fracture (ECF 7), Budapest, Hungary, Vol. II: 2661-2681
8. Mauzac 0., Schirrer R. (1989): Effect of particle volume fraction on crack tip crazes in high
impact poly(methyl methacrylate). J. Appl. Polym. Sci. 38: 2289-2302
9. Guiu F., Stevens R. N. (1991): Physical interpretation of fracture-toughening mechanisms. J.
Mat. Sci. 26: 4375-4382
10. Pearson R. A., Vee A. F. (1991): Influence of particle size and particle size distribution on
toughening mechanisms in rubber-modified epoxies. J. Mat. Sci. 26: 3828-3844
11. Chen T. K., Jan Y. H. (1991): Toughening mechanism for a rubber-toughened epoxy resin
with rubber/matrix interfacial modification. J. Mat. Sci. 26: 5848-5858
12. Gupta A. K., Ratnam B. K. et al. (1992): hnpact toughening of polypropylene by ethylene
vinyl acetate copolymer. J. Appl. Polym. Sci. 45: 1303-1312
13. Michler G. H. (1992): Kunststoff-Mikromechanik. Carl Hanser, Munich, Vienna
14. Galli P. (1994): The breakthrough in catalysis and processes for olefin polymerization:
innovative structures and a strategy in the materials area for the twenty-first century. Prog.
Polym. Sci. 19: 959-974
15. NeiBi W., Ledwinka H. (1993): Polypropylene - the future has only just begun. Kunst. -
Ger. Plast. 83: 577-583
16. Seiler E. (1995): Polypropylene is able to do more. Kunst. - Germ. Plast. 85: 1109-1116
17. Borggreve R. J. M., Gaymans R. J., Eichenwald H. M. (1989): hnpact behaviour of nylon-
rubber blends: 6. Influence of structure on voiding processes; toughening mechanism. Poly-
mer 30: 78-83
18. Borggreve R. 1. M., Gaymans R. J., Schuijer R. J., Ingen Housz J. F. (1987): Brittle-tough
transition in nylon-rubber blends; effect of rubber concentration and particle size. Polymer
28: 1489-1496
19. Ouederni M., Phillips P. 1. (1995): Influence of morphology on the fracture toughness of
isotactic polypropylene. J. Polym. Sci., Polym. Phys. 33: 1313-1322
20. Kotter I., Grellmann W., Seidler S. (2000): Crack toughness behaviour and morphology of
PP-blends. In: Proceedings of Polymeric Materials (P 2000), Halle, Germany, September
25-27:368-371
21. Seidler S., Koch T., Kotter I., Grellmann W. (2000): Crack initiation processes in polymers.
In: Proceedings of the 13th European Conference on Fracture (ECF 13), San Sebastian,
Spain, September 6-9, CD-ROM Polymer and Composites No. 17: 1-8
San Sebastian, Spain, September 6-9, CD-ROM Polymer and Composites No. 24: 1-8
23. Grellmann W., Seidler S., Jung K., Kotter I. (2001): Crack-resistance behavior ofpolypro-
24. Seidler S., Grellmann W. (1999): Application of the crack resistance concept to the tough-
ness characterization of high impact thermoplastics. In: Kahovec I. (Ed.) Mechanical Be-
haviour of Polymeric Materials. Wiley-VCH, Weinheim: 63-72
gener bruchmechanischer WerkstoftkenngroBen bei dynamischer Beanspruchung.
Fortschritt-Berichte, VDI Series 18: MechaniklBruchmechanik No. 18, VDI, Diisseldorf
27. Will P. (1994): R curves for energy dissipative materials. 1. Mat. Sci. 29: 2335-2340
300 S. Seidler, W. GreIlmann
28. Kim G. M., Michler G. H. (1998): Micromechanical deformation processes in toughened

and particle-filled semicrystalline polymers: Part 1. Characterization of deformation proc-
esses in dependence on phase morphology. Polymer 39: 5689-5697
29. Starke J. D., Michler G. H., Grellmann W., Seidler S., Gahleitner M., Fiebig J., Nezbedova
E. (1998): Fracture toughness of polypropylene copolymers: influence of interparticle dis-
tance and temperature. Polymer 39: 75-82
30. Wu, S. (1983): Impact fracture mechanisms in polymer blends: rubber toughened nylon. J.
31. Wu S. (1985): Phase structure and adhesion in polymer blends: a criterion for rubber tough-
ening. Polymer 26: 1855-1863
32. Margolina A. (1990): Toughening mechanism for nylon/rubber blends: the effect of tem-
perature. Polym. Commun. 31: 95 ff
33. Hobbs S. Y., Bopp R. C;, Watkins V. H. (1983): Toughened nylon resins. Polym. Engng.
Sci. 23: 380-389
34. Sjoerdsma S. D. (1989): The tough-brittle transition in rubber modified polymers. Polym.
Commun.30: 106-108
35. Margolina A., Wu S. (1988): Percolation model for brittle-tough transition in nylon/rubber
blends. Polymer 29: 2170-2173
B
Morphology-Property
Correlations
B1 Homopolymers
B2 Blends
B3 Copolymers
B 3.2 Crack Toughness Behaviour
of ABS Materials
R. Lach and W. Grellmann, Merseburg,
and P. KrUger, Leverlrusen
1 Introduction
It is well known that toughness, defined in the engineering sense as the material
resistance against stable and WlStable crack propagation or fracture, is a highly
important material property. Since excessively small values of toughness often
restrict the application fields of polymeric materials, various ways derived from
materials science of toughening brittle polymers (for instance SAN) have been
developed (such as blending with or without compatibilizer, and copolymeriza-
tion). This toughening is achieved by heterogenization of the material, i.e. incorpo-
ration of finely dispersed rubber particles into the matrix material [1]. Morpho-
logical parameters such as particle size and distance, matrix-particle adhesion, and
the internal structure of particles, which vary with production conditions (i.e. syn-
thesis and processing), strongly affect the morphology-toughness correlations
since the processes of stable and WlStable crack initiation and propagation and of
energy dissipation are influenced in different ways.
2 Experimental
ABS materials from Bayer AG, Leverkusen, Germany were used as model materi-
als for the investigations, in which the matrix was fonned from a styrene-acryloni-
trile copolymer (SAN) with a weight-average molecular weight of 85000 glmol. An
acrylonitrile-butadiene-styrene copolymer was used as the rubber phase; it was
synthesized by emulsion polymerization and was finely dispersed into the matrix.
The rubber content was varied from 0 wt. % (SAN) to 28 wt. % in steps of 4 wt. %.
An ABS material with 36 wt. % rubber was also produced. No further additives
were incorporated into the materials except for a very small amount of special
helping agents required for better processing.
Specimens were prepared by injection moulding according to ISO 179 and ISO
3167. For fracture mechanics investigations, single-edge-notched bend (SENB)
specimens (according to ISO 179, with length L = 80 mm, width W = 10 mm and
thickness B = 4 mm) were cut with a razor blade to produce a sharp initial notch
with a depth of a = 2 mm, i.e. the a/W ratio was set to 0.2.
Dynamic mechanical analysis (DMA) was used for the determination of relaxa-
tion and microflow phenomena and for interpretation of the deformation and frac-
ture behaviour.

302 R. Lach, W. GreUmann, P. KrUger
Additionally, the mechanical behaviour as a fimction of temperature Wlder im-

pact-loading conditions was quantified. Fracture mechanics experiments using an
instrumented Charpy impact tester (ICIT) with a pendulwn hammer speed of 1 mls
and a span of s = 40 rom (pp. 71-86) formed the main part of the investigations.
On the basis of an energetic interpretation of the J-integral, J values were de-
termined using the evaluation method developed by Merkle and Corten (MC) (pp.
71-86, [2,3]). Crack-tip-opening displacement (CTOD) values were computed
according to the plastic-hinge model (pp. 71-86, [4,5]). Fracture mechanics mate-
rial parameters related to resistance against unstable crack propagation were meas-
ured in a temperature range between -100°C and 95°C. In cases of no unstable
crack propagation, these parameters were approximated using the maximwn load.
Since the extent of stable crack propagation increases with temperature and/or
rubber content, the description of the material's crack resistance required the de-
termination of crack resistance (R-) curves (T= -20,23 and 60°C). These curves
enabled the determination of the resistance against stable crack initiation and
propagation. J-Aa and 8-Aa R-curve data can be evaluated by means of different
material laws (pp. 71-86, [6]). In the present study, a linear relationship between
the loading parameters (J, b) and the amoWlt of stable crack growth Aa was as-
swned because of the observed distribution of data points (see Sect. 3). The stable-
crack-initiation parameters JO.2 and ~.2 were defined at an amoWlt of crack growth
of Aa = 0.2 rom (the technical crack initiation point). The values of the tearing
modulus TJ and of JTJwere determined according to A 2.2 (pp. 71-86); the deter-
mination of To and oro was analogous. Here, the product JTJ is a sensitive indica-
tor of the energy dissipation. The product oro can be interpreted as the extent of
the plastic deformation of the crack surfaces during crack propagation. This allows
an estimation of the dissipated energy in the crack surface and in the plastic zone.
The results obtained were interpreted by analysing the material composition and
fracture surfaces using microscopes (transmission electron, scanning electron and
optical). Qualitative characterization of the morphology, quantification of particle
sizes for a rubber content of ¢ = 16 wt. % and fracture surface analysis to quantify
selected fracture surface phenomena were performed. In-situ tensile tests were
carried out in a high-voltage electron microscope (HVEM) in order to clarify the
micro-mechanical deformation processes.
3 Results
3.1 Basic Characterization Using DMA
The glass transition temperatures of the matrix (SAN) and of the rubber phase
were experimentally measured using DMA. Both of these temperatures correspond
to local maxima in the temperature dependence of the mechanical loss factor
tan 8 (Fig. Bl12). The glass transition temperature of the SAN phase was found to
Crack Toughness Behaviour of ADS Materials 303
be Tg = 105 DC independent of the rubber content. Thus, no significant influence of

grafting could be detected from the DMA results. By measuring tan t5 as a function
of frequency, relaxation and microflow phenomena activated in various ways
could be detected. Consequently, the material behaviour under impact-loading
conditions could be derived. The glass transition temperature of the rubber at 1 Hz
was determined to be -85 DC, while at a loading frequency of about 1 kHz, which
is characteristic for fracture mechanics experiments under impact-loading condi-
tions, it was nearly -70 DC. This means that up to temperatures close to -70 DC,
athermal brittle fracture behaviour should be observed, i.e. the rubber does not
play the role of a toughening agent. The results of the fracture mechanics experi-
ments provide evidence to confirm this assertion.
10
Tg(SAN)'=
....
I
~
- Tg (rubber)
I
0.1
-
:--0.1 Hz
~0.316 Hz
~\. 1 Hz
\ 3.16 Hz
10 Hz
0.01
-200 -150 -100 -50 0 50 100 150 200
T (OC)
Fig. B112. Mechanical loss factor tan t5 as a function of temperature in a frequency range of
0.1-10 Hz
3.2 Dynamic Flexural Modulus and Dynamic Yield Stress
Mechanical parameters such as the dynamic flexural modulus Ed and the dynamic
yield stress O"d are useful for calculating and interpreting fracture mechanics pa-
rameters. Both of these quantities were computed using the procedure described in
A 2.2 (pp. 71-86).
The mechanical parameters Ed and O"d decrease with increasing rubber content.
This correlates with the modification of a matrix having a relatively high modulus
and strength (SAN) by a dispersed rubber phase possessing a lower modulus and
strength (Fig. B 113). Because neither agglomeration nor percolation of the rubber
phase can be expected, especially at low rubber content (see Fig. B 117), the ob-
served reduction in flexural modulus and yield stress can be described by Niel-
sen's porosity model [7]:
(B18)
304 R. Lach, W. Grellmann, P. KrUger
100
-
80
-cu
D..
~
6OD..
cu
~
uf ~
40
20
0 0
0 10 20 30 40
4» (wt.%)
Fig. BI13. Dynamic flexural modulus Ed and dynamic yield stress O"d as a function of rubber
content at room temperature (points, experimental data; lines, model of Nielsen [7])
In this model, the rubber particles are treated as widely separated holes. Eo and
(To are the flexural modulus and yield stress, respectively, of the matrix. The pa-
rameters mE and mer depend on particle shape.
As is evident from Fig. B 1l3, the experimental results are in very good agree-
ment with Nielsen's model. Deviations from this model at rP = 36 wt.% (and most
probably after this point, too) are caused by the increasing interaction among par-
ticles, i.e. the increased tendency towards agglomeration and percolation of the
rubber phase.
3.3 Load-Deflection Diagrams
The experimental determination of fracture mechanics material parameters was

based on recording load-deflection (F-j) diagrams using SENB specimens. The
principle of evaluation of these diagrams can be found in A 2.2 (pp. 71-86).
The strong geometry dependence of load-deflection diagrams allows only a
qualitative description of the material behaviour.
The load-deflection diagrams plotted in Fig. B 114 demonstrate the variation of
material behaviour as a function of rubber content at T= 23 °e. A transition from
linear elastic to elastic-plastic behaviour with predominantly unstable crack propa-
gation can be found in the concentration range from rP = 0 wt.% to rP = 16 wt.%.
The F-fdiagram schematically shown in Fig. B114 for rP = 16 wt.% corresponds
to elastic-plastic behaviour with a greater amount of stable crack propagation, i.e.
of crack propagation energy. For rP = 20 wt.% and rP = 24 wt.%, predominantly
Crack Toughness Behaviour of ABS Materials 305
stable crack propagation can be observed, although a small amount of unstable

crack propagation also exists. However, for tP = 28 wt.% and tP = 36 wt.%, only
stable crack propagation can be detected. Specimens with tP = 20 wt.% show a so-
called hinge-like fracture. Above tP = 24 wt.%, the specimens do not break at all
under the experimental conditions chosen.
Owt.% 16wt.%
t
LL
20wt.% 36wt.%
1 mm
I I
Fig. B114. F-fdiagrams of materials investigated containing 0, 16,20 and 36 wt.% rubber at
room temperature
3.4 Fracture Mechanics Material Parameters as a Function

of Temperature and Rubber Content
By analysing F-f diagrams, fracture mechanics material parameters can be easily

evaluated. An important parameter, the J-integral, plotted in Fig. Bl15a as a func-
tion of temperature and rubber content, makes it possible to interpret the material
behaviour energetically. J values can be calculated using different evaluation
methods (pp. 71-86).
Because of the relatively large plastic part of the general deformation energy,
the evaluation method suggested by Merkle and Corten [2] was used. In contrast to
other procedures, this approach considers more precisely the influence of plastic
energy on Jvalues [2,3].
Predominantly athermal brittle fracture behaviour, corresponding to small J
values, can be observed at temperatures up to -60 °C. Therefore, this temperature
306 R. Lach, W. Grellmann, P. Kruger
range was identified as the lower shelf of toughness. Moreover, the toughness was
not significantly improved up to rp = 12 wt.%. A significant increase in toughness
was achieved after the rubber content exceeded 12 wt.%, the J values reaching a
maximum at rp = 28 wt.%. The evaluation of .J..dMC and J QdMC as a function oftem-
perature showed that a high rubber content (rp = 36 wt.%) has a negative influence
on the toughness behaviour. It is especially noteworthy that the J values decrease
at higher temperatures after passing through a maximum, i.e. a high level of tough-
ness. This is caused, on the one hand, by a decreasing flexural modulus and, on the
other hand, by formation of so-called 'intrinsic crazes' [8], which results in em-
brittlement of the SAN matrix near its glass transition.
16 140
b
14
120
12
100
E
~10 E80
2:
~ .., 8 'i:!
00
-
""')
~
~ 00
:2 6
""')
40
4
2 20
... •
0 0
-100 -50 0 50 100 -100 -50 0 50 100
T (0G) T(°G)
Fig. B115. J values JldMC and Jd MC (a) and CTOD values ~dk and bdk (b) as a function of tem-
perature: 4 wt.% (squares), 6 wt.% (circles) and 28 wt.% (triangles) rubber (filled
symbols, JldMC , ~dk; open symbols, JdMC , bdk)
Crack-tip-opening displacement values bidk and 8dk were calculated using the
modified plastic-hinge model, replacing the maximum load by the notch part of the
maximum load. In principle, these values, when plotted against temperature
(Fig. B 115b), should show the same dependence on rubber content and temperature
as the J values. Values of 8dk smaller than 0.03 mm correspond to predominantly
athermal fracture behaviour, where mostly small-scale yielding can be observed.
This observation is in agreement with earlier investigations on other amorphous
polymeric materials [9,10]. Considering the toughness reduction at higher tem-
peratures, a brittle-to-tough transition for the materials investigated can be for-

mally derived by averaging the lower and the upper shelf of toughness.
The differences between the lower and the upper shelf of toughness, for exam-
ple the increase in the J values M shown in Fig. B 116, can be used to characterize
the toughness increase resulting from rubber modification. A toughness enhance-
ment relevant to applications of the materials cannot be observed up to ¢ =
12 wt.%. After this point, M increases up to a rubber content of ¢ = 28 wt.%.
However, further addition of rubber causes no more toughness improvement.
1~---------------------------------r20
• o 6
1
~
E
.E
8
...,~
4
•
• •
o+---r-----~-------.------~------+~O
o 10 20 30 40
+(wt.%)
Fig. B116. Toughness increase M as a result of rubber modification (filled circles) and brittle-
to-tough transition (BTT) temperatures corresponding to JdMC (open squares) as a
function of rubber content
Figure B 116 also shows a plot of the brittle-to-tough transition (BIT) tem-
perature versus rubber content. These values were calculated from the temperature
dependence of the J-integral. For rubber contents lower than ¢ = 16 wt.%, BTT
temperatures do not exist because of the brittle behaviour of these materials.
On the basis of results for toughened polymers [11,12] and our own investiga-
tions on modified nylon (pp. 209-228 and pp. 243-256), it can be concluded that
the BTT temperatures decrease in most cases with increasing rubber content.
However, in order to understand the increasing BTT temperatures observed in this
study, we must take into account that toughness is dependent on particle size and
distance because of the varying rubber volume content. For this reason, morpho-
logical aspects of rubber modification will be discussed in the following section.
308 R. Lach, W. GreUmann, P. KrUger
3.5 Morphological Investigations
For the purpose of qualitative and quantitative morphology analysis, transmission

electron microscope investigations were performed for ABS with 16 wt.% rubber
using ultra-thin Os04-stained sections. A trimodal particle size distribution, with a
relatively high number of small spherical particles about 130 nm in size without
inner structure, a high number of star-like particles about 280 nm in size with an
inner structure because of micro-phase separation, and a relatively low number of
large spherical particles without clear inner structure with a size of about 480 nm
were observed (Fig. B 117).
Fig. B117. Transmission electron micrograph of ultra-thin section of ABS with 16 wt.% rubber
Each size group of particles was reflected in a clear maximum in the particle
distribution measured by means of ultra-centrifugation (Fig. Btt8). On the basis of
Fig. B 118, it was found that the large particles are relatively rare, but they should
not be ignored, because of their large volume. This assertion can be confirmed by
examining the particle volume distribution. Particles larger than 370 nm contain
more than 67 % of the whole rubber volume. These large particles are important in
the formation of crazes and in the initiation of brittle crack propagation.
3.6 In-Situ Tensile Tests by HVEM
In-situ tensile tests using a high-voltage transmission electron microscope (HVEM)

were performed to characterize the micro-mechanical deformation processes.
1 ... Small Particles (- 0.13 Ilm)

2 ... Medium-Sized Particles (- 0.28 Ilm)
3 ... Large Particles (- 0.48 Ilm)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Particle Size (\.1m)
Fig. B118. Particle size distribution of ABS with 16 wt.% rubber measured by means of ultra-
centrifugation
On the asstunption that the thermal expansion coefficient of the matrix is

smaller than that of the rubber phase at temperatures above the glass transition tem-
perature of the rubber, a force-locked connection between the phases is formed as
a consequence of tensile residual stresses in the rubber particles and compressive
residual stresses in the matrix. If the adhesion between the matrix and the rubber
particles is small (effective interfacial energy < - 1 kJ/m2), the formation of holes
close to the interfaces and the formation of crazes between particles, the so-called
multiple crazing, can be observed. Crazes are microflow zones filled with highly
stretched matrix fibrils. Larger particles initiate the formation of crazes with a
greater probability than smaller particles, which are more likely to initiate shear
flow mechanisms. If stress fields interact because of a high amount of rubber,
shear flow is preferred to crazing. For lower rubber contents, extended crazes in
the matrix can be observed that are comparable to notches, and these can initiate
early brittle crack propagation.
Holes are formed inside the rubber particles if there is good interfacial adhe-
sion, whereby the stress level in the matrix is reduced, so that shear flow is initi-
ated in the matrix material. For the ABS materials investigated, this mechanism
dominates at room temperature (Fig. B 119). However, fracture surface phenomena
induced by formation of crazes were observed at lower temperatures and/or for
lower rubber contents.
310 R. Lach, W. GreUmann, P. Kruger
Fig. B119. Deformation behaviour of ultra-thin sections observed using an in-situ tensile test in
the HVEM: formation of holes inside the rubber particles (a); plastic deformation of
matrix between cracked particles combined with formation of fibrils and further
formation of holes (b, c); matrix is nearly completely plastically deformed (d)
3.7 Analysis of Fracture Surface and Fracture Frequency
Some selected fracture surface phenomena were quantified by means of optical

and scanning electron microscopy to interpret the fracture mechanics behaviour.
The changes in fracture morphology are schematically shown in Fig. B120 as a
function of rubber content. Close to the notch (1), a stable-crack-growth area (2)
corresponding to stable crack propagation and a damaged area (3) resulting from
formation of crazes were found. Both the damaged area and the stable-crack-
growth area mark the extension of the plastic zone over the fracture surface. At
T= 23°C and rubber contents up to ¢ = 12 wt.% a rough brittle-crack-growth area
was detected close to these two areas. The specimens did not break at rubber con-
tents above ¢ = 24 wt. %.
(6) (6)
(4) (4) (2) ~

....-----....
..-(3) ___ (2) (2)
---(3) - =-;2) -
(1) (1) (1) (1) (1)
Fig. B120. Scanning electron micrographs of fracture surface for ABS (top): (1) notch, (2) area
of stable crack, (3) damaged area, (4) area of brittle crack growth, (5) broken after
test; schematic illustration of fracture surface morphology as a function of rubber
content (bottom)
Characteristic fracture frequencies 'F, plotted in Fig. B121, were calculated

from time to fracture or to maximum load. The fracture frequencies of nearly brit-
tle behaviour are typically about 1 kHz. The materials investigated exhibit brittle
fracture behaviour at lower temperatures. Corresponding to the decrease in frac-
ture frequency, they become more ductile if the temperature increases. However, a
more or less clear minimum that is dependent on rubber content occurs, where the
embrittlement of SAN near its glass transition temperature has a decreasing influ-
ence on the fracture frequency as the rubber content increases. On the one hand,
this phenomenon is caused by a decreasing matrix content with increasing degree
of modification. On the other hand, this phenomenon should be interpreted in
terms of changes in the deformation mechanisms. Analogous results were found
for the amount of stable crack growth, the length of the damaged area and the
geometry of the stretch zone.
1000
800
700
600
-
N
~
;:r
500
400
300
200
100
-100 -50 0 50 100
T (0C)
Fig. B121. Fracture frequency 'F as a function of temperature and rubber content: 4 wt.%
(squares), 16 wt.% (circles) and 28 wt.% (triangles)
3.8 Crack Resistance (R-) Curves
For ABS with rubber contents ¢ ~ 16 wt.%, crack resistance curves characterizing
stable crack propagation were measured under dynamic loading and at various
temperatures (T= -20,23 and 60 DC). From these curves, fracture mechanics ma-
terial parameters related to resistance against stable crack initiation and propaga-
tion were estimated. No arrest of stable cracks was achieved at rubber contents
¢ < 16 wt. %, because the stable crack growth was too small, i.e. the determination
of R-curves was not possible. The J-l1a and t5--l1a data must be independent of
specimen geometry if material-specific geometry-independent fracture mechanics
parameters are to be determined. Because of the limited specimen size, especially
the limited ligament length, a predominant plane strain state is transformed into a
predominant plane stress state. Corresponding to this transition, a strong increase
of the J and 0 values was found because the external energy was not dissipated in a
greater volume but instead caused a greater plastic deformation in a limited vol-
ume. The meaning of this behaviour is that useful, 'valid' data can be only meas-
ured up to the beginning of rapid changes in the J and 0 values (Fig. B 122), i.e. for
l1a ~ l1amax (Table B 18). The geometry independence of the fracture mechanics
parameters for l1a ~ l1amax was confirmed by means of generally valid a-J and ~-o
functions (pp. 3-26, pp. 71-86, [6,9]) using the experimentally determined maxi-
mum amount of stable crack growth l1amax and J(l1amax) or ~l1amax), where & and ~
are proportionality constants of the size criteria in the J-integral and COD concepts
(Fig. B123).
16
o
14 o
o
12 o
o
10
--
z
E
E 8
6
4
2
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
~a (mm)
Fig. B122. Evaluation of a J-M crack resistance curve (,p = 24 wt.%, T= 23°C); valid (filled
circles) and invalid (open circles) data points; M max, maximum valid amount of
crack growth
1000 o
100
10
1
0.1 1 10 100
J (N/mm)
Fig. B123. Proportionality constant & of the size criterion in the J-integral concept as a func-
tion of J; experimental data from A 2.2 (pp. 71-86) (open circles); data from R-
curves measured in this study: -20°C (filled diamonds), 23°C (filled circles), 60°C
(filled squares)
TableB18. Fitting functions for J-R and 8-R curves and maximum valid amount of crack
growth&lmax
Tee) ¢(wt.%) fitting functions, fitting functions, L'.amax &lmax

J(&I) ~&I) forJ(mm) for 8(mm)
-20 16 J=3.4&1-0.2 8=0.03&1 0.62 2.67

20 J= 6.4 L'.a - 0.9 8=0.05&1 2.12 2.12
24 J= 1.6 L'.a + 0.3 8= 0.01 &I + 0.01 3.09 3.09
28 J= 2.1 L'.a + 0.2 8= 0.01 L'.a + 0.03 1.25 3.32
36 J= 4.9 L'.a + 0.1 8= 0.07 L'.a + 0.D1 0.82 2.65
23 16 J= 0.4 L'.a + 0.1 8=0 3.56 3.56
20 J=6.2 &1-0.2 8=0 2.25 2.25
24 J= 10.7 &1-1.2 8=0.03 L'.a 0.60 0.60
28 J= 11.7 &1-0.8 8= 0.07 &I + 0.01 0.64 0.73
36 J = 11.2 &I - 0.6 8=0.03 L'.a 0.66 0.66
60 16 J= 0.4 L'.a + 0.3 8=0 0.83 0.63
20 J= 1.4 L'.a + 1.0 8= 0.01 L'.a + 0.01 0.75 4.31
24 J= 11.7 L'.a-1.1 8= 0.10 L'.a - 0.02 0.71 1.00
28 J= 7.4 L'.a- 0.4 8=0.09 L'.a 0.46 0.61
36 J= 9.3 L'.a - 0.2 8=0.11 L'.a 1.10 1.10
For the materials investigated, on the basis of valid data evaluated by the proce-
dure described above, fracture mechanics parameters related to resistance against
stable crack initiation and propagation (JO.2, ~1.2' TJ , T;;, JTJ and 01'8) were deter-
mined at the stable-crack-growth point tJ.a = 0.2 mm at various temperatures.
The technical crack initiation parameter J O.2 increases with increasing rubber
content, and is nearly independent of temperature at rjJ = 16 wt.% (Fig. BI24a).
The CTOD parameter ~.2 at room temperature shows a concentration dependence
that is different from that of J O.2• At rjJ = 28 wt. %, a toughness maximum resulting
from higher plastic deformation was found (Fig. B 124b). On the basis of a geomet-
rical interpretation of the crack-tip-opening displacement, small ~.2 values are
reflected in small values of the stretch zone height of about 5 11m.
Earlier investigations demonstrated that the crack initiation parameters reflect
mainly changes in the matrix, while the crack propagation parameters are deter-
mined by the dispersed phase (pp. 3-26, [13]).
The observed dependences on rubber content of the technical crack initiation
parameters are related to changes in the microstructure of the SAN because of
formation of grafted shells around the particles, corresponding to chemical and
physical interactions between the matrix and the particles. This means that the influ-
ence of rubber on the crack initiation behaviour should be greater if the rubber con-
tent increases. An example of this behaviour is demonstrated in Fig. B124b. The ~.2
values are constant up to 24 wt.%. For higher rubber contents, individual concentra-
tion dependences can be observed.
Considering both deformation- and energy-determined toughness values related

to resistance against stable crack initiation and propagation, it can be said in swn-
mary that the ABS materials investigated show, at room temperature, an optimwn
toughness at 28 wt. % rubber. The material behaviour is determined either by the
strength or by the deformation, but the influence of deformation increases with
increasing temperature.
2.0
1.5
E
E
--
Z 1.0
N
ci
..., 0.5
0.0
15 20 25 30 35 40
cI> (wt.%)
b
40
........
E 20
E
'"b
--
.....
N
r-8
0
15 20 25 30 35 40
cI> (wt.%)
Fig. B124. Stable-crack-initiation parameters JO.2 (a) and ~.2 (b) measured at technical initia-
tion point (&l = 0.2 mm) as a function of temperature and rubber content: -20°C
(open squares), 23°C (filled circles), 60°C (crosses)
Acknowledgements
The German Research Foundation (DFG) is gratefully acknowledged for financial

support for making possible investigations on the subject 'Influence of morphol-
ogy for polymeric multiphase systems on stages of crack propagation'. For the
opportunity to perform in-situ tensile tests using a high-voltage transmission elec-
tron microscope (HVEM), the authors express their thanks to the Max-Planck-
Institut fUr Mikrostrukturphysik, Halle, Germany.
316 R. Lach, W. GreUmann, P. Kriiger
References

2. Merkle J. G., Corten H. T. (1974): ASME Paper 74. J. Pressure Vessel Techno!. 96,4: 286ff.
ries, No. 17, Institut fUr Mechanik, Berlin, Chemnitz: 48-72
Charpy impact test. In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Se-
ries, No. 37, Institut fUr Mechanik, Berlin, Chemnitz: 186-192
5. Leevers P. S. (1999): Fracture mechanics. In: Swallowe G. M. (Ed.) Mechanical Properties
and Testing of Polymers - An A-Z Reference. Kluwer Academic, Dordrecht: 96-101
determination on polymers. In: Winkler T., Schubert A. (Eds.) Material Mechanics - Frac-
ture Mechanics - Micromechanics. DDP Goldenbogen, Dresden, Germany: 336-341
7. Nielsen I. E. (1967): Mechanical properties ofpartic1e-filled systems. J. Compos. Mater. 1:
100-119
8. Michler G. H. (1992): Kunststoff-Mikromechanik. Morphologie, Deformations- und Bruch-
9. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour of poly(methyl meth-
acrylate), polystyrene and polycarbonate. App!. Macromo!' Chern. Phys. 253: 27-49
10. Lach R., Grellmann W. (2000): Estimation of the resistance against stable crack initiation
and unstable crack propagation, R-curves and stability assessment diagrams in ductile poly-
meric ABS materials. In: Proceedings of the 13th European Conference on Fracture (ECF
13), San Sebastian, Spain, September 6-9, CD-ROM Polymer und Composites No. 20: 1-8
11. Gaymans R. J., Borggreve R. J. M., Oostenbring A. J. (1990): Toughening behavior of
polyamide-rubber blends. Macromo!. Chern. Macromo!' Symp. 38: 125-136
12 Borggreve R. J. M., Gaymans R. J., Schuijer J. (1989): Impact behaviour of nylon-rubber
blends: 5. Influence of the mechanical properties of the elastomer. Polymer 30: 71-77
13. Seidler S., Grellmann W. (1993): Fracture behaviour and morphology of PC/ABS blends.
J. Mat. Sci. 28: 4078-4084
B 3.3 Fracture Mechanics Characterization
of ABS Materials -Influence
of Morphology and Temperature
R. Lach and W. Grellmann, Merseburg, Y. Han,
Changchun, and P. KrUger, Leverkusen
1 Introduction
With the growing application field of polymer materials, not only for common but
also for technological purposes, the optimization of toughness behaviour, in addi-
tion to the improvement of stiffness and strength, has been become a major goal in
polymer research [1-4]. The incorporation of a well-dispersed rubber phase into
the matrix material has been shown to increase the toughness of glassy amor-
phous, macroscopically monophase polymers such as SAN or PS that break in an
extremely brittle manner [5]. In this connection, ABS materials are an important
material group with a broad range of engineering applications [6]. Normally, de-
pending on the particular application, commercial products contain 15-20 wt.%
rubber, which corresponds to a balance between toughness, stiffness or strength
and price. Depending on the manufacturing process, ABS materials containing
rubber particles with an average particle diameter of 0.1-1 flm show the most
pronounced influence on toughness behaviour at room temperature (optimum
particle size) [7-9]. This observation was verified in [9] by use of a model to esti-
mate the craze volume. Whereas particles with sizes of 0.1-0.5 flm are usually
effective in craze initiation, particles < 0.05 flm are not able to initiate crazes (10].
In the case of emulsion-polymerized ABS, the optimum particle size is about
0.3 /lm. ABS materials with a multi-modal particle size distribution are often
fabricated because other properties besides toughness have a great influence on
the overall product quality, for instance hardness, scratch resistance and surface
finish. Although ABS materials with broad or multi-modal distributions contain
particle fractions that have a lower effectiveness for plastic deformation of the
matrix (pp. 335-352), it has been found that their impact resistance is higher than
that of materials with narrow distributions [11]. However, the long-time failure
behaviour can also be affected negatively (pp. 335-352). ABS materials with a
particle diameter of 0.05 flm are scarcely able to dissipate any energy in the im-
pact test, although these materials show a high energy dissipation in the tensile test
that is comparable to that of ABS materials with 0.5 flm particles [9].
Using an example of high-impact polystyrene (HIPS) with various particle
sizes (0.3 flm, core-shell morphology, and 3 flm, salami structure), it could be
demonstrated using the instrumented Charpy impact test that the 'initiation en-
ergy' (deformation energy up to the maximum load) is not influenced by the mor-
phology but is characteristic of the matrix [12]. In contrast to this result, the 'crack
propagation energy' (energy after passing through the maximum load) is strongly
morphology-dependent [12]. However, it must be pointed out that the crack

318 R. Lach, W. Grellmann, Y. Han, P. Kriiger
propagation energy is a function of the specimen geometry. Therefore, only by use

of fracture mechanics parameters can a significant improvement of the quality of
the structure-toughness relationships be obtained, because of
- the possibility of separate evaluation of different parts of the energy and de-
formation
- the geometry independence of the fracture mechanics parameters
- the existence of direct correlations with defined phases of crack propagation
(an event-related interpretation of the fracture process).
However, only a small number of fracture mechanics investigations in which
the main concern is the crack resistance behaviour of ABS materials can be found
in the literature [13-17]. Most of these investigations were performed using quasi-
static loading conditions [13]; impact-like loading conditions were very seldom
used [13]. For adequate quantification of the toughness behaviour of ABS as a
typical impact-toughened polymer material that shows pronounced elastic-plastic
behaviour and predominantly stable crack propagation, the various concepts of
elastic-plastic fracture mechanics that reflect different aspects of the mechanical
behaviour should be used. Therefore, in future, the main way to gather informa-
tion should be toughness assessment under impact-like loading conditions, for
example the determination of so-called 'dynamic' crack resistance curves. In some
publications, decreasing slopes of R-curves and lower J O.2 values have been ob-
served as a function of the test speed [18,19], and increasing slopes and values as
a function of the rubber content [14,16]. The J O.2 values for two materials showed
opposite behaviour in the temperature range from -50°C to 20°C [13]: whereas
JO.2 increased for a bulk-polymerized ABS with increasing temperatures, J O.2 de-
creased for an emulsion-polymerized ABS over the same range of temperature;
this is caused by the different microstructures. From the dependences of the crack
initiation parameters and the tearing moduli on temperature (-20 to 80°C) for
different emulsion-polymerized ABS materials, the crack initiation parameters
were found to have a minimum value and the tearing moduli a maximum value at
80°C [16].
2 Experimental
2.1 Materials and Microstructure
The materials mostly used in this study were ABS materials with rubber contents
varying from 0 wt.% (pure SAN) to 36 wt.%, with two different particle size dis-
tributions (Figs. Bl25 and BI26). These ABS materials were manufactured by
mechanical compounding (extrusion) of various amounts of rubber particles (ac-
rylonitrile-butadiene-styrene copolymer) made by an emulsion process [6] with a
separately produced commercial SAN (Mw = 85000 glmol, Mv/Mn = 4, ratio of
styrene to acrylonitrile 72.5:27.5, content of oligomers about 1 %). After extrusion
at 180°C (during which the rubber particles were finely dispersed into a SAN
Fig. B111. SEM micrographs of fracture surfaces of the random heterophasic copolymer 2 with
a particle distance of A = 2.9 11m (damage zone formed during unstable crack
propagation) (a) and a particle distance of A = 1.4 11m (regime of stable crack
propagation) (b)
This means that the first transition, from brittle to tough, is connected with a ba-
sic change in defonnation mechanism, from crazing to cavity fonnation with local
plastic defonnation of the matrix ligaments. The second transition, from tough to
high-impact, appears if the whole matrix material is included in the defonnation
process, i.e. most of the matrix ligaments are oriented in the stress direction and
are able to receive load, and 'croidbands' are fonned. In consequence, this process
leads to a strong energy dissipation capacity of the material. According to this
picture, the second transition can be described as a transition to a higher quality of
the general defonnation process above the brittle-to-tough transition.
The aforementioned theories about the nature of the brittle-to-tough transition
are based on conventional notched impact strengths. In conventional impact tests
specimen fracture must occur, otherwise no values can be determined. This means
that only one part of stable crack growth is included in the value for 'notched im-
pact strength', and on this basis it is impossible to separate the stable and unstable
parts of the crack propagation process. There is a possibility that controversy is
due to the fact that the crack propagation behaviour is unknown. The results of the
fracture mechanics investigations presented above, including the discussion of the
relation between morphology and toughness, indicate the validity of Wu's perco-
lation theory [31] for the brittle-to-tough transition. If the region of predominantly
stable crack propagation is reached, the micro-mechanical defonnation mechanism
follows the assumptions of Margolina [32,35], i.e. the critical stress for matrix
yielding is reduced owing to a change from plane strain to plane stress conditions
in the matrix ligaments between the particles with decreasing interparticle distance.
By this mechanism, the stress on the fibrils oriented in the loading direction is
relieved and the energy dissipation capacity of the material increases.
320 R. Lach, W. Grellmann, Y. Han, P. Kruger
475 run, Fig. B126b). For comparative purposes, results from an ABS with an
average particle diameter d= 270 run were also used [19]. The styrene-acryloni-
trile ratio in the grafted SAN, forming extended shells nearly 10 run thick around
the rubber particles, was nearly the same as in the matrix SAN.
2.2 Fracture Mechanics Investigations
Single-edge-notched bend (SENB) specimens produced by injection moulding

according to ISO 179, with a length L = 80 mm, width W = 10 mm and thickness
B = 4 mm, were used. As a result of an optimization process combined with FEM
simulations [20,21], the specimens were sharply notched with a razor blade (notch
tip radius ~ 0.2 !lm); the initial crack length a was 2 mm (a/W= 0.2) and the sup-
port span s was 40 mm (s/W= 4). Thus, elastic oscillations of the specimen during
the test could be minimized, and fracture mechanics material parameters that were
independent of specimen geometry could be utilized (pp. 71-86, [22,23]).
An instrumented Charpy impact tester of 4 J maximum work capacity was used
for the measurements, and load-deflection (F-j) diagrams were recorded. The
pendulum hammer speed was set to VH = 1.0 m/s to realize quasi-static loading
conditions inside the material close to the crack tip. Only if such loading condi-
tions exist and no oscillations of stress intensity are detected during crack propa-
gation (dynamic crack propagation) can the methods of quasi-static fracture me-
chanics be successfully applied, for example the J-integral and the crack-tip-
opening displacement (CTOD) concept of elastic-plastic fracture mechanics. On
the basis of the energetic interpretation of the J-integral, J values were computed
according to the method of Merkle and Corten [24]. This evaluation method is
especially well suited for the quantitative fracture analysis of materials that show
pronounced changes in their behaviour, i.e. changes from linear elastic to pre-
dominantly plastic behaviour as a function of morphology and temperature. Com-
pared with other methods, for instance the procedure of Sumpter and Turner [25],
the plastic parts of the total energy are consid(!red to a much greater extent in the
method of Merkle and Corten. Because a direct measurement of the CTOD was
not possible, CTOD values were calculated using the plastic-hinge model intro-
duced by Kobayashi [26] and extended by Jungbluth (see A 2.2, pp. 71-86, [27]).
The CTOD values bdk computed from the notch part of the maximum deflection
are sensitive indicators of the amount of predominantly plastic deformation close
to the crack tip (pp. 181-192).
The fracture mirror length aBS and the plastic-zone size were measured on the
fracture surface by light microscopy. Further information about the fracture proc-
ess was derived using SEM investigations of fracture surfaces sputtered with gold.
The standard deviation of all quantities in this study was about 10 %.
Fracture Mechanics Characterization of ADS Materials 321
3.1 Morphology of Fracture Surfaces
The quantitative and qualitative analysis of fracture surfaces is necessary in order

to calculate fracture mechanics parameters, especially crack resistance curves, and
to derive generalized information about the fracture process from structure-tough-
ness relationships. With this in mind, it must be pointed out that the CTOD can be
interpreted geometrically, in the sense that direct correlations between the extent
and form of the blunting zone and of the plastic zone and the values of Odk exist
[28]. However, it must be taken into account that fracture surface phenomena are
strongly geometry-dependent: the behaviour becomes more brittle if the specimen
thickness increases.
ABS 110 breaks in a brittle manner over the whole ranges of rubber content
and temperature; this corresponds to only small amounts of plastic deformation,
and the dominant crack mechanism is unstable crack propagation [29,30].
Whereas ABS 330 with 4 wt.%, 16 wt.% and 28 wt.% rubber at T:s -40 °C shows
the same behaviour as ABS 110, semiductile behaviour can be observed in
ABS 330 with 16 wt.% rubber at temperatures higher than O°C [30], and highly
ductile behaviour can be observed in ABS 330 with 28 wt.% rubber at tempera-
tures higher than -40 °C [30,31]. This material shows intense stress whitening of
the entire fracture surface. At room temperature for ABS 330, a transition point
from predominantly unstable to stable crack propagation is found at 16 wt.% rub-
ber [30,31].
Figure B 127 shows a typical micrograph of a fracture surface at the transition
from brittle to ductile behaviour. In Fig. B128 the internal structure of the plastic
zone, which is reflected on the fracture surface, is schematically illustrated (pp.
301-316, [29,30]).
As a result of rising temperatures in areas of the material in front of the crack
tip because of the transformation of plastic-deformation energy into heat and be-
cause of the growing stable crack, stress relaxation close to the crack and stress
rearrangement into material far from the crack occur. This produces a separation
of the plastic zone into two parts:
- In the areas close to the crack, shear deformation mechanisms such as shear
yield are dominantly stimulated because of the higher temperatures and lower
stresses. At the microscopic scale, the shear yield process follows cavitation of
rubber particles. That is a main cause of the stress whitening, besides multiple
crazing (pp. 301-316). The cavitation is connected with a further reduction in
the stress level, which results in a thermodynamic stabilization of the shear
yield process. This is the main reason for the dominance of the shear yield pro-
cess at high temperatures and high rubber contents, which means small particle
distances.
322 R. Lach, W. Grellmann, Y. Han, P. KrUger
- In areas far from the crack, the stress level in the equatorial regions of rubber
particles must be high enough to initiate plastic deformation. Only crazes are
formed, i.e. mUltiple crazing.
As a consequence of the observed separation of the plastic zone, stable crack
propagation is observed only in the zone of shear deformation. Thus, this part of
the plastic zone correlates with the stable fracture surface (fracture mirror length,
aBS), which shows a typical flow pattern (Fig. BI27). If the crack grows to the
beginning of the second part of the plastic zone (the craze zone), the crazes inside
this zone act as initiators of unstable crack propagation. The fracture surface phe-
nomenon that is formed during crack propagation is called the damage area
(Fig. B 128). On the fracture surface, structures are identifiable that indicate the
formation of holes or crazes in the range of several micrometers in size (Fig.
BI27).
Fig. B127. SEM micrographs of a fracture surface: detail of stable fracture surface close to the
notch (a), detail of damage area far from the notch (b)
(4) Unstable Crack Surface

r·-·-·-·····!·········-·····················
d,
I I i
1
Craze Zone Damaged Area
····· -r··z~;~~~~~:·~ ··· Stable Fracture Surface

··········l····················_·······_·.......-(-1-)-1104 Notch
Fig. B128. Schematic diagram showing formation of various zones in front of the crack tip, and
their correspondence with the fracture surface [29,30]
Fracture Mechanics Characterization of ABS Materials 323
0.6 r-----------r-::-1
b
d= 110nm d=110nm
... 23 DC • 4 wt.%
+ 16 wt.%
0.4
e- o 28 wt.%
.sgJ
C1I 0.2
0.0
10
d = 330 nm l£ d = 330 nm d
8
...23 DC • 4wt.%
+ 16 wt.% :
e- o 28wt.% :
: Stable
.s
CI)
6
,
,, -
co'" ,
4 Stable , ~
,
~+
2
0 ~~
0 5 10 15 20 25 30 35 40 -100 -50 0 50 100
Rubber Content (wt.%) Temperature ("C)
Fig. B129. Fracture mirror length aDS as a function of rubber content (a, c) and temperature
(b, d) for ABS 110 (a, b) and ABS 330 (c, d) [29,30]
On the basis of microscopic investigation methods, selected fracture surface

phenomena such as the fracture mirror length aBS (Fig. B129) and the plastic-zone
size (Fig. B130) could be quantified in order to derive information about the mate-
rial behaviour.
Generally speaking, the fracture mirror length aBS and the plastic-zone size re-
flect the material behaviour, especially the increasing toughness, as a function of
rubber content, particle size and temperature. However, it was usually observed
that the values of aBS and the plastic-zone size decreased with increasing tem-
perature after passing through a maximum value. This effect, which does not re-
sult from physical ageing [30], was exceptionally evident for ABS with 16 wt.%
rubber. A possible explanation is matrix embrittlement by formation of so-called
intrinsic crazes [9,32,33].
Using the plastic-zone model of Dugdale, the plastic-zone size rplCal can be cal-
culated from the following equation [34-36]:
2
K
r cal -1r- - - (BI9)
pi - 8 mad'
where K is the stress intensity factor derived from the maximum load and m is the
constraint factor. According to ASTM 813 [37], m should be a constant, equal to
I, but experimentally it has been found that m is dependent on the material and
other parameters (for polymers m = 0.1-3, summarized in [28]). A comparison
between the experimental and calculated plastic-zone sizes (with m = 0.5 and
m = 1) suggests that the plastic-zone size of the more ductile materials, i.e. ABS
110 with higher rubber contents or at higher temperatures, can be described using
m = 0.5 because of the greater constraint of these materials, corresponding to
smaller m values (Fig. B 130). Otherwise, a rather good estimation of the plastic-
zone size is possible using m = 1.
2.0,....---------------r---.
d=110nm.T=23'C a
Experimental Calculated
• -m=1
_._.- m = 0.5
.~
~ 1.0
c:
~ 0.5
II)
cu
0:
o 5 10 15 20 25 30 35
Rubber Content (wt.%)
2.0.------'1--,
h
Exp. cal. ~
• - - 4wt.%
• --16wt.% 0
E o ._. 28wt.% 0 0
-
1.5
E
1m =11 00
IS
Ci5 1.0
Q)
~~ 0.5
0:
0.0 -100 -50 0 50 100 -100-50 0 50 100

Temperature eC} Temperature (DC)
Fig. B130. Comparison of experimentally detennined and calculated (using (B I 9» plastic-zone

size as a function of rubber content (a) and temperature (b, c) for ABS 110
Because ABS 330 is more ductile than ABS 110, one would expect that, as a
matter of principle, the constraint factor m = 0.5 would describe the size of the
plastic zone for all rubber contents and temperatures quite well. It could be hy-
pothesized that materials that show exponential stress-hardening behaviour in
accordance with the theory of Ramberg and Osgood would have a constraint fac-
tor m = 1, corresponding to an exponent in the crack resistance curves of 0.5, as
assumed by Will et al. [38,39]. However, if no stress hardening can be observed,
as in this study for ABS materials at higher rubber contents and higher tempera-
tures, the R-curves should be straight lines (i.e. their exponent is equal to 1). In
this case the constraint factor is equal to 0.5. A validation of the last part of this
assumption is provided by the shape ofthe measured R-curves for the example of
ABS 330 (see B 3.2 (pp. 301-316 and [31,40] and the following pages).
3.2 Crack Resistance Curves Under Impact-Like Loading Conditions
The general practice used to determine crack resistance curves under impact-like
loading conditions is described in A 2.2 (pp. 71-86) in detail. In B 3.2 (pp. 301-
316), the R-curve procedure was used for ABS materials.
Because ABS 110 materials show macroscopically only linear elastic behav-
iour, corresponding to predominantly unstable crack propagation, no crack arrests
and, thus, no R-curves could be produced. Analogously, an R-curve determination
of the ABS 330 materials was also impracticable at low temperatures and/or low
rubber contents because these materials show predominantly unstable crack
propagation at temperatures < -40 °C and rubber contents < 16 wt. %.
Figure B131 shows J-R and 8--R curves of ABS 330 as a function of rubber con-
tent (at room temperature) and temperature (for 24 wt.% rubber) [31,40]. From Fig.
B 131 a,c it is clearly visible that the slopes dJ/da and d&'da of the R-curves increase
with increasing rubber content [41]. The resistance against stable crack propagation
dJ/da and dada passes through a maximum value in the region of room temperature
(Fig. B 131 b,d) [41]. In this connection, the pronounced toughness decrease at 95°C
accompanied by a clearly visible increase in the scatter of the data should be pointed
out (this means the behaviour is more brittle at 95°C). These results indicate that an
increasing rubber content can increase the resistance against stable crack propaga-
tion. An increasing temperature has the same effect before the toughness reaches a
maximum. Below this maximum, the J-R and 8--R curves can show different be-
haviours. This indicates that a single-parameter description of the toughness behav-
iour is impossible (pp. 3-26, [30]). In the present case, the J-R and 8--R curves show
a qualitatively similar behaviour as a function of rubber content and temperature.
3.3 Fracture Mechanics Material Parameters Related

to Resistance Against Stable and Unstable Crack Initiation
Figure B 132 gives a survey of the J values as a function of rubber content and tem-
perature for the example of ABS 330. The critical J values Jd,Jd related to resistance
against unstable crack initiation were measured at the beginning of unstable crack
326 R. Lach, W. Grellmann, Y. Han, P. Kriger
20
a d =330 nm,
b
d = 330 nm,
T=23·C ~ ,=24wt.%
15 • 1Swt.% • -20 ·C
•c. 20wt.% • 23 ·C
c. SO·C
...... 24wt.%
•
-..,
95 ·C
~ 10
~
$
Z
.......
200
d = 330 nm, d
d = 330 nm,
T = 23·C '=24wt.%
150
• 1S wt.%
c. 24wt.%
• -20·C
• 23·C
E 100 c. SO·C
..,E • 95·C
b
.....
.......
50 •
...
.....,
c.o
$ •
0
0.0 0.5 1.0 1.5 2.00.0

•
0.5 1.0 1.5 2.0
M(mm) M(mm)
Fig, B131. Crack resistance curves (J-R and §-R curves) of ABS 330 as a function of rubber
content at room temperature (a, c) and as a function oftemperature for ¢= 24 wt.%
(b, d) (&I, amount of stable crack growth)
propagation or at the maximum load. In addition to these quantities, JO.2 values

related to resistance against stable crack initiation were measured at the technical
crack initiation point of fla = 0.2 mm.
The two quantities Jd,Id and JO.2 form, versus rubber content, a pronounced two-
level function that is characterized by a lower and an upper shelf of toughness.
Thus, 'brittle-to-tough' transition (BIT) contents tPaTT can be formally ascertained
by averaging the lower and the upper shelf of toughness.
The J values generally increase up to maximum values at 23-60 °C and de-
crease after passing through this maximum point (Fig. B132b,c) [29,31,40].
Whereas the causes of the increasing toughness versus temperature, especially at
temperatures higher than the glass transition temperature of the rubber, are well
established for ABS, the cliuses of the decreasing toughness that can be observed
close to the glass transition temperature of the SAN matrix are fundamentally un-
Fracture Mechanics Characterization of ABS Materials 327
20 d = 330 nm, <!lBTT (Unstable) a

T = 23 DC
10 ~ J'd,Qd
• J O.2
t
5 Upper Shelf
. J--
E of Toughness
E <!lBTT (Stable) "
~ 2
,0"i 1 Lower Shelf
'i!
.",
,'0 0 ,5 of Toughness
0.2
0.1 ~~ __ ~ __ ~~ __L -_ _ ~ _ _~_ _~_ _-L~
0 5 10 15 20 25 30 35
Rubber Content (wt.%)
25 2.5
d = 330 nm b d = 330 nm C
20
• 4wt.%
~ 16 wt,%
••
16wt.%
20wt.%
2.0
0 28wt.% t::. 24wt,%
~ 15
• 28wt.%
~ 36wt.%
1.5
E
E
z E
E
-..
--; 10 1.0 ~
, '0
, ,;'"
5 0.5
0 0.0
-100 -50 0 50 100 -100 -50 0 50 100
Temperature (DC) Temperature (DC)
Fig. B132. J values Jd.ld (a, b) and J02 (a, c) related to resistance against unstable and stable
crack initiation versus rubber content (a) and temperature (b, c) for ABS 330
resolved. This phenomenon, which was detected most clearly at a rubber content
of 16 wt.%, is connected with a reduction in the amount of stable crack growth
and the plastic-zone size (see above and [29,30]). Besides the decrease in the
Young's modulus at higher temperatures, other possible causes are matrix embrit-
dement effects due to initiation and growth of intrinsic crazes [30]. From the J
versus temperature plots, TBrT values relating to the 'brittle-to-tough' transitions
were formally deduced.
328 R. Lach, W. Grellmann, Y. Han, P. Krtiger
The dependence on rubber content and temperature of dl/da and the tearing
modulus TJ (Fig. B 133a) related to resistance against stable crack propagation are
qualitatively comparable to that of JO.2 [31,40,41], but JO.2 reflects the behaviour of
the toughness of the matrix to a high degree, and TJ gives morphology-sensitive
information related to the particle-matrix structure (pp. 3-26, [28]). However, J O.2
does not correspond to the physical initiation process, because of its empirical
defmition. An explanation of the increasing JO.2 values as a function of rubber con-
tent is the interaction between the SAN matrix and the SAN surrounding the rubber
particles in the form of graft shells about 100m thick. The resulting changes in the
molecular structure of the matrix, which increase with increasing rubber content,
have a great influence on JO.2 (pp. 301-316).
8 4
a b
d=330nm d=330nm
• 16wt.%
• 16wt.%
6 • 20wt.%
• 20wt.% 3
..,
6 24wt.%
28wt.% ..,
6 24wt.%
28wt.%
1-~4 • 36wt.%
• 36wt.%
2
I-~
2
6
0 0
-100 -50 0 50 100 -100 -50 0 50 100
Temperature (OC) Temperature (0C)
Fig. B133. Tearing moduli TJ (a) and To (b) related to resistance against stable crack propaga-
tion versus temperature for ABS 330
From a comparison between TJ (Fig. B133a) and T6 (Fig. B133b) as a function

of temperature, the statement that a single-parameter description of the toughness
behaviour gives information only about one single site of the fracture process can be
substantiated. The results should be discussed using the quantities JTJ and 8I'/h
formally determined as the product of JO.2 and TJ or ~.2 and T/h respectively. For
the materials discussed, JTJ and 8I'6 show dependences that are qualitatively com-
parable to those of TJ and T/h respectively, as described in [41]. JTJ, as a measure
of the material-specific energy dissipation capability, and 8I'l" as a measure of the
plastic deformation of the fracture surfaces formed during crack propagation,
make it possible to give different physical interpretations of the fracture process in
the plastic zone and in the zone of energy dissipation processes. The increase of
8I'6 with rubber content and temperature up to room temperature indicates that
higher rubber contents and higher temperatures are beneficial in maintaining crack
growth stability.
The parameter 01'6 is an expression of the deformation capability in the plastic

zone and the parameter JTJ is an expression of the energy dissipation capability in
the zone of energy dissipation processes. A quantitative comparison between these
two characteristic parameters is impossible because of their different physical
backgrounds (they have different units, rom and N/mm).
3.4 Brittle-to-Tough Transitions
In principle, two kinds of fracture mechanics parameters can be used to estimate

BITs: firstly, the crack toughness values JO.2 related to resistance against stable
crack initiation and, secondly, the crack toughness values Jd,Id related to resistance
against unstable crack initiation. As a result of this, some impact-resistant polymer
materials, such as ABS [30,31] and heterogeneous PP materials [42-44], show
two BITs: the BIT of Jd,Id related to resistance against unstable crack initiation
(BIT 1: the conventional 'brittle-to-tough' transition) and the BIT of JO.2 related
to resistance against stable crack initiation (BIT 2: the 'tough-to-high impact'
transition). Because stable crack propagation is followed by unstable propagation,
the rubber content corresponding to BTT 1 is less than or equal to that for BTT 2
(Fig. B134). In agreement with investigations of heterophase PP materials [44],
for ABS [30,31] the 'tough-to-high impact' transition is due to the matrix and,
thus, is nearly independent of the temperature, particle size [30,31] or particle
distance [44]. The BIT content [30,31] and the critical ratio of interparticle dis-
tance to diameter [44] measured at the transition point BTT I converge on the
values of the 'tough-to-high impact' transition (BIT 2) from both higher and
lower temperatures. In summary, BIT 1 is more sensitive to the structure and the
deformation mechanisms.
40
270nm 330nm
••
o 'ITT (Unstable)
<> 'OTT (Stable) High Impact
30
<>
~
!
0
20 Tough
j
10
•
Brittle
0
-100 -50 0 50 100
Temperature (OC)
Fig. D134. 'Brittle-ta-tough' transition content 9Brr [9'Brr (unstable), conventional 'brittle-ta-
tough' transition; 9Brr (stable), 'tough-to-high impact' transition] as a function of
temperature for ABS 270 and ABS 330
From a combination of f/Jarr and TBrr values, content-temperature windows for

application of the materials can be constructed (Fig. B 135). The main purpose of
such constructions is to give a general survey of the fracture behaviour for engi-
neering applications. For the two materials, ABS 270 (d= 270 nm) and ABS 330,
which have nearly the same average particle size (300 nm) but different particle
size distributions (for ABS 270 see B 3.2, pp. 301-316), the areas of 'high impact'
are nearly the same, because the extent of this area is determined by the matrix
material, which is the same in both materials. In contrast to the 'high-impact' area,
the 'tough' area, being a sensitive indicator of the microstructure, is much greater
for ABS 330 than for ABS 270. Summarizing, technological errors in the manu-
facturing process and changes in temperature should influence the mechanical
behaviour of components made from ABS 330 less severely than the behaviour of
components made from ABS 270, because of the greater toughness reserve of
ABS 330.
100
d=270nm a
75
••
Stable Unstable
<> TIITT <+)
50 0 'IITT(T)
25
6
t.- O
I-
I-
~ -25
-50 Brittle
-75
-100
100
75
50
25
6
t.- O
I-
I-
~ -25 Brittle
-50 d=330nm
••
Stable Unstable
-75 <> TIITT<+l
0 +IITT(T)
-100
0 4 8 12 16 20 24 28 32 36
','BIT (wt.%)
Fig. B135. Content-temperature windows for application of the materials, for ABS 270 and
ABS 330
4 Conclusions
I. From a comparison of experimentally determined and calculated plastic-

zone sizes, a constraint factor for higher contents and temperatures equal to
0.5 is found. This value of the constraint factor correlates with an absence of
stress-hardening behaviour, which is expressed in straight-line crack resis-
tance curves.
2. The fracture toughness increases with increasing temperature until it reaches
a maximum value; after that, it decreases with further increase of tempera-
ture. This is due to the reduction in Young's modulus at higher temperatures
and intrinsic-craze formation in the SAN matrix at temperatures close to the
glass transition temperature of SAN. This matrix 'embrittlement', which re-
duces the effectiveness of rubber modification significantly, can be dis-
cussed in terms of the temperature dependence of the fracture mirror length
and plastic-zone size because there are quantitative correlations between
fracture surface phenomena and energy dissipation processes.
3. Because stable crack propagation is followed by unstable propagation, two
'brittle-to-tough' transitions are observed for high-impact polymers such as
ABS. Whereas the 'tough-to-high impact' transition is due to the matrix, the
conventional 'brittle-to-tough' transition is more sensitive to the structure
and the deformation mechanisms. With increasing temperature, the 'brittle-
to-tough' transition shifts to higher values.
4. The construction of content-temperature windows for application of the
materials gives a general survey of the fracture behaviour for heterogeneous
polymers. Its application to the ABS materials allows a differentiation of the
materials on the basis of their toughness reserves.
5. A multi-parametric description of the fracture process is necessary if the
physical dependences of the fracture mechanics parameters show different
behaviours as a consequence of the different influences of load and deflec-
tion on the toughness behaviour.
Acknowledgements
The German Research Foundation (DFG) is gratefully acknowledged for fmancial

support which made possible the investigations on the subject 'Influence of mor-
phology for polymeric multi-phase systems on phases of crack propagation'. Dr.
Y. C. Han wishes to thank the DAAD-K.C. Wong Foundation for her research
fellowship at the Martin-Luther-University of Halle-Wittenberg.
References
1. Swallowe G. M. (Ed.) (1999): Mechanical Properties and Testing of Polymers. An A-Z

Reference. Kluver Academic, Dordrecht
2. Friedrich K. (1989): Application of Fracture Mechanics to Composite Materials. Elsevier,
Amsterdam
3. Kausch H.-H. (1987): Polymer Fracture. Springer, Berlin, Heidelberg
4. Williams J. G. (1984): Fracture Mechanics of Polymers. Ellis Horwood, Chichester
6. Eichenbauer H., Ott K.-H. (1992): Polystyrene and styrene copolymers, Chap. 5: Acryloni-
trile--butadiene-styrene (ABS) polymers. In: Evers B., Ullmann F. (Eds.) Ullmann's Ency-
clopedia of Industrial Chemistry, Vol. A21. VCH, Weinheim: 63~63
7. Stabenow J., Haaf F. (1973): Morphology of ABS-graft rubbers. Appl. Macromol. Chem.
Phys. 29/30: 1-23
8. K!!mpf G., Schuster H. (1972): Structure and properties of ABS-resins, IX. Quantitative
investigations of morphology of ABS. Appl. Macromol. Chem. Phys. 27: 81-98
9. Michler G. H. (1992): Kunststoff-Mikromechanik - Morphologie, Deformations- und Bruch-
verhalten. Carl Hanser, Munich, Vienna
10. Michler G. H. (1993): The Role of interparticle distance in maximizing the toughness of high
impact polystyrene (HIPS) blends. Acta Polym. 44: 113-124
II. Hobbs S. Y. (1986): The Effect of rubber particle size on the impact properties of high
impact polystyrene (HIPS) blends. Polym. Engng. Sci. 26: 74-81
12. Bucknall C. B. (1992): The relevance of impact testing in the materials science in polymers.
Plast. Rubber Compos. Process. Appl. 17: 141-145
13. Castellani L., Frassine R., Pavan A., Rink M. (1996): Rate and temperature dependence of
fracture toughness in ABS resins in relation to dispersed-phase structure. Polymer 37: 1329-
1338
14. Giaconi G. F., Casttelani L., Maestrini C., Ricco T. (1998): Development of toughness in
ABS resins. Polymer 39: 6315-6324
IS. Bernal C. R., Frontini P. M., Sforza M., Bibbo M. A. (1995): Microstructure, deformation,
and fracture behavior of commercial ABS resins. 1. Appl. Polym. Sci. 58: 1-10
16. Bernal C. R., Cassanelli A. N., Frontini P. M. (1995): A simple method for J-R curve deter-
mination in ABS polymers. Polym. Test. 14: 85-96
17. Lu M.-L., Lee C.-B., Chang F.-C. (1995): Fracture toughness of acrylonitrile--butadiene-
styrene by J-integral methods. Polym. Engng. Sci. 35: 1433-1439
using the J-method. Polym. Test. II: 271-288
19. Rico T., Rink M., Caporusso S., Pavan A. (1985): An analysis of fracture initiation and crack
growth in ABS resins. In: Proceedings of the International Conference on Toughening of
Plastics II, London, UK, July 2-4: 27/1-27/9
ries No. 17, Institut fUr Mechanik, Berlin, Chernnitz: 48-72
21. Grellmann W., Sommer J.-P., Hoffinann H., Michel B. (1987): Applications of different J-
ceedings of the 1st Conference on Mechanics, Prague, Czech, June 29-July 3, Vol. 5:
129-133
of tough polymers. Int. J. Fract., Lett. Fract. Micromech. 96: Ll7-L21
ics values of polymers. Int. J. Fract. 68: RI9-R22
24. Merkle J. G., Corten H. T. (1974): J integral analysis for the compact specimen, considering
axial force as well as bending effects. J. Pressure Vessel Technol. 96: 286-292
26. Kobayashi T. (1984): On the information about fracture characteristics obtained from in-
strumented impact test of A533 Steel for reactor pressure vessel. Engng. Fract. Mech. 19:
67-79
charpy impact test. In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Se-
ries No. 37, Institut fiIr Mechanik, Berlin, Chemnitz: 186-192
28. Seidler S. (1998): Anwendung des Ri6widerstandskonzeptes zur Ermittlung strukturbezoge-
ner bruchmechanischer Werkstofikenngro6en bei dynamischer Beanspruchung. Fortschritt-
Berichte, VDI Series 18: MechaniklBruchmechanik, No. 231. VOl, DUsseldorf
29. Han Y., Lach R., Grellmann W. (1999): The Charpy impact fracture behaviour in ABS
materials. Appl. Macromol. Chern. Phys. 270: 13-21
30. Han Y., Lach R., Grellmann W. (2001): Effect of rubber content and temperature on unsta-
ble fracture behaviour in ABS materials with Different particle sizes. J. Appl. Polym. Sci.
79: 9-20
31. Lach R., Grellmann W. (2000): Estimation of the resistance against stable crack initiation
and unstable crack propagation using R-curves and stability assessment diagrams in ductile
polymeric ABS-materials. In: Proceedings of the 13th European Conference on Fracture
(ECF 13), San Sebastian, Spain, September 6-9, CD-ROM Polymer and Composites,
No. 20: 1-8
32. Dettenmaier M., Kausch H.-H. (1980): New type of crazes in oriented polycarbonate. Poly-
mer 21: 1232-1234
33. Dettenmaier M. (1983): Intrinsic crazes in polycarbonate: phenomenology and molecular
interpretation of a new phenomenon. In: Kausch H.-H. (Ed.) Crazing in Polymers. Advances
in Polymer Science 52153. Springer, Berlin, Heidelberg: 57-104
34. Rice J. R. (1967): Mechanics of crack tip deformation and extension by fatigue. In: ASTM
STP 415: 247-309
35. Schwalbe K.-H. (1980): Bruchmechanik metallischer Werkstoffe. Carl Hanser, Munich,
Vienna
36. Anderson T. I. (1995): Fracture Mechanics. Fundamentals and Application. CRC Press, Boca
Raton
37. ASTM 813-81 (1981): Standard test for J[c, a measure of fracture toughness. Annual Book of
ASTM Standards, Philadelphia, Vol. 03.01
38. Will P. (1994): R-Curves of Energy Dissipative Materials. 1. Mat. Sci. 29: 2335-2340
334 R. Lach, W. Grellmann, Y. Han, P. Kriger
Fract. Mech. 37,2: 275-281
namic fracture toughness of ABS materials. J. Appl. Polym. Sci. 75: 1605-1614
tearing stability in ABS materials. Appl. Macromol. Chem. Phys. 270: 5-12
San Sebastian, Spain, September 6-9, CD-ROM Polymer and Composites, No. 12: 1-8
43. Kotter I., Grellmann W., Seidler S. (2000): Crack toughness behaviour and morphology of
PP-blends. In: Proceedings of Polymeric Materials (p 2000), Halle, Germany, September
25-27: 207-210
B 3.4 Brittle Fracture of ABS - Investigation
of the Morphology-Failure Relationship
B. Moginger, Stuttgart, G. H. Michler, Merseburg,
and H.-C. Ludwig, Boblingen
1 Introduction
The polymers belonging to the class of acrylonitrile-butadiene-styrene copoly-

mers (ABS) are toughened modifications of styrene-acrylonitrile (SAN) copoly-
mers. The toughening modification is achieved by dispersing submicroscopically
small elastomeric particles - mainly derived from butadiene - in the coherent,
brittle SAN matrix.
To improve the phase interaction of the elastomeric particles with the SAN ma-
trix, they are grafted with styrene and acrylonitrile. For the grafting reaction, the
double bonds of polybutadiene are used [1] (Fig. B136).
I
2 !
CH-CH-CH-CH~"''''''''--'CH-CH-CH-CH
2J j \ 2 2 I
Fig. B136. Chemical structure of ABS consisting of a SAN-grafted polybutadiene backbone
The mechanical performance of such ABS polymers is determined by the prop-

erties of the matrix and of the elastomeric phase, the mixture proportion and the
interaction between the two phases. The toughness and the surface fmish are con-
trolled by choosing appropriate size distributions of the elastomeric particles [2].
By the variation of the parameters elastomer content, degree of grafting and parti-
cle size, the properties of ABS type materials can be optimally adapted to the ap-
plication [3].
The typical two-phase structure of ABS (Fig. B137) depends on the specific
types of polymers, their compatibility and processing effects [4]. Only an optimal
coordination of all these parameters guarantees a 'good morphology' in the sense

336 B. Miiginger, G. H. Michler, H.-C. Ludwig
Fig. B137. Typical phase structure ofa commercial ABS grade
of optimized macroscopic properties. In particular, if the synthesis conditions are

not optimized, elastomer particles that are too big may be generated, leading to
undesirable mechanical behaviour.
In many commercial ABS grades the elastomeric particles have a bimodal parti-
cle size distribution. In addition to many small particles with diameters between
100 nm and 200 nm, there are bigger particles having diameters ranging from
300 nm to 1 !lm. The bimodal particle size distribution is mainly a compromise
between high toughness (big particles) and high surface fmish (small particles).
2 Failure Mechanisms in ABS
By introducing a soft rubbery phase into a brittle matrix material, the impact
toughness of the pure matrix may be significantly improved. The reason for this
are many simultaneously occurring dissipative deformation processes, such as
• multiple crazing and/or
• shear deformation processes
initiated and localized at the particles. These deformation processes are caused
by stress concentrations in the equatorial regions of the elastomeric particles. With
increasing deformation, the stresses are restricted and more mechanical energy can
be dissipated before fracture [2]. The ABS grades considered in this investigation
exhibit mainly multiple crazing (Fig. B 138). Only if the temperature exceeds 70°C
does shear deformation start to dominate the deformation behaviour [5].
The pure SAN matrix is brittle and has a tensile strength on the order of 75 to
80 MPa [1]. The stress concentration around the dispersed elastomeric particles
causes localized flow or crazing of the matrix if the macroscopic external stress
approaches 35 to 50 MPa, depending on the ABS grade. The difference between
Brittle Fracture of ABS 337
the moduli of the dispersed particles and the matrix is decisive for structurally
initiated multiple crazing. For ABS the matrix modulus is typically 100 or 1000
times larger than the modulus of the dispersed elastomeric particles. Therefore, a
heterogeneous stress distribution occurs in the phases if the material is mechani-
cally loaded.
Stress analyses of a homogenous material containing a spherical inclusion show
that the stress concentration depends on the modulus ratio Gp/GM [6]. The softer
the inclusion, the larger is the stress concentration in the equatorial region of the
particle (Fig. B 139).
Fig. B138. Phase structure of ABS showing multiple crazing
2.0
I.S
JI b
1.0
o 0.001 0.01 0.1

G
X = G~
Fig. B139. Stress concentration avviao as a function of modulus ratio Z

338 B. Mtiginger, G. H. Michler, H.-C. Ludwig
For a hole in a linear elastic material the calculation yields a theoretical stress
intensity factor of 2.04 in the matrix at the interface with the hole. With increasing
distance from the interface, the stress intensity decreases and reaches the overall
stress at a distance of 3 or 4 hole diameters [7]. The loading time and level affect
the growth of crazes significantly. If the craze-initiating stress level is exceeded at
only a few particles, crazes will be generated there. For a continuous external
stress, these crazes grow more and more until a crack is generated. Fracture with a
brittle failure behaviour is the consequence.
In the case of a high loading and deformation rate, crazes are generated at
nearly every particle. However, their dimensions are limited. In general, the craze
concentration is very high for styrene polymers toughened with elastomeric parti-
cles, and therefore the craze dimensions remain small. Because of this the crazes
unload each other, and their growth to critical dimensions is prevented [8].
With the appropriate particle size and particle size distribution, the deformation
behaviour of ABS can be steered in a defmed way. If there are many very small
butadiene particles dispersed in the matrix, the craze initiation energy has to be
supplied at many different places and the yield stress is high. On the other hand,
the crazes more often hit another, neighbouring particle, leading to a compara-
tively small yield strain. Furthermore, the deformed volume and, as a consequence,
the energy absorption remain smaller than for bigger particles [9].
3 The Phenomenon of 'Brittle Fracture'
Technical ABS grades have a well-developed morphology in the sense that they
are optimized with respect to strength, modulus and impact behaviour. For short-
time loading, e.g. in a tensile test according to DIN 53455, all ABS grades exhibit
the expected behaviour of toughened elastomer-modified thermoplastics. How-
ever, some ABS grades show a very unusual brittle fracture behaviour in a tensile
test if pre-loaded in a creep test according to DIN 53444 [10].
Using the stress-strain curve, the behaviour in the tensile tests can be classified
- depending on the pre-loading conditions - into three classes with respect to the
behaviour of unloaded specimens:
1. The stress-strain curve does not change significantly.
2. The elongation at break does not exceed the yield strain.
3. The shape of the stress-strain curve is changed completely.
The effects of pre-loading on the stress-strain behaviour of an ABS are shown
in Fig. B140. The pre-loading conditions were 100 hours with stresses of 10,20,
30, 40, 50, 60 and 70 % of the corresponding yield stress determined in a tensile
test. At 80 % of the yield stress the creep time had to be limited to 0.8 h because of
necking.
50
40
'OS ," "

~ 30
6 "
"
b _---"'80 % / 0.8 h
'"
'" 20
~
<ZJ
10
0
0 2 4 6 8 10
Strain E (%)
Fig. B140. Stress-strain curves of ABS after pre-loading in creep tests
4 Experimental
4.1 Materials Investigated and Strain Rate Dependence

of the Mechanical Properties
Three commercial grades of ABS were investigated: Terluran 955F (BASF), No-
vodur P2K and Novodur P2X (Bayer). In Table B19, some mechanical properties
are summarized.
Table B19. Mechanical properties of ABS grades investigated
Property Standard Material

NovodurP2K NovodurP2X Terluran 955F
Young's modulus (MPa) DIN 53457 2000 2700 2500

Yield stress (MPa) DIN 53455 36 47 47
Yield strain (%) DIN 53455 2.5 2.3 2.8
Elongation at break (%) DIN 53455 >15 >15 >15
Charpy impact strength
DIN 53453 19 II II
(kJ/mm2 )
Ball-indentation hardness
DIN 53456 70 115 96
(MPa)
Vicat heat resistance tem-
DIN 53460 98 98 87
perature (0C)
340 B. Mllginger, G. H. Michler, H.-C. Ludwig
The strain rates for long-time pre-loading in a creep test are very small. The
mechanical properties of a viscoelastic material such as ABS are dependent on
time and strain rate. The strain rate dependence of the mechanical properties, es-
pecially of the yield stress uzS, can be determined with tensile tests at different
strain rates.
Therefore, tensile tests according to DIN 53455 were performed at room tem-
perature with a Zwick 1476 instrument using constant strain rates. All three ABS
grades show distinctly strain-rate-dependent mechanical properties (Table B20).
Table B20. Strain-rate-dependent mechanical properties of the three ABS grades
Property Strain rate Material

Novodur Novodur Terluran
(o/crlmin) P2K P2X 955F
Young's modulus 1 1900 2680 2700

(MPa) 10 1960 2750 2760
100 2100 2830 2770
Yield stress 1 32.2 39.3 38.1
(MPa) 10 35.1 45.7 43.3
100 39.8 49.4 45.2
Yield strain 1 2.6 1.8 2.1
(%) 10 2.9 2.1 2.3
100 3.3 2.9 3.2
4.2 Morphology and Size of the Butadiene Particles
The three ABS grades also differ significantly with respect to the morphology and
size distribution of the butadiene particles, as shown by transmission electron
microscope investigations (Figs. Bl41 and BI42).
Novodur P2K contains relatively many butadiene particles having a diameter of
about 400 nm, leading to distinctively tough behaviour and a comparatively low
Young's modulus. Novodur P2X has a lower content of the elastomeric phase, and
the structure is dominated instead by small particles having a diameter of about
100 nm. The size distribution is clearly bimodal, with maxima at particle diameters
in the ranges of 100 to 150 nm and 400 to 500 nm. Therefore, the Young's
modulus is relatively high. Terluran 955F has a trimodal size distribution with
maxima at 100 nm, 250 nm and 550 nm.
Fig. B141. Transmission electron microscope images of the phase structure of the ABS grades:
(a) Novodur P2X, (b) Novodur P2K, (c) Terluran 955F
342 B. Moginger, G. H. Michler, H.-C. Ludwig
9
8
'""'
::R
0
7
'-'
..... 6
§
..... 5
§
u 4
0
§ 3
2
~
1
0
0 0 0 0 0 0 0 0 0 0 0 0 0
V) 0 V) 0 V) 0 V) 0 V) 0 V) 0
N N M M '<t '<t V) V) \0
Particle Diameter (nm)
--
::R
~
.....
8
7
6
~
£ 5
§
u 4
0
§ 3
~ 2
0
0 0 0 0 0 0 0 0 0 0 0 0 0
V) 0 II"> 0 V) 0 V) 0 V) 0 V) 0
N N M M '<t '<t V) V"') \0
12
,-., 10
~
.....
'-'
8
~
£
s::
0 6
u
0
E 4
:::3
~ 2
0
0 0
V) 8..... 0
II">
0
0
N
0
II">
N
0
0
M
0
M
II">
0
0
'<t
0
II">
'<t
0
0
V)
0
II">
II">
0
0
\0
Fig. B142. Volwne-related size distributions of the butadiene particles of the three ABS grades:
(a) Novodur P2X, (b) Novodur P2K, (c) TerIuran 955F
5 Mechanical Behaviour
After Long-Time Mechanical Pre-loading
In order to investigate the effects of long-time static pre-loading, creep tests ac-
cording to DIN 53444 were performed at different load levels with tensile test
specimens having the geometry No. 3 specified in DIN 53455. The yield stress
was determined in a tensile test at a strain rate of 10 %/min; the load levels ranged
from 10 to 70 % of the yield stress in steps of 10 %, and the pre-loading times
ranged from 3 to 6000 hours. The pre-loading experiments were performed in a
Zwick 4211 automated creep apparatus with optical strain measurement [11].
The creep strains during pre-loading are shown for Novodur P2X in Fig. B143.
Up to load levels of 50 % of the yield stress only a small amount of creep occurs.
At load levels beyond 60 % of the yield stress ternary creep occurs, and the strain
increases rapidly.
4 .
80% 70%
I
60%
L
V
3
I / ,j
~
L- l/ ~
/
I
~
V
...... ~ ,..,. ~ 50% 40%
I ~
""~
_ 30%
~ 20%
I.. 10%
o
10 -2 10 -1 [0° 10' ){f
Time t (h)
Fig. B143_ Creep strain curves of ABS Novodur P2X tensile-test specimens
After unloading, the mechanical behaviour was characterized with a tensile test.
The effects of the pre-loading on the mechanical behaviour, especially the elonga-
tion at break CzR, differ significantly for the three ABS grades and depend highly
on the pre-loading level and time (Table B21).
Table B21. Elongation at break of the three ABS grades after different pre-loadings
Load Time Material

NovodurP2K NovodurP2X Terluran 955F
(% CTzS) (h) (%) (%) (%)
30 10 2.2 6.9
100 1.4 2.6
1000 22 1.7 2.2
40 10 2.5 6.3
100 2.0 6.5
1000 26 1.8 1.7
50 1 20.8 3.8
10 11.3 4.9
100 22 1.3 5.6
The investigations show that the ABS grades Novodur P2X and Terluran 955F
tend towards brittle fracture failure after a certain amount of pre-loading. How-
ever, for Terluran 955F this behaviour is much less distinct than for Novodur P2X.
It is important to note that this behaviour occurs only within a certain interval of
pre-loading. This is clearly expressed by considering the work capacity WR of
Novodur P2X (Fig. B144). Long-time pre-loading with load levels ranging from
25 to 55 % of the yield stress seems to be very critical.
12
Loading Time: 100 h
10
'"'"'"
]
.....,
g 8
..:
~
6
l;-
'u
'"
0..
4
u'"
~
0 Brittle Fracture
~ 2
0
10 20 30 40 50 60 70
Load as % of Yield Stress CTzS
Fig. B144. Work capacity of Novodur P2X after 100 hours of pre-loading in a creep experiment
6 Deformation Mechanisms and Defect Structures
6.1 Micro-Mechanical Deformation Structures

of Unloaded Specimens
Micro-mechanical deformation processes can be directly investigated by in-situ

deformation in an electron microscope. This method allows one to determine the
temporal succession of specific deformation processes and defect structures during
straining of the specimens [5]. Of special interest is the question of whether there
are differences in the mechanical short-time behaviour of the ABS grades and
whether these may be related to the mechanical behaviour after long-time static
pre-loading.
Thin sections, typically having a thickness of 1.5 /lm, were taken from test
specimens of the three ABS grades and uniaxially deformed in the HVEM (high-
voltage transmission electron microscope) without any staining. As the thin sec-
tions are considerably thicker than the diameter of the butadiene particles, the
micro-mechanical processes observed in the HVEM are comparable to those of the
bulk material.
The HVEM investigations showed that the three ABS grades differ significantly
with respect to their micro-mechanical deformation behaviour. For Novodur P2X,
crazing dominates the deformation behaviour (Fig. B 145). The crazes are fibril-
lated roughly, and in the first deformation stage they grow independently of each
other perpendicular to the applied stress. Ifhigher strains are reached, the individ-
ual crazes unify into craze bands. Usually, crazes are initiated at the biggest buta-
diene particles and therefore are the reason for crack generation when the speci-
men is overstrained.
Fig. B145. HVEM image of the deformation structure ofNovodur P2X

For Terluran 955F, the defonnation structure also shows a low contribution of
shear defonnation; however, a clear necking zone is not visible. The defonnation
starts at the biggest butadiene particles, and many short and highly branched crazes
are initiated. The branching of the crazes is caused by contributions from shear to
the defonnation mechanism. In the regions of plastic defonnation the molecules of
the matrix are strongly oriented. Craze growth in such regions is hindered in its
further progress or stopped. New crazes are initiated and possible stress peaks can
be distributed over larger regions. This process impedes the creation of extended
crazes strongly (Fig. BI46). For large strains the micro-mechanical defonnation
structure of Terluran 955F is similar to the fibrillated craze bands of Novodur
P2X.
Fig. B146. HVEM image of the deformation structure ofTerluran 955F
The micro-mechanical defonnation processes of Novodur P2K are significantly

more complex. At small strains a diffuse shear defonnation is initiated between the
big butadiene particles. However, this defonnation lacks a clear shear band struc-
ture and necking of the thin sections. The main mechanism is cavitation of the
butadiene particles and, connected with this process, a progressive generation of
rows of holes (Fig. BI47). The matrix material between the cavitated butadiene
particles is continuously defonned, and the cavitation holes remain separate until
the fracture of the specimen. Only at high stresses are a few short crazes generated.
In general, crazing is of minor importance in the fracture ofNovodur P2K.
Fig. B147. HVEM image of the defonnation structure of Novodur P2K
6.2 Micro-Mechanical Deformation Structures

of Pre-loaded Specimens
All specimens, broken in a brittle manner in the tensile test after pre-loading, exhibit
a common feature in the fracture surfaces: at least one smooth comer (Fig. B 148).
Morphologically, these smooth regions correspond to 'macrocrazes' . The area
of these macrocrazes increases with pre-loading time and stress level, and it seems
that this increase is further promoted by substances that enhance stress cracking
[12, 13]. However, the size of the macrocrazes has no significant effect on the
brittle fracture behaviour (Fig. B149).
Fig. B148. SEM image of the fracture surface of a pre-loaded Novodur P2X specimen
348 B. Milginger, G. H. Michler, H.-C. Ludwig
)(
)(
• •
x )(
• •
x
)(
• x • •
5 IO 15 20
Craze-Area (mm2)
Fig. B149. Correlation of the work capacity with the size of the macrocrazes
Under optimal observing conditions, macrocrazes are visible in transmitted light

if thin, translucent specimens are used. The specimens for investigation by trans-
mission electron microscopy were taken from regions containing macrocrazes in
order to determine the structure of the macrocrazes. Staining with osmium tetrox-
ide hardened the butadiene particles. Subsequently, semi-thin sections for HVEM
investigations having a thickness of about 1 11m and ultra-thin sections for TEM
investigations having a thickness of about 100 nm were prepared using the stained
specimens.
The micro-mechanical deformation structures of the semi-thin sections of
specimens that were been pre-loaded in the critical stress range differ significantly
Fig. B150. Macrocraze in a pre-loaded Novodur P2X specimen; pre-loading, 166 hlO.4O"zs
from the structure of the specimens described above. The macrocrazes have a width
of about 200 nm, which is double the width of the crazes generated in the usual
tensile test. These macrocrazes are mainly oriented perpendicular to the direction
of the external load. Furthermore, their growth direction is hardly affected by the
local stress fields generated by the butadiene particles (Fig. B 150). The only dif-
ference is that the craze tip tends to be split into several small crazes [12].
As long as the craze borders are linked by fibrils, their width is small compared
with their length. If the fibrils break because of overstraining, the macrocraze
broadens distinctly. In this phase the interaction between the butadiene particles,
especially the big ones, and the macrocraze is minimal, and the growth direction of
the macrocraze is hardly affected. Sometimes the macrocrazes pass directly
through big butadiene particles (Fig. B 151).
Fig. B151. Suppressed craze stop mechanism ofa butadiene particle
Fig. B152. Macrocraze branching

The TEM investigations of ultra-thin sections revealed that no multiple crazing

was present in Novodur P2X specimens if they were pre-loaded in the critical
stress range. In some places, however, the macrocrazes are split into several
branches (Fig. BI52).
7 Results and Conclusions
The atypical brittle fracture behaviour of some commercial ABS grades after a
long-time static pre-loading is a result of a complex interaction of morphological
and mechanical parameters which is not completely understood. In principle, one
can say that it depends on
• the morphology of the material, especially the size distribution of the butadiene
particles and the resulting mean distances of particles of the same size class
• the pre-loading stress, which has found to be critical for stresses between 25
and 55 % of the yield stress determined in the tensile test, typically
• the existence of macrocrazes, because the deformation remains localized on the
tip of the macro craze if the specimen is loaded.
It is very serious that atypical brittle fracture behaviour seems to be linked
mainly to the existence of macrocrazes and not to their size. This means, in an
extreme case, that only a single macrocraze may be sufficient to initiate this atypi-
cal brittle fracture behaviour.
Furthermore, the craze stop mechanism of the large butadiene particles seems to
be nearly completely suppressed in the case of very slow macrocraze growth dur-
ing the pre-loading phase. Mechanical tests show that large butadiene particles
have a lower modulus than small ones. Theoretical considerations would lead us to
expect an improved craze stop mechanism. One reason for this completely differ-
ent behaviour might be the very small growth rate of the macrocrazes.
The properties of viscoelastic materials depend highly on deformation rate. The
tensile tests show clearly that the yield stress of the ABS grades decreases signifi-
cantly if the strain rate decreases (Table B20). This means that the yield stress
decreases during the pre-loading time to a relatively small value. Therefore, a few
crazes are initiated at only very big butadiene particles or other matrix defects, and
they grow very slowly. Tough material behaviour after initiation of crazes may be
guaranteed only if the craze growth is stopped immediately by other appropriate
butadiene particles in the direct neighbourhood. If the crazes can reach a critical
length, the craze stop mechanism is suppressed and the craze can become a macro-
craze.
In the craze tip region the matrix is softened because of stress concentrations
and the decrease in the yield stress in the course of time. To allow high plastic
deformation in the craze tips, the stress has to be close to the yield stress. Each
very small further deformation does not depend on the Young's modulus any more
but on the tangent modulus of the actual deformation state e".. For stresses close to
the yield stress Ex the tangent modulus is much smaller than the Young's modulus
(Fig. B153).
If the stress in the craze tip approaches the yield stress of the brittle SAN ma-
trix, the difference between the Young's modulus of the butadiene particles and
the tangent modulus at Ex tends towards zero. Because of the slow growth, the
craze tip moves in quasi-single-phase material. The consequence is that macro-
crazes can grow perpendicular to the external stress field in an undisturbed way
and also pass through butadiene particles.
,
40 , , ,"
, .-
30 ,, .-
,,
--
-- -- -- -- -- ,, "
,-.. .-_7'--
~
~
'" "
'-'
20
-... ,"
til
til
CI)
~
r Jl
- Experimental Curve
10
-----. E-Modulus
- - - Modulus at Ex
0
0 2 4
Strain (%)
Fig. B153. Stress-strain curve with detennination of the Young's modulus and the tangent
modulus at &x illustrated
8 Summary
The micro-mechanical processes and the corresponding mechanical properties of

three commercially available toughened ABS grades have been investigated with
respect to the effects of morphological parameters on the deformation behaviour
after pre-loading. After a long-time static pre-loading in a creep test, certain ABS
grades show atypical brittle fracture behaviour in a subsequent tensile test. Con-
ventional tensile tests and micro-mechanical in-situ deformation experiments in the
HVEM have been performed to describe and to investigate this special failure
behaviour. In the HVEM images very long and wide crazes are visible; these are
typical of specimens exhibiting brittle fracture behaviour. In particular, those ABS
grades which have a low content of elastomeric phase, a bimodal particle size
distribution and a large distance between the big elastomer particles are very sen-
sitive with respect to pre-loading.
Acknowledgements
The authors would like to thank the German Research Foundation (DFG) for fi-
nancial support of the projects Ey 13/34-1 and Mi 358/2-1 and the companies
Bayer AG, Leverkusen, Germany and BASF AG, Ludwigshafen, Germany for
supplying the ABS grades.
References
1. Platt A. E. (1989): Rubber modification of plastics. In: Comprehensive Polymer Science.

The Synthesis, Characterization, Reactions & Applications of Polymers (Allen G., Beving-
ton J. C., Chairmen) Vo!. 6: Polymer Reactions (Eastmond G. C., Ledwith A., Russo S.,
Sigwatt P., Eds.) Pergamon, Oxford, New York: 437-449
3. Domininghaus H. (1998): Die Kunststoffe und ihre Eigenschafien. 5th edition, Springer,
Berlin, Heidelberg
4. Michler G. H. (1986): Determination of the morphology and mechanical microprocesses in
polymer combinations by electron microscopy. Polymer 27: 323-328
5. Starke J. U., Godehardt R., Michler G. H., Bucknall C. B. (1997): Mechanisms of cavitation
over a range of temperatures in rubber-toughened PSAN modified with three-stage core-
shell particles. J. Mat. Sci. 32: 1855-1860
6. Goodier J. N. (1993): Concentration of stress around spherical and cylindrical inclusions and
flaws. Trans. ASME 55: 39
7. Ondracek G. (1978): Zum Zusarnmenhang zwischen Eigenschafien und Gefiigestruktur
mehrphasiger Werkstoffe, Teil 1lI: Zur Theorie des Zusarnmenhangs zwischen Ge-
fiigestruktur und Elastizitatsmodu!. Z. Werkstofftechnik 9: 31-36
8. Oswald I. A., Menges G. (1995): Materials Science of Polymers for Engineerings. Carl
Hanser, Munich, Vienna
9. Retting W. (1977): Deformations- und Bruchmechanismen in mehrphasigen Polymersyste-
men. Angew. Makromo!' Chern. 58/59: 133-174
10. Gust H. (1993): Material behaviour of prestressed ABS. In: Proceedings of the 13th Stutt-
garter Kunststoff-Kolloquium, Stuttgart, Germany, March 17-18: 353-358
II. Pollet P. (1985): Automatisierte Zeitstandpriifung - Verfahren mit beriihrungsloser
Dehnungsmessung. Kunstst. - Germ. Plast. 75: 829-833
12. Gust H., Starke J.-U. (1995): Correlation between changed material behaviour and structure
of prestressed ABS. In: Proceedings of the 14th Stuttgarter Kunststoff-Kolloquium, Stutt-
gart, Germany, March 22-23: 6.3/}-{i.3/12
13. Michler G. H. (1998): Microstructural construction of polymers with improved mechanical
properties. Polym. Adv. Techno!. 9: 812-822
c
Hybrid Methods
of Polymer Testing
and Polymer Diagnostics
C.1 Defect-Selective Imaging
A. Dillenz, N. Krohn, R. StoBel and G. Busse, Stuttgart
1 Introduction
Non-destructive inspection of a component means that one has to look for devia-
tions from structural integrity [1]. So there is a need to know what the structure
should look like if it were good - and what the real situation is. This comparison
can be difficult for complex structures because small deviations may not be obvi-
ous. So a method is required which responds only to the characteristic properties
of a defect, while it ignores the potentially complicated 'good' structure. In this
way the non-relevant information is suppressed, while all attention can be focused
on the defect.
The question of interest is whether defects have properties which allow one to
distinguish them in a reliable way from good material. If a defect is characterized
by stress concentration around it (since it is a source of mechanical weakness),
then the local stress-strain diagram might be closer to its non-linear regime, where
it reacts mechanically by overtone generation. There also might be a crack, causing
friction or hysteresis and thereby local damping of mechanical waves in addition to
overtone generation. These are examples of effects that can be caused by defects.
Any method responding to such effects allows selective imaging with respect to
the specific contrast dictated by the intended application.
In this chapter we concentrate on mechanical defects and show how they can be
detected by their reaction to certain inputs.
2 Methods
Visual inspection is highly attractive since it is coupled to our optical input,

through which most information is fed to us from the outer world. Therefore, tech-
niques delivering images are usually preferred to those which provide only data,
though it may be difficult to interpret the images in a reliable way. So, after an
image is obtained, the next important task is its correct interpretation to make sure
that all essential information is evaluated properly.
In technological applications, the component to be inspected is subjected to a
certain input, e.g. a wave. The local response (consisting of the amplitude and
phase of the output wave) results in data which can be presented as an image. The
detection of defects is achieved by comparison with an image obtained from a

356 A. Dillenz, N. Krohn, R. StoOel, G. Busse
standard component which is known to have no defects in it. Besides the technique
being time-consuming, there is too much information involved, so that specially
trained personnel are required for interpretation.
Defect-selective techniques, however, suppress the potentially confusing intact
structure and display only the defect itself so that it stands out clearly. Such tech-
niques are very much like dark-field imaging in optical microscopy, where, for
example, transparent objects with internal stresses are inspected between crossed
polarizers. Defect-selective imaging is a convenient and reliable technique for
systems aimed at a reduction of inspection time and at automatic defect detection.
In this chapter we report on two techniques. In both cases the basic idea is that
structural defects are characterized by their mechanical properties:
- there can be non-linear behaviour resulting in local overtone generation
friction under oscillating load causes local heat generation that can be detected
either by thermal expansion or thermal infrared emission.
The principles and methods of application are presented below for both of these
techniques.
2.1 Overtone Imaging
The input/output aspect of the characterization of materials can be described in a

simplest way by a linear relation. For acoustic waves, this consists of Hooke's law
and the elastic constants. However, the stress-strain diagram may display a devia-
tion from a straight line at large strains, and there may also be an asymmetry, e.g.
at a crack, which can transmit only positive displacements. The curvature of the
transfer function has the consequence that a sinusoidal input (a mechanical excita-
tion injected, for example, by an attached transducer) generates an output (a vibra-
tion which can be monitored interferometrically with a laser vibrometer) contain-
ing not only the original frequency but overtones of this frequency as well. If the
intact material is linear, local mechanical defects may cause a stress concentration
Linear Behaviour
Non-Linear Behaviour
Fig. Cl. Scanning laser vibrometry for selective overtone imaging of mechanical defects
Defect-Selective Imaging 357
ranging into the non-linear regime, while a crack causes an asymmetry and thereby
also a curvature of the transfer function [2,3]. In any case, the mechanical defect
generates overtones, allowing its detection by vibrometry, which can be performed
as an integral method so that one monitors only frequencies, regardless of the
location where they come from. Then the appearance of overtones in the response
to a single-frequency input indicates that there is a defect somewhere in the in-
spected sample. This might be used for a rejection criterion. However, the avail-
ability of commercial scanning vibrometers allows one to perform such measure-
ments locally, so that one can image the source of overtones or the defect (Fig.
CI). Therefore, mechanical defects can be revealed and selectively imaged owing
to their non-linear reaction to an elastic-wave input.
2.2 Ultrasound Lock-in Thermography
While conventional thermography generates images where the contrast is provided

by local thermal emission, lock-in thermography [4-7] responds only to a coded
heat flow, by analysing the temperature modulation which is induced by periodic
heat deposition. Absorption of modulated optical radiation results in a temperature
modulation that propagates as a thermal wave into the component being inspected.
As the thermal wave is reflected at the boundaries of subsurface features, its su-
perposition on the original thermal wave causes a change in the signal, which is
described by its magnitude and phase. Correspondingly, one obtains a magnitude
image and a phase image of thermal features within a certain depth range, de-
pending on the modulation frequency and thermal diffusivity. However, lock-in
thermography cannot distinguish the boundaries of good areas from those of new
and significant defects.
Instead of injecting heat from the outside, one can generate it inside the sample.
Vibrothermography is a method where a modulated load induces temperature
changes owing to the thermoelastic effect [8]. Stress concentrations can be visual-
ized this way. Photothermal inspection of cyclically loaded polymer samples al-
lows one to distinguish thermoelastic effects from hysteresis effects [9]. At higher
frequencies, the efficiency of heating is improved. Therefore, the injection of am-
plitude-modulated ultrasound provides an efficient means of periodic heating, by
which the defect acts as a thermal-wave transmitter operating at the low modula-
tion frequency [10].
Figure C2 shows the principle and its difference from optical excitation. With
modulated optical illumination the thermal wave is generated externally in the
surface, from where it propagates to be reflected at the hidden defect (Fig. C2,
left). With acoustic excitation (Fig. C2, right) the ultrasonic transducer attached to
the sample launches modulated acoustic waves into the sample, where they propa-
gate until they disappear when they are converted into heat in the high-loss-angle
areas of defects. Thereby, the defect reveals itself by an internally generated and
emitted thermal wave, which is phase-sensitively monitored when it arrives at the
sample surface. So only the defect is detected and not the intact areas of the com-
358 A. Dillenz, N. Krohn, R. StoOel, G. Busse
ponent being inspected. As the phase difference between the modulation and the
thermal wave is proportional to the depth at which the defect is located, local
depth information is obtained [11].
Instead of a modulated acoustic input, one could use an acoustic pulse, which
would be the counterpart of optical-pulse thermography [12]. However, this re-
quires a much higher power density on the sample. It should also be mentioned
that loss-angle heating can be detected selectively by correlated thermal expansion
[13].
'\
1
/~' .... ,If
Ultrasonic
\~
Defect
Sample
Optical Wave Thermal Wave Elastic Wave Thermal Wave

.......)$<- ------. • ------~
Fig. C2. Comparison of optical and acoustic excitation. Left - optical lock-in thermography
(OLT). Thermal waves are excited in the surface by optical absorption; sup.erposition
of the incoming and outgoing thermal waves allows imaging of hidden boundaries.
Right - ultrasound lock-in thermography (ULT). The absorption of modulated ultra-
sound turns a defect into a local thermal-wave source, which is selectively monitored.
3 Examples of Defect-Selective Imaging
The mechanical change induced by a defect can result in an enhancement of the

non-linearity or of the local hysteresis. We performed feasibility studies on both
methods in order to see how sensitively they responded to various kinds of defects
and processes. Examples are now presented, with an emphasis on polymer materials.
3.1 Monitoring of the Bonding Process
Monitoring of the acoustic changes in epoxy resin during curing allows one to
characterize the bonding process and even to predict the strength finally achieved
[14]. If a piezoelectric transmitter is attached with resin to a carbon-fibre-
reinforced polymer (CFRP), the overtone spectrum will change during the curing
process (Fig. C3, left). So the observed spectrum is an indicator of the curing pro-
cess. In Fig. C3 the overtones reach asymptotically a level which differs from zero
since a flaw was built in on purpose. This area of unbonded material stands out
clearly if higher harmonics are used for imaging (Fig. C3, right).
-30 .-------------~
1 kHz
4 6 8 10 12 14 16 18 20 22
li me (min)
Fig. C3. Change of overtone spectrum during the curing process (left) and overtone image
(taken after curing) of a transducer, half of which is disbonded (right)
3.2 Detection of Disbonded/Delaminated Areas

in Polymers, Wood and Coatings
The generation of harmonics in bonding defects has been investigated with ultra-
sonic waves [15]. The image in Fig. C4 is an example showing how the overtone
spectrum in a delaminated area and in an area next to it differ. This technique
might also provide a means of detecting kissing bonds.
m~ r-----------'
"0 Bonded
';;'-60
"0
~ ·70
~-80
<{
....- - CI:
i ·~: 1Ul.-''-:4:--____,....,.,.-~",______,~.
. 110
o 200k
Delaminated
Fig. C4. Overtone spectra of a delamination (bottom graph) and of an area next to it (top
graph); acoustic excitation at 50 kHz
360 A. Dillenz, N. Krohn, R. Sto8el, G. Busse
Fig. CS. Phase-angle image obtained by ultrasound lock-in thermography from a local disbond
in zirconia-coated metal; carrier frequency 20 kHz, amplitude modulation at 0.06 Hz
Delamina tion
Fig. C6. Amplitude image obtained by ultrasound lock-in thermography from a bonding defect
between wood panels
The detection of disbonds can also be performed using acoustic excitation and
thermographic loss-angle imaging [16], as is shown in Fig. C5 for a metal sample
coated with ceramic (zirconia) and in Fig. C 6 for a delamination between wood
panels.
3.3 Impact Damage
Impact damage in CFRP may be a source of serious failure. However, this type of
damage is difficult to detect since the damaged area is hidden underneath a seem-
ingly undamaged impact area. For this reason, OLT may not reveal such an im-
pact, while ULT can: with OLT one sees mainly the fibre orientation inside the
laminate, while with ULT only the impact is imaged (Fig. C7). A CFRP sample
with seven impact points was investigated with both ULT and overtone imaging
(Fig. C8). In both cases the impacts stand out clearly since the boundaries of the
damaged zones rub or clap against each other during cyclic loading, thereby gen-
erating both overtones and heat.
Fig. C7. Lock-in thennography (modulation frequency 0.03 Hz) of impact in CFRP (thickness
8 mm) with inspection from impact side: ULT (left) and OLT (right)
Fig. CS. CFRP with seven impact points: ULT (left, [16]), overtone imaging (middle) and air-
coupled ultrasound C-scan (right) give very similar results
3.4 Cracks in Metals and Ceramics
Most of the previous examples were related to polymers. The potential for the
detection of cracks is demonstrated here with some examples of other materials. In
the stress field of an elastic wave, the sides of a crack move against each other,
either in a shear or in a tensile mode. In a ductile material, the sides come into
contact mainly near the crack tip, while in a brittle material the crack does not
open, so that heat and overtones are generated along the whole crack. Therefore, in
a metal only the crack tip can be detected, while in a ceramic material the whole
crack becomes visible, as demonstrated in Fig. C9 for a ceramic cup [16].
362 A. Dillenz, N. Krohn, R. Sti)Oel, G. Busse
Fig. C9. ULT of a crack in a cup [16]: thermography image (left) and ultrasound lock-in ther-
mography image (right)
3.5 Corrosion
Metal components may deteriorate not only by cracking but also by corrosion.
Therefore, early detection of corrosion is important in inspection for the purpose
of maintenance, e.g. in aircraft. Corrosion products are loose, thereby allowing
ooo o o o o oo~
Rivets
'i - .... _--
Ultrasound Lock-in
Thermography O.03Hz
Lock-in Thennography O.03Hz
Fig. CIO. Results of remote lock-in thermography inspection of riveted aluminium structure
using ULT (middle) and OLT (bottom)
relative motion, which causes selective heating when ultrasound is injected [16].
Figure CIO shows the difference between the information provided by OLT and
ULT, both performed at a modulation frequency of 0.03 Hz. OLT responds to the
effect of the injected, local modulated heat flow, so it displays mainly the rivets
pressing the two aluminium plates together. However, ULT responds to the effi-
ciency of acoustic heating by internal friction, so the pattern of the same area looks
very different, since it displays most of all the corrosion. The results would be the
same if the sample were painted, thereby making the rivets invisible. So one needs
both images: OLT to display the hidden structure and ULT to see where corrosion
is located with respect to the internal structure.
4 Conclusions
We have presented examples offast (for ULT the required inspection time is typi-
cally 3 minutes) defect-selective imaging of various materials and their defects.
The advantage is the capability to provide clear information on- defects even in the
presence of complicated intact structures. Reference images, whose structure is
used for image comparison, are no longer required, since the background is zero
('no defect - no signal'). Therefore, the inspection of critical components becomes
safer and less sensitive to human factors. This makes the technique better suited
for robotic inspection aimed at avoiding catastrophic failure. Hence the applica-
tions in the field of quality control of safety-critical systems (e.g. in the aerospace
and car industries) are obvious.
However, we would also like to point out that some problems still need to be
solved, e.g. the input coupling of ultrasound limits the object size to about 0.5 m2
at present. Also, it is not known what level of acoustic power can be applied to
intact components without affecting their quality.
Acknowledgements
The authors are grateful to Dr. H. Richter (WIWEB, Erding), Dipl.-Ing. R. Aoki
(DLR Stuttgart) and Dipl.-Ing. M. Flock (lFB, University of Stuttgart) for kindly
providing the samples used in our experiments.
References
1. Summerscales J. (1987): Non-destructive Testing of Fibre-Reinforced Plastics Composites.

Elsevier Applied Science, London, New York, Vols. I and II (1990)
2. Richardson J. M. (1979): Harmonic generation at an unbonded interface -I. Planar interface
between semi-infinite elastic media. Int. J. Engng. Sci. 17: 73-85
364 A. Dillenz, N. Krohn, R. StiSBel, G. Busse
3. McCall K. R., Guyer R. A. (1994): Equation of state and wave propagation in hysteretic
non-linear elastic materials. J. Geophys. Res. Vol. 99 B12: 23887-23897
4. Carlomagno G. M., Berardi P. G. (1976): Unsteady thermotopography in non-destructive
testing. In: Proceedings of the 3rd Biannual Exchange, St. Louis, USA, August 24-26:
33-39
5. Beaudoin J. L., Merienne E., Danjoux R., Egee M. (1985): Numerical system for infrared
scanners and application to the subsurface control of materials by photothermal radiometry.
In: Infrared Technology and Applications, SPIE 590: 287
6. Kuo P.K., Feng Z. J., Ahmed T., Favro L. D., Thomas R. L., Hartikainen J. (1987): Parallel
thermal wave imaging using a vector lockin video technique. In: Hess P., Pelzl J. (Eds.)
Photoacoustic and Photothermal Phenomena. Springer, Berlin, Heidelberg: 415-418
7. Busse G., Wu D., Karpen W. (1992): Thermal wave imaging with phase sensitive modulated
thermography. J. Appl. Phys. 71: 3962-3965
8. Henneke E. G., Reifsnider K. L., Stinchcomb W. W. (1979): Thermography - a NDT
method for damage detection. J. Metals 9: 11-15
9. Busse G., Bauer M., Rippel W., Wu D. (1992): Lockin vibrothermal inspection of polymer
composites. In: Balageas D., Busse G., Carlomagno G. M. (Eds.) Quantitative InfraRed
Thermography. QIRT 92, Paris. Editions Europeennes Thermique et Industrie: 154-159
10. Rantala J., Wu D., Busse G. (1996): Amplitude modulated lockin vibrothermography for
NDE of polymers and composites. Res. Nondestruct. Evaluat. 7: 215-218
11. Busse G. (1980): Photothermal transmission probing ofa metal. Infrared Phys. 20: 419-422
12. Salerno A., Dillenz A., Wu D., Rantala J., Busse G. (1998): Progress in ultrasonic lockin
thermography. In: Balageas D., Busse G., Carlomagno G. M. (Eds.) Quantitative Infrared
Thermography. QIRT 98, Lodz, Poland. Akademickie Centrum Graficzno - Marketingowe
Lodart S. A.: 154-160
13. Salerno A., Danesi S., Wu D., Ritter S., Busse G. (1997): Ultrasonic loss angle with speckle
interferometry. In: 5th International Congress on Sound and Vibration, Adelaide, Australia,
December 15-18, CD-ROM
14. Geisse D., Nixdorf K., Busse G. (1997): Cure monitoring of thin adhesive layers. In: 5th
International Congress on Sound and Vibration, Adelaide, Australia, December 15-18, CD-
ROM
15. FaSbender S.U., Arnold W. (1996): Measurement of adhesion strength of bonds using non-
linear acoustics. In: Thompson D.O., Chimenti D. E. (Eds.) Rewiews of Progress in QNDE
15, Plenum, New York: 1321-1328
16. Rantala J., Wu D., Salerno A., Busse G. (1997): Lockin-thermography with mechanical loss
angle heating at ultrasonic frequencies. In: Busse G., Balageas D., Carlomagno G.M. (Eds.)
Quantitative Infrared Thermography. QIRT 96, Stuttgart, Germany. Edizione ETS, Pisa:
389-393
C.2 Determination of Local Deformation
Behaviour of Polymers
by Means of Laser Extensometry
C. Bierogel and W. Grellmann, Merseburg
1 Introduction
On the one hand, the polymer industry is characterized by an increasing speciali-

zation and variety of products. On the other hand, there is also an increasing range
of applications for plastics. Chemical and physical modification of the basic poly-
mer materials, often as a result of customers' demands, has led to a confusing
variety of materials on offer. The choice of a specific polymer is often influenced
by an insignificant change in the material properties. In addition to such small
changes, the cost is decisive in practice.
As a result of the wide range of polymer materials, the range of thermal and
electrical properties, and the many modifications of strength and toughness, exten-
sive requirements are placed on modem methods of material testing. This applies
particularly to the sensitivity and precision of the test methods and the quality of
the material parameters used. For the purposes of material development, quality
assurance and failure analysis, the important relationships between properties and
structure must be well known in order that the lifetime and reliability can be simu-
lated.
In spite of the many microscopic, spectroscopic and defectoscopic tests using
new or better physical principles of operation, the mechanical testing of polymers
still has high importance. Suitable material parameters are required, especially for
the purposes of differentiation of polymers, the evaluation of strength and tough-
ness modifiers, and the dimensioning of polymer components by users or develop-
ers. These parameters must be independent of influences of the methods used to
obtain them. This applies to both conventional testing methods, such as tensile and
flexural testing, and methods based on fracture mechanics for the characterization
of the toughness of polymers and blends.
Independently of the test method or the specimen geometry (specimens of the
geometry specified in ISO 3167 or fracture mechanics compact tension (CT)
specimens), the measured parameters related to strength, deformation or toughness
reflect influences of manufacturing and loading. The geometrical influences on the
material parameters are minimized if the appropriate geometrical criteria are satis-
fied. On the other hand, factors associated with the production process, such as
molecular weight, degree of crystallinity, residual stress and orientation influence
the deformation and fracture behaviour directly. In the case of composites, these
effects are overlapped by the those of the adherence, anisotropy, geometry and
filler content achieved. For blends and copolymers, the stabilization, interparticle

366 C. Bierogel, W. Grellmann
distance and particle diameter are the more important material parameters. The
deformation, strain rate and kinetics of damage processes are decisively influenced
by the test conditions. It is clear that material parameters suitable for material
development and optimization that are independent of production conditions are
not achievable with conventional testing methods. The structure and morphology,
in combination with the test conditions, determine not only the local and tiroe-
dependent deformation behaviour, but also the nature and extent of the pre-
damage. For the interpretation or evaluation of the results of mechanical investiga-
tions, experimental data on the residual stress, orientation, filler content, fibre
length and fibre distribution are necessary [1-3].
As a result of these increasing requirements on the mechanical and fracture me-
chanics parameters determined with quasi-static loading, some new methods are
being used in material testing. These methods include
• hybrid test methods such as ultrasonic methods, acoustic-emission analysis
and thermography, used simultaneously with mechanical test methods [2]
• improvement of mechanical test methods by using digital and closed-loop
techniques, new physical principles of measurement [4], and optical tech-
niques for strain measurement [5].
Independent of the methods and loading conditions chosen, the most important
aim of all such work is an event-related interpretation and evaluation of deforma-
tion and fracture behaviour.
The load cells of modem universal testing machines have a low temperature
drift, high linearity and a high dynamic range for the signal. So, a wide range of
load cells with high precision is available for material testing. This means that the
results of mechanical tests are strongly influenced by the extensometer used. It is
well known that various problems can occur with the use of mechanical exten-
someters for polymer testing. These problems are caused by the non-negligible
inherent weight of the sensor and the high notch sensitivity of polymers. As a re-
sult of these disadvantages, optical extensometers such as video extensometers [6],
field-measuring systems (scanning methods, moire techniques, holography [7] and
laser speckle interferometry [8]) and also laser extensometers are being increas-
inglyused.
2 Experimental
The techniques of strain measurement by means of a laser extensometer use re-

flection or transmission modes. The specimens used with these methods need re-
flecting targets for the laser beam or, alternatively, the edges of the specimen are
used to interrupt the beam. In tensile or compressive testing, these extensometers
are used in a direct or a diffuse reflection mode with gauge marks (fringes) applied
to the specimen [9]. For application to fracture mechanics experiments, e.g. with
CT specimens, the transmission mode is preferred. The measurement is based on
Determination of Local Deformation Behaviour 367
the scanning time ofa rotating mirror or prism (Fig. Clla). The laser extensometer
shown in Fig. Cll works with a constant laser power of 4 to 6 mW and an object
distance of 200 mm. With these values and a total initial gauge length Lo of 160
mm, the maximum resolution of the extension is 1.5 11m. For the determination of
the local deformation behaviour, it is necessary to apply reflecting fringes to the
specimen surface. The maximum number of fringes in this case is 63, whereby 62
strain-time curves are observable. In case of polymers, the application of fringes is
possible by tampon-printing or silk-screen printing or by making simple bright-
dark marks with pens or an airbrush. For the measurements to be accurate, a
minimum distance of the fringes of 1 mm should be ensured. To define the initial
position of the applied fringes, 20 scans in the unloaded state are necessary. This
means also a self-calibration of the laser extensometer, because the initial gauge
length is not exactly the specified value in all cases. The start and stop diodes
allow compensation of speed fluctuations and synchronization with the system time
of the universal testing machine. If a laser extensometer is coupled with the
crosshead of a testing machine (Fig. C 11 b), the most important things to ensure are
low vibration and an exact alignment of the optical axis. The experience of many
investigations shows that there must be precisely parallel grips for reliable results,
too. As part of the measurement of the of local deformation, the system allows also
the measurement of transverse strain until the fracture of the specimen (Fig. Cllc).
By means of movable vertical and horizontal apertures, a measuring window 20
mm long is created. The transverse strain is measured with a shadow-image
method, and a resolution of 0.15 11m can be achieved. The disadvantage of this
type of test is the manual calibration that is required for every specimen before
starting a measurement. The recorded local extensions between the fringes are
available for the calculation of the strain in relation to the measured stress, the time
signal or a simultaneously recorded signal such as acoustic emission. The integral
stress-strain diagram in the case of ten fringes (Fig. C 11 b) is the same as that
obtained using a mechanical extensometer but is complemented by the nine local
diagrams. So, many possibilities for the presentation of the stress-strain behaviour
exist. Often, a high variation of yield strain or breaking elongation can be detected
(Fig. C12a). If the ultimate failure of the specimen occurs in the middle of the
initial length, there is a maximum of local strain between fringes 5 and 6. The
computerized off-line evaluation of the data allows the determination of various
relationships from the local and integral stress-strain diagrams. There is also the
possibility to calculate local parameters such as the local tensile modulus, yield
strength, Poisson's ratio (Fig. C12b) and strain rate, as well as a simple monitoring
of the measured data. Furthermore, for the representation of the results, an
extensive graphics unit that can calculate 3D diagrams is included (Fig. C12c)
which offers the possibility to determine the maximum and minimum of the local
strain, independent of the fringe position in the range investigated. These diagrams
also give an impression of the dynamics of the deformation process for the
specimen investigated. A new feature is that the strain values between two fringes
can be taken into account in the control of the testing machine, so that closed-loop
experiments using the local strain can be performed.
368 C. Biertigel, W.Grellmann
a)
Photo diode
b) Specimen
5-6
1-\0
6-10
Reflector
c) Specimen
Transmitter
Light Beam
Photorecei ver
Junction Diode
Fig. Cll. Principle oflocal strain measurement (a) and application ofiongitudinai/laterallaser
extensometer in a universal testing system (b); device for measurement of transverse
strain (c)
a)
70
60
....-- 50
t':S
~ 40
Fring7 Fringe
b 30
6 -10 5-6
20
10 local
0
0 2 40 2 4 6 8 0 2 4 6 8 10 12 14
e(%)
61 == f(t) 61 == f(t)
b)
F == f(61,) F == f (6Ii)
cr == f(t) cr == f(t)
cr == f (e,) cr == f (ei)
ei == f(t) e, == f(t)
e, == feE;) e, == f (Eq)
Eq == f (E;) Eq = feel)
Eli == f(ei) Eli == f(Ei)
v, == f (Ei) Vi = f(ei)
Ei = f(t) E, = f(t)
c)
25
20
....-- 15
if-
~ l~~~illIlIlIlIlI~~
2 3 4 5 6 7 8 910
Reflecting Fringes
Fig. e12. Stress-strain diagrams of PP/GF with 30 wt.% short glass fibres (a), ways of evalu-
ating and presenting the data from the longitudinaVtransverse laser extensometer
(b), and local deformation behaviour ofPP/GF with 30 wt.% GF (c)
370 C. Bierigel, W. Grellmann
The laser double scanner shown in Fig. Cl3 and described in A2.1 (pp. 51-70)
and in [10-12] works in transmission mode with two parallel laser beams. For
adaptation to different geometries of specimens, the beams are adjustable in the
range from 0 to 50 mm. In contrast to the 10ngitudinaVtransverse laser extensome-
ter, this system is favoured for experimental fracture mechanics investigations on
CT specimens. Because mechanical extensometers such as COD sensors influence
the measurement of crack-mouth-opening displacement or load-line displacement,
a contactless optical extensometer is necessary. MechanicaLextensometers that can
be used with a temperature chamber are not generally available. With the two
beams of the laser-double scanner, experiments using a temperature chamber can
be carried out successfully on various polymers. The object distance is again 200
mm, but when a temperature chamber is used, the power of the laser must be con-
trollable. If the conditions of measurement are perfect, a resolution of 0.5 J.lm is
possible. The edges of the CT specimens must be of a high quality because the accu-
racy of the measurements depends on the roughness of the outline. For every meas-
ured point, ageometrical correction of the displacement is necessary. This is because
of the rotation and translation of the CT specimen while the laser beams remain
fixed. Before the test can be started, an unloading period of 20 scans is necessary to
determine the position of the edges as in a self-calibration.
Evaluation and
Graphics
Mirror
Fig. Cl3. Schematic representation of principle of measurement and function of the laser
double scanner for static fracture mechanics experiments
The extensive equipment that makes up the laser multi-scanner, shown in Fig.
C14, is based on the same technical principle of local strain measurement as that of
the 10ngitudinaVtransverse laser extensometer. However, this system works with
six semiconductor laser diodes, providing a parallel laser field. So, these time-
displaced lasers produce, in addition to information about the local deformation,
information about the deformation across the width of the specimen. Control of the
time displacement and rotation speed of the scanner is realized with the help of the
start and stop diodes. Because this multi-scanner uses the basic techniques of the
longitudinaVtransverse laser extensometer, there is also the possibility of self-
calibration. The resolution of this scanner is 1.5 Ilm at an object distance of 200
mm and at a rate of 40 ms per scan. All laser beams are freely adjustable over a
range from 0 to 50 mm. The beams can be adjusted to one point; the advantage of
this is that the roughness of the fringes can be corrected if extremely high precision
is required. As a result of the flexibility of this laser multi-scanner, new applica-
tions in polymer and component testing are possible.
Fig. C14. Principle of measurement of the laser multi-scanner
3 Applications of Polymer Testing
3.1 Longitudinal/Lateral Laser Extensometer
Simultaneous experiments with laser extensometry and acoustic-emission analysis

under tensile loading show that a connection exists between the acoustically meas-
ured damage kinetics and changes of the local deformation behaviour, especially
for short-fIbre-reinforced thermoplastics [2]. An inspection of the dependences
shows a correspondence between the acoustic onset and the time of splitting in the
local deformation. Independent of the amount of integral strain, an effect of local-
ization seems to be present in the distribution of local deformation. Using a linear
372 C. Bieriigel, W. GreUmann
o A.·
17~--~-rr-----------------------------~~16
a
/
Effects of
Track ofehnax in the Gauge Length
Clamping
....
..
!
,-... •• • . . r:•••
V2 • I. • .,::~
...
., .., .. ..-
II .'
<i O+---t--_J ·_.....•·...!!. - - .'•

:. •• •
""if'"., 0;;;;.: .....
"e.I!'.•. • . ., .
..••••• ,.
• ••
- Localization of Damage by Local Strain

••• Localization of Damage by Acoustic Emission
I • •
11~-+---r--.---.---.-~.--,---.---r--~--+ 1
o 0.5 1.0 1.5 2.0 2.5
e(%)
::r:11~ __
1 Fringes
Fig. CIS. Comparison of damage localization by means of acoustic-emission analysis and

laser extensometry on notched PA 6 with 10 wt.% GF (a) and localization by means
of heterogeneity on unnotched PA 6 with 20 wt.% short GF (b)
mode of localization with two resonant sensors at a distance I apart, it is possible

to measure the propagation time of acoustic signals. For a notched polyamide
specimen with 10 wt. % short fibres that had a suitable velocity of surface waves, a
total time difference of 34 ms was measured. In the case of a notch (razor blade
notch with a radius of 0.2 J.lm and a depth of 0.3 nun) in the middle of the speci-
men, a propagation time of 17 ms for each sensor is due to a deformation process at
the crack tip (Fig. elSa). The friction and sliding of the specimen in the grips cause
Bonding Seam
Injection Injection
~ ~~~~~g5~~~
Point
Specimen
Cross Section
25
1425
20
~ 15
~1O
tA)
5
o
1 Fringes 16
Fig. C16. Influence of microstructure on local deformation behaviour for PA with 10 wt.%
short glass fibres
localization signals at the start of the experiment. A wide localization area occurs
after the observed onset of acoustic emission. This effect is due to the damage
processes in the area around the crack tip. The projection of the maximum local
deformation into the plane of acoustic localization shows a good correlation be-
tween these two parameters. After the crack initiation, a concentration of acoustic
signals and local strain in the middle of the specimen was recorded. The increasing
signal dispersion and decreasing propagation time cause the acoustic localization
to fail if the content of filler is higher than 15 wt.% short fibres or if unnotched
specimens are used. If the heterogeneity function H(&; ), schematically shown in
Fig. C19, is used and the results are evaluated taking account of the position of the
fringes, the localization shown in Fig. CI5b is obtained for unnotched polyamide 6
with 20 wt.% glass fibres . With increasing integral strain, the localization function
reflects the orientation of the specimen clearly and detects the area of ultimate
374 C. Bieriigel, W. Grellmann
failure early (Fig. C 12c). It is also visible that the heterogeneity suddenly increases
after a defmed amount of integral strain. The splitting of local the strain-time
curves is obviously due to the initiation of damage processes or the disturbance of
the equilibrium between adaptive and damaging processes. With unnotched speci-
mens, local deformations on the surface reflect internal structural effects in the
investigated volume. Concerning this structural sensitivity, industrial use of this
technique for investigation and evaluation of bonded and welded seams is possi-
ble; because these effects are not avoidable in structures [9]. Beside a decrease in
tensile strength to 30 % of that of the non-influenced material, a change of the local
deformation behaviour can be observed. Extremely brittle fracture or plastic instabil-
ity occurs. In comparison with the integral strain, a region containing a bonding seam
shows zones of higher or lower local strain, and the deformation occurs in a hetero-
geneity (Fig. C16).
The heterogeneity shown in the 3D strain-time diagram depends on the type of
material, conditions of production, loading conditions and factors of environ-
mental loading. An increase of the heterogeneity parameter generally occurs with
increasing filler content for thermoplastics. On the other hand, the material and the
conditions of adhesion influence the behaviour determined in this way. Investiga-
tions of this kind show that polyamide 6 with 10 wt.% glass fibres has a homoge-
nous distribution of local strain over the whole range of integral strain. For PBT
with the same content of fibres, the local and time-dependent deformation behav-
iour is irregular and non-uniform (Fig. C 17). The cause of this heterogeneous local
stress-strain behaviour is the considerably smaller coupling of the fibres in the
PBT material investigated.
7
/ilf--r300
Fig. C17. Distribution oflocal deformation for PBT with 10 wt.% glass fibres
Similar effects on local deformation are observed for polyamide 66 when the
coupling conditions are varied (Fig. CIS). The results for uncoupled PA compos-
ites show higher absolute values of heterogeneity and noise (Fig. C ISb). The local
strains at 40 s (Fig. CISa) reflect the differences in local deformation caused by
various deformation and damage mechanisms. For moisture-free and normal-
moisture injection-moulded specimens of nylon, similar effects are observable.
The decrease in heterogeneity with increasing moisture content can account for the
influence of plasticizing, conditioning and heat treatment. In addition to these
influences of the material, the loading conditions during tensile testing also have
consequences for the heterogeneity. A comparison between results of conventional
tensile tests and of closed-loop tests on different polyamides with various filler
contents shows that using deformation control considerably increases the hetero-
geneity of the strain [13].
o~__~~~~~~~----------~
PA 66 without Coupling Agent
2~--------~~----------------~
Fringes 20
0.5
0.4
0.3
:I:
0.2
0.1
0.5 1.0 1.5 2.0

Ei (%)
Fig. CIS. Heterogeneity as a function of integral strain for polyamide 66 with 30 wt.% short
glass fibres: heterogeneity at 40 s (8) and heterogeneity function (b) with and with-
out coupling agent
7 70
5 56
~ 42 ~ 42
~ ~
'-'
\:)
28 '-'
\:)
28
14
0 0
0 0.8 1.6 2.4 3.2 4.0 0 0.8 1.6 2.4 3.2 4.0
E) (%) Ej (%)
4.0 70
3.2 5
,-...
2.4 ~42
'#. ~
';;;' 1.6 '-'
\:)
28
0.8 14
0
0 14 28 42 56 70 0.8 1.6 2.4 3.2 4.0
t (s) E (%)
2.0
I I lJ:..
rHe=H(EMi)
Elmax - Elmio
1.6,--
/
H=
Ei
1.2
/
o. 8 /
o. A - ~ "--- H=f(Ei)
~
o ..,-/
o 0.8 1.6 2.4 3.2 4.0
Ei (%)
Fig. C19. Local stress-strain behaviour (a), integral stress-strain diagram (b), local and inte-
gral strain-time curves (c), maximum and minimum stress-strain diagram (d), and
heterogeneity function of polypropylene with 30 wt.% glass fibres (e)
For the quantification of the heterogeneity by laser extensometry using the local
strain or the local stress-strain curve (Fig. C19a), the maximum and minimum
values of the local deformation, independent of the fringe position on the speci-
men, can be calculated. So, it is possible to determine the heterogeneity H as the
difference between the maximum and minimum of the local deformation (Fig.
C 19c) at each system time.
Because the integral strains at fracture of the specimen used are different, the
calculated strain difference was related to the integral strain over the whole meas-
uring range (Fig. C 1ge). From the related stress-strain diagrams (Figs. C 19b, d),
the parameters of the tensile test can be determined for the extremes of the local
curves and for the integral curve. The stress-strain diagram with the smallest local
tensile strain at break corresponds to a possible brittle fracture, whereas that dia-
gram with the maximum strain indicates a plastic instability of the material (pp.
549-560). The heterogeneity function so determined reflects the state ofthe mate-
rial, the conditions of production and the actual loading conditions. In dependence
on the factors described above, a buckling or splitting of the local strain-time
curves under tensile loading is often observable. These effects correlate with
changes of the damage mechanism or deformation behaviour, as well as with the
results of non-destructive methods such as acoustic-emission analysis. As a com-
parative parameter for different states of specimens or test conditions, the pa-
rameter He at fracture is practicable.
25
20
15
)t:
;; 10
C.
<iJ
0
0 so 100 150 200 250
Fringes t (s)
25
d
20
E....
15
Eo
)t:
:;- 10
>f
~
<iJ
100 150 200 250

t (s)
Fig. C20. Distribution of local defonnation for PP with 30 wt.% GF in the initial state (a) and
after 1000 hours of detergent exposure testing in I % detergent at 95 °C (b); inte-
gral, maximum and minimum local strain as a function of time for the initial state
(c) and for the material after 1000 hours of detergent exposure (d)
The laser extensometer was used to investigate PP composites intended for sub-
stitution of the steel lye tubs of top-loading washing machines by polymer materi-
als. The application of this method to the evaluation of detergent resistance dem-
onstrated a change of the specimen state for PP/GF composites at different times
of exposure (pp. 549-560). Figure C20 shows the significant influence of the ex-
posure time on the local deformation behaviour and on the heterogeneity. The
initial state of the PP/GF composite is marked by a high heterogeneity He of nearly
37 % (Figs. C20 a, c) because of the conditions of production. During the first 100
hours of the exposure test, the heterogeneity sometimes increased. The reason for
the decrease in homogeneity at 1000 hours, visible in the local strain distribution
(Fig. C20b) and the heterogeneity (Fig. C20d), is the reduction of residual stress
caused by internal changes. With higher exposure times, an equilibrium is reached.
This effect is demonstrated by the homogeneous distribution of local deformation
and by a heterogeneity that reaches only 26 %.
3.2 Laser Double Scanner
A comparison between the crosshead measurements and the laser data determined
under quasi-static loading conditions shows that a difference occurs even at
small crack-opening displacements. These errors occur because of the compliance
or stiffuess of the universal testing machine. The main variables of importance
here are the clamping conditions and the change of crosshead speed on starting the
experiment. These errors also influence the calculation of fracture mechanics pa-
rameters (Fig. C2la). For the characterization of the toughness of polymers by
means of fracture mechanics parameters, it is preferable to use the energy-
determined J-integral. The calculation of J values is based on the determination of
load-load-line displacement curves. Figure C2lb shows that the displacement
obtained from the crosshead measurements is higher than the values obtained from
of the laser. The crack resistance curve obtained from the crosshead displacement
leads to overestimation of the toughness parameters. This means that the crack
resistance against stable crack growth of the PVC material investigated is found to
be 25 % higher than the result of the laser measurements. This effect increases if
the rotational factor is taken into consideration. A change in the values for crack
initiation and crack propagation is then also observable.
The industrial application of fracture mechanics values for the optimization of
toughness for the evaluation of structures basically depends on the geometry inde-
pendence of these parameters. For this reason, additional experiments were carried
out to characterize the influence of specimen thickness on the J-Aa and 8--Aa
curves. By means of the laser double scanner, PVC specimens with thicknesses
ranging from 4 to 20 rnrn were investigated at a crosshead speed of 1 rnrn/min. The
evaluation of J-Aa curves was performed by the multiple-specimen method ac-
cording to ESIS TC4 [14]. The J-Aa curves so determined, taking account of the
validity limits Aamim Aamax and Jmax, are shown in Fig. C22. All points outside
these limits are shown by open symbols. In accordance with the literature on PVC,
1000
800 Laser
600
g
~
400
Crosshead
200
0
0 0.5 1.0 1.5
v(mm)
Crosshead
40
30 Laser
i 20 J=
aU
b...... B (W-a)
10 a = 2 + 0.552 (1 - a/W)
U= IF dv
0
0 0.3 0.6 0.9 1.2 1.5
~a(mm)
Fig. e2l. Measured data (a) and toughness parameters (b) of PVC specimens
geometry-independent crack initiation parameters JO.2 and tearing moduli TI2 were
found at a thickness B of 10 mm (Fig. C22b). There also exists a transition from a
plane stress to a plane strain state with a local maximmn of the crack initiation
parameters. Simultaneous measurements of t5-t1a curves with the laser double
scanner lead to the conclusion that this maximmn is deformation-determined [10].
For PVC, geometry-independent fracture mechanics parameters of J O.2 = 8.3 Nmm- I
and TJ = 34 are significant. A similar critical specimen thickness was obtained by
Huang [15], where the requirements on the specimen dimensions for the crack
initiation parameters were satisfied.
50 ~--------------~J~_-~~=~2~Omm~)~----~--'
i&a..-:-----------------~I a
I ISmm I
40 ~-----------------~-~
I 06. I
I ~_~
6. 0 * 0
L _ _ _ _ _ I~ ~m __ _ m_ .P_---... - I
o ~.. i;:f~ I
" B(mm) I
4 --+-
6 .........
8 ,-'.... -,-,
IO 10 ._*".-
IS - - . - -
20-......
O~~~--r_--~~r_~--_.--~~~
o 0.5 1.0 1.5 2.0
Aa(mm)
10
8 (j
,,-
.._.... I
I
• W
60
J O.2
.......
/~:
§
..
40
6 ~ t
~ . To.2 ....
o -.
......0'" ] f-<
4
State of Plane Stress State of Plane Strain 20
I
2 I
Compact Tension (CT) Specimen
0
0
.
5 IO 15 20
0
B(mm)
Fig. C22. Effects of specimen thickness on J-Aa curves for PVC (a) and dependence of
JO.2and TP (b) (evaluation in accordance with ESIS TC4; filled symbols, valid
data points; open symbols, invalid data points)
3.3 Laser Multi-Scanner
As a result of the flexibility of the laser multi-scanner, the adjustment to different

specimen dimensions at fixed positions on the surface is possible. The unprevent-
able change of measurement position caused by the transverse strain is approxi-
mately measurable by means of special reflecting targets. So, information about
the position of the laser beam relative to the specimen width is available, and new
applications in polymer and component testing are possible. This means that exten-
sive technological applications, until now only possible with laser interferometry or
strain gauge measurements, can be realized by means of a laser multi-scanner.
With this method studies of the deformation field around the crack tip of a CT
specimen are possible, as well as investigations of multi-adhesive layers (Fig.
C23). This measurement method is also applicable to the simulation of deforma-
tion and fracture behaviour in the transition from a state of plane stress to a state of
plane strain.
The predominant application fields of the laser multi-scanner are investigations
of special components, multi-layer composites, foils and welded structures [9,16].
Figure C24 shows, for example, the local deformation behaviour of a multi-layer
polyester foil. This foil was prepared with 32 fringes, and the six laser beams were
calibrated symmetrically with respect to the load line and specimen width
(Fig. C24f). The local strain distribution for different integral strain states is shown
in Figs. C24a-C24e.
At the beginning of the tensile test the local deformation behaviour is nearly
homogeneous, but at about 6 % integral strain a maximum in the local strain dis-
tribution occurs. This increased local deformation was caused by a thickness dif-
ference of 0.1 mm in the middle of the specimen. As a result of yield processes in
the area of this defect, an unloading outside this zone was registered by means of
the laser multi-scanner.
I
~~~~ ~ ~ ~
Position of Laser Beam t F
Layer I I 2 3 4 5 6
Layer 2
I Layer 3
Layer 4
W
} Layer 5 G
I 2 3,
~~ ~ ~ ~ ~ ~
I
~ 5 6 - - Position of Laser Beam
Fig. e23. Examples of application of laser multi-scanner to polymer testing

81 a 81 b
(%) (%) .
20 20
10 10
6 6
01 01
Fringes Fringes
8; = 2 % 8; = 4 %
81 c 81 d
(%) (%)
20 20
10 10
6 6
01
Fringes Fringes
8; = 6 % 8j= 8 %
81 e
(%)
20
10
6
Fringes
V'>
N 8j = 10 %
I 32
~ 6
~
65 Scan
105
Fig. C24. Local defonnation of a multi-layer polyester foil with a thickness of 0.35 nun as a
function of the measuring position of the laser beams
4 Summary
The experimental possibilities of measurement and evaluation provided by the

laser extensometers described above demonstrate the versatility of these modem
optical extensometers. One of the main advantages, especially for plastics, is that
the method is contactless and can be applied to specimens of different dimensions.
The information that it can provide, about local and integral strain as well as strain
rate, and calculated parameters such as the heterogeneity, is very important for
material development and polymer testing. With regard to loading conditions, the
laser extensometer shows every error associated with clamping or the methods of
tensile testing. The observed distribution of local deformation at the specimen
surface correlates with the results of non-destructive methods such as acoustic-
emission analysis or thermography. By means of an extensive evaluation, the lo-
calization of active damage processes with respect to time or position is also pos-
sible. Because the results of in-situ experiments are always related to the integral
strain, there exists now the chance to correct the strain to that of the local damage
zone. Among the multiple possibilities of such laser extensometers for the deter-
mination and representation of local deformation behaviour, the control of engi-
neering strain and true strain is possible if the requirements of the appropriate
mathematical algorithms are satisfied. The independence of laser extensometers
from the specifications of universal testing machines allows the application and
coupling of laser extensometers to all testing systems.
From this point of view, these optical extensometers are an innovative tech-
nique for the testing and development of materials. For the enhanced evaluation of
materials and the simulation of reliability and lifetime, laser extensometry is a
valuable technique for the material testing of polymers.
References
1. Matsuoka T. (1998): Fibre orientation due to processing and prediction. In: Karger-Kocsis 1.
(Ed.) Polypropylene: A-Z Reference. Kluwer Academic, Dordrecht: 233-239
2. Bartnig K., Bierogel C., Grellmann W., Rufke B. (1992): Anwendung der Schallemission,
Thermographie und Dielektrometrie zur Bewertung des Deformationsverhaltens von Poly-
amiden. Plaste Kautschuk 39: 1-8
3. Thomason J. L. (1998): Mechanical and thermal properties of long glass fibre reinforced
polypropylene. In: Karger-Kocsis J. (Ed.) Polypropylene: A-Z Reference. Kluwer Aca-
demic, Dordrecht: 407-414
4. Markowski W. (1990): New fundamental of material testing system. Materialpriifung -
Mater. Test. 32: 144-148
5. Andresen K., Ritter R, Steck E. (1991): Theoretical and experimental investigations of
fracture by FEM and grating methods. In: Blauel J. G., Schwalbe K.-H. (Eds.) Defect As-
sessment in Components - Fundamentals and Applications. ESISIEGF9, Mechanical Engi-
neering Publications, London: 345-361
6. Pye C. T., Martin 1. (1993): Video extensometer principles and operation. Sensors 11:
10-14
7. Jones R., Wykes C. (1989): Holographic and Speckle Interferometry, 2nd edition. Cam-
bridge University Press, Cambridge
8. Ettemeyer A., Wang Z., Walz T. (1996): Applications of speckle interferometry to material
testing. In: Proceedings of the KSME Conference, Seoul, Korea, November 1: 461-466
9. Martinez J. A., Criado N., Eisenreich N., Kugler H. P., Drude H. (1993): Distribution of
mechanical properties in a 2014 T6 welded joint. Aluminium 69: 646-648
384 C. Bieroge1, W. GreUmann
Appl. Polym. Sci. 64: 1079-1090
II. Grellmann W., Seidler S., Bierogel C. (1997): Geometry-independent fracture mechanics
values as a requirement for toughness optimization of polymers. In: Proceedings of the 9th
International Conference on Fracture (lCF9), Sydney, Australia, April 1-5, Vol. 2:
1013-1020
12. Bierogel C., Fahnert T., Grellmann W. (2000): Evaluation of the deformation of polyamide
materials under tension using laser extensometry. In: Proceedings of the 13th European
Conference on Fracture, San Sebastian, Spain, September 6-9, CD-ROM Polymer and
Composites, No. 17: 1-8
13. Grellmann W., Bierogel C., Konig S. (1997): Evaluation of deformation behaviour of poly-
amide using laser extensometry. Polym. Test. 16: 225-240
Resistance Curves Tests on Plastics
toughened polymers. In: Proceedings of the 7th International Conference on Fracture, Ad-
vances in Fracture Research (lCF7), Houston, Texas, March 20-24: 2725-2732
16-.Grellmann W., Bierogel, C. (1998): Laser extensometry applied. Materialpriifung - Mater.
Test. 40,11-12: 452-459
C.3 Damage Analysis of Composite Materials
by Acoustic-Emission Examination
J. Bohse and T. Krietsch, Berlin
1 Introduction
The damage behaviour of composites is of interest for the purpose of development

and application of materials. The strength and lifetime of composite materials are
determined by the stress stages of the initiation of micro-failure processes and by
the transition of the damage from a subcritical stage to a critical stage. Inter-fibre
fracture, delamination and fibre failure are processes of special importance. The
damage behaviour of composite materials depends considerably on the crack re-
sistance of the matrix, on the debonding energy of the fibre-matrix interface and,
fmally, on the delamination resistance of the layers of the composite.
Acoustic-emission examination is a very successful tool for the sensitive detec-
tion and location of active damage in fibre composite materials [1,2]. The inter-
pretation of the signals and, hence, the identification of damage stages is a major
problem of the acoustic-emission (AE) method. There is much experience in inter-
preting correlations between conventional AE features, such as amplitude distribu-
tion, counts, duration, etc., and true damage mechanisms, as well as friction and
noise. However, the absolute values of all these features change because of wave
propagation through the test specimen, as a consequence of material and geometri-
cal effects such as attenuation, dispersion, reflection, etc. [3]. Modem tools for the
examination of AE source mechanisms include mode analysis of AE waves and the
use of neural networks [4-9]. At the Federal Institute for Materials Research and
Testing (BAM, Berlin, Germany) analytically based waveform treatment is pre-
ferred. Therefore, a new selection method for single transient acoustic emissions
has been introduced. In a second step, the selected emissions are classified into
matrix cracking, fibre breakage and interface processes such as fibre-matrix
debonding or fibre pull-out on the basis of their total power in defmed frequency
intervals of the spectral power density. Changes in acoustic-emission source
mechanisms are recognized from average power spectra for selected time or stress
intervals or from the classification of single emissions during damage evolution. A
distinction between subcritical and critical damage stages becomes possible.
Numerous failure theories are available for fibre-reinforced composites [10].
Almost all of the theories are able to discriminate between two or more modes of
failure. A few theories rely on micro-mechanics. For a realistic fracture analysis of
highly stressed composite laminates, only a layer-by-Iayer prediction of the two fail-
ure modes of (a) initial damage by inter-fibre failure (matrix or interfacial fracture)
and (b) fibre breakage as the ultimate failure criterion is suitable [11]. Most of the
theories use a degradation model to account for post-initial or first-ply failure of
the laminate.

386 J. Bohse, T.Krietsch
Tensile tests accompanied by acoustic-emission examination of inter-fibre and

fibre fracture mechanisms should show how well theoretical predictions are in
agreement with experimental results. On the other hand, the quality of the AE
results can be assessed and AE signals can be interpreted as a result of such a
comparison. Studies of this kind have been carried out earlier on short-fibre-
reinforced thermoplastics [2].
The aim of the fracture toughness tests performed here was to determine the in-
terlaminar fracture energy for initiation and propagation of delamination in con-
tinuous and unidirectional fibre-reinforced composites under mode I crack-opening
load [12]. However, some problems concerning the measurements required for the
calculation of fracture energies in such tests exist. For example, the visual deter-
mination of the initiation point of delamination growth and of the exact delamina-
tion length is experimentally difficult [13] and leads to inaccurate calculation of
fracture toughness parameters. In addition, a realistic calculation of the fracture
toughness on the basis of micro-mechanical ' dissipation mechanisms [2,14-16]
needs information about the failure mechanisms activated and about the size of the
damage zone near the delamination tip. AE monitoring offers a solution to these
experimental and theoretical problems by locating the failure on a microscopic
scale, by classifying the mechanisms involved and by quantifying the extent of the
damage zone. Further, AE results can be used to approximate the energy dissipa-
tion by means of a micro-mechanical fracture model.
2 Experimental
An overview of the method of recording and processing AE signals is shown in
Fig. C25.
ACOUSTIC EMISSION EXAMINATION
Test Object AE Measuring System

Data Data => AE Signal
Acquisition Evaluation Par.uneter
Wavcfonn ~ Arriva l Time
:::> Rise Timo
~ Peak:
Amplitude
=- Duration
=> Energy
::::) Wave Mode
=> Loc,ation
::::) Partia l Power

= Median I
Maximum
Frequency
.. AE Sources (Damage ; Noise.) .. Changes in Activity and Intcnsity of AE

tI Wave Propa~ lion Effec.l5 • Location of AE Active Zones
.. Sensor Mounting .. Identificat ion of AE SOUJU5. Mechan isms/
.. AE Sensor, Fi lter Modes of Failure
Evaluation of the Structural Integrity, Damage ' Fracture Stage etc.
Fig, C2S. Overview of acoustic-emission examination

Damage Analysis of Composite Materials 387
Model specimens with well-defined source mechanisms (Fig. C26) made from the
pure matrix (notched epoxy specimens) and unnotched single-fibre-matrix specimens
(glass fibre-epoxy, carbon fibre-epoxy) in the shape of a simple plate (55 mm x
20 mm x 3 mm), loaded at a slow crosshead speed (0.2 mmlmin), were used to cre-
ate acoustic-emission reference patterns in tensile tests, for use in identification of
AE source mechanisms. Complete AE waveforms were recorded with a PC-Scope
T512 transient recorder (Intelligente Messtechnik GmbH, Germany). A PAC WD
wideband transducer, a PAC 1220A preamplifier with a gain of 60 dB and an
HPIOO filter (100-1200 kHz), a fixed threshold of 100 mY and a transient-wave-
form recorder with a sampling frequency of 5 MHz were used.
Model Materials a) Composites b)

I I
Tensile Test Tensile Test Tensile Test Double Cantilever Beam Test
Matrix Single-Fibre-Matrix Fibre-Matrix Fibre-Matrix
1
Fig. C26. Tests and specimens for identification of AE source mechanisms in (a) model
materials and (b) composites
Tensile tests for analysis of inter-fibre and fibre fracture were carried out on
carbon fibre-epoxy (CFIEP) specimens (160 mm x 16 mm x 1.3/2.1 mm) with
(0°/90°) and (90 0/±45°/900) lay-ups, a fibre volume of 60 % and a fibre diameter
of 7 Jlm, at a crosshead speed of 1 mmlmin. The conventional parameters of the
AE signals were determined using a one-channel SEK 3243 AE system (Einrich-
tung fUr akustische Diagnose und Qualitatssicherung FhG, Germany). Waveforms
were recorded from selected stress intervals with the PC-Scope T512 transient
recorder. The AE signals were monitored with a PAC WD wideband transducer, a
PAC 1220A preamplifier with a gain of 60 dB and an HP 100 filter (100-
1200 kHz), a fixed threshold of 40 dB and a transient-waveform recorder with a
sampling frequency of 5 MHz. Tensile tests were also performed on glass fibre-
polypropylene composites (for the materials characterization, see the fracture
mechanics tests below) at a crosshead speed of 2 mmlmin with a three-channel
Mistras 2001 system (Physical Acoustics Corporation, USA) in a linear set-up. A
PAC WD wideband transducer, a PAC 1220A preamplifier with a gain of 40 dB
and an HP 20 filter (20-1200 kHz), a fixed threshold of 40 dB and a transient-
waveform recorder with a sampling frequency of 8 MHz were used.
388 J. Bohse, T. Krietsch
Fracture mechanics tests were performed on double-cantilever beam (DCB)

specimens with total length 1= 120 mm, thickness B = 20 mm, total height
2h = 4 mm and an initial delamination length ao = 30 mm to measure the critical
energy release rate under a mode I crack-opening load. Piano hinges were used to
apply the load, with a crosshead speed of 2 mmlmin. Initial cracks were produced
with a diamond saw blade. The delamination growth was measured visually on the
edge of the specimen as well as by the acoustic-emission method. Two series of
unidirectional continuous reinforced glass fibre-polypropylene (GFIPP) laminates
(0° lay-up), produced by Borealis AG, Linz, Austria, with a fibre volume of 18-
25 %, a fibre diameter of 18-21 J.1m and a variable fibre-matrix adhesion,
produced by physical (matrix PP 1) or chemical (matrix PP 2) coupling with
different agents, were the materials investigated. A two-channel Mistras 2001
system was used in a linear set-up for AE measurement in DCB tests (Fig. C27).
The AE signals were monitored with two PAC WD wideband transducers a
distance of 75 mm apart, a PAC 1220A preamplifier with a gain of 40 dB and an
HP 20 filter (20-1200 kHz), a fixed threshold of 40 dB and a transient-waveform
recorder with a sampling frequency of 4 MHz.
Sensor 1 Sensor 2
P'IO o I
P I Lengthx ~
Fig. C27. Sensor positions for location of AE sources in the DeB test
3 Results
3.1 Identification of AE Source Mechanisms
Waveform files measured and saved by commercial AE systems often contain

acoustic-emission signals that are useful but are irregularly variable in number and
vary in size, and include a lot of pure noise. All these emissions are recorded with
more or less fixed threshold values and fixed time intervals.
A new method has been introduced to select transient acoustic emissions auto-
matically from the measured data. The method works on the basis of moving stan-
dard deviations on waveform time series. In this method, the beginning and end of
an emission within the time series are determined numerically and are independent
of the amplitude, duration, rise time, energy and noise level of the waveform. The
procedure for the selection of the AE signals is discussed elsewhere [17,18]. In
addition, amplitude filters are used for selecting acoustic emissions. The minimum
amplitude avoids emissions with excessively small amplitudes and unfavourable
signal-to-noise ratios. The maximum amplitude avoids the recording of over-
modulated emissions, e.g. during the macroscopic disintegration of the material.
The aim is to classify the intrinsic source mechanisms of the selected emissions.
The possibility of classifying the source mechanisms into matrix cracking, fibre
breakage and interface debonding relies on the different viscoelastic relaxation
processes near the source itself. Generally, the source mechanisms cause wideband
excitation because of stochastic processes in the structure. However, the acoustic ve-
0.4
a
....... 0.3
e'"
~
i
'-'
u
0.2
Q.
til
~
0
0.1
Q.,.
o0 100 200 300 400 500 600 700

Frequency (kHz)
0.4
0.3
o0 100 200 300 400 500 600 700

Frequency (kHz)
Fig. C28. Average power spectrum of matrix cracks in pure epoxy (a) and of carbon fibre
fragmentation in epoxy matrix (b)
locity of the relaxation differs in the fibre and the matrix because of its relation to
the elastic modulus. Thus, matrix cracking creates lower wideband frequencies
than fibre breakage, in a ratio of about 1 to 3. Interface debonding should show
both ranges of frequency because of the fibre vibration in conjunction with the
growth of the debonding crack. Characteristic relaxation frequencies between
10kHz and 1 MHz and the measurability of acoustic waves in polymers over short
distances (:::;; 100 mm) in the ranges 100-350 kHz and 350-700 kHz by wideband
transducers give good candidates for intervals that could distinguish the mecha-
nisms.
The source mechanisms are analysed as a function of time. Firstly, average
power spectra in defmed time or stress intervals are calculated. These spectra
allow one to estimate the dominant mechanisms. Secondly, single emissions are
classified and the accumulated power associated with one mechanism is plotted
versus time. A large increase in the accumulated power should give a hint of a
critical damage stage associated with that specific mechanism. Experimental re-
sults obtained from model materials (Fig. C28) are in agreement with theoretical
expectations. In Sects. 3.2 and 3.3, information about failure mechanisms in com-
posite laminates obtained from AE power spectra is described.
3.2 Analysis of Inter-Fibre Fracture
Modem physically interpretable failure criteria should be based on specified

modes of failure [11]. The criterion considered in this section relates to the inter-
fibre fracture stress level of a laminate.
The inter-fibre fracture of laminates depends on the matrix strength, the fibre-
matrix adhesion and the volume content of fibres. There is no reliable way to ap-
proximate the intra-laminar failure of a layer analytically. Critical stresses for inter-
fibre failure parallel and perpendicular to the fibre axis as well as for shear failure
501fr,--------------------------------,-, 100
CF/Epoxy (90 o /±45°/900) a
400 80
.0 300
.;;:
"£
< 200
""'
<
100 20
~~~--~~~-.--~---r--~--,---~~O
o 50 100 150 200 250

Time (s)
10°.------------------------------------.-.
~
10'3 I-~--__r-~--_r--~--_r_--~--_.__--~____I
o 50 100 150 200 250
Time (s)
c 800
300
CFlEpoxy (0°/90°)
600
'£200
·S
~ - IFF Criterion
~
100
200
0 ~~~~--~~~~~~--r-~~O
0 50 100 150 200 250
Time (s)
101
W ..
. ..,....
10° ...~ '
;..
.-:::
til
10'1
~
~
10'2
... ••
... ..
•
-t • .......1- .. # ••+:
e# ••••
.. ....-:. .........
+.+ .....
..: . ..
• •• ~....: • .; 4f.:: ••-
: ft. _ •• 6' +. • •
10,3
.
:.#~ .~~~
~
0 50 100 150 200 250

Time (s)
Fig. C29. Analysis of initial failure of a CFIEP laminate by means of modelling inter-fibre
fracture (IFF) (arrows) and acoustic-emission behaviour; lay-up with dominant ma-
trix and interface shear failure (a,b); lay-up with transverse matrix-interface crack-
ing and fracture of fibre elements (c,d)
have to be determined experimentally for each composite (for each matrix, fibre
material and strength of adhesion). The results are then used to calculate the criti-
cal stresses oflaminates with specific lay-ups with regard to initiation of inter-fibre
processes.
Here, inter-fibre fracture means both interfacial failure (fibre-matrix debond-
ing) and matrix fracture. Figure C29 shows acoustic-emission test results com-
pared with the theoretical results of a layer-by-Iayer stress and failure analysis
(arrows) for carbon fibre-epoxy laminates obtained by use of the inter-fibre frac-
ture (IFF) criterion.
20.-------------.---------------~~
o~~~~~~~~~~~~~--~
3
Frequency (kHz)
0.15 + - - - - - - - - - r - - - - - - - - . -b-,
o 100 200 300 400 500 600 700

Frequency (kHz)
Fig. C30. Average power spectra of AE signals from (a) CFIEP (90 0/±45°/900) laminate and
(b) CFIEP (0°/90°) laminate
There is agreement between the theoretically estimated inter-fibre fracture

stresses and the onset of AE activity at different levels of intensity. AE examina-
tion is a very sensitive method and also gives information about the development
of micro-failure processes before ultimate failure. However, it is not possible to
differentiate failure mechanisms clearly by means of AE activity and intensity
parameters. This can be achieved by the methodology of power spectrum analysis
described in Sect. 3.1. Figure C30 shows average power spectra from two different
carbon fibre-epoxy laminates.
The spectra indicate matrix-dominated IFF mechanisms and delamination proc-
esses (the latter being observable macroscopically) in the (90 0/±45°/900) laminate,
and combined IFF and fibre breakage mechanisms in the (0°/90°) laminate. How-
ever, such an analysis gives information only about the dominant failure mecha-
nism. An analysis of the damage behaviour into individual mechanisms can be
performed by signal-by-signal analysis of the AE data for each mechanism. Figure
C31 shows the stress and AE activity versus time for a glass fibre-polypropylene
composite and the accumulated total AE power released from various individual
mechanisms under tensile loading.
For the assignment of AE waves to individual failure mechanisms, a 70 %
power criterion was applied. It was assumed that a matrix crack had a minimum of
70 % of its power in the 0-350 kHz frequency range and a fibre fracture had a
minimum of 70 % of its power in the 350-700 kHz range. Fibre-matrix debonding
processes should release power in an intermediate range. In the case shown in Fig.
C31, the highest accumulated total AE power comes from matrix and debonding
processes and only a small part comes from fibre failure. Another piece of infor-
mation about failure on the micro-scale is that fibre breakage has already started at
about 50 % of the strength of the composite. This results from different local
stresses in the fibres and their strength distribution. Hence, this stage characterizes
a transition from a subcritical to a critical damage situation. Such quantitative
information about the occurrence and energy release of single failure mechanisms
obtained from AE measurements confirms other information about mechanical
failure criteria and modes. With this information, a more objective evaluation of
damage stages becomes possible.
The procedure discussed above for the identification of damage mechanisms in
composites was successfully applied to specimens and small structures in labora-
tory tests with short distances between the AE sources and the AE transducer (~
100 mm). Its application to larger composite structures is problematic, since the
effects of wave propagation influence the frequency spectra. This makes the de-
tection of fibre failure more difficult.
394 J. Dohse, T. Krietsch
0.3 d
'Ui'0.25 Fibre Breakage

e
~ O.
~
&'0.15
"0
~
-= O.
~
'<0.05
0
5
Fig. C31. (a) Stress and AE activity and (b-d) accumulated total AE power from various
failure mechanisms in GFIPP (0° laminate) under tension up to ultimate specimen
fracture
3.3 Mode I Delamination Behaviour
The combination of a DCB fracture toughness test with simultaneous AE moni-

toring gives useful infonnation about microscopic and macroscopic aspects of the
delamination behaviour. Figure C32 shows the time-dependent fracture force and
AE response during DCB testing.
The first AE events, located near the delamination tip (ao = 30 mm), character-
ize the initiation of the delamination process. From the location plot, the motion of
the AE damage zone can also be observed. This can be used for the detennination
of the onset of macroscopic delamination growth before the maximum of the de-
lamination force is reached. The force versus time fimction becomes non-linear at
this time in the test.
Furthennore, the acoustic emission characterizes the kinetics of progressive
delamination. The quantification of the average damage zone length and the de-
tennination of the crack velocity are possible by means of time-dependent linear
location of the AE sources at the crack tip of the delamination. 'Stick-slip' effects
during delamination growth are recorded by an alternate decrease and increase of
the AE activity and intensity. Such stages of unstable delamination growth should
be excluded from a calculation of valid toughness values for crack propagation. As
the discussion will show, a correlation exists between the mechanical fracture
energy and the release of acoustic-emission energy. This correlation is demon-
strated by AE measurements of surface motions using stress waves from the re-
lease of elastic energy by fracture processes on a microscopic scale.
What failure mechanisms cause AE signals? The spectral analysis of acoustic-
emission waveforms can give a hint of the answer to this question.
1..------------------.1.0
o 400 800 1200 1600 2000

Time (s)
Fig. e32. Results of DCB fracture toughness test and AE analysis on GFIPP (0°) laminate:
delamination force and AE energy (top) and position and size of the damage zone
determined by location of AE sources with amplitude:?: 60 dB (bottom)
Figure C33 shows power spectra obtained from two composites with different
strengths of fibre-matrix adhesion.
The intensity in the spectra at frequencies f < 300 kHz is assumed to come from
matrix cracks, the intensity at f > 300 kHz from fibre breaks. Interface processes
should show intensity in both ranges because of interface-failure-induced fibre
vibration.
All DCB test results indicate dominant matrix cracking with changing contribu-
tions from the interface energy. Increasing adhesion causes stronger interface
debonding processes, resulting in higher-frequency contributions to the power
spectrum. Fracture of fibre bridges and fibre breakage in the beams are observed,
especially in composites with weak fibre-matrix adhesion. The first process gives
only a small energy contribution. The second energy-releasing process is clearly
detected by AE close behind the delamination tip and must be excluded from the
calculation of valid parameters for propagation of the delamination.
0.4....-------,--------,---,
a
o+-~~--~~~~~~~~~~
o 100 200 300 400 500 600 700
Frequency (kHz)
0.6.------.---------,---,
b
0.5
i
~ 0.4
~ 0.3
~
~
~ 0.2
~ 0.1
O~~~-T_~~--,---~--~~~
o 100 200 300 400 500 600 700
Frequency (kHz)
Fig. C33. Average power spectra of selected acoustic emissions obtained during DeB tests on
GFIPP composites with (a) weak adhesion and (b) strong adhesion
3.4 Micro-Mechanical Fracture Model
From the concept of stress intensity of linear elastic fracture mechanics [14], the
size rg) of a damage zone is determined by the critical stresses CT~i) for initiation
of energy-consuming processes (i) ahead of the crack tip:
r(i) = f3 GrcEc (e1)

D (CT~i))2 '
398 .T. Bohse, T. Krietsch
where Ec is the Young's modulus of the composite and the factor f3:S; ~.
1[
Figure e34 shows experimental values of the damage zone length determined
by acoustic emission. However, the length In, rather than the height hn perpen-
dicular to the fibre axis, is the interesting infonnation for the fracture model dis-
cussed later. For unidirectional composites we assume
21)) = hD = xlD , (e2)
with x < 1.
The critical stress of the composite for fibre-matrix debonding can be calcu-
lated from
(e3)
where the stress intensity at the fibre-matrix interface is a"., the Young's modulus
of the interface is EIn!> the debonding energy of the fibre-matrix interface is 2rInt
and the fibre diameter is dF. The fibre-matrix debonding energy was estimated
from force-displacement functions of single-fibre pull-out experiments perfonned
with a special test apparatus of high stiffuess [19] and an interface failure domi-
nated by the mode I fracture.
MatrixPPl
MatrixPP2
I
•
0.00 0.04 0.08 0.12 0.16

Interface Debonding Energy 2r
In. (N/mm)
Fig. C34. Length of the damage zone in DeB tests detemtined by AE vs. debonding energy of
the fibre-matrix interface, for GFIPP (0°) laminates
Figure e35 shows a comparison of the theoretical approximation with experi-

mental values for the initiation of damage processes, detennined from the point at
which a significant increase of the AE intensity with time was observed.
20~--------------------------------.
• Exper. PPI
• Exper. PP2
Theory
. •
•
20 40 60 80 100 120
Interface Strength (2r (ot E(ot / -p
112 2
~) (N/mm )
Fig. C3S. Dependence of critical stress of composite on interface adhesion for GFIPP (0°)
laminates
The deviation from linearity at strong adhesion indicates an embrittlement of

the behaviour of the composite. This means that matrix fracture occurs at a macro-
scopic stress level below the level at which interface debonding becomes possible.
The micro-mechanical fracture model equation
* 2tPF*
Gc ==(l-tPF)GM* +hD ( - 2
dF
tPF* D
rInt +-17M*
2
, J (C4)
where Gc is the interlaminar fracture energy of the composite, tho is the volume of
debonded fibres (0 S tPF" s tPF), tPF is the total fibre volume, GMo is the fracture
energy of the modified matrix MO, 17Moo is the energy dissipation for shear-
yielding, craze or micro-crack processes of the modified matrix near debonded
interface areas, 2Iint is the energy for fibre-matrix interface debonding, dF is the
fibre diameter, and ho is the height of the damage zone where interface-related
mechanisms are initiated, was developed to approximate the contributions of ma-
trix and interface mechanisms to the interlaminar fracture energy [20].
The term 'modified matrix' expresses the change in the micro-mechanical be-
haviour as a consequence of structural and morphological changes in the matrix
caused by the fibre inclusions and coupling agents. The energy contribution from
the fracture of fibre bridges is disregarded because it is negligible in practice.
Figure C36 compares the theoretical and experimental results. Only the experi-
mentally determined values for the propagation of delamination are used in the
following discussion. These values were estimated according to [12] for delamina-
tion growth beyond the maximum-force point up to Aa = 20-30 mm. The theoreti-
cal behaviour was calculated under the assumption that all fibre-matrix interfaces
take part in debonding processes (¢IF' = (IF). The heights of damage zones were fitted
by (C2) with x = 114.
8
TheoryPPl
~
b Theory PP2
~= • Exper. PPI
6
•
il
Exper. PP2
~e
.5iii ct 4
~ • ...
-
~
2+------r------~----_.----~----~
2. 4 • 6
Interface Debondmg Energy ~ (4 <l>F r Inl /~) (N/mm)
Fig. C36. Theoretical and experimental results for the contribution of interface debonding
processes to mode I interlaminar fracture energies of GFIPP (0°) laminates
Increasing the matrix toughness shifts Gc to higher levels. For the same matrix
type, the fracture energies of the composites theoretically depend only on the
debonding energy.
A discussed previously, however, the experimental fracture energy shows a
transition from a tough to a brittle fracture behaviour of the composite after the
optimum adhesion is exceeded. Above this optimum, fewer fibres take part in
debonding processes (¢IF' < (IF) and the related energy contributions decrease. The
modified matrix fails in a more brittle manner as a consequence of the strong inter-
action between fibres and matrix.
4 Conclusions
• Acoustic-emission examination is a very powerful method for online analysis

of active damage in composites.
• New procedures for selection and classification of AE signals on the basis of
their power spectra allow a distinction of the failure mechanisms in fibre-
reinforced materials, such as matrix cracking, fibre breakage and interfacial
failure (fibre-matrix debonding, fibre pull-out) during materials testing.
• A theoretical approximation of the inter-fibre fracture stresses in the various

laminae of laminates and the AE analysis of the onset and modes of micro-
failure processes are in good agreement.
• A signal-by-signal classification of AE sources permits the evaluation of dam-
age stages on the basis of failure mechanisms.
• The AE activity and AE intensity characterize the kinetics of the delamination
process in DeB fracture toughness tests.
• The characterization of the position and size of the damage zone near the
delamination tip by location of AE sources gives information about the micro-
scopic and macroscopic onset of delamination growth. From such results, the
length and velocity of a delamination crack can be derived more objectively.
• A prediction of the interlaminar fracture energies of composites using a micro-
mechanical fracture model is supported by AE analysis of the damage zone
and by the classification of the failure mechanisms on a microscopic scale.
References
1. CARP Aerospace/Advanced Composites Subcom.n'littee (1993): Guidance for development

of AE applications on composites. J. Acoust. Emiss. 11(3): CI-C24
2. Bohse J., Kroh G. (1992): Micromechanics and acoustic emission analysis of the failure
process of thermoplastic composites. J. Mat. Sci. 27: 298-306
3. Downs K. S., Ramstad M. A. (1996): Wave propagation effects relative to AE source dis-
tinction of wideband AE Signals from a composite pressure vessel. J. Acoust. Emiss. 14 (3-
4): 61-73
4. Gorman M. R. (1991): Plate wave acoustic emission. J. Acoust. Soc. Am. 90 (1): 358-364
5. Gorman M. R., Ziola S. M. (1991): Plate waves produced by transverse matrix cracking.
Ultrasonics 29: 245-251
6. Gorman M. R., Prosser W. R. (1991): AE source orientation by plate wave analysis. J.
Acoust. Emiss. 9: 283-288
7. Guo D., Ajit M., Ono K. (1996): Wave theory of acoustic emission in composite laminates.
J. Acoust. Emiss. 14: 19-46
8. Prosser W. R., Jackson K.E., Kellas S., Smith B. T., McKeon J., Friedman A. (1995): Ad-
vanced waveform-based acoustic emission detection of matrix cracking in composites. Ma-
ter. Evaluat. 53: 1052-1058
9. Prosser W. R. (1996): Advanced AE techniques in composite materials research. J. Acoust.
Emiss. 14: 1-11
10. Soden P. D., Hinton M. J., Kaddour A. S. (1998): A comparison of the predictive capabilities of
current failure theories for composite laminates. Compos. Sci. Technol. 58 (8): 1225-1254
11. Puck A., Schfumann H. (1998): Failure analysis of FRP laminates by means of physically
based phenomenological models. Compos. Sci. Technol. 58 (7): 1045-1067
12. Davies P. (Ed.) (1993): Protocols for Interlaminar Fracture Testing of Composites. ESIS,
available from Marine Materials Laboratory, IFREMER, Centre de Brest, BP 70, 29280
Plouzane, France
13. Brunner A. J., Flueler P., Davies P., Blackman B. R. K., Williams J. G. (1996): Determina-
tion of the delamination resistance of fibre-reinforced composites: current scope of test pro-
tocols and future potential. In: Proceedings of ECCM7, Seventh European Conference on
Composite Materials, London, UK, May 14-16, Vol. 2: 3-8
14. Lauke B., Schultrich B., Pompe W. (1990): Theoretical considerations of toughness of
short-fiber-reinforced thermoplastics. Polym. Plast. Technol. Engng. 29 (7&8): 607-806
15. Lauke B., Friedrich K. (1992): Fracture toughness modelling of fibre reinforced composites
by crack resistance curves. Adv. Compos. Mater. 2 (4): 261-275
16. Bohse J., Grellmann W., Seidler S. (1991): Micromechanical interpretation of fracture
toughness of particulate-filled thermoplastics. J. Mat. Sci. 26: 6715-6721
17. Krietsch T., Bohse J. (1998): Classification of damage mechanisms in fibre composites by
acoustic emissions. In: Proceedings of the 23rd European Conference on Acoustic Emission,
Vienna, Austria, May 6-8: 9-14
18. Krietsch T., Bohse J. (1998): Selection of acoustic emissions and classification of damage
mechanisms in fibre composite materials. In: Proceedings of Progress in Acoustic Emission
IX, International Acoustic Emission Conference, Big Island, Hawaii, USA, August 9-14:
IV-30-39
19. Hampe A., Marotzke C. (1997): The energy release rate of the fibre/polymer matrix inter-
face: measurement and theoretical analysis. J. Reinf. Plast. Compos. 16 (4): 341-352
20. Bohse J., Krietsch T. (1998): Interpretation of fracture toughness in unidirectional glass-
fiber/polypropylene composites by acoustic emission analysis of damage mechanisms. In:
Proceedings of Progress in Acoustic Emission IX, International Acoustic Emission Confer-
ence, Big Island, Hawaii, USA, August 9-14: V-55-62
D
Technological Test Methods

0.1 Polymer-Based Composites
for Friction and Wear Applications
K. Friedrich, P. Reinicke and J. Hoffmann, Kaiserslautem
1 Introduction
Composite materials, one of the most rapidly growing classes of materials, are
being used increasingly for technological applications, especially when the tribo-
logical behaviour of the material is of interest. The properties of polymer-based
materials can be varied and optimized over a wide range with regard to the re-
quirements of the particular application. Some tribological applications, e.g. slid-
ing elements and sliding bearings, require low friction and wear, whereas for other
applications, e.g. in clutches and brakes, high friction combined with low wear is
necessary (Table Dl). Tribological systems of polymer-based composites against
steel are of high interest for technological purposes. Yet much of the present
knowledge on the tribological behaviour of polymer-based composite materials,
especially high-temperature-resistant (> 150°C) thermoplastics, is empirical, and
very limited predictive capability currently exists. In particular, experimental re-
sults of friction and wear tests under conditions close to those occurring practice
and basic knowledge of the friction and wear mechanisms in the tribological inter-
phase of polymer materials are needed.
Table D1. Tribological applications of polymer composite materials
Field Examples Type of wear loading and requirements
Civil engineering Conveyors Abrasive, minimize wear

Mechanical engi- Seals Sliding, minimize friction and wear
neering Bearings Sliding, minimize friction and wear
Gears Sliding, minimize friction and wear
Turbine and pump Erosion, minimize wear
blades
Medicine Dental applications Abrasion, minimize wear
Hip replacements
-Shaft Fretting, minimize wear
-Cap Sliding, minimize friction and wear
Automotive Brakes Sliding, minimize wear, optimize friction
Tyres Abrasion, minimize wear, maximize friction

406 K. Friedrich, P. Reinicke, J. Hoffmann
It has been shown by now in the literature that the friction and wear behaviour
of polymer materials can be improved by lower adhesion and higher stiffness and
strength [1-9]. This can be achieved quite successfully by using special fillers. To
reduce adhesion, internal lubricants such as PTFE or graphite are frequently used.
Short aramid, carbon or glass fibres are used to increase the stiffness and strength
of the polymer system. Figure D 1 shows that good friction and wear properties
(range IV) can be achieved with a material system consisting of a high-
perfonnance thennoplastic matrix with lubricants and fibrous reinforcement.
High Wear
Realstance
High "'
Temperature
Resistance
\
Reduce the
Coefficient of
Friction
Self-Lubrication
High Resistance
Against Creeping
Fig. D1. Requirements for a composite with a low coefficient of friction and a low wear rate
(range IV)
2 Friction and Wear of Short-Fibre-Reinforced Polymers
Friction and wear are properties describing not only a single material but a whole
tribological system. The tribological behaviour of short-fibre-reinforced polymers
therefore cannot be characterized by material parameters only. The tribological
behaviour in actual operation is detennined by the structure and stress of the tri-
bological system [10].
To characterize the tribological behaviour of polymer materials, standard tests
based on the tribological system being considered for practical use are used. The
pin-on-disc test, one of the most frequently used test configurations, is described in
detail in [11-15]. The results of such a pin-on-disc test of three high-temperature-
resistant polymers (polyethernitrile (PEN), polyetheretherketone (PEEK) and
polyetheretherketoneketone (PEEKK)) with different internal lubricants and short
fibres are shown in Fig. D2 [16].
Polymer-Based Composites for Friction and Wear Applications 407
10 -4 .----.---=--....-----l • 0 PEN
eo PEEK
PEEKK
E
...~ 10~ 10 ~
E
§. ;
I
;
10.7 ·~-·-,·~--~·-~-,,·····--"-··t·~-··---··-···~-·-·····----_..
I I
I I T= 150· C
! ! p = 1 MPa
! !
1008 L _.J;_ _J.;....!::::=== :::!...I
v :: 1 mls
0.4 0 .6 0.1 o 0.2 0.4 0.6 O ..J

Jl Jl
Fig. D2. Friction and wear of various short-fIbre-reinforced composites
This illustrates, for two different testing temperatures, that the fibres primarily
reduce the wear rate, whereas the internal lubricants, such as polytetrafluoroethyl-
ene (TF) and graphite (Gr), result in an improvement of the coefficient of friction.
Considering both the coefficient of friction and the specific wear rate, a system
consisting of a PEEK matrix with 10 % of CF, TF and Gr each seems to be the
best compromise, because in the temperature range considered (20°C-50 0c) a
low coefficient of friction (II < 0.2) and a low wear rate
(Dl)
can be achieved at an applied pressure of p = I MPa and a sliding velocity of

v = I mls.
In relation to the choice of a particular material for an application, two ques-
tions must be considered:
1. What materials can bear the expected load?
2. How can a particular component be produced with this material?
Although the first question is of primary concern, the question of economy will
fmally determine the choice of the particular material [17].
For technological applications, there are always several requirements, a so-
called requirement profile, to be fulfilled. Within the scope of the requirement
profile, there are properties of primary concern, such as a low coefficient of fric-
tion and a low wear rate for sliding elements. A secondary requirement could be
high stiffness and strength. Further, the requirement for high hardness is often
combined with a requirement for sufficient fracture toughness. For sliding ele-
ments in textile-drying machines subject to extreme conditions, such as elevated
temperature and impact events, the following requirement profile must be fulfilled:
(a) Manufacturing aspects: Injection-mouldable into complex geometries

(b) Primary properties: Low coefficient of friction and low wear rate
Applicable at elevated temperatures
(c) Secondary properties: High fracture toughness.
An evaluation of various potential material systems led to the final choice of a
PEEK matrix composite (containing 10 % of carbon fibres, PTFE particles and
graphite flakes each) [18]. This system showed good tribological behaviour as well
as a high fracture toughness and, further, it is injection-mouldable into complex
geometries. Systems with a PBI or PI matrix were even better with respect to their
tribological properties, but nevertheless could not be utilized over such a wide
range of technological applications because of their insufficient properties related
to manufacturing and secondary properties.
In particular, the PEEK system performed quite well over a wide range of the
pressure (P) times velocity (v) product, at various testing temperatures. Figure D3
[18] shows the influence of temperature and pv product on the depth wear rate
(Fig. D3a) and on the coefficient of friction (Fig. D3b) of a material referred to
here as PEEK-CF/GrffFI0. Because of the linear relationship between the depth
wear rate and the pv product, the gradient is related to the specific wear rate, which
is frequently called the wear factor (k~, in mm3/N m) in the open literature
[11,12]. Figure D3a shows the following average values of the wear factor: at T=
20 °C, k~ = 0.27 X 10-6 mm3/N m; at T = 100 °e, k~ = 0.91 X 10-6 mm3/N m; at
T= 150 °e, k~ = 1.11 X 10-6 mm3/N m; at T= 220 °C, k~ = 4.63 x
10-6 mm3/N m.
Because of the linearity of the graph, k~ is constant for a given material and
temperature, for a pv product up to 6.6 MPa mls (for variation of the sliding ve-
locity between 1 and 3.3 mls and the pressure between 1 and 2 MPa). In contrast
to the increase of the depth wear rate Wt = k~(pv), the coefficient does not show
any explicit dependence on either the pv product or the temperature (Fig. D3b).
o 2O"C i
0.4 r.:1-r-r-,--"r===:::::::;,
D tOll"C
150 ... A tSOOC
Ir······_·_· --··_·-1-
,
1100 1-'=<>===r22O"C=:!ti_+~-b.·;;£--1-4--1
•• 50 0.1
oLL~~~
o 2 3 456 7 234 567
pv (MPlunla) pv (MPaml8)
Fig. D3. Depth wear rate (a) and coefficient of friction (b) of PEEK with 10 % of carbon
fibres, PTFE particles and graphite flakes each, as a function of the pv product
Assuming the following operating conditions for a sliding element made of this
material:
(a) T= 20°C or 150°C
(b) p=O.l MPaor 1 MPa
(c) v= 1 mlsor3 mls
and assuming an acceptable loss in height of !lhacc = 2 mm, the maintenance inter-
val !ltmaintenance can be calculated via the relationship
Llhacc = Wt = k~(pv). (D2)

Lit maintenance
Hence,
Llhacc
Lltmaintenance = -.- - . (D3)
ko(pv)
The maintenance intervals for the given operating conditions are shown in Ta-
ble D2. Under conditions close to those occurring practice (T = 150°C, p = 0.1
MPa, v = 1 mls), the maintenance interval for the chosen material is about 2.5
years, assuming that the machine is operated under these conditions for about 40
hrs per week.
Table D2. In-service lifetime of a sliding shoe made of PEEK-CF/GrffFIO under different
sliding conditions
Temperature Load Velocity Mlnspection In-service

eC) (MPa) (mls) (h) lifetime a
20 0.1 1 20576 ca. 10 years

Ws = 0.27 X 10-6 1 1 2058 ca. 1 year
0.1 3 6859 ca. 3 years
(rnm3/N m)
1 3 1686 ca. 4 months
150 0.1 1 5005 ca. 2.5 years
Ws = 1.11 X 10-6 1 1 501 ca. 3 months
0.1 3 1668 ca. 10 months
(rnm3/N m)
1 3 167 ca. 1 month
a In-service lifetime is calculated on the basis of 40 working hours per week
3 Sliding Wear
of Continuous-Fibre-Reinforced Polymer Composites
The main characteristics of high-performance polymer composites are a continu-

ous (endless) fibre reinforcement, a high fibre content (up to 60 vol.%) and an
ideal fibre orientation. The limiting tribological loading conditions of high-
performance composites are primarily determined by the type of fibres and their
orientation relative to the sliding direction. During the sliding motion, asperities of
the hard, metallic part sliding against the composite interact continuously with the
various components of the composite (Fig. D4). The latter, in turn, possess various
mechanical properties, e.g. hardness, strength and toughness. This leads to varia-
tions in the wear mechanisms, which result not only in wear of the composite ma-
terial, but also in wear of the surface of the metallic part.
Normal (N) Parallel (P) Antlparallel (AP)
Fig. 04. Tribological system of metal asperity vs. composite with different fibre orientations
Hard counterpart Composite Solt counterpart
&
(I)
Steel
Pure elastIC or elastic-plastic

behaviour; }
Various.~hanisms b~ome [ V"oseo-elastic to viscous
behaviour; surface deformation.
aC1lVe In a compoSite
crack lormation and breaking out ploughing, cutting,
01 particles material take off by adhesion
WlirDepth
(b)
=f (Proper1I•• 01 Find M; Rt}
H
Composite
(N - Orlentltlon) t (M)
~ Criticil Crack Length
m----
~ f (Flbr. Proper1I,.)
[[1]{ !~;l) T
(4) (3)
Fig. OS. Principal mechanisms that are active between various wear couples (a) and possible
sequence of wear steps at the surface of a composite (b)
Figure DSa illustrates schematically the principal mechanisms which can be-
come active between various sliding partners, and Fig. DSb illustrates the possible
sequence of wear steps at a surface of a unidirectional fibre-reinforced polymer
composite in the normal (N) orientation. Usually, the softer polymer matrix be-
tween the much harder fibres is removed first, and then cracking of fibre pieces
occurs, which leads to a repetition of the wear cycle. Figure D6 shows an interme-
diate stage of this wear cycle for an example ofa polished CFIPEEK specimen.
Jlm
Fig. D6. Intennediate stage of a wear cycle for a polished CFIPEEK specimen. Increased
matrix removal (dark) can be seen between the protruding fibre ends (light). (hnage
recorded with an atomic force microscope)
Experimental studies on the interaction of various fibre-reitiforced polymer

composites with steel counter-surfaces have shown that, depending on the type of
fibre (AF = aramid, CF = carbon, GF = glass), the optimum fibre orientation rela-
tive to the sliding direction can vary greatly [19]. Aramid fibres, which interact
less abrasively with the steel surface [20], exhibit, owing to their tendency to fib-
rillation and their poorer bonding to the polymer matrix, a better performance in
the N orientation, whereas carbon fibres have proved to be most beneficial for the
wear resistance of composites in the P orientation (Fig. D7). The reason for the
higher wear rate of CF composites in the N orientation than in the P or AP orien-
tation is that the upper load limit (strength) of the fibres (with the same extemal-
load enhancement) is reached inside the composite earlier for the N orientation.
This was verified by a finite-element simulation of the stress conditions in a
unidirectional composite subjected to contact with a moving asperity for different
fibre orientations (Fig. D8) [21]. If these stresses locally exceed the strength of the
fibre, fibre fracture occurs as shown schematically in Fig. DSb, which, in tum,
causes temporarily an enhanced contribution of abrasive wear from fibre fragments
in the contact region. The way in which these load limits, which in the end also
determine the application limits of these materials as tribological components,
change under the influence of temperature (owing to external effects or as a result

of frictional heating) is presently under investigation [22,23]. In these modelling
studies, the formation of a transfer film which can reduce friction and protect the
underlying material (both determining the wear behaviour of a wear couple in the
steady-state range) is also considered.
, 1.0 .--- - -
,,
,:-------,,----- - - . . ,
E ,,
_ ____ __ ..J ____ _ _ _ _
, ,,
z , -- - --- - , ,,
iiE '" 0.8
c:
.2 ,,
r-- -- ---~-----
,
-- -
U
,
k! ,,,
- - - - - - _ 1- _____ _ _ _ 1_ __ _ ___ _
,,,
;:: 0.6
..... E ,
,, ,,
II: ~ 0.4 -~-------~--------
5:
~u ~ 0.2
0;::
U
X-
VI
0.0
N P AP N P AP
Tesl lng Conditions: v= 1 mlsec T = RT

p =1 MPa 1 = 24 hI'S
Fig. D7. Experimental measurements of sliding wear for CFIPEEK vs. steel for different
fibre orientations
a) Normal (N) b) Antlparallel (AP)
Stress (N/mm')
1000
875
750
625
500
375
250
125
. ::' 0
c) Parallel (P)
Fig. D8. FE simulation of the stress conditions in a unidirectional composite with different
fibre orientations. Clearly, higher stresses (characterized by the darker colour) occur
in the N orientation than in the AP or P orientation
4 Filament-Wound Thermoplastic Sliding Bearings
In addition to the more sophisticated, very complex and expensive 3d woven

structures, a simpler but quite successful winding technology has been used to
produce polymer composite journal bearings on a commercial basis [24,25]. The
cross-sections of these bearings are characterized by their functions. In the inner
contact region, the bearings are designed for low-friction and low-wear perform-
ance, whereas the outer structure is designed to be resistant to creep and plastic
deformation, so as to give the component a high load-bearing capacity. Therefore,
these bearings are found to have advantages in the area of high contact pressures
and low velocities, and under conditions of dry lubrication or in extremely corro-
sive environments (Fig. 09) [24].
1000
Thin Layer Bearings with
PTFE Liner
Sinterbronze 011 Filled
1
0.01 0.1 1 10
Sliding Velocity (m/s)
Fig. D9. p-v diagram of various bearing materials, showing the load-bearing capacity of fila-
ment-wound bearings at low speed
A new group of filament-wound bearings is under development at present,

namely those produced from continuous-fibre-reinforced, high-temperature ther-
moplastic matrix composites, which, according to scientific knowledge about the
tribological behaviour of these materials, is expected to lead to a further increase
in efficiency [19]. This expectation is based mainly on the higher ductility and the
better tribological properties of the thermoplastic matrix. On the other hand, the
higher viscosity of the melted thermoplastic material leads to the necessity for high
pressures and forces during the manufacturing process. This requires, however,
new technologies to be developed, in particular with regard to consolidation con-
trol, and the incorporation of sensor elements for preventing unexpected failure
(i.e. wear control). Additional developments being considered for the future are
complete wear-resistant units, produced by a combination of (a) filament winding
[26-28] of highly loadable bearing segments, and (b) the injection-moulding insert
technique, which allows one to form complex structures around these segments
(Fig. OlD) [29].
- -......~ .. ~T"
~~
Filament
Fibre Preform Winding
,
Ring with Fibres In
Hoop Direction
Stable Insert
Complex Part Geometry
Inj ection Moulding
Fig. 010. Schematic illustration of the injection-moulding insert technique for the manufac-
ture of complex, wear-resistant components
Figure DII shows two thennoplastic filament-wound thennoplastic bearings,

which both contain aramid fibres in a polyamide matrix as the basic material. The
bearing on the left possesses, in addition, PTFE fibres in the inner surface layer,
which guarantee an excellent dry-sliding behaviour. A cross-section of the sliding
layer of such a bearing is shown in Fig. D12.
Fig. 011. Thermoplastic filament-wound composite bearings

Fig. D12. Cross-section of the sliding layer of a composite bearing
The actual content ofPTFE fibres (black colour in Fig. D12) in this layer influ-
ences the resulting coefficient of friction (Fig. D 13). The latter amounts, on aver-
age, to a value of 0.1, which is also typical for commercial dry bearings. With
regard to the wear of these bearings, it can be stated that under the testing condi-
tions used, no significant wear rates could be measured, i.e. the upper load limits
of the bearing could not be determined with the bearing tests available in the study
[30].
0.20 r----------.----------,
:1. ,
,,,
C 0.16
o -- - ---------- -- ----------~----------- - -------------
ts
if 0.12
'0 ,
1:
CD 0.08 ,,
----- -- ------------------.---------------- -- ---
u ,,,
io 0.04
,
-------- ----- -- ----- -----,------- ------ ------------
o
0.00 '--_ _ _ _ _ _ _-'-_ _ _ _ _ _ _---J
o 7 14
PTFE Content (wt. %)
Fig. Dl3. Coefficient of friction of a filament-wound composite bearing as a function of the
content of PTFE fibres
5 Conclusions
Over the past few years, interest in polymers and polymer-based composites for
technological applications in which low friction and low wear must be provided
has grown rapidly. Short-fibre-reinforced high-temperature-resistant thermoplastic
materials can now be used in sliding elements where formerly only metallic mate-
rials could be used. Their applications are limited to temperatures up to about
220°C, pressures of more than 10 MPa and sliding velocities of about 3 mis, the
pressure and velocity limits depending on the temperature. Continuous-fibre-
reinforced, high-performance composites can endure even higher pressures de-
pending on the fibre orientation relative to the sliding direction. The load limits
which finally determine the limits of application can be verified by fmite-element
simulation.
A new manufacturing technique, a combination of the thermoplastic winding
process and the injection-moulding insert technique, allows one to fabricate sliding
elements even with complex geometries, such as sliding bearings, flange bearings
or sliding cams.
Acknowledgements
Parts of these studies were carried out within the project 'Tribology of Composite
Materials', supported by the German Research Foundation (DFG FR 675/19-2).
Thanks are due to the Stiftung Industrieforschung for supporting (a) Petra Rei-
nicke with a grant for scientific research on the theme 'tribological surface char-
acterization of rollers coated with thermoplastic composites' (Forschungsprakti-
kum T-41196) and (b) Jiirgen Hoffmann on the topic 'high-temperature-resistant
thermoplastics for applications in sliding bearings and sliding elements' (For-
schungspraktikum T-12/97).
References
1. Zum Gahr K. H. (1987): Microstructure and Wear of Materials. Elsevier Scientific, Amster-
dam
2. Uetz H., Wiedemeyer 1. (1985): Tribologie der Polymere. Carl Hanser, Munich, Vienna
3. Erhard G. (1980): Zum Reibungs- und VerschleiBverhalten von Polymerwerkstoffen. PhD
thesis, University of Karlsruhe, Germany
4. Bartenev G. M., Lavrentev V. V. (1981): Friction and Wear of Polymers. Elsevier Scientific,
Amsterdam
5. Szameitat M. (1997): Reibung und VerschleiB von Mischungen aus hochtemperatur-
besti\ndigen thermoplastischen Kunststoffen. PhD thesis, University of Erlangen-Niirnberg,
Germany
6. Hager A. M., Davies M. (1993): Short-fibre reinforced, high-temperature resistant polymers

for a wide field of tribological applications. In: Friedrich K. (Ed.) Advances in Composite
Tribology, Composite Materials Series, Vol. 8. Elsevier Scientific, Amsterdam: 107-157
7. Friedrich K. (1997): Wear performance of high temperature polymers and their composites.
In: Luise R. R. (Ed.) Application of High Temperature Polymers. CRC Press, Boca Raton,
Chap. 12: 221-246
8. Hager A. M. (1997): Polyaryletherketone fUr den Einsatz in Gleitlagem und Gleitelementen.
D 386, Shaker, Aachen
9. Anderson J. c., Robbins E. J. (1976): The influence of temperature generation on the wear
of some polymers. In: Dowson D. (Ed.) The Wear of Non-Metallic Materials. MEP, London,
UK: 94-98
10. Beringer H.-P., Heincke G., StrickIe E. (1991): Polymere im VerschleiBtest. Technische
Rundschau 25: 46-50
11. Friedrich K. (Ed.) (1986): Friction and Wear of Polymer Composites. Elsevier Scientific,
Amsterdam
12. Czichos H., Habig K.-H. (1992): Tribologie Handbuch. Vieweg, Braunschweig
13. VDI-Richtlinien (1975): Gleitlager aus thermoplastischen Kunststoffen. VDI 2541, VDI-
Gesellschaft Kunststofftechnik, Diisseldorf
14. DIN 31680 (1985): Gleitlager, Priifung des tribologischen Verhaltens von Kunststoffen
15. DIN 7148 (1985): Part I, Gleitlager, Priifung des tribologischen Verhaltens von Lagerwerk-
stoffen, Priifung des Reibungs- und VerschleiBverhaltens von Lagerwerkstoff-Gegenkorper-
werkstoff-Ol-Kombinationen unter Grenzreibungsbedingungen
16. Lu Z. (1994): Geschmierte Hochtemperatur-Verbundwerkstoffe fUr Anwendungen als Gleit-
elemente. DHS 527, Hansel-Hohenhausen, Egelsbach
17. Hombogen E. (1993): Werkstoffe. Springer, Berlin
18. Lu Z., Friedrich K. (1997): Polymere Hochtemperatur-Verbundwerkstoffe fUr Anwendungen
als Gleitelemente. Materialwiss. Werkstofftech. 28: 116-123
19. Friedrich K. (1993): Wear models for multiphase materials and synergistic effects in poly-
meric hybrid composites. In: Friedrich K. (Ed.) Advances in Composite Tribology. Elsevier
Scientific, Amsterdam: 209-273
20. Braches E. (1996): VerschleiBfeste Thermoplaste. Kunstst. - Germ. Plast. 86: 1720--1724
21. Varadi K., Neder Z., Friedrich K., Flock J. (1997): Finite element stress and thermal analy-
sis of a composite fibre-matrix micro-system subjected to sliding metal asperity contact. In:
Proceedings of EUROMAT '97, Maastricht, Netherlands, April 21-23, Vol. 2: Polymers
and Ceramics: 2/63-2/66.
22. Varadi K., Neder Z., FlOck J., Friedrich K. (1998): Numerical and experimental contact
analysis of a steel ball indented into a fibre reinforced polymer composite materials. J. Mat.
Sci. 33: 841-851.
23. Varadi K., Neder Z., Friedrich K., Flock J. (1998): Numerical and finite element contact
temperature analysis of real composite-steel surfaces in sliding contact. Tribol. Int. 31: 669-
686
24. SKF (1992): Gewickelte Buchsen, Baureihe PWM. Brochure, SKF-Gleitlager GmbH, Piit-
tlingen, Germany
25. Garlock (1992): Gar-Fil and Gar-Max, Filament Wound, High Load Self-Lubricating Bear-
ings. Catalog 881, Garlock Bearings, Inc., Coltec Industries, USA
26. Goedel M. (1992): Moglichkeiten und Grenzen des Wickelns mit Infraroterwlirmung -
Herstellung endlosfaserverstarkter Wickelbauteile mit thermoplastischer Matrix. PhD thesis,
Rheinisch-Westflllische Technische Hochschule Aachen, Germany
27. Funck R. (1997): Entwicklung innovativer Fertigungstechniken zur Verarbeitung kontinuier-
lich faserverstarkter Thermoplaste im Wickelverfahren. Fortschritt-Berichte, VDI Series 2:
Fertigungstechnik, VOl, DOsseldorf
28. Haupert F. (1997): Thermoplast-Wickeltechnik: EinfluB der Verarbeitungstechnologie auf
Struktur und Eigenschaften kontinuierlich faserverstarkter Verbundwerkstoffe. Fortschritt-
Berichte, VOl Series 2: Fertigungstechnik, VDI, DUsseldorf
29. Haupert F., Reinicke R., Friedrich K. (1996): Thermoplast-Wickeltechnik: Ein neues Fer-
tigungsverfahren fUr Hochleistungs-Verbundwerkstoff-Gleitlager. Kunststoffberater 10:
37-39
30. Reinicke P., Haupert F., Friedrich K. (1997): Herstellung von selbstschmierenden, war-
tungsfreien Gleitlagern aus thermoplastischen Verbundwerkstoffen. In: Deutsche Gesell-
schaft fUr Materialkunde e.V., Frankfurt, Tagung Verbundwerkstoffe und Werkstoffver-
bunde, Kaiserslautern, Germany, September 17-19: 349-354
0.2 Modification of Polymers
by Means of Amorphous Carbon
for Optimization of Tribological Properties
I. Hyla and J. Myalski, Katowice, and W. Grellmann,
Merseburg
1 Introduction
Heterophasic polymer composites are often used for brake linings. The nature and
quantities of the components of the composite have a significant influence on its
properties, including, in particular, the tribological properties. Asbestos, which has
traditionally been used as an additive to frictional materials, had to be removed
from these composites because of its negative influence on human health. As a
result, an urgent need arose to develop a new material with properties similar to or
even better than those of asbestos. In particular, the new material must ensure an
appropriately high level of the friction coefficient and its stability as the tempera-
ture rises during brake operation. At the same time, it is necessary to maintain the
mechanical characteristics of the brake lining material, particularly its crack resis-
tance. The resistance of the material to micro crack formation and propagation has
a decisive influence on the life of facings and thus also on the safety and economy
of use ofa device.
The properties of amorphous carbon indicate that it could be suitable to replace
asbestos in composite materials designated for brake linings [1]. In order to obtain
the amorphous carbon, various grades and kinds of precursors can be used. Phe-
nolic resins were used in the tests described here. Considering both material and
economic requirements, tests were performed with the aim of determining which
phenolic resin should be used for producing the amorphous carbon.
2 Selection of Phenol-Formaldehyde Resin for Tests
On the basis of the results of initial tests on various kinds of phenolic resins and
after analysis of the results, five kinds of resins were chosen for further tests. The
decisive factors in this selection were the size of the mass change of the resin dur-
ing the carbonization process, the physical and mechanical properties of the amor-
phous carbon obtained, and the price and availability of the resin in the Polish
market. According to these criteria, the following phenolic resins were chosen for
further tests:
1. the thermoplastic phenol-formaldehyde resin Nowolak 18 (NI8)

420 I. Byla, J. Myalski, W. Grellmann
2. some phenol A-stage resins (FIIO, FllOM, FP-31, B-1)

3. waste resin obtained in the process of dephenolization of waste materials.
Table D3. Selected properties of the resin precursors of amorphous carbon examined
Resin type N 18 FIlO FIlOM BI FP-31
Maximum content of free 6 15 12 6.5 0.05

phenol (%)
Minimum viscosity of 50 0/0- 100 120--150 120--140 90--140 300--400
solution in ethanol (mPa/s)
Minimum content of solid 73 65 74-80 45
substance (%)
1250
,-..
U
0
I6'
'-'
0
tl t2 t3 Time (h) ~ t5
Fig. D14. Scheme of the carbonization process of resin precursors
Some selected properties of these materials are presented in Table 03. The pa-
rameters of the carbonization process, plotted schematically in Fig. 014, were
fixed on the basis of derivatographic tests and gas chromatography [2,3].
The process of decomposition of the resins tested (excluding FP-31 and the
waste resins) features a one-step mass loss which is connected with the emission of
gases such as methane, carbon dioxide, nitrogen and other gases such as oxygen
and carbon monoxide. The order of emission of the gases and their amounts de-
pend on the type of resin. The temperature range in which the maximum mass loss
occurs is also specific to the type of resin. Table 04 shows the total values of the
mass loss during heating of the resins tested and the temperature range in which it
occurs. The mass losses of the FIlO, FIlOM and FP-31 resins are comparably
low, which allows one to anticipate the possibility of obtaining large quantities of
amorphous carbon. The use of other sorts of resins to obtain amorphous carbon
does not pay. This refers particularly to the waste resin, where out of 1 kg of resin
one obtains merely 0.1 kg of amorphous carbon.
Modification of Polymers by Means of Amorphous Carbon 421
Table D4. Temperature ranges of maximum mass loss of the resin precursors of amorphous carbon
Resin type Temperature range Maximum mass loss

CC) (wt.%)
FIIOM 353-816 32
FllO 416-750 38
FP-31 374-670 29
B-1 395-751 45
NI8 353-880 54
Waste resin 450-750 78
3 Tests of Frictional Properties
A material denoted as FO 70 I was used as a comparative material for the tests

carried out. This material is currently used for producing brake linings and does
not contain an additive of amorphous carbon [4].
00
-
"1
~I~~------------~~------------~
max. 3.5 I
97
Fig. D15. Shapes and dimensions (millimetres) of samples used for measuring the friction
coefficient (I, 2, 3 - points of temperature measurement)
Amorphous carbon obtained from the five resin types mentioned above was
used to modify this basic material; mass fractions in the range 2-10 vol. % were
used. The thermal stability of the friction coefficient of the composites obtained
was tested over a temperature range from 100°C to 350°C. The determination of
the friction coefficient was carried out according to the BN-81/3612-21 standard
with a special inertial apparatus in which braking was performed by a Bendix disc
brake. This device had a measuring system which allowed recording of the tem-
422 I. Hyla, J. Myalski, W. Grellmann
perature of the brake disc during the braking process, the braking moment and the
pressure in the hydraulic system of the brake. The shape and dimensions of the
samples used for testing are shown in Fig. DIS.
The results for the temperature dependence of the friction coefficient are shown
in Fig. D16 for the example of composites containing 10 vol.% amorphous carbon,
obtained from the five types of phenolic resin examined, in comparison with the
FO 701 basic material without amorphous carbon.
0.4r--------------------,
0.3 ....
------~
~-..-.
-
0.2······· "-~~~:;S . lIE_~~p.
---F0701 ~-F110 -A-FllOM I

1 -.-B-l -lIE-NI8 -.-FP-31
O.ll..---,-=================--.J
100 150 200 250 300 350
Temperature eC)
Fig. D16. Dependence of the friction coefficient J.I on temperature for the reference material
FO 701 and composites with 10 vo\.% amorphous carbon
For the FO 701 basic material the friction coefficient p shows a strong decrease
starting at a temperature of approximately 200°C and reaches its minimum at a
temperature of 250 °C. The addition of amorphous carbon allows one to obtain a
significant stabilization of the friction coefficient p at higher temperatures. How-
ever, the general level of the friction coefficient obtained is somewhat lower than
that obtained with the basic material. In this respect, the best effects were obtained
from the composite containing amorphous carbon from the resin FllO. On the
basis of these results, this material (FIIO) was chosen for studies of the influence
of the amorphous-carbon content on the frictional properties.
Figure D 17 shows the influence of amorphous-carbon content on the friction
coefficient for the FO 701 basic material modified with Fll0 amorphous carbon.
A variation of the amorphous-carbon content in the range 2-10 vol.% influences
only the absolute value of the friction coefficient p, not its thermal stability [5].
This means that the thermal stability of the friction coefficient can be changed very
little by the use of different volume contents of amorphous carbon. However, the
friction coefficient itself depends significantly on the amorphous-carbon content. If
the friction coefficient is to be maintained, the amorphous-carbon content must be
low. A modification with 2 vol.% is considered the most favourable since it guar-
antees both high values of the friction coefficient and very good thermal stability.
Owing to the fact that, besides thermal stability and frictional properties, state-
ments about the strength and toughness of brake linings are necessary, fracture
mechanics tests were carried out in addition to the analysis for optimization of the
carbon content described above.
0.4.--------------------,
::I.
= 0.3 ..
.~
]
u
.!·c 0.2 ..................... .
~
- . - FO 701 - - 2% -.A.- 4%
-..,- 6% - . - 8% -:11-10%
0.1 L-----'-_ _--'-_ _--'-_ _---'-_ _---'-_ _---L-----l
100 150 200 250 300 350
Temperature (0C)
Fig. D17. Temperature dependence of the friction coefficient f.l for the FO 701 material modi-
fied with different amounts of amorphous carbon from the Fll 0 precursor
4 Crack Resistance Tests
Taking into account the operating conditions of brake linings and the fact that a
hard and fragile additive such as amorphous carbon has been introduced into the
composite, toughness characterization seems to be advisable. For this purpose, the
instrumented Charpy impact test was used [6].
For composites containing variable fractions of amorphous carbon obtained
from the FIIO resin, parameters related to crack resistance against unstable crack
propagation were determined, i.e. the fracture toughness KId, the J value and '''-IT
the critical crack-opening displacement bid,
A Charpy impact tester with a maximum work capacity of 4 J [7] was used in
the tests, and the following test conditions were applied:
• striker mass: 0.955 kg
• support span: 40 mm
• test velocity: I mls
• temperature: 21°C
• sample geometry: thickness B = 4 mm, width W = 10 mm, length L = 80 mm
• notch depth: a = I mm (the notch was cut with a razor blade).
424 I. Byla, J. Myalski, W. Grellmann
Figure DIS presents some examples of load-deflection (F-f) diagrams obtained

during the tests of the composites modified with various grades of amorphous
carbon. The basic material FO 70 I shows linear elastic material behaviour (Fig.
D 1Sa). With the addition of amorphous carbon, the material behaviour changes to
elastic-plastic with components of crack propagation energy (Fig. D ISb).
300.--------------------.-.
a
F0701
200
g
] 100
...:l
o
0.0 0.3 0.6 0.9 1.2 1.5
Deflection (mm)
3 0 0 . . - - - - - - - - - - - r :b....,
B-1
200
g
~
.3 100
o~----~----~------~--~
0.0 0.3 0.6 0.9 1.2
Deflection (mm)
Fig. D18. Load-deflection diagrams of two different materials tested
The influence of the amorphous-carbon content on the maximum impact load

Fmax and on the maximum deflection/max is shown in Fig. D19. From the measured
load-deflection diagrams, the characteristic values mentioned above, namely the J-
integral Jir (Fig. D20), the fracture toughness KId and the critical crack-opening
displacement ~d (Fig. D21), were determined [S,9] and recorded as a function of
volume content. The determination of the critical J values followed the procedure
described in A 2.2 (pp. 71-S6). The plots presented in Figs. D20 and D21 indicate
that an increasing amorphous-carbon volume content in the range examined (2-
10 vol.%) does not cause significant changes in the quantities measured. This
means that the modification of a composite by means of amorphous carbon does
not cause a decrease in crack resistance against unstable crack growth, which
would influence the frictional behaviour of the material.
300 0.30
•
250
• 0.25
~ S
0.20 .§.
Ii
~
200
J
• fmax 0.15
• F....
150 0 100 .10

2 4 6 8
Carbon Content (vol.%)
Fig. D19. Influence of the FIIO amorphous-carbon content on the maximum impact load Fmax
and the deflection/max
1.8,...------------------,
1.6
•
•
1.2
2 4 6 8 10
Carbon Content (vol.%)
Fig. D20. JIdST versus amorphous-carbon content (type FllO)
80 80
• • 60
S8
i
70
8
oS
~ •
J 60
• KId 20
• I>ld
50 0
2 4 6 8 100
Carbon Content (vol. %)
Fig. D21. Dependence of fracture toughness KId and critical crack-opening displacement 41 on
amorphous-carbon content (type FIIO)
426 I. Hyla, J. Myalski, W. Grellmann
2.5
2.0
1.5
Ie-
1.0 ....
...,
"'''
0.5
0.0
....~ ~ ~ ~ -...
~ ~ <' ~
q$
80
60
I
'i'
40
~
20 '"'"
~ ~
, .;:: t~ ~ ~ -...
" ... ""i<: q$
Fig. D22. J values ,,,IT (a), cn[lcal cracK-openmg alsplacemem Old {DJ ana rracrure toughness
Kid (c) of composites with 10 vol.% amorphous carbon obtained from different
precursors
However, it should be pointed out that the tests did not cover an evaluation of
the material resistance against stable crack propagation. From tests on other sys-
tems, it is known (see A 1.1 in pp. 3-26 and [10]) that a toughness characterization
by means of the R-curve concept is necessary if crack propagation energies occur.
However, the characteristic parameters determined using the R-curve, namely the
crack initiation toughness JO.2 or ~, the tearing modulus TJ and the product JTJ
allow one to obtain more information and to detect any correlation [11,12].
Figure D22 shows a comparison of crack toughness parameters related to resis-
tance against unstable crack growth, i.e. J IdST, bid and KId, for composites with a
constant carbon content (10 vol.%) but different precursor materials. Differences
in the deformation-determined material behaviour for different precursor materials
can be seen. Furthermore, it is obvious that a description of the toughness behav-
iour with only the fracture toughness KId does not allow a characterization of the
material properties.
5 Summary
The tests carried out in the study described here showed that amorphous carbon is
useful for the modification of polymer-based frictional materials. Using amor-
phous carbon provides a possibility for keeping the friction coefficient p at a high
level at higher temperatures (> 200°C) without a decrease in the fracture mechan-
ics properties. The content of amorphous carbon influences the level of the friction
coefficient p and therefore this form of modification can also be used for that
purpose.
The type of resin used for obtaining the amorphous carbon has a significant in-
fluence on the frictional characteristics of the composite.
Fracture mechanics allows a multi-parametric description of the mechanical be-
haviour of the materials. The inclusion of experimental methods for determination
of crack toughness related to resistance against stable crack growth is advisable in
any further investigations.
References
l. Hyla I. (1994): Entwicklung Asbestfreier Bremsbelagwerkstoffe fUr die Autoindustrie aus

Phenolharzen. In: Proceedings of Polymerwerkstoffe '94, Merseburg, Germany, September
27-29:441-449
2. Hyla I., Myalski J., Sieziona J. (1994): Amorpher Kohlenstoff als Asbestersatz in Brems-
beUigen.lnzynieria Materialowa 3-4: 70-73
3. Polish Patent: P-166256 (1995)
4. Polish Patent: P-294492 (1996)
428 I. Byla, J. Myabld, W. Grellmann
5. Hyla I., Myalski J., Sieziona J. (1996): Thermische Stabilisierung von Reibungswerkstoffen
durch die Zugabe von amorphem Kohlenstoff. In: Proceedings of Nowe Materialy Weglowe,
Ustron, Poland: 54--61
In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Series, No.3. Institut
fUr Mechanik, Berlin, Chemnitz: 102-111
7. ISO 179 (2000): Plastics - Determination of Charpy Impact Properties. Part 1: Impact
Strength
8. Grellmann W., Che M. (1997): Assessment of temperature-dependent fracture behaviour
with different fracture mechanics concepts on example of unoriented and cold-rolled poly-
propylene. J. Appl. Polym. Sci. 66: 1237-1249
9. Grellmann W., Sommer J. P. (1985): Description of toughness properties of polymer materi-
als with the J-integral concept. In: Fracture Mechanics, Micromechanics and Coupled Fields
(FMC) Series, No. 17. Insitut fUr Mechanik, Berlin, Chemnitz: 48-72
10. Heinrich G., Kramer T., Grellmann W. (1996): Crack initiation, wear and molecular struc-
ture of filled vulcanized materials. In: Proceedings of Kautschuk-Herbstkolloquium '96,
Hannover, Germany, October 24-26: B 1-8
11. Grellmann W., Seidler S. (1992): J-integral-analysis of fibre reinforced thermoplastics. J.
Polym. Engng. Sci. 11: 391-401
0.3 Mechanical Vibration Behaviour
of a Compressor Blade
Made from a High-Performance Composite
W. Grellmann, R. Steiner and 1. Kotter, Merseburg, and
M. Neitzel, M. Maier and K. von Diest, Kaiserslautem
1 Overview
Some components of rotating machines, for example axial-compressor blades, are

manufactured from composite materials instead of steel to reduce the weight, with
the aim of increasing the throughput (larger blades or a higher revolution rate).
Consideration of complex aspects of the transfer of forces into the composite and
the mechanical vibration behaviour is necessary before substituting steel by a high-
performance composite [I].
A compressor blade, as a rapidly rotating component, is loaded in a complex
way. Because of the high rate of revolution, the number of loading cycles during
the lifetime is very high. This can lead to component fatigue even if the amplitude
of the stress or strain is relatively low. The main mechanisms of damage are the
propagation of fatigue cracks between the different components of the heterogene-
ous blade and material destruction. However, the determination of the dynamic
loading is complicated. The difficulty lies particularly in the excitation of the
loading. Because of aerodynamic interaction during the use of these blades, peri-
odic and impact loads occur. These loads are disturbances of the vibrational sys-
tem 'compressor blade' [2], which lead to stationary and non-stationary vibrations
because of the stiffuess and damping of the blade. These vibrations may cause the
stresses and strains to exceed critical amplitudes locally.
It was not possible to carry out a component test under real conditions, because
the research project described here did not include prototype production. There-
fore, no compressor with a complete ring of compressor blades manufactured from
a composite material was available for investigation. Furthermore, the scope of
investigations under real conditions in a turbine is severely restricted.
As a result, the investigations were confmed to testing blades on a whirling test
stand. In each test, two prototypes of the composite compressor blade were tested
under real-load conditions (3000 rpm) and under maximum-load conditions (3600
rpm) in vacuum on the whirling test stand of the ABB company. The rotor diameter
of the whirling test stand was smaller than the diameter of the axial compressor.
Therefore, the rate of revolution was scaled; the rates used were 5700 rpm (real-load
condition) and 6769 rpm (maximum-load condition). During such a test, the rate of
revolution and the vibrations occurring during the test can be measured. The first
whirling test was interrupted because of very strong vibrations at 4639 rpm. Here the
fIrst cases of delamination occurred and, consequently, the torsional stiffuess de-
creased rapidly.

430 W. Grellmann, R. Steiner, I. Kotter, M. Neitzel, M. Maier, K von Diest
The vibrations could be prevented by small changes of the construction of the

compressor blade. By this means the real load and the maximum load could be
reached without any problems.
By simulation with the fmite-element method (FEM), the natural vibration fre-
quencies and modes can be calculated for the distorted blade. These natural vibra-
tion frequencies and modes do not show the complex vibration behaviour of the
compressor blades in operation, but they are a useful device for constructing and
dimensioning such components. In this way, failure at the resonant frequency can
be obviated. The calculated natural vibration frequencies and modes are the basis
for a simulation of the loading and stresses in operation. By this means it is possi-
ble to estimate the stresses in the blade and to make a prediction of the fatigue
loading. The calculated values can be used to optimize the design of the blade
(stiffness and damping) and the recommendations for its use.
The results of the FEM simulation were verified by experimental measurements
of the vibrations and modal analyses of a compressor blade made from steel and
another blade made from a high-performance composite material.
2 Design of the Axial-Compressor Blade
The axial compressor blade investigated here was a sandwich construction of fibre
laminates. It consisted of a lightweight core, epoxy-resin-pre-impregnated layers of
unidirectional carbon fibres (UD prepregs) and a shell of woven carbon fibre fila-
ments oriented at ±45°. FEM calculations showed high-pressure loads in the core
of the blade root. The loading in the core of the blade itself was low. Hence the
core consisted of two different materials: the root consisted of a glass-sphere-filled
epoxy resin and the blade consisted of an epoxy resin foam manufactured by cast-
ing or reaction injection moulding [1]. At the rootlblade transition the two materi-
als were shafted into one another. Because of the lightweight core, the overall
weight of the axial compressor blade is decreased, and therefore the centrifugal
force is also decreased and the damping properties can be improved.
The blade was manufactured from inside to outside in three steps (Fig. D23):
• Manufacture of the lightweight core with grooves for the unidirectional pre-
pregs. The core was manufactured by casting or reaction injection moulding
because of its complex design so as to realize it rapidly and economically.
• The unidirectional layers were conventionally laminated on the core and hard-
ened in an autoclave. They had a constant width and a thickness decreasing step
by step from root to tip.
• The shell of woven carbon fibre filament oriented at ±45° was intended to ab-
sorb the imposed torsional loads. These torsional loads occur because of the
tendency of the twisted blade to untwist during rotation. A woven hose was
pulled over the blade and drenched inside a negative mould by resin transfer
moulding (RTM). A good surface precision and structural stiffness could be
achieved with this thin compact coat at the leading and trailing edges of this
blade profile.
Mechanical Vibration Behaviour 431
The wedge-shaped root of the blade was designed to fit the contours of the core.
The various steps of the process could be controlled and partly automated in a
satisfactory manner.
~
Woven CF A.... tal Shell CF.UD !..yen
3 Numerical Calculations of the Natural Frequencies

and Modes by Application of the Finite-Element Method
To calculate the vibration behaviour, a parametric fmite-element model of the blade

was designed. The strength of the blade could be calculated by the same model
after a small modification. The different components of the blade and their layer
thicknesses are variable in this parametric model. The model consists of volume
elements which permit the input of anisotropic material parameters.
The blade was statically clamped in the region of the blade root (no rotation).
The vibration behaviour was simulated in an undamped manner, although this is
hardly close to reality. The natural vibration frequencies of the rotating blade were
calculated with estimated aerodynamic loads (a simple simulation of the working
load). The static pre-loading caused by the centrifugal load leads to an increase of
the various natural-vibration loads. The first eight natural vibration modes are
shown in Fig. D24.
The third mode (see Table D5) represents the torsional vibration. Because of its
low damping, this torsional vibration causes a very strong loading of the blade.
During the whirling test the damage could be qualitatively shown to be an inter-
laminar shear fracture between the UD layer and the shell of woven carbon fibre
filaments. This damage was an effect of this vibration mode. The failure occurred
at 4639 rpm in the first whirling test, which agrees with the FEM calculations. This
rate of revolution corresponds approximately to the resonance frequency: 156 rpm
"" 4639 x 2/60 = 154.6.
432 W. Grellmann, R. Steiner, I. Kotter, M. Neitzel, M. Maier, K. von Diest
Mode 1 Mode 2 Mode 3 Mode 4
Fig. D24. Natural vibration modes of the blade: mode 1 to mode 8
Table D5. Natural vibration frequencies of simulated blades
Mode Steel blade, Composite Composite Whirling test Whirling test Whirling test
clamped blade, blade,
(reference) clamped real load 4639 rpm 5700 rpm 6769 rpm
(damage) (real load) (max. load)
(Hz) (Hz) (Hz) (Hz) (Hz) (Hz)
1 101 101 129 156 166 178

2 337 303 336 411 419 429
3 465 370 377 447 457 471
4 755 558 575 742 751 763
5 932 771 799 938 947 958
6 1021 848 872 1066 1081 1101
7 1562 1240 1256 1375 1382 1391
8 1655 1357 1375 1564 1575 1589
4 Experimental Modal Analysis
4.1 Fundamentals
The fmite-element model was verified for the real blade by application of experi-
mental modal analysis. The experimental modal analysis was carried out on a
compressor blade made from steel (from ABB) and a compressor blade made from
a carbon fibre laminate (manufactured at the Institute of Composite Materials,
Kaiserslautern). A Bruel & Kjrer 3550 analysirig system was used, extended with
components to stimulate and measure the vibrations. This system is a real-time
Fourier analyser with two inputs and a programmable function generator. In this
way, it is able to carry out a dual-channel Fourier analysis (the minimum require-
ment for carrying out a modal analysis).
Two shells cast from polymer concrete were used to clamp the carbon fibre
laminated blade at the root during the vibration tests. In general, the investigation
of an unclamped blade is more favourable regarding the measuring technology, but
in this case that would be far from reality. Furthermore, it is difficult to handle the
fir tree root of the steel blade in the unclamped case. The fir tree root was not
implemented in the fmite-element mesh. The compressor blade, clamped in a
heavy table vice, was bedded in a soft, elastic material (the natural vibration fre-
quency of this system was roughly 6 Hz). Thus, the transfer of external vibrations
could be suppressed.
A 4 x 6 grid of measuring points was marked on the blade. The transmission
behaviour was measured from four starting points (at the upper edge of the blade)
to each of the other points and to the starting point (the so-called excitation point)
itself. Thus, there were 96 complex functions. These functions are called the 'fre-
quency response functions' and describe the ratio of the cross-spectrum to the
autospectrum in the frequency domain (see Fig. D25). Here, the auto spectrum is
the product of a measured complex signal and its complex conjugate. The cross-
spectrum is the product of one channel and the complex conjugate of another
channel [3]. A double logarithmic description of the reciprocal frequency response
functions (Fig. D26) shows significant resonances at various frequencies. This
description is sufficient to obtain a rough approximation to the frequencies where
resonances occur (frequencies of the damped vibration).
The frequency response function at the excitation point differs fundamentally
from the other frequency response functions (see Fig. D27). It characterizes the
damping behaviour.
The application of software for modal analysis supplied by LMSTM allowed the
calculation of a multi-channel modal analysis from these 96 complex functions.
Coupled vibrations occur in the twisted compressor blade. In this case a dual-
channel analysis is insufficient because it is unable to solve complex conjugate
modes [4].
dB
1\ f\ A 1\ /\ ~ .-..
4
\~ 7V "\f\:7 \~
.....
7'v-.J II
0,/
2 p
o
-20
o 0.5 1.0 1.5 2.0 2.5 3.0 kHz

Fig. D25. Typical frequency response function
dB
40
A First Resonance
20
I~-"t /
o / II
\(~ ~ \
-20
'f' \ I hI \
'v
~
V \/ "\ /\n ~
0 100 200 300 400 500 600 700 Hz
Fig. D26. Inverse frequency response function
100 Excitation Point

Real
1/kg 0
-100
200 No Excitation Point

Real
1/kg 0
-200
0 200 400 600 Hz
Fig. D27. Real part of the frequency response function

Imag.
1/kg
o
o 100
Real 1/kg
Fig. D28. Curve fitting in the Nyquist plane
If the complex frequency response function is plotted in the Nyquist plane (a

specific complex plane, Fig. D28), locus curves are obtained as the frequency is
varied, which encircle poles. Each circle corresponds to a resonance, for which
there is a frequency, damping and residue. By application of the program, locus
curves were fitted for all measured frequency response functions. At each point of
the structure the frequency and the damping are equal to those at other points but
the residue is different. A large residue means a high pole strength at this point
(this is related to the amplitude of the vibration).
Using the calculated natural vibration frequencies and modes, it was possible to
simulate each of the frequency response functions determined by the experiment.
Figure D29 shows a good agreement between the measured curve and the simu-
lated frequency response function.
Real 100
1/kg
0
-100
Imag.
0
1/kg
-100
-200
0 200 400 600 Hz

Fig. D29. Comparison of the experimental and simulated complex frequency response
Relative 1.6
Vibration
Amplitude
o 800 1600 2400 3200 4000 Hz
Fig. D30. Finite-element calculation of the vibration amplitude as a function of frequency
The calculation of these functions was possible by the FEM too. Thus, the
simulation of the working load can be simplified greatly. An example is shown in
Fig. D30 for one excitation point only.
4.2 Results of the Experimental Modal Analysis
The natural vibration modes shown in Fig. D31 were determined by application of
the multi-channel modal analysis. The natural vibration frequencies and damping
are qualitatively comparable to the corresponding FEM results (Table D6).
A quantitative comparison shows that each resonance frequency of the real
blade is lower than expected, especially for the higher modes. A decrease of the
frequencies occurs at resonances of the amplitudes, but this should only affect the
strongly damped first mode. It can be supposed that the material properties used in
the FEM calculations were overestimated. Furthermore, the various structures of
the blade have different influences on different modes. Thus, it is important to
model the blade geometry exactly (a precise surface stiffness is necessary). The
differences which occur between experiment and simulation show the importance
of an experimental verification of the simulated FEM results.
An adjustment of the simulation to the experiment is possible if suitable pa-
rameters are modified within a range of tolerance.
The experimentally determined natural vibration frequencies of the CFRP blade
are not much lower than those of the reference blade made from steel. The strong
third mode (torsional vibration) has little damping in both cases, but the composite
blade has five times the damping of the steel blade (Table D6).
Fig. D31. Experimentally determined natural vibration modes of the blade
Thus, this vibration mode is the main cause of dynamic loading of the structure.
The errors in the blade model are probably the main cause of the difference be-
tween the measured and calculated natural vibration frequencies.
Table »6. Comparison of the modal analyses for the two compressor blades
Composite blade Steel blade
Clamped Experimental FEMmodal Experimental FEMmodal

root modal analysis analysis modal analysis analysis
Mode Frequency Damping Frequency Frequency Damping Frequency

(Hz) (Hz) (Hz) (Hz) (Hz) (Hz)
I 90 4.7 101 109 5.2 101

2 212 4.8 303 238 5.0 337
3 334 1.5 370 396 0.3 465
4 429 2.6 558 475 2.6 755
5 648 2.1 771 519 1.1 932
5 Summary
The aim of this investigation was the determination of the vibration behaviour of
an axial-compressor blade designed with a carbon fibre sandwich construction.
The numerical results for the natural vibration frequencies and modes obtained
by application of the finite-element method could be verified by experimental
modal analysis.
Thus, a basis for a theoretical estimation of the behaviour of a material and

component under dynamic loading and of the fatigue behaviour has been given.
The results may be useful for prediction of the load limit and lifetime.
Acknowledgements
The authors gratefully acknowledge the fmancial support of the Arbeitsgemein-

schaft Industrieller Forschungsvereinigungen 'Otto von Guericke' e.V. (AIF)lFor-
schungskuratorium Maschinenbau e.V. We, especially Prof. Neitzel and Prof.
Maier, acknowledge the Institute of Composite Materials and the creative coop-
eration between the University of Kaiserslautem and the Martin-Luther-University
of Halle-Wittenberg. Furthermore, we thank the companies ABB, Mannheim,
Germany and ABB, Baden, Switzerland for their technical support and for the
carrying out of the whirling tests.
References
I. Maier M., Grellmann W., Leps G., Pfefferkorn W., Steiner R., Kotter I. (1995): Faser-
verstllrkte rotierende Bauteile. Forschungshefte FKM (Forschungskuratorium Maschinen-
bau), Part 202: 1-67
2. Farkas F. (1986): The development of a multi-stage heavy-duty transonic compressor for
industrial gas turbines. In: International Gas Turbine Conference and Exhibition, Dusseldorf,
8-12.6
3. Briiel & Kjrer (1987): Frequency Analysis. Larson & Son, Denmark: 227-265
4. LMS International (1992): LMS-CADA-PC User Manual. Eigenverlag: 25
E
Biocompatible Materials
and Medical Prostheses
E.1 Polymer Materials in Joint Surgery
J. Brandt and W. Hein, Halle
1 Introduction
The artificial replacement of joints opened a new era in the treatment of degenera-
tive diseases. It became possible to relieve patients from pain and to give them
back nearly normal mobility. Today more than 500000 prostheses are implanted
in the hip joint alone, all over the world every year. With the development of joint
replacement during the past 40 years, much research has been done to prolong the
survival of joint prostheses and to make them reliable for the rest of the patient's
life. Polymer materials have continued this development; they have become an
important constituent part of most prosthesis systems and have prolonged their
survival to a range of 12-15 years today. On the other hand, polymers seem to
contribute to an aseptic loosening of prostheses after this period, so that they need
to be improved further.
2 Development of Joint Replacement
With the development of orthopaedic surgery, a number of trials have been un-
dertaken to replace joints destroyed by degenerative diseases since the end of the
19th century. The large joints of the lower extremities, the hip and knee joint es-
pecially, undergo severe dynamic loading: 1-2 million load cycles per year with
loads of three to eight times the body weight occur in normal everyday motion [1].
The first trials of the replacement of destroyed joints by internal prostheses
were undertaken by the Berlin surgeon Gluck in 1890. He implanted a knee pros-
thesis made of ebony and glued in with putty made from plaster, colophonium and
pumice. Later, Smith-Peterson used a simple glass cup between the femoral head
and the acetabulum to restore the sliding surface. However, he was also not able to
achieve long-time recovery. Only at the end of the 1930s could the survival of
joint replacements be prolonged, when Wiles first used steel as a material for hip
joint prostheses.
With the rapid development of polymer chemistry since the beginning of the
20th century, an increasing amount of new materials has become available with
properties completely different from metals. In addition, polymers revealed a very
good biological compatibility in their first medical applications, e.g. the covering
of bone defects of the skull. In 1946 the French brothers Judet first tried to replace
a destroyed femoral head with a polymer material. They used a polymer that was
already very well known at that time, poly(methyl methacrylate) (plexiglas~).
Their first prosthesis was simply shaped like a hemisphere, with a stem for inser-
tion into the femoral neck through a central drill hole. Initially, the Judet prosthe-

442 J. Brandt, W. Hein
sis was an immediate success: the patients could walk soon after the surgical pro-
cedure and had a nearly completely restored joint mobility. However, after the
first, overwhelming effects, doubts soon arose as long-time results revealed a very
high incidence of loosening, breakage and, in many cases, severe chronic inflam-
matory reaction due to a massive amount of polymer wear in a few years. More
than ten years later, in 1958, a major step forward towards a joint prosthesis with
long-time survival was made by the British surgeon John Charnley: the rapid
failure of all former implants led to his concept of 'low-friction arthroplasty', a
principle that had been known for a long time in bearing technology - the superior
behaviour of sliding partners with different hardnesses. First, he chose Teflon, and
later the more wear-resistant PE-HD (high-density polyethylene) as the sliding
counterpart for his steel prostheses. From the middle of the 1960s Charnley addi-
tionally used self-curing polymer resins of poly(methyl methacrylate) to anchor
his joint prosthesis in the surrounding bone; later the resin was called 'bone ce-
ment'. This procedure had already been inaugurated by Harboush in 1951 to an-
chor femoral prostheses in the thigh bone. He took the material poly(methyl meth-
acrylate) over from stomatology, where it was already commonly used to fill
dental cavities. Until now, Charnley's 'low-friction arthroplasty' in combination
with the use of bone cement has been the most commonly used technique for joint
replacement. In the beginning of the 1980s the material PE-HD was substituted by
ultra-high-molecular-weight polyethylene (UHMWPE), a polyethylene with ex-
tremely long polymer chains and a molecular weight of 4-5 million glmol. The
use ofUHMWPE together with numerous variations in the design of joint prosthe-
ses prolonged the survival of artificial hip joint replacements to 10-12 years on
average. Sliding combinations of smoothly polished metal or ceramic surfaces
with UHMWPE have, consequently, become the material combination favoured in
clinical use nowadays. In comparison with all other polymers used before (poly-
amide, poly(ethylene terephthalate), polyacetal), UHMWPE unquestionably pos-
sesses the best mechanical properties: besides the low adhesive friction and resul-
tant low starting forces, it shows a high wear resistance owing to its chain length.
Other outstanding merits, such as high fatigue strength and high toughness, result
in the long survival ofUHMWPE implants in sliding components. The chemically
inert aliphatic chains and the electrically neutral polymer surface make the mate-
rial biocompatible and prevent adverse reactions of surrounding tissue; body cells
are barely able to attach themselves to polyethylene surfaces.
With the increasing reliability and survival of the implants, the number of re-
placed joints increased. Encouraged by very good clinical results, the time for first
implantation shifted to younger and younger patients. The only alternative to arti-
ficial joint replacement in cases of severe osteoarthritis or of joints destroyed by
other diseases is a stiffening of the joint (arthrodesis) - that means a total loss of
its function. As a result of these facts, the number of first joint replacements has
grown to 500000 per year for hip joints alone. With the increasing use of prosthe-
ses, the number of failures because of wear and loosening increased, of course,
and deficiencies that diminished the long-time survival became obvious after a
time.
Polymer Materials in Joint Surgery 443
3 Problems of Polymer Joint Replacement
With the great improvement in construction and material properties of the ceramic
and metal components of artificial joints in the past 10 to 15 years, the UHMWPE
has become the weakest part in the implant systems, despite its numerous advan-
tages and the unchanged or even improved quality of the material itself. Nowa-
days the continuous wear of UHMWPE is regarded as the main reason for aseptic
implant loosening and the crucial factor limiting the long-time survival [2-5]. In
studying the threats presented to artificial joint prostheses by PE, we must take
into account two independent aspects: on one hand, the failure behaviour of PE,
and on the other, the reaction of the biological host tissue to the material itself and
to abrasive wear and eluted substances that arise during long-time loading. The
material may provoke foreign-body reactions of varying degrees in the surround-
ing biological tissue depending on the shape, amount and chemical nature of the
degradation products.
Therefore, we should look at the development of failure in the polyethylene
used in joint replacement from the viewpoint of materials science: UHMWPE
shows very little plastic deformation (creep) that changes the geometry of the
polymer components, but nevertheless it is important after long periods of pros-
thesis life. Under the enormous pressure of the harder sliding counterpart (metal or
ceramic), UHMWPE shows plastic deformation in the main loading zones, which
are mostly very small. The UHMWPE material quickly reaches critical limits
because implant components can have only a low wall thickness (6-10 rom); ad-
ditionally, this thickness is continuously diminished by abrasion due to wear, as
will be explained in more detail later on.
The most important problem in connection with failure and loosening of endo-
prostheses is the abrasive wear of the polymer generated even under normal load-
ing conditions [4,6]. In studies with joint balls of 22 mm head diameter as nor-
mally used today, Clarke et al. [7] found a mean wear rate of23.2 rom3/10 6 load
cycles (equal to one year of normal walking); with balls of 28 rom diameter they
found a mean wear rate of32.8 rom3/10 6 load cycles. The abrasive particles had a
mean diameter of 0.3 !lm. In comparing these results with the wear calculated
from clinical application conditions, an unexpected difference became evident: in
retrievals from hip cups Schmalzried et al. [8] calculated a mean wear rate of
393 rom3 per year, and Semlitsch and Willert [9] reported on linear penetration
rates of steel balls into the cup between 20 and 500 !lmlyear from retrieved pros-
theses. In addition to these results, one should take into account that UHMWPE
has relatively low hardness. In the presence of abrasive, harder particles (bone
cement, bone fragments, metal wear) between the sliding surfaces, the abrasion
rate grows like an avalanche and may quickly lead to implant failure. Figure El
shows a hip cup retrieval destroyed by abrasion wear after an implantation period
of eight years.
The combination of chemical and mechanical loading leads to a time- and load-
dependent decrease in average molecular weight, to crystallization phenomena and
to an increase in material density. The increase in density is larger in unloaded
parts of the implant, presumably owing to free wetting with synovial fluid contain-
444 J. Brandt, W. Rein
Fig. El. Hip cup retrieval ofUHMWPE, eight years after implantation
ing hydrolytic enzymes. The increase in density, especially in the surface layer,
leads to increased brittleness, i.e. higher Young's moduli, up to the double the origi-
nal value after ten years' load [2,10,11]. Corrosion by the biological surroundings
plays an important role: various body-specific enzymes catalyse oxidation reac-
tions at polymer chains that lead to hydrolytic changes, embrittlement, breakage of
chains and finally to a diminished survival time. Quantitative measurements of
surface damage and wear debris revealed that their amount increases with time of
survival and with the body weight of the patient. The extent of surface damage
depends not only on the material itself but also on its processing, the conditions of
the polymerization process (catalyst-dependent molecular-weight distribution,
pressure, temperature), the method and parameters of sterilization and the degen-
eration in vivo. Common sterilization procedures of polymer implants with
gamma radiation (2.5-3.5 Mrad) induce physical and chemical changes in the
interior of the material such as polymer chain scission, crosslinking and generation
of free radicals. The radicals remain after irradiation, form peroxide radicals and
contribute, by oxidative reactions and crosslinking of chains, to the degradation
process. Free radicals remain trapped in some places, mainly in the crystalline
regions, and after their migration to amorphous regions cause further crosslinking,
oxidation and embrittlement [12,13]. Ambrosio et al. [14] showed by IR spectros-
copy in retrieved hip cups after survival of up to nine years that the extent of the
oxidation process depends on the local loading: with a gauge that allows focusing
on very small areas down to 200 x 200 J.lm, they found reaction products typical
of oxidative processes (keto carbonyl groups, aldehyde carbonyl groups), indicat-
ing an increased amount of chain ends in the sliding surface and in the adjacent
layers. It is of special interest that such indications of degradation were found in
exactly the same zones that appear in the microscope as darkish, discoloured spots
of increased abrasion, so that Ambrosio et al. [14] presumed them to be spot-
shaped cores ofbiocorrosion.
The second main reason for implant loosening relates to the reactions of bio-
logical tissue to wear. Particles of PE are not resorbable, because of their inert
chemical nature. According to the very wide range of sizes of PE wear particles,
from 0.5 to 200 /lm, the ways of their cellular elimination vary: small particles can
be enclosed in single-nucleated phagocytes (macrophages), be carried away from
the location of their origin and migrate along the lymphatic vessels or in the adja-
cent connective tissue. Cells loaded with abrasion particles can be found even far
from the location of their generation by wear, especially in the regional lymph
nodes, where they may provoke an inflammatory reaction in cortical zones [15] . If
the particles are too large for a single cell, multi-nucleated phagocytes (giant cells)
are formed; these enclose the debris in their interior and settle down in the adja-
cent tissue of the interface. In the connective tissue next to a loosened joint pros-
thesis, huge amounts of particles, 108_10 11 particles per gram of tissue, could be
detected [16]. Figures E2 and E3 show our own fmdings using microscopy eight
years after implantation of a hip replacement: in the connective tissue of the bone-
prosthesis interface, giant cells which enclose foreign material are visible. Wear
particles of PE are conspicuous in polarization microscopy because of their bire-
fringent properties.
Fig. E2. Multi-nucleated giant cells contain- Fig. E3. Multi-nucleated giant cells contain-
ing particles of PE (arrow); haema- ing particles of PE (arrow) in polar-
toxylinleosin-stained ized light; unstained tissue
Cells containing wear particles liberate mediator substances which not only ac-
tivate other inflammatory cells but also concurrently cause the resorption of sur-
rounding bone tissue [17,18]. In the middle of the 1970s systematic histological
investigations of the resorption zones typical of long-time loosening regularly
revealed PE particles in the bone around implants in consequence, the hypothesis
was suggested that there may exist a balance between the wear debris generated
and the capacity of the surrounding host tissue to eliminate it by cellular scavenging
mechanisms [19-21]. If the amount of debris exceeds this capacity, it is deposited
in the joint capsule and the tissue adjacent to the interface; the balance is disturbed
and the implant will be disintegrated. Later investigations suggested that the inci-
dence of loosening is of a more complex nature. Howie et al. [22] described an
experiment in which PMMA plugs were implanted in unloaded locations of rat
thigh bones outside the joint. After repeated injection of PE particles into the knee
joint cavity, resorption zones arose around the plugs. Howie et al. [22] concluded
that osteoclast-stimulating factors were released by stimulation of macrophages and
giant cells. These fmdings were confirmed by the results of Murray and Rushton
[23] and those of Hennan et al. [24], who even found activation of macrophages
by large implants regarded as biologically inert. Santavirta et al. [25-27] con-
cluded, from investigations of connective tissue membranes adjacent to loosened
prostheses, that fonnation of foreign-body granulomas is a specific feature of the
loosening process. These theories were supported by numerous later results that
immunological reactions participate in the genesis of the particle-induced long-
time loosening. Fmkjrer et al. [2S] showed, by immunohistochemical methods, the
liberation of the monocyte-activating factor and the mediator substance interleu-
kin-S. Takagi et al. [29] detected in the extracellular matrix the activation of vari-
ous metalloproteinase subspecies which regulate the resorption of collagen and the
resorption of bone. Nordsletten et al. [30] discovered that macrophages containing
wear particles induce activators of the plasminogen system and probably maintain
the stimulus for the activation of the proteolytic cascade in this way. Goodman et
al. [31] found, in animal experiments, even an inhibition of new bone fonnation in
the presence ofPE particles in suspensions.
The cells containing wear debris migrate in the tissue border along the implant,
fonn islands of connective tissue which enlarge and in this way finally lead to the
loosening of the implant. In cases of loosened hip joint prostheses the PE wear
generated at the surface of the cup could be found deep in the thigh bone and
probably contributed in an important way to the loosening process of the stem
[3,S]. After examination of autopsy preparations from asymptomatic hip joint
prostheses, Fornasier et al. [32] even suggested a direct transcellular transport
mechanism in the direction of the zone with the highest bone resorption, which
they called 'directed exocytosis'. In contrast to these theories, Aspenberg and
Herbertsson [33] suggested that the wear debris was not able to cause loosening in
the case of strongly integrated implants, e.g. if implant micromotions have not
generated a gap in the interface.
4 Developments and Prospects

With the recognition of the problems of UHMWPE in joint replacement, some
attempts were started to improve its properties by modification of the manufac-
turing and processing parameters. Attempts were made to fmd alternative poly-
mers that would be suitable for a sliding component in artificial joint replacement.
Owing to its high melt viscosity, UHMWPE is usually processed by a melt-
sintering procedure, beginning at a temperature of 200°C under a pressure of 10-
20 MPa. At these pressures grain boundaries remain, which act as notches and can
become the origin of crack fonnation. Similar damage can result from remnants of
the supporting agents of the polymer processing (release agents, calcium stearate).
By using an experimental device, Kurth et al. [10] produced UHMWPE which
showed obviously improved properties after consecutive steps of very-high-
pressure sintering at pressures of 300 MPa-l GPa: the crystallinity and density
increased, the Young's modulus was higher and, as a result, the pressure creep
behaviour was also improved. Table El presents an overview of the most impor-
tant properties of these materials.
Table El. Properties ofUHMWPE after processing at different sintering pressures [10]
Sintering pressure IOMPa 300 MPa-1 GPa
Density (glcm 3) up to 0.935 0.965--0.980

Crystallinity (%) 50-80 80-90
Melt temperature COC) 135-138 145-155
Young's modulus (N/mm2) 900-1200 to 1800
However, the improvement of the mechanical properties is limited, especially

after gamma radiation, by an increased abrasion rate; this was higher than after
processing at normal sintering pressures, despite a reduced sliding coefficient.
An improvement of the properties could also be achieved by means of gamma
radiation, but the radiation dose and the atmosphere during irradiation are of cru-
cial importance. Only if the oxygen contained in normal air is excluded does the
generation of free radicals lead to crosslinking reactions, especially in amorphous
regions of the polymer [34,35]. The mechanical properties are changed as a result
of an increase of crystallinity and density; the extent of the change depends on the
radiation dose, the thickness of the material and its molecular weight. UHMWPE
with a higher molecular weight reacts with a clearly stronger increase in density
[12,13,36]. More recent investigations reveal that defined crosslinking by means
of gamma radiation can also increase the wear resistance of UHMWPE; however,
it decreases the toughness as well [37]. An alternative method, to overcome these
disadvantages, could be crosslinking by means of an electron beam. Streicher [38]
reported on an enhancement of mechanical properties such as the creep resistance
after irradiation-induced crosslinking, but he pointed out that a loss in toughness
and reduced cyclic bending resistance also result. To reduce post-oxidation effects
due to free radicals, the material should be annealed or stored in a nitrogen atmos-
phere after irradiation at least for three months. The first clinical and hip-joint
simulator results confirm the improved properties of crosslinked polyethylene:
Wroblewski et al. [39] found, in studies on the performance of ceramic femoral
heads of 22.225 mm diameter, an initial penetration rate of 0.2-0.4 mm/million
cycles experimentally and 0.23 mm/year clinically, during the first 1.5 million
load cycles (corresponding to 18 months of walking). The average subsequent
bedding-in diminished to 0.022 mm/year in clinical studies and was very small in
simulator testing. The first acetabular cups made of radiation-crosslinked
UHMWPE (Durasul® cup, Sulzer Medica, Switzerland) are undergoing clinical
trials at the present time.
A further attempt to improve the mechanical properties of UHMWPE was
made by means of reinforcement with several types of fibres. Deng and Shalaby
[40] reported that the incorporation of UHMWPE fibres in a UHMWPE matrix
(self-reinforced composite) significantly increased the tensile strength and
modulus as well as the creep resistance. Carbon fibres were also considered be-
cause of their excellent reinforcement properties and their good biological com-
patibility. However, experiments with hip cups (Poly Two® Cup, Zimmer Corp.,
USA) failed because of the poor adhesion between the matrix and fibres. The hard
carbon fibres were pulled out of the composite, moved into the joint space and
consequently destroyed the head and cup of the prosthesis [2].
Beside these trials to improve the properties of UHMWPE, some efforts have
been undertaken to find alternative materials with reliable mechanical properties.
Among the possible candidates, carbon-fibre-reinforced polymers containing
carbon fibres of high modulus in a matrix of epoxy or triacine resin (Endocarbon®,
Orthoplant Endoprothetik Corp., Bremen, Germany; Caproman®, MAN Ceramics
Corp., Deggendorf, Germany) are of special interest. These materials show, be-
sides favourable sliding properties, a high wear resistance and loading capability,
have an elastic behaviour similar to bone and allow a great variability in the me-
chanical properties of components by means of variation of the fibre content, fibre
length, fibre orientation and kind of fibres chosen [41,42]. From the beginning of
the 1980s, carbon fibre implants have been used in humans, after preclinical bio-
logical testing, without any adverse effects [43,44]. Carbon-fibre-reinforced im-
plants were first used in tumour surgery for special implants; later, hip prosthesis
stems were investigated in series of clinical tests. Owing to their lower stiffness in
comparison with traditional metal materials, better integration in the surrounding
bone is expected. The long-time behaviour of carbon-fibre-reinforced hip stems
(endless-fibre laminate) is still not known, but clinical results depend of course on
the design and the correct implantation procedure also. Adam et al. [45] reported
on a study with a very high loosening rate of more than 30 % after five years, but
they believed the reason to be in the poor design of the prosthesis, not in the mate-
rial itself. Our own experience, in a clinical study with 15 hip stems made from
Caproman, showed better results; only one prosthesis failed by loosening after
four years of survival [46].
References
1. Paul J. P. (1976): Loading on nonnal hip and knee joints and on joint replacements. In:
Schaldach M., Hohmann D. (Eds.) Advances in Artificial Hip and Knee Joint Technology.
Springer, Berlin, Heidelberg
2. Eyerer P. (1986): Kunststoffe in der Gelenkendoprothetik. Z. Werkstoffiechnik 17: 384-391
3. Maloney W. J., Smith R. L., Stanford T. P., Schmalzried P., Chiba J., Huene D., Rubash H.
(1995): Isolation and characterization of wear particles generated in patients who have had
failure of a hip arthroplasty without cement. J. Bone Joint Surg. Am. 77: 1301-1310
4. Bosdorf K., Mollenhauer I., Willmann G., Boenick U. (1995): Werkstoffe fUr Hiiftendo-
prothesen. Biomed. Technik 40: 356-362
5. Schneider R. (1990): Der Mechanismus der Schaftlockerung. In: Ascherl A., Lechner F.,
Siebels W., Tensi H. M., Bliimel G. (Eds.) Die gelockerte Hiiftprothese. Schattauer, Stutt-
gart, New York: 112-116
6. McGee M. A., Howie D. W., Neale S. D., Haynes D. R., Pearcy M. J. (1997): The role of
polyethylene wear in joint replacement failure. Proc. Instn. Mech. Engrs. 211: 65-72
7. Clarke C., Gustafson A., Jung A., Fujisawa A. (1996): Hip-simulator ranking of polyethyle-
ne wear. Acta Orthop. Scand.67: 128-132
8. Schmalzried T. P., Kwong L. M., Jasty M., Sedlacek R. C., Haire T. C., O'Connor D.O.,
Bragdon C. R., Kabo J. M., Malcolm A. J., Harris W. H. (1992): The mechanism ofloosen-
ing of cemented acetabular componenets in hip arthroplasty. Clin. Orthop. 274: 60-78
9. Semlitsch M., Willert H. G. (1997): Clinical wear behaviour of ultra-high molecular weight
polyethylene cups paired with metal and ceramic ball heads in comparison to metal-on-metal
pairings of hip joint replacements. Proc. Instn. Mech. Engrs. 211: 73-88
10. Kurth M., Ellwanger R., Madler H., Eyerer P. (1990): Werkstoffkundliche Untersuchungen
an UHMWPE-Prothesenkomponenten: Von der Schadensanalyse zur Werkstoffoptimierung.
In: Ascherl A., Lechner F., Siebels W., Tensi H. M., Bliimel G. (Eds.) Die gelockerte Hiift:-
prothese. Schattauer, Stuttgart, New York: 98-111
11. Crugnola A. M., Radin E. L., Rose R M., Paul I. L., Simon S. R, Berry M. B. (1976): UHMW
polyethylene as used in articular prostheses. J. Appl. Polym. Sci. 20: 809-812
12. Nusbaum H. J., Rose R. M. (1979): The Effects of radiation sterilization on the properties of
ultrahigh molecular weight polyethylene. J. Biomed. Mat. Res. 13: 557-576
13. Roe R. J. (1981): Effect of radiation sterilization and ageing on ultrahigh molecular weight
polyethylene. J. Biomed. Mat. Res. 15: 209-230
14. Ambrosio L., Carotenuto G., Marietta G., Nicolais L., Scandurra A. (1996): Wear effects in
retrieved acetabularUHMW-PE cups. J. Mat. Sci. Mat. Med. 7/12: 723-729
15. Basle M. F., Bertrand G., Guyetant S., Chappard D., Lesourd M. (1996): Migration of metal
and polyethylene particles from articular prostheses may generate lymphadenopathy with
histiocytosis. J. Biomed. Mat. Res. 30: 157-163
16. Kobayashi A., Bonfield W., Kadoya Y., Yamac T., Freeman M. A., Scott G., Revell P. A.
(1997): The size and shape of particulate polyethylene wear debris in total joint replace-
ments. Proc. Instn. Mech. Engrs. 211: 11-15
17. Kim K. J., Itoh T., Tanahashi M., Kumegawa M. (1996): Activation of osteoclast-mediated
bone resorption by the supernatant from a rabbit synovial cell line in response to polyethyl-
ene particles. J. Biomed. Mat. Res. 32: 3-9
18. Goodman S. B., Davidson J. A., Song Y., Martial N., Fornasier V. L. (1996): Histomor-
phological reaction of bone to different concentrations of phagocytosable particles of high-
density polyethylene and Ti-6AI-4Valloy in vivo. Biomaterials 17: 1943-1948
19. Willert H.-G., Semlitsch M. (1977): Reactions of the articular capsule to wear products of
artificial joint prostheses. J. Biomed. Mat. Res. 11: 157-164
20. Vernon-Roberts B., Freeman M. A. R (1977): The tissue response to total joint replacement
prostheses. In: Swanson S. A. V., Freeman M. A. R (Eds.) The Scientific Basis of Joint Re-
placement. Pitman Medical, Tunbridge Wells: 86-129
21. Willert H.-G., Bertram H., Buchhorn G. H. (1990): Osteolysis in alloarthroplasty of the hip:
the role of ultra-high-molecular-weight polyethylene wear particles. Clin. Orthop. 258: 95-
107
22. Howie D. W., Vernon-Roberts B., Oakeshott R., Manthey B. (1988): A rat model ofresorp-
tion of bone at the cement-bone interface in the presence of polyethylene wear particles. J.
Bone Joint Surg. Am. 70: 257-263
23. Murray D. W, Rushton N. (1990): Macrophages stimulate bone resorption when they phygo-
cytose particles. J. Bone Joint Surg. Br. 72: 988-992
24. Herman J., Sowder W. G., Anderson D., Appel A. M., Hopson C. N. (1989): Polymethyl-
methacrylate-induced release of bone-resorbing factors. J. Bone Joint Surg. Am. 71: 1530-
1541
25. Santavirta S., Konttinen Y. T., Bergroth V., Eskola A., Tallroth K., Lindholm T. S. (1990):
Aggressive granulomatous lesions associated with hip arthroplasty. J. Bone Joint Surg. Am.
72:252-258
26. Santavirta S., Konttinen Y. T., Hoikka V., Eskola A. (1991): Immunopathological response
to loose cementless acetabular components. J. Bone Joint Surg. Br. 73: 38-42
27. Santavirta S., Hoikka V., Eskola A., Konttinen Y. T., Paavilainan T., Tallroth K. (1990):
Aggressive granulomatous lesions in cementless total hip arthroplasty. J. Bone Joint Surg.
Br. 62: 29-32
28. Fr"kja:r J., Deleuran B., Lind M., Overgaard S., Soballe K., Biinger C. (1995): Polyethylene
Particles stimulate monocyte chemotactic and activating factor production in synovial
mononuclear cells in vivo. Acta Orthop. Scand. 66: 303-307
29. Takagi M., Konttinen Y. T., Santavirta S., Sorsa T., Eisen A, Nordsletten L., Suda A
(1994): Extracellular matrix metalloproteinases around loose total hip prostheses. Acta Or-
thop. Scand.65: 281-286
30. Nordsletten L., Buo L., Takagi M., Konttinen Y. T., Yamakawa M., Santavirta S., Aasen A. O.
(1996): The plasaminogen activation system is upregulated in loosening of total hip prosthe-
ses. Acta Orthop. Scand. 67: 143-148
31. Goodman S., Aspenberg P., Song Y., Regula D., Lindgren L. (1996): Polyethylene and
titanium alloy particles reduce bone formation. Acta Orthop. Scand. 67: 599--605
32. Fomasier V., Wright J., Seligman J. (1991): The histomorphologic and morphometric study
of asymptomatic hip arthroplasty. A Postmortem Study. Clin. Orthop. 271: 272-282
33. Aspenberg P., Herbertsson P. (1996): Periprosthetic bone resorption. J. Bone Joint Surg. 78-
B: 641-646
34. Bohm G. A. (1967): Diffusion theory of the post-irradiation oxidation of polyethylene. J.
Polym. Sci. 5: 639--652
35. De Vries K. L., Smith R. H. (1980): Free radicals and new end groups resulting from chain
scission. I. Gamma irradiation of polyethylene. Polymer 21: 949-955
36. Bhateja S. K. (1983): Radiation-induced crystallinity changes in linear polyethylene. J.
Polymer Science 21: 523-536
37. Oonishi H., Ishimaru H., Kato A (1996): Effect of cross-linkage by gamma-radiation in
heavy doses to low wear polyethylene in total hip prosthesis. J. Mat. Sci. Mat. Med. 7/12:
753-763
38. Streicher R. M. (1989): Investigation on sterilization and modification of high molecular
weight polyethylenes by ionizing irradiation. Beta-Gamma 1: 34-43
39. Wroblewski B. M., Sidney P. D., Dowson D., Collins S. N. (1996): Prospective clinical and
joint simulator studies of a new total hip arthroplasty using alumina ceramic heads and cross-
linked polyethylene cups. J. Bone Joint Surg. 78-B: 280-285
40. Deng M., Shalaby S. W. (1997): Properties of self-reinforced ultra-high-molecular-weight-
polyethylene. Biomaterials 18: 645-55
41. Lancaster I. K. (1972): Lubrication of carbon fibre-reinforced polymers. Wear 20: 315-351
42. Kaddick C., Ascherl R., Siebels W., Bruller 0., BIUmel G. (1996): Mechanische Sicherheit
von HUftgelenkendoprothesenschiiften. Z. Orthop. 134: 111-116
43. Williams D. F., MeN
E.2 Material Parameters
and ESEM Characterization
of Functional ENT Prostheses
During Ongoing Degradation
E.-J. Haberland, A. Berghaus and M. FUting, Halle,
and I. Bethge and W. Grellmann, Merseburg
1 Introduction
Voice prostheses and tracheal prostheses are functional prostheses that are im-
planted to copy a natural function. Voice prostheses are implanted in patients who
have had their larynx removed for medical reasons. With a voice prosthesis, fast
voice rehabilitation is possible. The prosthesis generates the voice only indirectly.
It makes possible an airflow that can cause vibrations of the entrance musculature
and mucous membrane of the oesophagus. It is a pharyngo-tracheal shunt valve.
Because of the overpressure in the trachea, air streams into the oesophagus or the
low pharynx area. On the other hand, the passage of saliva and undigested food
into the trachea from the oesophagus is prevented because of the valve effect.
Perfect valve function is necessary for correct functioning: dysfunction necessi-
tates a prosthesis change. There is a range of products on the market that differ
mainly as far as the valve is concerned. The material of choice is silicone rubber.
Polyurethane has also been tested as a material. Details of clinical use are given in
[1-3].
The use of a tracheal prosthesis is necessary if several tracheal rings must be
removed because of disease and an end-to-end suture is no longer possible. The
condition for correct function here is, on the one hand, an adequate reaction to the
alternating mechanical loads because of motions of the head and neck and, on the
other hand, support of secretion transportation from the bronchi. The imitation of
mucociliar transportation has not been achieved satisfactorily so far. A hydrophilic
inside surface of the prosthesis supports the coughing up of secretions. As with the
voice prosthesis, silicone rubber is a suitable material for the realization of the
required mechanical properties. The silicone rubber is reinforced additionally with
metallic rings, which are surrounded by porous polyethylene. Details of the con-
struction of such a tracheal prosthesis are described in [4].
Both prosthesis types lose effectiveness as a result of the fluid medium. The
shaft of the voice prosthesis is in contact with the tissue lymph of the non-
epithelian shunt between the trachea and pharynx. The shaft should remain mov-
able in the shunt so that it can be removed easily when no longer functioning. Its
valve side is in constant contact with saliva and so also with mouth flora. Biocor-
rosion can be caused by this, in particular. The outside of the tracheal prosthesis is
surrounded by non-epithelian tissue, which should preferably grow into the sur-
face so that the mechanical stability is increased and the connective sutures are

452 E.-J. Haberland, A. Berghaus, M. Fliting, I. Bethge, W. Grellmann
relieved of stress. The inside is exposed to bronchial secretion, as is also the tra-
cheal side of the voice prosthesis. Both surfaces should possess a high self-
cleaning ability. In the following, investigations into the coarse structure of the
surface changes due to biocorrosion of voice prostheses and a pilot examination of
the behaviour over time of the plastic hardness of such surfaces during water stor-
age are described. The effect of different stiffness values of the material on the
aerodynamic behaviour of the valve mechanism in vitro is described with respect
to several voice prostheses with varying degrees of hardness.
2 Materials and Methods
The normally available voice prostheses of the Provox® and ESKA-Herrmann types
are made of silicone rubber. Further details of the materials are not available. For
comparison, samples of the Halle voice prostheses, which are produced from
polyurethane [5], were examined. The surfaces were investigated with light mi-
croscopy and with conventional and environmental scanning electron microscopy.
The plastic hardness was measured with the Fischerscope HIOO test device, which
measures micro-hardness. For measurement of the pressure/flow characteristics,
an aerodynamic measuring device was used [6].
3 Results
3.1 Corrosion and Degradation
The voice prostheses shown in Figs. E4 and E5 had to be replaced after four to six
months of use because of leakage or impaired functioning of the valve tab. Both
materials - polyurethane and silicone rubber - were covered with a biofilm several
tenths of a millimetre thick. In the case of the polyurethane prosthesis, this layer
had grown unevenly into the valve split and had thus caused a leak. The delicate
tab valve of the silicone prosthesis has become immovable because of the layer
thickness is critical.
After the Provox® prosthesis had been cleaned with warm water and a soft
brush, there remained on the surface of the prosthesis edge, on the oesophagus
side, clusters with a typical relief structure which shows radial growth from a
central point. Surfaces which are seemingly unaffected by biocorrosion lie between
these clusters (Fig. E6).
The scanning electron microscope image (Fig. ES) shows that the edges of the
relief areas are irregular. The remains of the valve-tab layer which remained stuck
after cleaning have a different structure (Fig. E7). Here the growth clusters appear
to be distributed around multiple centres in the form of lumps. Between, there are
node lines like those of plate vibrations. How deep the biofilm reaches into the
material cannot be determined with this procedure without sectioning. That in
Material Parameters of ENT Prostheses 453
addition to this superficial corrosion, changes also take place in the depth of the
material is shown by Fig. E9. An ESKA-Herrmann prosthesis which is noticeably
corroded only in the area of the oesophagus side of the lip valve has a gaping
valve split. Greater degradation of the material can be seen here.
Fig. E4. Halle voice prosthesis made of Fig. E5. Provox'" voice prosthesis made of
polyurethane, after destructive silicone rubber, directly after non-
removal because of impaired destructive removal because of
functioning, subsequently kept in impaired functioning
air
Fig. E6. Surface patterns of biocorrosion Fig. E7. Fixed biofilm on the valve tab of
on the oesophagus-side edge of a a Provox'" prosthesis (bright
Provox'" prosthesis (bright field) field)
454 E.-J. Haberland, A. Berghaus, M. Filting, I. Bethge, W. Grellmann
This becomes even more evident with the prosthesis shown in Fig. EIO. While
the lip valve of the prosthesis in Fig. E9 is occupied only by small, evenly distrib-
uted growth clusters, in Fig. E 10 the lip valve of the prosthesis shows, in the ad-
vanced stage, a few large clusters with a dense overgrowth. The prosthesis shaft
and the two flanges are apparently not corroded. In the area of the valve lips, clear
degradation has led to blockage of the valve split. It is not known what type of
mouth flora and saliva composition were present in the case of this patient that
such degradation of the material could take place within a duration of four months.
Fig. E8. SEM image of the boundary of Fig. E9. Gaping valve split in an ESKA-
the cluster shown in Fig. E6 Herrmann prosthesis
Fig. EIO. Degradation of the material of the valve lips (silicone rubber) of an ESKA-
Herrmann prosthesis. BiocoITosion and biodegradation are visible only in
the area of saliva effect
Material Parameters ofENT Prostheses 455
3.2 Environmental Scanning Electron Microscopy
In addition to the extensive examinations of biocorrosion and biodegradation of

prostheses by conventional scanning electron microscopy, investigations using an
environmental scanning electron microscope (ESEM) have now started. This new
type of scanning electron microscope allows the specimen chamber of the micro-
scope to be filled with gas even when it is fully working. As a result, non-
conducting materials can be observed without preparation. This is possible be-
cause the gas is ionized during the electron bombardment, so that the resulting
positive ions compensate disturbing negative charges at the specimen surface.
Using water vapour as the filling gas, environmental scanning electron micro-
scopes are able to prevent or retard the drying of specimens. According to the
temperature and the (total) water vapour pressure inside the specimen chamber,
the specimen is kept in an environment of defined humidity. These conditions
correspond directly to those in air of the same relative humidity (rh). The partial
water vapour pressure in the air is equal to the total pressure of the water vapour
surrounding the specimen in the microscope. Measurements of the relative hu-
midity inside the oral cavity gave values of 75 % rh and 95 % rh in the open and
closed states, respectively. It seems to be useful to transfer these values to the
microscope to stabilize the specimen under investigation in its original state.
Fig. Ell. Provox® voice prosthesis (silicone rubber), parts of the shaft and the valve; freshly
removed, original state (only rinsed in warm water), without drying or coating; the
shaft appears to be without a biofilm, but the valve is nearly fully covered with a
biofilm (left, at 75 % rh; right, wet state; the biofilm is partially saturated with
condensed water)
Compared with investigations by conventional scanning microscopy, a great

advantage of the ESEM technique is that the drying artefacts can be recognized
and identified in the nascent state. Under these circumstances it becomes increas-
ingly important to guarantee the conservation of the original state between the re-
moval of a prosthesis and its transport to the microscope laboratory (Figs. Ell-14).
456 E.-J. Haberland, A. Bergbaus, M. Fllting, I. Betbge, W. Grellmann
Fig. E12. Characteristic biofilm fonnation on Fig. E13. Details of biofilm fonnation with
the valve surface granulocytes and mycelium of
Candida albicans
Basically, the advantage of the ESEM technique is the possibility of observing

dynamic processes influencing and/or modifying the specimen in situ [7].
For example, it becomes possible to simulate the rhythmic drying and wetting
of the epithelium and of the biofilm inside the trachea and the prosthesis with
expiration and inspiration, by variation of the water vapour pressure inside the
specimen chamber.
Additional condensation of water on the specimen surface can be induced if the
pressure of water vapour is varied around the equilibrium vapour pressure. This
provides an opportunity to study the wetting and drying behaviour of both the
original and the deteriorated prosthesis surfaces, which is strongly influenced by
the formation ofbiofilms. Vice versa, the in-situ observations of the condensation
provide information that can be used to differentiate morphological features of the
biofilm (Fig. E 15). To characterize the wetting behaviour of water or other solu-
Fig. E14. Granulocyte

Fig. E15. Edge of a valve plate at 75 % rh (a); edge of the valve plate (same location as in (a»
with condensation of water droplets at strongly hydrophilic biofilm components (ar-
row) (b)
tions, for example Ringer's solution or saliva, they can be poured onto the speci-
men by a microinjection device with a fine capillary terminating inside the speci-
men chamber. The capillary can be positioned from outside the chamber. The
droplets coming out of the capillary are some hundred microns in diameter, and
they can be moved onto the surface of the sample under investigation.
The field of microscopically controlled in-situ experiments inside the ESEM is
still in progress, and new developments can also be expected in combinations of
the well-established cryo-techniques with the ESEM technique.
3.3 Micro-hardness of the Material
The influence of the effects of the medium on the surface hardness of the material
was examined with a micro-hardness measurement technique. Only the properties
of the uppermost layer can be determined with an impression of a few tenths of a
millimetre. Figure E 16 shows the scatter of the results for ten voice prostheses
from different batches, before implantation. Test No. 10 (Provox®) serves as a
reference, as it appears to be a suitable comparison because of its smooth surface
and low creep rate. There are considerable differences in the plastic hardness of
the various prostheses. The fluctuation of the same parameter in the case of an
ESKA-Herrmann prosthesis after it had been in the patient for ten weeks and sub-
sequently kept in water at room temperature did not show the expected result.
Despite careful experimental design, including drying-off of the specimen after
water storage and immediate measurement of the marked flange piece, significant
differences appeared for almost equal removal times, presumably caused by proc-
esses at the border between the biofilm layer and the base material. These differ-
ences have not been verified yet. In addition, systematic errors can be excluded
458 E.-J. Haberland, A. Berghaus, M. FUting, I. Bethge, W. Grellmann
only by further experiments, for example with variation of the load values. The
aim of such investigations must be to determine the hardness of the layer as a
function of the hardness of the base material and to elucidate the question of what
influence this layer has on the change of the hardness at the surface. Detailed
considerations with regard to the characterization of voice prosthesis materials are
to be found in the chapter by Bier<>gel et al. (pp. 471-476).
,.-..
]
N
b 2
'"'"
a)
.§
1;;
::r::
....
.~
'"
E:'"
o 2 7 8 10
9
Prosthesis Number
Fig. E16. Plastic hardness of a selection of ten voice prostheses (1-7 and 9, silicone rubber,
Halle; 8, PUR, Halle; 10, silicone rubber, Provox~
3.4 Flow Resistance
In [8] the physical basis for the calculation of the airflow resistance of the ESKA-
Herrmann prosthesis is explained. The dependence on the elasticity of the material
is not taken into account in that calculation. This dependence is especially clear in
the case of the Halle voice prosthesis because the tab, viewed from the split, be-
comes thicker in the direction of the base. Figure E 17 shows pressure/flow curves
for Halle prostheses of different shaft lengths and measured hardness. Only the
polyurethane prosthesis, which is classified as soft, is clearly distinguished from
the others by its low resistance of 6 Pa s/ml.
An investigation of the resistance using a dynamic method with a half-sinusoidal
flow signal shows that the resistance increases in the flow peaks for some prosthe-
ses by up to 30 %, while for others it remains approximately constant (Fig. E 18).
4 Discussion
Experience shows that even strongly corroded voice prostheses can still work
satisfactorily, provided the valve fulfils its task. The degree of corrosion depends
on the individual oral milieu. More precise details regarding this problem are to be
found in the chapter by Sebova (pp. 461-470) and a paper by Everaert et al. [9].
Substitute prostheses, including the ESKA-Hemnann prosthesis, can be cleaned by
5~--------~--------~------~
'OJ'
~3+-------~~--~---H~----~
~
£2+-------47~~~_r----=-~
O+---~----r-~~--~--~T---~
o 100 200 300
Flow (mIlsl
Fig. E17. Flow-pressure relations for four different Halle prostheses made of silicone rubber (T,
..., e) and PUR (II)
21
--NO.5
No.2
-oQo-
20 ........ No.6
fe::.
- - - No.1
19 +-----J~--t-+\;_--r___l ...... No.4
.... .Q .... NO.7
--No.3
18
..,1
~ 17
~
~
16
15
0 200 400 600
Time (ms)
Fig. E18. Flow resistance of seven HaIle prostheses obtained by use of haIf-sinusoidal flow
pulses as test signal
460 E.-J. Haberland, A. Berghaus, M. Filting, I. Bethge, W. Grellmann
the patient, with some skill. The resulting increase in the device life is known. For
indwelling prostheses, such as the Provox® prosthesis, the lifetime is increased by
application of an anti-mycotic agent on the surface. The mobility of the flap is
restricted later [10-13]. Every avoidance of biocorrosion causes an extension of
the lifetime of the prosthesis. After approximately one year, the long-time changes
in the material properties can become a problem. Distortions which lead to inade-
quate performance of the valve or an increase in resistance through hardening are
a consequence of such changes. Here, only water storage was examined as an
influencing factor. As the first images show, in addition to physical and chemical
influences, biological influences are also important in any model for testing corro-
sion and degradation. A systematic analysis of prostheses which no longer func-
tion belongs to further research programmes based on applied materials science.
References
1. van den Hoogen F. l A., Oudes M. I., Hombergen G., Nijdam H. F., Manni I. I. (1996): The
Groningen, Nijdam and Provox voice prostheses: a prospective clinical comparison based on
845 replacements. Acta Otolaryngol. (Stockh.) 116: 119-124
2. Aust M. R., McCaffrey T. V. (1997): Early speech results with the provox prosthesis after
laryngectomy. Arch. Otolaryngol. Head Neck Surg. 123 (9): 966-968
3. Herrmann I. F. (Ed.) (1986): Speech Restoration via Voice Prostheses. Springer, Berlin,
Heidelberg
4. Berghaus A. (1992): Eine neue Trachealprothese. Experimentelle und erste klinische Er-
gebnisse. Otorhinolaryngol. Nova 2: 232-237
5. Haberland E.-I., Neumann G., LObe L.-P., Voigt K. (1990): Stimmprothese. German Patent
DD 275183 AI, Aktenzeichen WP A 61 F, January 17
6. Haberland E.-I., Grellmann W. (1996): Material- und Funktionskenngr06en von pharyngo-
trachealen Shuntventilen ('Stimmprothesen'). In: Proceedings of Polymerwerkstoffe '96,
Merseburg, Germany, September 18-20: 252-255
7. Hopfe I.; Ftiting M. (1995): Fundamentals and applications of environmental scanning
electron microscopy. In: Wetzig K. (Ed.) In situ Scanning Electron Microscopy in Materials
Research. Akademie-Verlag, Berlin: 219-240
8. SchOn F. (1986): Physical aspects oftracheo-oesophageal shunt prosthesis. In: Herrmann I.
F. (Ed.) Speech Restoration via Voice Prostheses. Springer, Berlin, Heidelberg: 59-62
9. Everaert E. P. l M., Mahieu H.E, Wong Chung R P., Verkerke G. l, van der Mei H. C.,
Busscher H. l (1997): A new method for in vivo evaluation of biofilms on surface-modified
silicone rubber voice prostheses. Eur. Arch. Otorhinolaryngol. 254: 261-263
10. Hilgers F. l M., Balm A. I. M., Gregor R. T. (1995): Stimmrehabilitation nach Laryngekto-
mie mit der Provox\l)-Stimmprothese - Chirurgische und technische Aspekte. Teil!. RNO
43: 197-201
11. Hilgers F. I. M., Balm A. I. M., Gregor R T. (1995): Stimmrehabilitation nach Laryngekto-
mie mit der Provox\l)-StimIilprothese ~ Chirurgische und technische Aspekte. Teil II. RNO
43:263-267
12. Natarajan B., Richardson M. D., Path M. R c.,Irvine B. W. H., Thomas M. (1994): The Pro-
vox voice prosthesis and Candida a1bicans growth: a preliminary report of clinical, myco-
logical and scanning electron microscopic assessment. l Laryngol. Otol. 108 (8): 666-668
13. de Carpentier l P., Ryder W. D. l, Saeed S. R, Woolford T. I. (1996): Survival times of
Provox valves. l Laryng. Otol. 110 (1): 37--42
E.3 Microbial Corrosion
of Pharyngo-Tracheal Shunt Valves
('Voice Prostheses')
1. Sebova, Bratislava, and E.-J. Haberland
and A. Stiefel, Halle
1 Introduction
Singer and Blom published in 1980 their technique of the tracheo-oesophageal

puncture (T-E puncture), including the application of the fIrst commercially manu-
factured valve prosthesis. With this method, they started an ever-broadening inter-
est in the voice rehabilitation of patients without a larynx. In the last 20 years
different variations of this prosthesis have been developed, which have been tested
successfully with several thousand patients [1,2].
A valve prosthesis, internationally known as a 'voice prosthesis' (or voice but-
ton), is made from a biocompatible plastic and consists of a tubular body with two
flanges. A valve is incorporated into the flange on the oesophageal side. This
makes it possible for air to flow from the trachea into the oesophagus after the
insertion of the prosthesis in the T-E shunt; it is impermeable, however, to food
and saliva. The flanges provide a solid seat for the valve in the shunt.
The biggest disadvantage of currently used valve prostheses is their limited de-
vice life. It amounts on average to fIve to eight months and is limited by biocorro-
sion and biodegradation. For reasons of simplicity, only biocorrosion is discussed
in this chapter. The exposure to a moist, enzymatically active and non-sterile mi-
lieu and the high mechanical loading lead to biocorrosion of valve prostheses
made from elastomers. Their surface is quickly populated with bacteria and germs.
Owing to the microflora, the self-cleaning ability of the prosthesis is decreased in
a short time. The prostheses are at present mostly manufactured from silicone
rubber. Polyurethane is used only in a few cases. These materials have especially
good chemical and physical qualities for this purpose: biocompatibility, high elas-
ticity, long durability and resistance against chemical influences [3]. These quali-
ties result from the molecular structure of these elastic polymer materials with
limited hardness. The molecules are displaced with respect to one another as a
result of externally applied forces. The elastic properties are dependent on the
length of the molecules, the extent and connection of the side chains, and the so-
called 'inner' or 'outer' softening agents or on substances incorporated as a fIller
between the molecules. The loosened-up molecular structure allows incorporation
of water, by which means the material can swell. The softer an elastomer is, the
more easily it will absorb water if this is not prevented by other additives. Both
surface corrosion and the growth of bacterial flora can be explained with this
structure.

462 I. §ebova, E.-J. Haberland, A. Stiefel
The surface of the elastomer can be smoothed only with great difficulty, be-
cause of the high degree of local ·softness. Polishing can be carried out only by
cooling the material to extremely low temperatures. An uncontrolled property of
all elastomers from the point of view of implantation into the human body is the
bacterial and mycotic adherence of the surface.
2 Material and Methods
2.1 Materials for Attempts to Achieve Bacterial Settlement
Experiments on the settlement of micro-organisms were carried out to examine the

influence of micro-organisms on the biocorrosion of valve prostheses. Eight
prostheses, taken after going out of function in vivo, were separately examined by
scanning electron microscopy from the direction of the trachea and also from the
direction of the oesophagus.
The following valve prosthesis materials were included in the examinations:
- silicone rubber, type 'Wtlrzburg'
- silicone rubber, type 'Bratislava'
- silicone rubber, type 'Groningen'
- polyurethane type, 'Halle'.
Test germs were isolated from the main collection of the Institute for Medical
Microbiology of the Martin-Luther-University of Halle-Wittenberg, as follows:
- Staphylococcus aureus ATCC 6538
- Staphylococcus epidermidis 433/6/89
- Pseudomonas aeruginosa A TCC 15442
- Klebsiella pneumoniae 462/4/89
- Candida albicans ATCC 10231
- mixed culture from all five germ species.
The germs were cultivated on ovine blood agar (5 %) in pure culture. Bouillon I
from the SIFIN company served as the nutritional liquid.
Incubated test specimens consisting of the elastomers mentioned above were
pre-fixed in glutaraldehyde using solutions of the following compositions [4]:
- stock solution (a): 0.1 M Na2HP04 (= 7.1 g dehydrated Na2HP04 in 500 ml
distilled water)
- stock solution (b): 0.1 M KH2P04 (= 0.8 g KH2P04 in 500 ml distilled water)
- stock solution (c): 7-8 parts solution (a) + 2-3 parts solution (b), to give a 0.1
M phosphate buffer of pH 7.2-7.4
- stock solution (d): 25% glutaraldehyde (freshly purified).
To prepare the fixative solution, 2 parts of (c) were mixed with 1 part of (d) + 1
part of distilled water.
Microbial Corrosion ofPharyngo-tracheal Shunt Valves 463
To wash out the excess fixative, 6.85 g of sucrose were dissolved in 100 ml of
phosphate buffer.
For the post-fixation of the specimens, an OS04 fixative of the following com-
position was used [4]:
- stock solution (a): NaH2P04'H20, 2.26 % in distilled water (for buffer)
- stock solution (b): NaOH, 2.52 % in distilled water (for adjustment of the pH
value)
- stock solution (c): glucose 5.4 % in distilled water (for plasma isotonicity)
- stock solution (d): mixture of solutions (a) and (b) in the ratio 41.5: 8.5 ml.
The fixative solution was prepared by mixing 45 ml of solution (d) with 5 ml of
solution (c) and adding 0.5 g ofOs04 dry substance. The pH of the resulting solution
was 7.3 to 7.4.
2.2 Methods of In Vitro Investigation of Microbiological Colonization

on Silicone Rubber and Polyurethane
The elastomer materials were cut into cubes of approximate size 5 x 5 x 5 mm 3

and autoclaved for 20 minutes at 221°C. For each combination of material and
germ, five cubes of the material were put into an Erlenmeyer flask with 10 ml of
bouillon I. Afterwards, five colonies of the relevant germ or of the mixed culture
were introduced per flask. A sterile control cube was kept for each material. Then
the specimens were incubated for a period of24 hours at 37°C. Subsequently, all
specimens were stored for 120 days at room temperature (22°C). The bouillon
was drained from the flask weekly, leaving a residual volume, and was replaced
by sterile bouillon I. By this means, the growth conditions of the micro-organisms
were optimized. At the same time, smears were made of the used bouillon on
ovine blood agar in order to recognize any possible contamination in time.
2.3 Methods of Investigation by Scanning Electron Microscopy
The SEM examinations were carried out with a TESLA BS301 scanning electron
microscope made by the Tesla company, Bmo, Czech Republic.
A test cube of each material was used on days 0, 30, 60, 90 and 120 for the
purposes of this examination. The cubes were rinsed with distilled water directly
after removal, cut vertically with sterile razor blades into parts A and B and fixed.
Part A was fixed for 4 hours in buffered glutaraldehyde, then rinsed in distilled
water and stored in 13 ml of70 % ethanol.
Part B was treated in 10 % sodium hypochlorite for 10 minutes, rinsed for a
short time with distilled water, treated in an ultrasonic cleaning bath (70 % etha-
nol) and then stored in 70 % ethanol.
464 I. ~ebova, E.-J. Haberland, A. Stiefel
The removal of the organic substances of the biofilm was achieved in part B in
this way. Possible interactions between micro-organisms and the surface of the
elastomers would thus be made visible for evaluation, and the depth to which they
penetrated would also be visible.
The preparation technique before the SEM investigation had the goal of pre-
serving the morphology of the micro-organisms. By dehydration in alcohol in
steps with increasing concentrations of 60 %, 70 % and 100%, the subsequent
evaporation of carbon and gold onto the specimens was made possible. The thick-
ness of this layer of carbon and gold was 30 nm. Specimens prepared in this man-
ner were examined in the SEM and images were recorded at magnifications be-
tween 38 : 1 and 10000: 1. The beam voltage was 15 kV and the beam current was
50 ~A.
3 Results
Macroscopic changes were observed in the prostheses used. Besides a change in

colour· of the material (silicone rubber, yellowish; polyurethane, dark brown),
microbial biofilms that were stuck solidly to the surface and could be mechani-
cally removed from the material only incompletely were observed, principally on
the oesophageal flange. The macrostructure of the surface changes is described in
the chapter by Haberland et al. in E.2 (pp. 451-460).
For the in vitro experiments, germs from smears of the pharynx and tracheos-
toma were used, as listed in Sect. 2.1.
The dates for the removal of the preparations in the in vitro experiments were
determined according to data from the literature and statements from the manu-
facturers concerning the indwelling duration of the prostheses. In each case we
evaluated the SEM images of the surface and of a section of the material of the
prosthesis. Material that was fixed in the way used for part A showed continuous
biofilms on the surface of all samples in the images.
The typical growth of Candida albicans could be identified only in the pure
culture. We were not able to identify Candida albicans as a component of a mixed
flora. A roughened surface with innumerable scratches and cracks was found in
the samples fixed by method used for part B. We assume these cracks are artefacts
originating as a result of the vacuum during the SEM ~xamination.
The colour of the elastomers did not change in the in vitro experiments, except
that in the case of cultivation with Pseudomonas aeruginosa, the material turned
light green. By means of cross-sections through the test specimens it was deter-
mined that the microflora did not penetrate into the material but adhered to the
surface. No differences were observed between the silicone rubber and polyure-
thane elastomers used. Characteristic SEM images are shown in Figs. EI9-E26.
Fig. E19. In vivo colonization of silicone rubber ('Groningen') with mixed flora after 10
months indwelling. Corrosion and residues ofbiofilm are visible on the surface of the
flange on the oesophageal side. The grooves on the surface are caused by the method
of production. (Type B fixation; SEM image)
Fig. E20. In vitro colonization of silicone rubber ('Bratislava') with Candida a1bicans after 2
months. The grooves on the surface are caused by the method of production. (Type A
fixation; SEM image)
466 I. ~ebova, E.-J. Haberland, A. Stiefel
Fig. E21. In vivo colonization of silicone rubber ('Groningen') with mixed flora after 5 months
indwelling. Corrosion is visible on the surface of the flange on the oesophageal side.
(Type B fixation; SEM image)
Fig. E22. In vitro colonization of silicone rubber ('Wiirzburg') with Candida albicans after 3
months indwelling. Besides the remains of biofilms, erosion is visible. The cracks are
caused by the process of preparation. (Type B fixation; SEM image)
Fig. E23. In vivo colonization of polyurethane ('Halle') with mixed flora after 3 months in-
dwelling. Flange on the tracheal side with biofilm. (Type A fixation; SEM image)
dwelling. The cavity at the cut edge is probably a manufacturing error. (Type A fixa-
tion; SEM image)
468 I. Sebova, E.-J. Haberland, A. Stiefel
Fig. E25. In vitro colonization of silicone rubber ('Bratislava') with Candida aIbicans after 2
months indwelling. Blastomyces with pseudomyceJium are visible. (Type A fixation;
SEM image)
dwelling. (Type B fixation; SEM image)
4 Discussion
The mechanism of bacterial and mycotic adherence to elastomer and polymer

surfaces is not completely explained yet. It is suspected that this process is similar
to bacterial adherence on epithelium. It occurs by means of specific interactions
between wall components (adhesins) and epithelial receptors. The bacteria cannot
be eliminated by the cleaning mechanisms of secretions after interaction; they

penetrate into the tissue and begin to multiply [5-7].
The irreversible mechanism of adherence of Staphylococcus epidermidis, for
example, on the surface of elastomers is well known. The germs are able to multi-
ply there and can form extracellular mucus. The Staphylococci are then obviously
protected against the host's defence mechanisms, but are also protected against the
effect of chemotherapeutic agents. Staphylococcus epidermidis dominates as a
pathogen in infections and is associated with foreign bodies and intravaSal cathe-
ters made from elastomers.
Mycotic adherence differs from bacterial adherence in that there is more fre-
quent coadherence. The mycotic cells adhere on the one hand to the elastomer,
while on the other hand they also adhere to one another. Colonization with Can-
dida albicans takes place most frequently. The pathogens Candida albicans, Can-
dida tropicalis and Candida stellatoidea stick more quickly than non-pathogenic
germs [8]. It is suspected that the adherence of Candida albicans to epithelial cells
results from the action of extracellular membrane proteins and protein components
of the adhesins. The adherence of Candida albicans is of significance, for instance,
in the pathogenesis of endocarditis or sepsis caused by Candida. Because of the
adhesion to endothelial cells, the pathogens can no longer be engulfed. In people
with dentures, a stomatitis can be caused by the fungus sticking onto the material
of the prosthesis. The adherence to traumatized tissue is stronger than to healthy
structures [9]. The same situation occurs in the non-epithelialized T-E shunt,
where the valve prosthesis must be tolerated for a relatively long time as a foreign
body. From clinical experience, it has been found that chronic inflammations
originate locally about the T-E shunt with development of granulation, and only
exceptionally does a restricted regional inflammation or a systemic spread, for
example sepsis, occur [10,11].
The methods described here for the examination of bacterial and mycotic ad-
herence are a useful tool when looking for a material that prevents adherence. In
the literature, coatings of the surface with Teflon, titanium and other materials are
reported on. None of these procedures has brought a breakthrough in the signifi-
cant prolongation of the indwelling duration of valve prostheses, however. In
addition to the methods described here, biophysical measurements of the adhesion
forces of cells should be made in vitro.
References
1. Blom E. D., Hamaker R. C. (1996): Tracheoesophageal voice restoration following total

laryngectomy. In: Myers E. N., Suen J. (Eds.) Cancer of the Head and Neck. W. B. Saun-
ders: 839-852
2. Leder S. B., Erskine M. C. (1997): Voice restoration after laryngectomy: experience with the
Blom-Singer extended-wear indwelling tracheoesophageal voice prosthesis. Head Neck 19:
487-493
470 I. Sebova, E.-J. Haberland, A. Stiefel
3. Nijdam H. F. (1991): The development of an indwelling voice prosthesis. In: Hilgers

F. J. M., Schouwenburg P. F. (Eds.) Teaching Course on Indwelling Voice Prostheses for
Voice Restoration after Total Laryngectomy. The Netherlands Cancer Institute, Amsterdam:
8-10
4. Ohnsorge J.,Holm R. (1978): Rasterelektronenmikroskopie. Eine EinfUhrung fUr Mediziner
und Biologen. Thieme, Stuttgart: 157
5. Centano A. C., Davis C. P., Cohen M. S., Warren M. M. (1983): Modulation of Candida
albicans attachment to human epithelial cells by bacteria and carbohydrates. Infect. Immun.
39: 1354-1360
6. Quadripur S. A. (1988): Theoretical and practical foundations of fungus adherence to host
cells. Git. Suppl. 6: 39-40
7. Everaert E. P. J. M., Mahieu H. F., Wong Chung R. P., Verkerke G. J., van der Mei H. C.,
Busscher H. J. (1997): A new method for in vivo evaluation ofbiofilms on surface-Modified
silicone rubber voice prostheses. Eur. Arch. Otorhinolaryngol. 254: 261-263
8. Trouchin G., Bouchara J. P., Robert R., Senet J. (1988): Adherence of C. albicans genn
tubes to plastic. Ultrastructural and molecular studies of fibrillar adhesins. Infect. Immun.
56: 1987-1993
9. Tomsikova A., Kotal L. (1990): Vyznam adherencni schopnosti kandid pro patogenezi
aliznicnich kandidoz. Cesk. Epidemiol. Mikrobiol. Imunol. 39: 182-190
10. Manni J. J. (1995): MHgliche Komplikationen bei der Verwendung von Stimmprothesen. In:
Maier H., Johannsen H. S. (Eds.) Stimmrehabilitation nach Laryngektomie. Symposium
Bundeswehrkrankenhaus, U1m: 23-26
11. van den Hoogen F. J. A., Oudes M. J., Hombergen G., Nijdam H. F., Manni J. J. (1996): The
Groningen, Nijdam and Provox voice prostheses: a prospective clinical comparison based on
845 replacements. Acta Otolaryngol. (Stockh.) 116: 119-124
E.4 Deformation Behaviour
of Voice Prostheses-
Sensitivity of Mechanical Test Methods
c. Bierogel, 1. Bethge and W. Grellmann, Merseburg, and
E.-J. Haberland, Halle
1 Introduction
Polymer materials are of importance as substitute materials in medicine because
their individual properties are specifically adjustable. Their compatibility with the
biological system (biocompatibility) and the functionality and stability of the mate-
rials used are decisive factors for medical practice, especially for the purposes of
implants. The materials commonly used in medicine include PET (textile vascular
prostheses), PTFE, PUR (vascular replacements, blood pumps, cardiac valves),
polyolefines such as PP or PE (threads, joint replacement), silicone elastomers
(prostheses in the ear, nose and throat field, and special orthopaedic items) and poly-
amide. Pharyngo-tracheal valve prostheses, also referred to as voice prostheses, are
used for the rehabilitation of patients who have had their larynx completely removed.
2 Experimental
To determine the sensitivity of mechanical test methods for the purpose of the
modification of materials for voice prostheses, ten different prostheses were inves-
tigated. Materials consisting of silicone rubber, blends of different silicone rub-
bers, PUR (nine voice prostheses of type 'Halle' [1] and one reference prosthesis
of type 'Provoxil>,) were prepared for micro-hardness measurements, creep investi-
gations and technological tests for the characterization of dimensional stability.
Because of the different sizes of the prostheses, the dimensional-stability tests could
not be used for all specimens. In addition, for the characterization of the surface and
roughness, all materials and prostheses were investigated by microscopy.
For a comprehensive hardness measurement of a polymer, it is necessary to rec-
ord the indentation load and the penetration depth during the whole experiment
[2,3]. For this purpose, the penetration depth is measured at all times and therefore
the evaluation of loading and unloading curves is possible. Quantification or analysis
of the viscoelastic-plastic behaviour of the polymer can be performed by comparison
of the loaded and unloaded states. For this, measurements are normally performed in
a micro-hardness range lower than 2 N. However, a micro-hardness measurement
allows in principle only the analysis of the region near the surface, because the
maximum load is very small. For the studies of voice prostheses described here, a
commercial instrumented Fischerscope H100 micro-hardness system was used in
the range from 0.4 to 1000 mN. Figure E27 shows schematically the measuring
cycle for the determination of the load-penetration diagram.

472 C. BieriSgel, I. Bethge, W. Grellmann, E.-J. Haberland
Indenter
Specimen
Depth at Maximum Load
Fmm. F (N) Fmax

Fig. E27. Schematic illustration of detennination of the load-penetration diagram
The landing of the indenter on the specimen surface and the penetration are
controlled by a microprocessor forming part of the measuring system. The result-
ing penetration depth and the system time are measured simultaneously, as the load
increases. After the maximum load has been reached correctly, the penetration
depth hI can be obtained from the measured values of the plastic deformation hpl
and the elastic deformation hel of the material. Subsequently, a reduction of the
load to the predicted value Fmin is performed automatically by the measuring sys-
tem. At this point, the hardness is determined only by the residual plastic deforma-
tion of the material.
The hardness parameter H is calculated for the last load point used and corre-
sponds to the universal hardness [4]. Elastic and plastic parameters, including the
Young's modulus E/(1- J), the plastic hardness Hpl and the plastic deformation
energy of the material, were determined. To optimize the influence of the indenter
geometry, a ball indenter with a diameter of 1 mm was used.
An investigation of the creep behaviour was also performed by means of an in-
strumented hardness measurement using a Vickers indenter. Figure E28 shows the
measuring cycle schematically. After loading up to Fmax , the system records the
creep over a given time period at a constant load. For the materials investigated the
maximum load was chosen as 250 mN because the hardness of these elastomers is
low. To compare the creep at maximum and minimum load, the specimen was
unloaded down to Fmin and the creep of the penetration was observed. The time
increments for the determination of the penetration depth in the creep investiga-
tions were calculated as equal logarithmic steps.
The basic requirements for the 'pull-through' and 'lid elongation' tests is the
existence of sample holders suitable for the load cell of the universal testing sys-
tem. Hence, an adapter with a pin indenter and special clamps for the voice pros-
theses were manufactured as accessories for the Zwick 1464 testing system. Figure
E29 shows the arrangements developed for the two test methods schematically.
Deformation Behaviou r of Voice Prostheses 473
..c:
F = 250 mN
max
t.hL .et)
mm
t=b
~----------------~-------
Fmm. F F (N)
max
Fig. E28. Detennination of creep at maximum and minimum load
I: Maximum Load
Friction 2: Transition of Pull
Through to
Friction of Flange
b
I : Maximum Load
Friction 2: Transition of Lid
Elongation to the
Friction of Lid
3 : Friction
Fig. E29. Schematic illustration of (a) pull-through test and (b) lid elongation test
For the pull-through test of the voice prostheses, the end opposite to the lid was
passed through a clamp with a defined diameter adapted to the size of the prosthe-
sis. All prostheses were tested several times at a crosshead speed of I mmlmin to
determine the load and deformation. The lid elongation test using a pin indenter
was applied to determine the quasi-static lid motion at a crosshead speed of 2
mmlmin. The measured load and deformation are caused by elongation of the lid
and, above a maximum load, by the friction between the lid and the pin indenter. For
these investigations on different voice prostheses, a pin indenter diameter of 2.5 mm
was used.
474 C. Bierogel, I. Bethge, W. Grellmann, E.-J. Haberland
3 Results
3.1 Instrumented Micro-Hardness Tests
For comparison and evaluation of different voice prostheses by means of micro-

hardness measurements, the Young's modulus and plastic hardness are suitable
parameters. Both parameters were calculated from the measured load-penetration
diagram in the loading and unloading stages. Figure E30 shows the Young's
modulus for the ten voice prostheses investigated as a bar chart. It can be clearly
seen that prostheses 8 and 10 have the lowest levels of elasticity and numbers 4
and 5 have the highest levels.
Prosthesis Number
Fig. E30. Young's modulus obtained from micro-hardness measurements (prostheses 1-7,
silicone elastomer blends, 'Halle'; prosthesis 8, PUR, 'Halle'; prosthesis 9, silicone
rubber, 'Halle'; prosthesis 10, silicone rubber, 'Provoxill ,)
The differences in properties of the elastomer blends are important because they
show that the morphology has direct consequences for the micro-hardness. The
plastic hardness of prosthesis 4 is twice as big as the values for prostheses 6 and 7.
Figure E31 shows the material behaviour measured in the creep tests at the maxi-
mum load of 250 mN. A comparison between the micro-hardness parameters and
those obtained from the creep tests permits a correlation of these material parame-
ters. As expected, prosthesis 8 (PUR) shows a maximum for the creep because
there exists an inverse proportionality to the elastic properties. Indeed, prostheses
4 and 5, with the maximum plastic hardness, are also the materials with the lowest
tendency to creep (Fig. E3l).
Deformation Behaviour of Voice Prostheses 475
15
\0
o
Prosthesis Nwnber
Fig. E31. Results of creep tests at a load of250 mN and a hold time of 64 seconds
3.2 Technological Test Methods
The load that is necessary for the pull-through test under the chosen conditions can
be interpreted as a measurement of the strength of the shape of the voice prosthe-
ses. Analogously, a measure of the dimensional stability is obtained from the lid
elongation diagram. Therefore, the maximum load for each test method was de-
termined from the load-deformation curves (Fig. E32). For both test methods,
prosthesis 8, produced from PUR, requires the lowest load, which corresponds to a
poor dimensional stability. The highest load is necessary for the voice prostheses 4
and 5, whereby they have a high stability, but also a high resistance against im-
plantation (Fig. E32a). These results correspond directly to the micro-hardness
measurements.
3.3 Light Microscopy
The investigation of the voice prostheses by light microscopy showed that the
quality of the surface was influenced by the conditions of production and the
roughness of the production tool. The properties of the surface are important to the
transport of secretions and to the lifetime of the prostheses. Because of the lathe
tracks on the surface and the porosity found, a material optimization is recom-
mended.
476 C. Bierogel, I. Bethge, W. Grellmann, E.-J. Haberland
20r-----------------~
.-..15
Z
~
1 2 345 67 8 9 5 6 7 8 9
Prosthesis Num ber Prosthesis Number
Fig. E32. Maximum load for pull -through test (a) and lid elongation test (b)
4 Conclusions
These comprehensive studies show that the material parameters chosen allow the
differentiation and evaluation of voice prostheses regarding the materials and
service conditions used.
The determination of parameters describing the surface of the prostheses and
microscopic investigations are of importance if the surface is to be modified or
coated. Such modifications of the silicone rubber surface are necessary to improve
the transport of secretions and extend the lifetime.
The characterization of the loading behaviour by means of technological test
methods allows one to draw conclusions about the strength of the shape and about
the lifetime of voice prostheses, for example, the function of the lid.
References
1. Haberland E.-J., Neumann G., Lobe L. P., Voigt K. (1990): Stimmprothese. German Patent
DD 275183 AI , Aktenzeichen WP A 61 F, January 17
2. Frohlich F., Grau P., Grellmann W. (1997): Performance and analysis of recording micro-
hardness test. Phys. Stat. Sol. (a), 42: 79-89
3. Lopez I. (1993): Microhardness testing of plastics: literature review. Polym. Test. 12: 437-
458
4. Oliver W . c., Pharr, G. M. (1992): An improved technique for determining hardness and
elastic modulus using load and displacement sensing indentation experiments. J. Mater. Res.
7,6: 1564-1583
F
Special Materials
F.1 Crack Initiation, Wear and Molecular
Structure of Filled Vulcanized Materials
W. Grellmann, Merseburg, O. Heinrich, Hannover,
and T. Casar, Merseburg
1 Introduction
In their numerous areas of application, elastomer materials are subject to a number

of complex wear factors. A typical example of this phenomenon is the abrasion of
the tread of a tyre or the pattern of chipping and chunking damage observed under
certain circumstances on truck tyre treads [1].
Given that wear occurs as a result of contact between at least two surfaces be-
tween which friction is occurring, the wear properties of elastomers are determined
not only by their tribological properties, but also as a result of the taxonomy of the
circumstances of that wear, namely the tribo-structure and the set of tribological
load factors. The wear behaviour of the material itself is specified by characteris-
tics associated with its viscoelasticity, mechanical strength and mechanical-chemi-
cal degradation. These material-specific aspects are, fundamentally, determined by
the molecular structure of the polymer substances used [2].
Given that wear processes are related to material failure, fracture mechanics
properties such as crack toughness (resistance to crack initiation and crack propa-
gation) have a major importance for wear properties. Using a fracture mechanics
evaluation of crack toughness properties, under trial conditions relevant to the
material application being tested, it should be possible to characterize the material-
specific aspects of the wear properties of an elastomer [3,4].
This article looks into the wear behaviour and crack toughness of various filled
vulcanized materials. In the case of fracture toughness, an attempt is made to discuss
the results on the basis of molecular models of the structure of filled natural-rubber
networks. The point of departure in this respect is the link between the crack expan-
sion energy triggering the mechanically induced crack growth present, and the mo-
lecular structure and strength of the filled network.
2 Wear and Fracture Characteristics of Elastomers
The tribological behaviour of elastomers is influenced by the mechanical proper-

ties. of this group of materials. The term 'wear' is generally understood as referring
to the progressive loss of material from the outer surface of a solid body, brought
about by mechanical causes, i.e. due to contact with - and the relative motion of -
a solid, liquid or gaseous opposing body, with the degree of wear being propor-
tional to the frictional energy dissipated on the frictional contact surface [2]:

480 W. Grellmann, G. Heinrich, T. Cisar
(Fl)
where A is the amount of wear (e.g. volumetric abrasion rate), W is the frictional
energy (density) and n is a material and load-specific exponent.
In the consideration of the wear process, a basic distinction can be made among
four different wear mechanisms:
• adhesion (formation and destruction of adhesive bonds)
• abrasion (microscopic cutting, tearing and micro-ploughing)
• wear by fatigue (surface destruction caused by cyclic deformation)
• tribo-chemical wear (e.g. degradation due to frictional heat).
There are in fact usually several different mechanisms acting simultaneously,
although in varying contents, in the wear process. Elastomer materials exhibit pri-
marily mechanical abrasion and fatigue. The dominance of one or more mecha-
nisms depends on the application conditions, the material characteristics (Table
F 1) and the change that the friction triggers in the structure and properties [5].
It is evident from the factors influencing the wear behaviour (see Table FI) that
wear processes are systemic in nature. This circumstance complicates the quantifi-
cation of material-specific influences on the observed wear properties of elastomer
materials.
Table Fl. Material and load-specific factors influencing the wear properties of elastomers
Material-specific factors of influence
Polymer type • Chemical constitution, configuration, conformation

Influence on dynamic-mechanical characteristics
(glass transition Tg), crystallization, crosslinking
Filler content and • Volume content, structure, particle size

filIer characteristics Influence on strength, rigidity, hysteresis properties, crack
propagation behaviour
Density and type of • Influence on dynamic-mechanical characteristics, strength,

cross linking rigidity, hysteresis, degradation stability
Load-specific factors of influence
Intensity and distribution of forces on the contact surface

Relative speed of the items being rubbed together
Contact surface temperature
Surface roughness of items in frictional contact and shape of micro-protrusions
(e.g. sharp, rounded)
Crack Initiation, Wear and Molecular Structure 481
As a result of the combined effects and interaction of the aforementioned fun-

damental mechanisms, the following characteristic wear processes can be ob-
served, under different conditions, in elastomer materials [5]:
• patterned abrasion (formation of stationary abrasion patterns: Schallamach'
waves)
• abrasion in the form of incisions (furrows, grooves)
• microscopic particulate wear (chipping of particles, diameter 1-5 JlDl)
Microscopic particulate wear is the main process in wear occurring in the case
of sliding friction in alternating directions. If the direction of sliding friction is
constant, this mechanism is overridden, with increasing wear duration, by intense
patterned frictional wear. The formation of stationary patterns of frictional wear is
thus linked to a marked rise in the friction rate [6].
The models of the wear process mechanism are based on a model of frictional
contact between an elastomer and a rough surface. According to this model, rough
microscopic surface protrusions penetrate the frictional contact area of the surface
of the elastomer, deforming the elastomer and leading to the formation ofa stress-
strain field in the elastomer in the vicinity of the microscopic protrusions. The
cyclic deformations and dynamic stress fields caused by the sliding motion of the
microscopic protrusions over the surface of the elastomer can cause the formation
and propagation of cracks - in other words, damage and subsequent destruction of
the surface material [7].
In the case of a stress on the material caused by a low normal load (contact
pressure), there is a low-intensity stress field in the material in the area of the mi-
croscopic protrusion. This means that dynamic crack propagation gives rise to the
formation of (microscopic) frictional particles only after a large number of defor-
mation cycles. The abrasion rate (e.g. the volume of abrasion measured over time)
is low in this case, corresponding to wear by a fatigue mechanism. An abrasion
mechanism, on the other hand, is dominant when a high load is present, in which
case the material experiences high-intensity stress at the level of the material's
tensile stress, so that frictional particles are tom away after just one or a very few
deformation cycles. The abrasion rate is high in this case [7-10].
It can be observed that, because of the conditions in the frictional contact area,
wear by fatigue is influenced predominantly by the material's dynamic crack
propagation properties, while abrasion is influenced by the material's ultimate
properties, such as tensile stress, tensile strain at break and fracture energy [8].
Owing to the viscoelastic nature of the mechanical behaviour of elastomers, a
clear relationship can be observed between the frictional properties (and thus the
wear properties) of the material, on the one hand, and the sliding speed and tem-
perature of the contact surface, on the other. This is because the sliding motion of
the microscopic protrusions over the surface of the elastomer corresponds to a
dynamic-mechanical excitation of the material.
The dynamic stiffness increases with increasing sliding speed and decreasing
temperature, so that the reaction of the material to mechanical contact with the
microscopic protrusions changes from predominantly plastic to primarily elastic. In
parallel with this transition, there is normally a change in the wear mechanism from
one of wear by fatigue to abrasion, along with an increase in the abrasion rate [8-10].
482 W. Grellmann, G. Heinrich, T. Casar
Fillers such as carbon black are generally used to increase the strength and
stiffness of elastomer materials. The addition of filler also affects wear properties.
For example, an increase in tensile stress and tensile strain at break can, in the case
of mainly abrasive wear, cause a lower rate of abrasion [II]. In the case of wear
that is mainly due to fatigue, adding carbon black increases the filler-to-filler and
polymer-to-filler interactions, resulting in intensified hysteresis and, consequently,
a drop in the rate of abrasion, owing mainly to a decrease in the rate of dynamic
crack propagation. Adding carbon black also promotes crack-branching and crack-
blunting mechanisms and influences the material's degradation behaviour [12].
As mentioned in the introduction, prediction of the general damage behaviour
of elastomer materials should be performed using a fracture mechanics evaluation
of the crack toughness behaviour. For this purpose, certain aspects of the fracture
behaviour of elastomers are discussed here.
Elastomers have a molecular structure with long-chain, wide-mesh crosslinked
molecular chains and thus they behave in a highly elastic fashion. Adding fillers
(e.g. carbon black) initiates a transition to non-linear viscoelastic and viscoplastic
material behaviour. Depending on the properties of the filler and its amount, a
marked hysteresis can be observed in a dynamic experiment.
The stress-strain characteristics of unfilled and filled polymer networks are de-
scribed by the constitutive equations of a theory of rubber elasticity [13].
The concepts of elastic-plastic fracture mechanics are suitable for describing
the crack toughness behaviour of these material systems. Energy-based parameters
provide an especially suitable measure of fracture mechanics characteristics, in-
cluding, for example, those based on the J-integral concept [14,15]. Parameters
determined by deformation (e.g. the CTOD) are less accessible, owing to experi-
mental difficulties, but are also usable [14].
Fracture mechanics parameters based on the concept of the stress intensity fac-
tor are, on the other hand, not appropriate for describing crack toughness behav-
iour, because the stress field in front of the crack tip is not characterized by the
stress intensity factor, i.e. there is a clear deviation from the r- 1I2 stress character-
istic [15,16].
On the basis of an overall energy balance for a growing crack, Rivlin and
Thomas [17] identified the crack energy factor T (F2) as a fracture mechanics
parameter for elastomer materials, with T being the change of the potential energy
U in a test specimen caused by an amount of crack propagation Aa. This defmition
is formally identical to Rice's defmition of the J-integral [18], which is based on
an energetic interpretation. The two parameters should thus be regarded as
equivalent, although experiments may produce differences in the parameters as a
result of the different methods of approximation applied in determining T and J.
The equation defming Tis
T = J = - ( :) = - ~ ( :~). (F2)
where B is the specimen thickness.

Due to the materials' viscoelastic defonnation behaviour, the crack toughness

behaviour of elastomer materials is temperature- and speed-dependent [19]. With
increasing crack propagation speed, the crack propagation behaviour undergoes a
transfonnation, which manifests itself in the fonn of an alteration in the fracture
surface morphology, which evolves from rough, through a fonn typical of stick-
slip behaviour, to smooth. The rough fracture surface occurs in this case owing to
crack propagation along the cavities fonned in the multi-axis stress field in front of
the crack tip. As the crack tip expansion rate rises, the occurrence of cavitation is
increasingly hindered, producing a smooth fracture surface [20].
Alongside these external factors, there are also material-specific factors - such
as polymer type (e.g. the molecular structure of the material), crosslinking or filler
addition - that have an influence on the crack toughness behaviour. The level of
the glass transition temperature Tg, which is detennined by molecular structure,
influences, in this connection, the extent of the internal damping (hysteresis) of the
material, which in turn exercises a marked influence on the crack toughness prop-
erties.
While hysteresis has only a marginal influence on crack initiation behaviour, it
does clearly affect the crack propagation characteristics. A crack propagation step
leads to an unloading of the crack edge, which causes a time-delayed reverse de-
fonnation of the crack edge. Consequently, a brief drop in the stress intensity oc-
curs and, thus, a reduction in the energy release rate [19]. An increase of hysteresis
consequently leads to an increase in crack toughness against crack propagation. In
line with this, a reduction in crack growth rate, with increasing hysteresis, can be
observed in a dynamic crack growth experiment with constant loading AT [12].
In crosslinked elastomers, the type of crosslinking can be observed to have an
influence on crack toughness, with polysulphide-crosslinked systems enjoying
advantages over peroxide-crosslinked systems. The high degree of flexibility in
comparison with C-C links and the low amount of fracture energy in sulphur links
pennit an efficient recombination of broken molecular chains (fault points). This,
in turn, allows for a reduction of stress peaks and thus leads to a more homogene-
ous stress state in the material [21,22].
The addition of fillers (especially carbon black) greatly influences the crack
toughness of elastomers. Here can one generally observe two significant influences
on fracture behaviour:
1. Adding filler can be observed to trigger an enlargement of the process area in
front of the crack tip, with an increase in the volume of intensely defonned ma-
terial and thus also in the amount due to hysteresis present in the general defor-
mation energy [15]. Depending on the magnitude of the defonnation, the poly-
mer-to-filler and filler-to-filler interactions produce hysteretic effects.
2. A drastic increase in the tearing energy Tc resulting from distinctive crack-
deviation and crack-blunting mechanisms ('knotty tearing') can be detected in
the case of filled elastomers. This leads to a reduction in the intensity of the
stress field in front of the crack tip. These mechanisms are triggered by the
strong anisotropy usually present at the crack tip as a result of a high stretching
and orientation of the molecular chains and the formation of bunched mole-
cule/carbon black aggregate strings (super-networks). The occurrence of blunt-
ing mechanisms is normally limited to a narrowly defined range of temperatures
and speeds, and is influenced by the density and type of crosslinking, the type
of carbon black and the stress state [12].
The higher the filler (carbon black) content, the greater the crack toughness
against crack initiation, whereby - in line with what has been said above - the
crack initiation behaviour is influenced much more by the filler's reinforcement
effect relating to UB and E t than by any increase in hysteresis (no load release until
crack initiation). This is also shown by investigations in which the critical defor-
mation at the crack tip during crack initiation is both constant and independent of
the increase of carbon black content of the material. The increase in crack tough-
ness relative to crack initiation observable in this case thus results primarily from
the increase in tensile stress at break and the concomitant augmentation of the
deformation energy in the process zone in front of the crack tip [15].
Three different ranges can be distinguished with regard to the dynamic crack
growth behaviour of elastomers as a function of loading:
1. Below a certain critical stress level (denoted as the 'threshold tearing energy
To'), no mechanically induced crack propagation can be observed.
2. Once To is exceeded, the rate of crack propagation immediately increases line-
arly with the difference T-To.
3. In the presence ofa moderate crack energy, there is, in line with the Paris equa-
tion, an exponential relationship between the rate of crack propagation and the
tearing energy present (F3). This applies prior to attainment of the critical tear-
ing energy level Te for the commencement of unstable crack propagation (in the
case of a high level of tearing energy) [12]. The equation is
da = CilTfJ (F3)
dN '
where P is a material-specific exponent, C is a constant and N is the number of
deformation cycles.
The threshold tearing energy To lies within a characteristic range of approxi-
mately 50 Jm-3 to 100 Jm-3 [21,22] for different elastomers. At the same time, To
shows no significant dependence on loading conditions.
According to the LakelThomas theory [23], the value of To is determined solely
by characteristic features of the molecular structure of the material, such as the
chemical molecular weight of a network chain Me:
TO f':SkJM c . (F4)
An attempt is undertaken in Sect. 4 to expand on the view stated here concern-

ing (F4). This is done in the light of more recent knowledge of the characteristics
of filled natural-rubber networks and with reference to the influence of filler-
polymer junctions.
3 Experimental
The materials used in the experimental investigations described in this chapter

were filled vulcanizates based on the polymers SBR 1500 (statistical styrene-bu-
tadiene copolymer containing 23 wt.% styrene), NR (natural rubber) and BR
(high-cis butadiene rubber), with carbon black contents of 20, 40, 60 and 80 phr
(parts per hundred of rubber). A silica-filled SBR compound was used to compare
the influences of carbon black and silica fillers (Table F2) on the mechanical be-
haviour and wear characteristics.
An overall basic mechanical classification was performed for all materials (ten-
sile test, etc.) and an abrasion test corresponding to DIN 53516 was carried out.
Table F2. Test programme for tribological and mechanical investigations
Compound 2 3 4
Polymer SBR 1500 BunaCB 10 NR(TSR) SBR 1500

Carbon black 20,40,60,80 20,40,60,80 20,40,60,80 14
N339'
Silica VN3' 70
Activator X50S' 12
Aromatic 20-60 phr 20-60 phr 20-60phr 35
mineral oil' carbon black: 5 carbon black: 5 carbon black: 5
80phr 80phr 80phr
carbon black: 35 carbon black: 35 carbon black: 35
'All quantities shown are in phr (parts per hundred of rubber)
Table F3. Specimen geometry and procedure used to determine J c
Specimen • Single-edge-notched tension specimen (SENT)

Geometry • Length L: 75 mm
Width W: 30 mm
Thickness B: 2 mm
Span Lo: 40 mm
Notchdepth • 8,10,13,15, 17 mm (a/W= 0.27-0.57)
Crosshead speed Vr • 10 mrnImin
Conditioning • 3 deformation cycles
of the specimen (max. strain: 100 %; Vr = 200 mrnImin)
Observation of • Notch previously sprinkled with white powder;
crack initiation observation of the notch until a new crack becomes visi-
ble along the entire specimen width
External conditions • 23°C and 50 % air humidity
The parameter used in the fracture mechanics characterization of the crack

toughness properties of the vulcanized rubbers was the J value at the point of crack
initiation Je, as determined by the Begley/Landes method for single-edge-notched
tension specimens [24]. The crack initiation was recorded by observation of the
growth of a prepared notch. Stable crack propagation occurred - without exception -
after crack initiation, so the crack toughness value Je can be regarded as a measure
of resistance against stable crack initiation. Further data concerning the procedures
used for determining the crack toughness against crack initiation J e are given in
Table F3 [3].
Figures F I-F4 show the results of the experimental investigations carried out on
these vulcanized materials. Figure Fla indicates the tensile stress as a function of
carbon black content as determined in a test on standard rings. As the content of car-
25~--------------------------~~
20
/
~15
~
g'10
~- SBR+Carbon Black
5 l),.~ -~ SBR+Silica
____ -l),.-BR
l),. -'\1- NR
O~~--~--~--~----~--~--~~
20 30 40 50 60 70 80
Filler Content (phr)
500
400
~300
t3
200
100
20 30 40 50 60 70 80
Fig. Fl. Tensile stress 011 (a) and tensile strain at break t'll (b) as a function of filler content,
detennined in a tensile test on standard rings
bon black increases, the tensile stress increases for all materials. In the cases of the
SBR and NR compounds, however, the maximum stress is observed at a carbon
black content of60 phr. This is caused by the high content of mineral oil present in
the 80 phr compounds (softening effect). The markedly greater tensile stress at
break of the NR compounds with lower carbon black content can be attributed to
the ability of NR to crystallize under the influence of strain and thus provide self-
reinforcement of the material. This reinforcement effect disappears as the content
of carbon black increases; this is caused, on the one hand, by increasingly hindered
crystallization and, on the other hand, by a coming together of the losses due to
hysteresis in the SBR and NR compounds. The values of tensile strain at break
show, compared with tensile stress at break, a low dependence on carbon black
content. In this case, the tensile strain at break drops for the SBR and NR com-
pounds (Fig. FIb). A different process can be observed in the BR compounds,
where the tensile stress and strain at break steadily increase in line with the in-
creasing content of carbon black. The greater tensile strain at break of the
SBRIsilica compound compared with the SBRIcarbon black compound is attrib-
uted to the more pronounced filler/filler interaction in the silica compound, which
hinders phase separation between the filler particles and the polymer matrix and
thus helps prevent the formation of secondary cracks.
Figure F2 shows the dependence of the measured rebound resilience on carbon
black content for these materials. The rebound resilience represents the elastic part
of the deformation energy to which the material is subject (expressed as a percent-
age) and therefore shows a behaviour opposite energy dissipation caused by hys-
teresis. Figure F2 thus clearly shows that an increase in the content of carbon black
is accompanied by an increase in the dissipative part of the deformation energy, as
the rebound resilience drops. With a decrease in the glass transition temperature Tg
of the material (SBR, -49 DC; NR, -64 DC; BR, -104°C; Tg measured using DSC
and operating with a heating rate of 10 Klmin [25]), there is a steady increase in
the elastic energy component, caused by a decrease in the dissipative component.
80.----------------------------.
6___ -0- SBR+Carbon Black
~70 '\l
~
---~. SBR+Silica
-6-BR
~ -'\l-NR
0 ___________ v~~.
0 ______ ~
o~
o
20~~--~--~--~--~---L--~~
20 30 40 50 60 70 80
Fig. F2. Rebound resilience as a function of filler content

175 -0- SBR+Carbon Black
""g
E 150 ~ • SBR+Sllica
-t::.-BR
-V-NR
Q) 125
~~
E
-= 100
~
a 75
'iii
~ 50
~ 25 ~t::. t::.
o 20 30 40 50 60 70 80
Fig. F3. Volwne loss due to abrasion in the DIN frictional-wear test as a function of filler content
Figure F3 shows the results of wear investigations based on DIN abrasion tests.
As the content of carbon black increases, an immediate and clear drop in abrasion
loss is detected for all materials. At carbon black levels in excess of 60 phr, how-
ever, the abrasion loss increases slightly once more; this is attributable to the influ-
ence of the increased oil content. The DIN abrasion test shows a high level of wear
load, accompanied by the observation of mainly abrasive wear.
Using the data from Sect. 2, one can thus identify a reciprocal relationship, for
the SBR and NR compounds, between the tensile stress at break and the abrasion
loss, Le. a decrease in abrasion loss related to an increase in tensile stress at break.
Completely different behaviour is detected in the BR compounds, which - de-
spite their low tensile stress at break (Fig. Fla) - show a remarkably low level of
abrasion loss. This implies that the mechanism of mild wear by fatigue is primarily
active. In order to explain this observation, it can be supposed that, owing to the
significantly lower glass transition temperature of BR, the dynamic stiffening of
the material related to the wear factors (load, frequency, temperature) is much
lower than that present in the SBR and NR compounds. This means that BR re-
sponds to contact with the microscopic protrusions primarily in an elastic manner,
while the other materials show mainly 'plasticity-based' responses. This may ex-
plain the wide difference between the abrasion loss rates for SBRINR and BR.
The fracture mechanics investigations using quasi-static loading produced the
results shown in Fig. F4 for the crack toughness related to resistance against stable
crack initiation as a function of filler content. The large standard deviation of the
results obtained (ll.Jc '" 1.5 N mm- I ) allows only conclusions in terms of minimum
and maximum Jc values. This means that the lowest and highest Jc values were
measured at 40 phr and 60 phr for all materials.
The decrease in Jc in the case of vulcanized materials filled to 80 phr can be ex-
plained in terms of the increased oil content, which - owing to its softening effect
- reduces the critical intensity of the stress-strain field in front of the crack tip. At the
minimum level (40 phr), the BR compound shows the least resistance against crack
initiation. This minimum occurs in each series of compounds, so it should be possi-
5
-0- SBR+ Carbon Black
4
• SBR+Silica
-A-BR
-V-NR
E g
.€ 3 0
~
~2 0
A
•
0
20 30 40 50 60 70 80
Fig. F4. Critical Jvalue Jc at the point of visually detennined crack initiation in the quasi-static
tensile test
ble to rule out any experimental error, and - in terms of the elastomers examined
here at least - one can assume that the effect is generally valid. As an explanation
of the observed relationship between crack toughness and carbon black content,
we have formulated a supposition that there exist two different physical ranges,
which we designate 'I' (low filler content: up to approximate 40 phr) and 'II'
(filler content above 40 phr). In range II, it is apparent that the critical percolation
threshold for the occurrence of one or several fractal filler networks has been
crossed [26]. The related factors of hysteresis loss, crack branching and crack
deviation mechanisms [27-29] explain why the resistance against both crack ini-
tiation and crack growth increases with increasing filler content. These 'mecha-
nisms' vary considerably relative to the filler content. Their effect is based on the
drop in stress concentration in front of a crack tip and/or the division of the tearing
energy among various potential crack tips.
In a highly filled silica-vulcanized material, the hysteresis losses are much
lower in comparison with carbon black-filled vulcanized material, which can be
attributed to the stronger silica-to-silica interaction in the filler network. This can
account for the lower Jc value of the silica-vulcanized material relative to carbon
black-filled materials.
In range I, the filler agglomerates linked to the polymer matrix can be consid-
ered as elastically active, multi-functional junction areas in the polymer matrix.
For further general discussion of this case, reference should be made to the Lake-
Thomas theory for the tearing energy To related to mechanically induced crack
propagation in a network of Gaussian chains [23]:
-1I2
(
To ~ _1_+_1_+_1_ J (FS)
R2c R2e R2f '
where Rc is the average end-to-end distance of a network chain between two

chemical cross links, Re is the average end-to-end distance of a network chain be-
tween two elastically active entanglements and Rf is the average end-to-end dis-
tance of a network chain between two filler particles.
The original Lake-Thomas theory considers only the influence of network
chains between chemical crosslinks on crack tip enlargement and crack propaga-
tion (Sect. 2). For this case, one can obtain from (F5), assuming Gaussian chain
statistics, the expression
". _ R - b·",112 _ 11- 1/2 _ MII2
~0 - c - He - roc - C , (F6)
where b is the statistical segment length, Ne is the number of statistical segments in

a chain, Pc is the chemical-crosslink density and Me is the molecular weight of a
chain network.
The influence of polymer chain entanglements on To has been discussed by
Mazich and Samus [30). One can use the tube model of rubber elasticity theory as
a basis and thus defme the contour length of a section of a Gaussian chain between
two successive entanglements. In this case Le = do2/b (maximum stretching of the
chain section), where do is the average end-to-end distance of the section of chain,
which is identified with the diameter of the conformational tube [13,31). This pro-
cedure provides the necessary explanation of the influence of entanglements on the
apparent correlation between
(1) the stretching at fracture Acraze, which can be determined, for example, by a
transmission electron microscope (TEM) examination of different polymer
films below the glass transition temperature, along with the mechanical stress
ratio in the fibrils and the surface stress of the crazes [32,33), and
(2) the maximum stretching A.n.x = Lefdo = dolb of a Gaussian chain section between
two chain entanglements. The statement applicable is Acraze - Amax, where the
factor of proportionality is somewhat greater than one.
From (F6) it is immediately apparent that with an increasing content of filler in
range I (increasing filler crosslink density), the average end-to-end distance of a
network chain between two filler particles Rf , the effective network chain length
and thus the threshold tearing energy decrease. This then indicates that a reduced
crack toughness will be found as a result of the diminished local stretching ability
in the area of a stress concentration. It must be assumed here that the chemical-
crosslink density and entanglement density remain approximately constant with
increasing carbon black content [34).
5 Summary
The decisive factor in the case of the wear properties in the DIN abrasion test is
the mechanical characteristics of the frictional contact surface. If, under these cir-
cumstances, the reaction of the vulcanized material to contact with the rough sur-
face can be characterized as 'plastic', then the abrasion loss is inversely propor-
tional to the tensile stress at break of the material. If the reaction is mainly elastic,
the abrasion loss is proportional to the resistance against dynamic crack propaga-
tion (fatigue). The molecular structure and dynamic-mechanical characteristics of
the vulcanized material determine the type of reaction. This means that under these
circumstances, with an increasing glass transition temperature, there is a transition
from an elastic to a plastic reaction to frictional contact.
The dependence of the crack toughness related to resistance against stable crack
initiation on carbon black content follows two patterns (ranges). The addition of
carbon black reduces crack toughness (range I) by increasing the node density of
the filler network. Once the percolation threshold for the occurrence of one or
more filler networks is crossed, the crack toughness increases with carbon black
content (range II), owing to an increase in hysteresis. An interpretation of the re-
sults of fracture mechanics investigations was made on the basis of the molecular
performance regarding the structure and strength of filled natural-rubber networks.
This was centred on extending the Lake-Thomas theory of the threshold tearing
energy To for the case of mechanically induced crack growth.
References
I. Udagawa Y., Kawakami (1990): Structural changes in rubber vulcanizates caused by differ-
ent types of defonnation. In: Proceedings of the Rubber Division Meeting, ACS, Washington
DC, USA, October 9-12, Paper No. 62
2. Ward I. M. (1971): Mechanical Properties of Solid Polymers. Wiley-Interscience, London
3. Heinrich G., Kramer T., Grellmann W. (1996): Crack initiation, wear and molecular struc-
ture of filled vulcanized materials. In: Proceedings of 'Kautschuk-Herbstkolloquium '96',
Hannover, Gennany, October 24-26: B 1-8
4. Grellmann W., Casar T., Heinrich G. (1999): Mechanical properties, crack initiation and
molecular structure of filled polymeric materials. Kautsch. Gummi Kunstst. 52: 37-43
5. Gent A. N. (Ed.) (1992): Engineering with Rubber. Hanser, Munich, Vienna
6. Gent A. N. (1989): A hypothetical mechanism for rubber abrasion. Rubber Chern. Techno!.
60:750--756
7. Schweitz J.-A., Ahman L. (1986): Mild wear of rubber-based composites. In: Friedrich K.
(Ed.) Friction and Wear of Polymer Composites. Composites Material Series, 1, Elsevier,
Amsterdam
8. Veith A. G. (1992): A review of important factors affecting treadwear. Rubber Chern. Tech-
no!. 65: 601-658
9. Zhang S. W. (1984): Mechanisms of rubber abrasion in unsteady state. Rubber Chern. Tech-
no!. 57: 755-768
10. Gent A. N., Pulford C. T. R. (1983): Mechanisms of rubber abrasion. J. App!. Polym. Sci.
28:943-960
11. Thavamami P., Bhowmick A. K. (1994): Universal relationship between abradability and
breaking energy, shear loss modulus and ridge spacing. Rubber Chern. Techno!. 67: 129-136
12. Lake G. J. (1995): Fatigue and fracture of elastomers. Rubber Chern. Techno!. 68: 435-460
13. Heinrich G., Straube E., Helmis G. (1988): Rubber elasticity of polymer networks: theories.
Adv. Polym. Sci. 85: 33-88
14. Lee R F., Donovan J. A. (1987): J-integral and crack opening displacement as crack initiation
criteria in natural rubber in pure shear and tensile specimens. Rubber Chem. Techno!. 60:
674-688
15. Liu H., Lee R F., Donovan J. A. (1987): Effect of carbon black on the J-integral and strain
energy in the crack tip region in a vulcanized natural rubber. Rubber Chem. Techno!. 60:
893-909
16. Fukahori Y., Seki W. (1994): Stress analysis of elastomeric materials at large extensions
using the finite element method Part ID. J. Mat. Sci. 29: 2767-2774
17. Rivlin R S., Thomas A. G. (1953): J. Polym. Sci. 10: 291ff
18. Rice J. R. (1968): Trans. ASME, Ser. E 35: 4ff
19. Andrews E. H. (1963): Reinforcement of rubber by fillers. Rubber Chem. Techno!. 36:
325-336
20. Kadir..A., . Thomas A. G. (1981): Tear behavior ofrubbers over a wide range of rates. Rubber
Chern. Techno!' 54: 15-24
21. Chun H., Gent A. N. (1996): Strength of sulphur-linked elastomers. Rubber Chem. Techno!.
70:577-590
22. Bhowmick A. K., Gent A. N., Pulford C. T. R. (1983): Tear strength of elastomers under
threshold conditions. Rubber Chern. Techno!. 56: 226-232
23. Lake G. J., Thomas A. G. (1967): Proc. R. Soc. London, Ser. A 300: 108ff
24. Begley 1. A., Landes J. D. (1972): The J-integral as a fracture criterion. In: ASTM STP 514:
1-20 and 24-29
25. Heinrich G. (1992): Dynamics of carbon black filled networks, viscoelasticity, and wet skid
behaviour. Kautsch. Gummi Kunstst. 45: 173-180
26. Kitlppel M., Heinrich G. (1995): Fractal structures in carbon black reinforced rubbers. Rub-
ber Chern. Techno!. 68: 623-651
27. Lake G. J., Lindley P. B. (1964): Ozone cracking, flex cracking and fatigue of rubber. Rub-
berJ. 146, 10:24-30; 11:30-36
28. Gent A. N., Pulford C. T. R (1984): Micromechanics of fracture in elastomers. J. Mat. Sci.
19: 3612ff
29. Hamed G. R. (1994): Molecular aspects of the fatigue and fracture of rubber. Rubber Chern.
Techno!. 67: 529-536
30. Mazich K. A., Samus M. A. (1990): Role of entanglernent couplings in threshold fracture of
a rubber network. Macromolecules 23: 2478ff
31. Edwards S. F., Vilgis T. A. (1988): The tube model of rubber elasticity. Rep. Prog. Phys. 51:
243ff
32. Donald, A. M., Kramer E. 1. (1976): 1. Polym. Sci.: Polym. Phys. 14: 170lff
33. Dettenmaier M. (1983): Intrinsic crazes in polycarbonate. Phenomenology and molecular
interpretation of a new phenomena. Adv. Polym. Sci. 52153: 57-104
34. Heinrich G., Vilgis T. A. (1993): Contribution of entanglernents to the mechanical properties
of carbon black filled polymer networks. Macromolecules 26: 1109-1119
F.2 Investigation of Crack Propagation
Behaviour of Unfilled and Filled
Vulcanizates
K. Reincke, R. Lach, and W. Grellmann, Halle,
and G. Heinrich, Hannover
1 Introduction
Elastomer materials are used in tyres, conveyor belts, engine bearings and isolation
materials, for example. Definite requirements for the material properties depending
on application field exist. For tyres, in addition to safety, economic and ecological
aspects, wet-skid resistance, rolling resistance and wear resistance are important
properties. A predictable lifetime of products is necessary; this is, however, restricted
by an inadequate control of wear phenomena.
In the case of passenger car tyres, the main wear phenomenon is fatigue wear,
leading to abrasion losses within the tyre tread. This is caused by initiation and
propagation of cracks, which is the reason why fracture mechanics concepts are
used for material assessment in order to obtain an increase of lifetime. Therefore,
the use of fracture mechanics methods has a very high economic significance, e.g.
in modem transport technology. Because the application of fracture mechanics
concepts has been proven to be useful for establishing morphology-property cor-
relations for thennoplastic materials and resins (pp. 3-26, [1,2]), the use of such
methods for the evaluation of elastomers would seem to be promising.
With the help of the examination described below, it should be possible to as-
sess elastomer materials by various fracture mechanics methods.
2 Experimental
2.1 Examination Methods
2.1.1 Dynamic Loading (Fatigue)
Examinations concerning the fatigue behaviour of vulcanizates were perfonned

using a 'tear and fatigue analyser' (TFA) measuring system [3,4]. This is a servo-
hydraulic testing machine specially developed for the examination of crack growth
behaviour of elastomer materials under cyclic loading conditions. The tests were
carried out at a frequency of 30 Hz, with four pulses per second and constant pre-
strain of 22.5 %. The recording of measured parameters such as the crack length
aeff, the number of cycles N, and the elastic and dissipated energy densities Wei and

494 K. Reincke, R. Lach, W. Grellmann, G. Heinrich
Wdiss , respectively, took place with an interval of 2 minutes. As a result of these

tests, crack propagation curves for each material were determined (Fig. F5). In Fig.
F5, range I is connected with stable crack propagation, and range II corresponds to
accelerated stable crack propagation which changes to an unstable state at a mate-
rial-dependent critical point, from which the critical value of the tearing energy Tc
was estimated. The fracture mechanics parameter T, derived from a global energy
balance for a propagating crack, represents the change of the potential energy U of
a test specimen caused by an amount of crack growth !l.a. This defmition is for-
mally identical to the energetic interpretation of the J-integral, and therefore both
parameters can be considered as equivalent (pp. 479-492, [5]).
60000
N
Fig. F5. Example of a crack propagation curve recorded in a TFA test
2.1.2 Impact-Like Loading
Elastomers are highly flexible materials. Therefore, the fracture mechanics be-
haviour cannot be characterized by the instrumented Charpy impact test (ICIT, pp.
71-86) that has been successfully applied to such materials as thermoplastic poly-
mers. However, instrumented tensile-impact tests can be used to investigate even
such flexible materials as elastomers under impact-like loading conditions. Figure
F6 shows schematically the experimental set-up of the instrumented Zwick pen-
dulum used and also the analysis of the load-extension diagrams recorded. The
principle is that a specimen is fIxed between the unsecured crosshead and a se-
cured clamp. Then the specimen is loaded by the pendulum hammer, which hits
the unsecured crosshead, and so the specimen is strained in the direction of its
longitudinal axis until it tears. At the same time the load-time curve is recorded
and, afterwards through double integration, the load-extension curve is calculated.
The tests were carried out with a pendulum hammer speed of 2.9 mls and an
initial gauge length of 30 mm at room temperature. For each vulcanizate the load-
extension diagrams of six specimens were recorded and analysed.
Crack Propagation Behaviour ofVulcanizates 495
In analogy to the ICIT, characteristic parameters, such as the time to fracture fB,
the maximum load Fmax with its related maximum extension [max, and the two ener-
gies Amax and A p, are used for analysis. Using the integral energy, which consists of
Amax plus A p, the J values J dBL were determined according to the evaluation method
of Begley and Landes (F7):
BL 1]{Amax + Ap )
Jd = B(W-a) , (F7)
where
1] = ~[tan(;r
;ra W
a) + O.lsin (~)l.
O.5W
(F8)
Fig. F6. Schematic representation of the 'instrumented tensile-impact test' measuring system
and analysis of recorded F-/ diagrams
2.1.3 Quasi-Static Loading
Besides the examinations described above, the possibility of recording crack re-
sistance (R-) curves using a quasi-static fracture mechanics test was examined. For
this purpose, ten double-edge-notched tension (DENT) specimens per material
were used. Following the stop-block method (pp. 71-86) of the ICIT, the tests
were stopped after reaching different strain values to produce different amounts of
crack growth. In some cases unstable fracture of the specimens occurred; in other
cases, subsequent splitting of the specimens using the tensile-impact test with a
high speed was necessary so that the fracture surfaces could be examined with
light microscopy to measure the amount of stable crack growth. From the recorded
load-extension curves, energies for calculation of the loading parameter J for use
in the R-curve were determined. Here, the same evaluation method (F7) was ap-
plied as in the analysis of the impact tests. Finally, the J-D.a data were plotted on a
496 K. Reincke, R. Laeh, W. Grellmann, G. Heinrich
graph, and regression functions were fitted and then used for subsequent analysis.
Ascertainable parameters of the crack resistance curves include, for example, the
technical crack initiation values JO.2 determined according to ESIS TC 4 recom-
mendation [6], and the slope of the R-curve dJ/dAa, which supply quantitative
criteria for a comparison of materials.
2.2 Materials
The composition of the vulcanizates examined is shown in Table F4. The basis of
the materials is the statistical styrene-.:butadiene copolymer SBR 1500 with a sty-
rene content of about 23 wt. %. Crosslinking was performed with a sulphur-accel-
erator system. The sulphur content was varied in the range from 0.8 to 2.4 parts
per hundred rubber (phr) to investigate the influence of the crosslink density on the
properties. For assessment of the influence of the filler, vulcanizates with a con-
stant sulphur content of 1.6 phr but with different carbon black contents in the
range from 0 to 50 phr were produced. Vulcanizates of this kind are used in vari-
ous fields, for example in tyres and conveyor belt materials.
Table F4. Elastomer materials examined
Mixture SBR 1500 Carbon black S Accelerator

N330 (pbr) (pbr)
(pbr) (pbr)
1 100 0.8 0.8

2 100 1.2 1.2
3 100 1.6 1.6
4 100 2.0 2.0
5 100 2.4 2.4
6 100 15 1.6 1.6
7 100 30 1.6 1.6
8 100 40 1.6 1.6
9 100 50 1.6 1.6
,/
/
?'B L al2
W
I al2
L
Fig. F7. Specimen geometry
The test specimen shape used for the TFA tests corresponds to a sing1e-edge-
notched tension (SENn specimen geometry with dimensions L = 64 mm,
W = 15 mm and B = 1.5 mm, and a initial crack cut with a depth of about 1 mm. For
the tensile-impact tests and the quasi-static fracture mechanics examination, speci-
mens of the geometry shown in Fig. F7 with the following dimensions were punched
from plates about 1.5 mm thick and subsequently cut in two (DENT specimens):
B ~ 1.5 mm, L = 64 mm, W = 15 mm, and a ~ 4 mm for tensile-impact tests or a
~ 6 mm for quasi-static fracture mechanics tests.
2.3 Basic Mechanical Characterization
Within the scope of the basic mechanical characterization of the materials, various
parameters were determined, including the tensile stress at break O'B, tensile strain
at break "B and conventional tensile-impact strength En (Figs. F8 and F9).
2.0 800
a
~
~m
cu
-
e 1.5 600 .:.:
'"
III ra
cu ......
ra I!!
III
ilia..
-....."
:g
~
~
en
'iii
~1.0
In
0.5 CJ
400 iii
200 .J!!
~
en
I:
I:
Q) 'iii
I:
l-
~
0.0 0
1.0 1.5 2.0 2.5
Sulphur Content (phr)
30
~
l!,...
~ m
-
ra
e
III
20
400 .:.:
ra
'"
cu ......
cu 2!
ilia..
-e ....."
IXl
III~
10 iii
I:
en In
200
~
'iii
~
c .J!!
0 'iii
~ c
Q)
0 I-
a 10 20 30 40 50 60
Carbon Black Content (phr)
Fig.F8. Parameters measured in the quasi-static tensile test as a function of (a) sulphur
content and (b) carbon black content
With increasing sulphur content the tensile stress at break decreases linearly,
but the tensile strain at break is lowered from a value of about 600 % to 200 %
because of the increasing crosslink density (Fig. FlO), from which a decreasing
network mobility follows. The conventional tensile-impact strength remains at
approximately the same level up to a sulphur content of2 phr. For a higher sulphur
content, a reduction in En is observed. An increase of the carbon black content
leads to a continuous increase of the tensile stress at break and of the conventional
tensile-impact strength over the whole range investigated and the tensile strain at
break shows a maximum at 30 to 40 phr (Fig. F8).

o 10 20 30
300.-.---.----,---,----.---.---~
40 SO 60
£Cl 2S0
c:
~
en .r200
t) .E
~ ~ 1S0
E
ji w" 100
'iii
c:
{!!. SO
OL---~-----J------~-----L __~
1.0 1.S 2.0 2.S
Fig. F9. Conventional tensile-impact strength as a function of sulphur and carbon black
contents

o 10 20 30 40 SO 60
2.0 .--,--..,-----.-----,-----,.---,---,----,
Physical Network Density for:
• Increasing Sulphur Content (1)
","'" 1.S . Increasing Carbon Black Content(2 a, 2 b)
.E • ;0 . Primary Network 2b
~ 1.0 • . Secondary (Filler) Network
o......
'-'
c: O.S
o.oLt~~~~r~L~
1.0 1.S 2.0 2.S 3.0
Fig. FlO. Network density of the vulcanizates examined as a function of sulphur and carbon
black contents
In the literature, various methods for the determination of network parameters of

rubber-like materials are described. Here, network densities were calculated from
stress-strain diagrams recorded in quasi-static tensile tests with a crosshead speed of
0.18 mmlmin on the basis of the Mooney-Rivlin equation. The physical network
density shown in Fig. FlO contains the primary network density as a result of chemi-
cal crosslinking as well as the molecular entanglements, and, in the case of filled
vulcanizates, the secondary network density of the overlying filler network. Secon-
dary and tertiary structures initiated by highly active filler particles form this overly-
ing network [7,8].
For the unfilled vulcanizates, an increase of the primary network density be-
cause of an increased number of net nodes is observed. The filled vulcanizates
have a constant primary network density because they have the same sulphur con-
tent of 1.6 phr. With increasing carbon black content, the secondary network den-
sity increases owing to development of the carbon black network, with many inter-
actions between the carbon black and the polymer, as well as between the carbon
black agglomerates themselves.
The results of the examinations are shown in Figs. FII-F14. The values of the criti-
cal tearing energy Te determined from the TFA measurements are shown in Fig.
FIl. Because Te remains nearly constant up to a sulphur content of2.0 phr, it can
be said that the increasing network density (see Fig. FlO) in this sulphur content
range has no influence on Te. Only a higher sulphur content decreases the critical
tearing energy.

0 10 20 30 40 50 60
4
3
E
.E Carbon Black
~ 2
...." Sulphur
c
I
~
CI
0
1.2 1.6 2.0 2.4
Fig. Fll. Critical tearing energy To of the vulcanizates examined detennined from TFA tests,
as a function of sulphur and carbon black contents
500 K. Reincke, R Lach, W. Grellmann, G. Heinrich
Similar results were provided by the instrumented tensile-impact tests. The

shape of the load-extension diagrams recorded using the instrumented tensile-
impact test (Fig. F12) is only slightly influenced by an increase of the sulphur
content, and the Jvalues (Fig. F13) for these unfilled vulcanizates indicate a con-
stant decrease of toughness, e.g. the resistance against unstable crack propagation
is decreased with increasing network density.
The Te values, as well as the shape of the F-/ curves and the J values, of the
filled vulcanizates are decisively influenced by the changes in structure, i.e. by the
increasing carbon black content. This is attributed to reinforcement effects, the
development of a filler network, and increasing interactions between filler and
polymer and between filler and filler. At 30 phr carbon black the critical tearing
energy shows a maximum value, which means that an increase of filler content to
50 phr does not improve the fatigue properties. Possible causes are a reduction of
the effective network chain length due to the high filler content and a reduced
extensibility in the areas of stress concentration within the material, which de-
crease the energy absorption capacity. Even for the Jvalues of the filled materials,
a maximum was found. From the maximum loads and maximum extensions it was
derived that this maximum is strength-determined; this shows, just as with the
maximum of Te, that only a filler content of up to 40 phr leads to an improvement
ofthe crack growth behaviour of the material system studied here.
150.----------------------, 150.---------------------~
\1.2 phr Sulphur \2.4 phr Sulphur

~100 100
~
'0
g
.3'"
'0
50
.3"'50~
o~ o
o 50 100 150 200 o 50 100 150 200
Extension (mm) Extension (mm)
150.----------------------, 1oolr==========~------,
100 100
g g
-g50~
.3
'050
~
o o
o 50 100 150 200 o 50 100 100 200
Extension (mm) Extension (mm)
Fig. F12. Examples of load--extension diagrams recorded using the instrumented tensile-
impact test
The results of the quasi-static fracture mechanics examination are shown in

Figs. FI4-FI6. Figure FI4 presents the J-!l.a curves of the filled materials, and a
sig~ificant increase of the slope with increasing carbon black content is visible.
For the unfilled vulcanizates, only small differences were found. This is reflected
by the technical crack initiation parameter JO.2 (Fig. F 15) and the values of the
slope of the crack resistance curves dJ/d(!l.a) at the point of maximum experimen-
tally determined crack growth !l.amax(exp) (Fig. FI6).

0 10 20 30 40 50 60
200
CarbOnBI~
E 150
-E
z
........ 100
...,
al1:l
50
0
1.0 1.5 2.0 2.5
Fig. F13. J values Jl L related to the resistance against unstable crack as a function of sulphur
and carbon black contents
400r----------------r======~
• •
•
50 phr
40 phr
300 • 30 phr
E • ••• • 15 phr
.§ 200 •• y 0 phr
~
..., •
100
O~~~~~~~~--~
0 0.2 2 3 4
da (mm)
Fig. F14. Influence of carbon black content on the crack propagation behaviour under quasi-
static loading conditions
The crack initiation and crack propagation behaviour is influenced only little up
to a sulphur content of2.0 phr; a higher sulphur content of2.4 phr leads to a lower
crack initiation parameter, but to a higher resistance against crack propagation.
With increasing filler content up to 30 phr, JO.2 is slightly increased at first, and
then a strong increase of the resistance against stable crack initiation appears.
Also, the resistance against stable crack propagation, characterized by dJ/d(Aa) at
Aamax(exp), increases somewhat up to 30 phr filler, and shows a maximum value at
40 phr. This means that more energy is necessary for initiation of a stable crack
within the 50 phr filled vulcanizate, but at the same time the resistance against
stable crack propagation is lower in comparison with the 40 phr filled material.

1.2 1.6 2.0 2.4
150 16
•
0
Carbon Black
Sulphur
12
100
E E
-
~
E 8 .E
~
-,~
"!
50 -,0
4
0 0
0 10 20 30 40 50
Fig. F15. Influence of sulphur and carbon black contents on the technical initiation parameter

1.2 1.6 2.0 2.4
100 8
cor
E •0 Carbon Black cor
E
.E 80 Sulphur
~ 6~
-;;:
t60
_i
!
i 4
~
!
~
40
!
-, 2~
-,
20
" "
0 0
0 10 20 30 40 50
Fig. F16. Slope of the R-curves at the point &Imax (exp)
4 Summary
With the help of the examination methods described here, it is possible to describe
the toughness behaviour of the elastomer materials investigated as a function of
structural parameters. An increase of the sulphur content leads, in some cases, to a
considerable decrease of the toughness parameters determined under cyclic, im-
pact-like and quasi-static loading conditions. This results from an increasing net-
work density, with reduced chain mobility. The critical tearing energy Te, the J
values Jl L and the technical crack initiation parameter JO.2 remain at approxi-
mately the same level up to a sulphur content of 2.0 phr. After that, the levels of
these parameters are decreased. Additionally, differences in crack initiation and
propagation behaviour under quasi-static load were found.
The addition of filler entails a strong increase of these parameters through rein-
forcement effects and the development of a secondary carbon black network, with
increasing interactions between polymer and filler and between filler and filler.
Possible explanations for the decreased Te and Jd BL values and the decrease in
resistance against stable crack propagation of the vulcanizate with 50 phr filler
content, in comparison with the 40 phr filled material, are an incomplete distribu-
tion of the carbon black agglomerates, and the formation of larger filler agglomer-
ates which can be considered as inhomogeneities and function as crack initiators
owing to unfavourable stress circumstances nearby.
As a function of filler content, maximum values were found for Jl L as a pa-
rameter of resistance against unstable crack propagation, for Te as a parameter of
fatigue crack propagation and for the resistance against stable crack propagation.
The addition of 30-40 phr carbon black enhances the crack initiation and propa-
gation behaviour of the rubber in the filler content range investigated.
The differences observed in the crack initiation and propagation behaviour
make it clear that a multi-parametrical description of the fracture behaviour is
necessary for an optimal material characterization.
References
1. Lach R., Grellmann W. (1997): Correlation between toughness and molecular relaxation
behaviour of amorphous polymers. In: Proceedings of Deformation Yield and Fracture of
Polymers, Cambridge, UK, April 7-10: 490--493
2. Beerbaum H., Grellmann W. (1999): The influence of morphology and structure on the
mers, Composites and Adhesives. ESIS Publication 27, Elsevier Science Ltd., Oxford:
163-174
3. Gerber G., Struve J. (1999): EinfluI3 der Mischungszusammensetzung und Belastungsart auf
das Versagensverhalten von Elastomeren. Kautsch. Gummi Kunstst. 52: 400--405
4. Eisele U., Kelbch S., Engels H.-W. (1992): The tear analyzer - a new tool for quantitative
measurements of the dynamic crack growth of elastomers. Kautsch. Gummi Kunstst. 45:
1064-1069
5. GreI1mann W., Clisar T., Heinrich G. (1999): Mechanische Eigenschaften, Rissinitiierung

und molekulare Struktur geflUlter Vulkanisate. Kautsch. Gummi Kunstst. 52: 37-43
6. Standard Draft ESIS TC 4 (1995): A testing protocol for conducting J-crack growth resis-
tance curve tests on plastics.
7. Keirn J., Tobisch K., Leisen J. (1998): Comparative investigations of the crosslink density of
elastomers. Kautsch. Gummi Kunstst. 51: 364-369
8. Eisele U. (1990): Introduction to polymer physics. Springer Verlag, Berlin, Heidelberg
F.3 Characterization of Deformation Behaviour
of Modified Polymer Concrete
H. Wehner and W. Grellmann, Merseburg,
and T. Hildebrandt, Rendsburg
1 Introduction
Polymer concrete is a relatively young material that has been used in the building
industry since the 1950s (known then as reactive resin concrete). Since the begin-
ning of the 1980s, polymer concrete has been applied in mechanical engineering.
Here, machine bases and components of fast-running machines have been made in-
creasingly of polymer concrete because of its high damping properties for sound and
mechanical vibrations and because it is simple to make complicated shapes with it.
The excellent damping behaviour is often a decisive factor in the application of
polymer concrete in the building industry, where it is used in sound-absorbent
walls and sound-absorbent components. In some cases, for example when they are
used in joints of window frames during renovation work, polymer concrete com-
ponents must also show high strength and good thermal isolation, besides their
sound-damping properties [1,2].
An additional aim of the optimization of these materials, besides the achieve-
ment of the above properties, is a reduction in density while maintaining the level
of properties. Therefore, for lightweight fillers, fundamental research on the type
of particles and their size and volume content is required for establishing quantita-
tive morphology-property correlations.
2 Experiments
2.1 Polymer Concrete Materials Tested
Polymer concrete is, like cement concrete, a brittle and heterogeneous material. It
consists of quartzitic fillers of varying particle size (up to 32 mm) and particle size
distribution, and a binder matrix such as unsaturated polyester resin, epoxy resin or
poly(methyl methacrylate).
The investigations were carried out on polymer concrete with a binder matrix of
unsaturated polyester resin and quartzitic fillers with a maximum particle size of 8
mm. The particle size distribution corresponded approximately to the particle size
distribution given by Fuller [3] (Fig. FI7):
S = 100 g. , (F9)

506 H. Wehner, W. Grellmann, T. Hildebrandt
where D is the maximum particle size of the total mixture and d is the maximum
particle size of the fraction considered.
Modification of the initial mixture was performed by partial substitution of dif-
ferent fractions of the quartzitic filler by lightweight fillers (Table F5; lightweight
fillers in italic); 25, 50, 75 and 100 vol.% of the quartzitic filler were substituted
by lightweight fillers (Vd so that 12 modifications of the initial mixture were
formed.
12
10 25 vol.% ofTotal Filler Content

~ Corresponds to Each of the Ranges
-'
> 8 I to IV
I/)
c:
0 6
nm
....
u.
....
0
4
t
m
+ II
tl. 2
0
0 2 3 4 5 6 7 8
Classification of Fraction (mm)
Fig. F17. Particle size distribution given by Fuller for lightweight fillers
Table F5. Modification of polymer concrete mixtures: substitution of quartzitic fillers by light-
weight fillers
Distribution given by Fuller a

Range I Range II Range III Range IV
Initial mixture Quartz Quartz Quartz Quartz
Series I, Hollow glass Quartz Quartz Quartz

'hollow glass spheres
spheres'
Series 2, Quartz Quartz Expanded clay Quartz

'expanded clay'
Series 3, Hollow glass Expanded clay Expanded clay Quartz

'light' spheres
a Range I, 0-0.5 mm; range II, 0.5-2 mm; range III, 2-5 mm; range IV, 5-8 mm
Deformation Behaviour of Polymer Concrete 507
2.2 Test Methods
• Conventional Bending Test Coupled with Acoustic-Emission Analysis

Three-point bending tests were performed in accordance with [4] on specimens
with the following dimensions: thickness B = 40 mm, width W = 40 mm, length
L = 200 mm and support span s = 120 mm. The crosshead speed was chosen in
such a way that the maximum load was reached within one minute.
At a distance of 30 mm from the expected fracture surface, an acoustic-emission
sensor was attached so that the load-displacement signal and the acoustic signal
could be recorded simultaneously. An evaluation of the acoustic-emission signal
was performed to determine the beginning of marked material damage and to ob-
tain information about the amplitude distribution of the acoustic signal.
• Study of Damping Behaviour by Drop Weight Tests
In order to test the mechanical damping behaviour, multiple-impact tests were
undertaken. Polymer concrete plates of dimensions width W = 100 mm, length
L = 100 mm and thickness B = 20 mm were examined, using a falling weight with
20 mm ball diameter and a test speed VH = 1.2 mls. This drop test was repeated
until the fracture of the specimen. The load and displacement were recorded in
each test and the potential work and the dissipated work were determined by inte-
gration.
• Fracture Mechanics Investigations Under Impact Loading
Instrumented Charpy impact tests were carried out with specimens of dimensions
thickness B = 40 mm, width W = 40 mm and length L = 200 mm, under test condi-
tions with s/W= 4 and pendulum hammer speed VH= 1 mls. The depth of the initial
notch, produced by sawing, corresponded to the optimized notch depth of a/W = 0.2
[5], which was determined by the minimum requirement for the dimensions of a
concrete specimen ofeW - a) = 3 x maximum particle size [6].
• Modal Analysis
The modal analysis was performed with a Brtlel & Kjcer measuring system, type
2035. The configuration and the mechanical excitation of the specimen corre-
sponded to the Charpy configuration of an impact bending test, so that the funda-
mental natural frequencies measured in bending were transferable to these tests.
An impact hammer with a rubber calotte proved to be advantageous for stimulation
of vibration. In addition, the damping capacity was investigated during these tests.
• Examination of Crack Propagation Under Fatigue Loading
Fracture mechanics examinations under fatigue loading were performed to deter-
mine da/dN-MC curves, which make possible the description of stable crack ini-
tiation and propagation by the threshold value MCo or by the constants of the
Paris-Erdogan equation. The experiments were carried out with an Instron 8501
testing machine. A sine-wave load with stress ratio R = 0.1 and frequency f = 2 Hz
was chosen for the tests. The dimensions of the compact tension specimens used
were thickness B = 40 nun, width W = 100 nun and height H = 98 nun, and the
ratio a/Wwas maintained at 0.45. The notch radius was 50 J.1m. The determination
of the amount of crack propagation da was performed with the help of compliance
measurements.
3 Results
3.1 Examination of Damage Process in Polymer Concrete

Modifications
The results of the conventional bending test show that the flexural strength is re-
duced with decreasing concrete bulk density (Fig. FI8). However, series I, 'hol-
low glass spheres', has the most favourable specific flexural strength CTspee, i.e. the
most favourable ratio of flexural strength to bulk density, of the modifications
examined. This statement is valid for the specific modulus of elasticity Espee (ratio
of flexural modulus to bulk density) too (Fig. FI9).
The investigations on series 2, 'expanded clay', show that the influence of ex-
panded-clay content on the strength and bending stiffness is relatively low.
30
2.2
~:~ ~
;;- 2.0
E '~=======,~o 20
~ • ~o IV
1-·-som'l ~
1.B Q.
'"
.><
::;::
"- 1.6 -0- Series 2 .~ ~ 10

b
-~Series 3
1.4
0 25 50 75 100 0
0 25 50 75 100
V,( vol.% ) V,( vol.% )
Fig. F18. Bulk density (a) and flexural strength (b) of polymer concrete
.
~ 14
~ .;::6 ~
-------. ~~~
E E
~ 12 ~o~ "~5
'"
.--.
.>< .><
~o~~o
::::- 10
IV ~:====:~o IV ...
0- -0- Series 1 • ~ 4 -o-Series 1 ~
~ 8 -o-Series 2 - 0 - Series 2 •
--Series 3 --<-Series 3
b" 6 W 3
o 25 50 75 100 o 25 50 75 100
V,( vol.% ) V L ( vol.% )
Fig. F19. Specific flexural strength (a) and specific modulus of elasticity (b)
Initial MixtLJ'e
30~---~~-~ 5 > ~
e::.
15 ~--------~
4';' c::
3 C) ~ 10
~
2~ E 51--------------l
1 w ::l
0« ~ 0 WIioJoJIJoUJU
0.1 0.2 0.3 o 20 40 60 80

Strain (%) AE Amplitude (cB)
Series 1 'Hollow glass spheres'
~ 15 ~--------~
e::.
ic::
10
E 5 I--------~IUI_
::l
~ 0 i..w..J..u.LLJ..LLl-'J..LLJ..u..J.J.JololIIIIBLlBIIL
o 20 40 60 80
Series 2 'Expanded day'
In
30~--~-------, 5 > ~ 15 ~--------~
e::.
4';'
i
In c::
~
- ....... 20
Was
3 C) 10
-a.. ~
I!! :!: 10 2~ E 5 I---~----,-=_-----l
~- 1 w ::l
Q)
u:: « ~ 0 Lu....w..J..IUoIoIIJIIillULUBJ.
0.1 0.2 O.:P o 20 40 60 80
Strain (%) AE Amplitude (c18)
Series 3 'Ugtt'
In
~15~--------~
e::.
i
In c::
-~ .......
Was
20 10
-a..
5~ 10 E 51----------
::l
~ OLu..w..............................,.,....w
Q)
u::
0.1 0.2 o 20 40 60 80
Fig. FlO. AE signals and amplitude distribution of different polymer concrete modifications
Figure F20 shows typical results from the acoustic-emission analysis. Although
superposition of the AE signal with single events as a result of fractures of hollow
glass spheres occurs, it can be seen from the graphs that series 1 shows no clear
increase in acoustic emission, and so no increase in damage, until 80 % of the
maximum flexural strength (TtM is reached. However, the distribution of amplitude
shows differences in the form of the existence of one or two maxima, which are at-
tributed to different types of damage behaviour. Values of a parameter describing the
onset of damage (pp. 385-402), determined with the help of mathematical algo-
rithms, are shown in Fig. F21.
In contrast to the initial mixture, series 2 and 3, containing expanded clay, show
already an onset of damage at between 40 and 60 % of the flexural strength. This
leads to a decrease in the maximum flexural capacity. Comparing series 2 and 3,
no differences are noticeable, though series 3, unlike series 2, contains both ex-
panded clay with a particle size of 1-2 rom and hollow glass spheres. Therefore,
independent of the kind of fme filler, the damage behaviour is influenced only by
the large-size fractions of the polymer concrete mixture.
-
100
0~
•
~:
80 .~
"0
~
0
•
60
~+--.
0
..-- +
-l·
b
w
(J)
40
- . - Series 1
A _______ A
20 -+- Series 2
- A - Series 3
0
0 25 50 75 100
V L (voL%)
Fig. F2t. Onset of damage for different polymer concrete modifications
3.2 Damping Behaviour in Drop Weight Tests
During the impact loading of the polymer concrete plates in the drop weight test,
90 % of the energy supplied was used up in dissipation processes in the specimen.
However, this examination method reveals no clear differences in the damping
behaviour of the polymer concrete modifications used (Fig. F22).
The relative damping index An given by
(FlO)
where A di • is the dissipated energy and Apot is the potential energy, is equal to 9 on
average. A comparison of the amounts of energy in the fust and second impacts
shows only a minor increase in the dissipated energy.
However, the load-penetration behaviour shows clear differences between the
first and second impacts, as is shown for the example of the initial mixture in
Fig. F23. The fust impact initiates pre-damage in the material. During the second
impact load, the initial cracks grow and the strength is reduced.
100r---------------------------------------~
Dissipated Energy 1st I~

"+-. Potential Energy 1st I~
-(}- Dissipated Energy 2nd I~
--<r- Potential Energy 2nd IIllBi
~ ~ --==t===. =! ~
'#. '#. ~ ~ ~
8
~ ~ ~ ~ ~
fu
~
§
0
LO
r-..
&,
N ~
&,
r-.. 8 ~ § &,
r-.. T""
T""
Series 1 Series 2 Series 3
Fig. F22. Dissipated energy and potential energy of different polymer concrete mixtures
16
1st Impact
....... 12
z
..I<:
'-'
'0 8 2nd Impact
ro
0
...J
4
a
0.0 0.5 1.0 1.5
Penetration Depth (mm)
Fig. F23. Load-penetration behaviour in first and second impacts

3.3 Damping of Mechanical Vibrations
A clear classification of the damping behaviour of the various polymer concrete

modifications is possible by means of the method of modal analysis. The logarith-
mic decrement 8[6] is given by
8=_1_ln~, (Fll)
I-n An
where A 1 is the first amplitude considered, An is the nth amplitude considered and
n is the number of cycles considered. This represents a quantitative measure for
the damping of mechanical vibrations. The results in Fig. F24 show that the
damping capacity increases with an increase of lightweight-filler content.
L -.- ----~
100
- 75
~._ Series 1 A
"'0
.... Series 2
-A-Series 3
-~ 50
25
• .----:::::::= =&
o~~--~~~~--~--~~~~--~~
o ~ w ~ 100
VL (vol.%)
Fig. F24. Logarithmic decrement of different polymer concrete modifications
3.4 Fracture Mechanics Examinations Under Impact Loading
Load~eflection and load-time diagrams of polymer concrete modifications re-

corded during impact testing under mode I load are shown in Fig. F25.
The recorded load~eflection curves simulate stable crack growth after the
maximum load is passed. Polymer concrete is a relatively brittle material [7].
Therefore, the real reasons for the seemingly stable crack growth are, on the one
hand, the reduction in hammer velocity during separation of the relatively large
specimen and, on the other hand, the friction between the specimen and the sup-
port.
The following experimental conditions must be satisfied for a successful
evaluation of fracture mechanics parameters:
1. Fmax > F" where Fmax = maximum impact load and Fl = inertial load.
2. AH> 3 AG, where AH = nominal impact energy of pendulum hammer and AG =
general deformation energy up to Fmax'
3. ts > 1.5-31; where ts = time to fracture and f' = oscillation period (pp. 71-86,
[8]).
~2
u.
o
o 0.25 0.5 0.75 o 0.25 0.5 0.75
t (ms) f(mm)
Fig. F25. Typical load-time (a) and load-deflection (b) diagrams recorded in impact tests
under mode I load
An examination of whether these experimental conditions were satisfied for the

various polymer concrete modifications resulted in the following conclusions:
1. No marked inertial load was recorded, as is usually the case for thermoplastic
polymers [8-10], because polymer concrete is a material with a very high
damping capacity for mechanical vibrations.
2. The experimental requirement AH > 3AG was satisfied.
3. Average times to fracture of ts = 0.2 ms were determined for all polymer con-
crete modifications.
The oscillation period of the specimens, determined by means of modal analy-
sis, had a value of f'= 0.33 ms on average. One of the conditions for calculation of
fracture mechanics parameters is not fulfilled, since ts < f'. This could be the rea-
son for the overestimation of the crack resistance when the fracture mechanics
parameters are calculated. It is also possible that, in the case of polymer concrete,
this experimental condition for fracture mechanics measurements is subject to a
criterion other than that applicable to common polymers.
The stress intensity factor Kid was determined for all of the modifications of
polymer concrete, as a typical fracture mechanics parameter of linear elastic be-
haviour (Fig. F26) (pp. 71-86).
80.---------------------------- ,
70
I:!
~E
E 60
CO
a..
~ 50
......- - .- Series 1
..,
';:1;- - A- Series 2
40
- .- Series 3
30
o 20 40 60 80 100
VL (vol.%)
Fig. F26. Reduction of crack resistance against unstable crack propagation with increasing
lightweight filler content
Fig. F27. Fracture surface of an expanded-clay particle in polymer concrete
The results in Fig. F26 show that the modifications 1 and 2 put up the greatest
resistance to the beginning of unstable crack propagation, with values of
KId = 63 MPa mm I/2 to 72 MPa mm I/2, in comparison with KId = 74 MPa mm 1/2 of
the initial mixture. The hollow-glass-sphere modifications of polymer concrete
(series 1 and 3) have a considerably lower crack resistance than the initial mixture.
The reason for this behaviour is the morphology of the lightweight fillers .
Whereas expanded clay possesses an outer layer approximately 50 Ilm thick with a
relatively dense structure (Fig. F27), the outer layers of the hollow glass spheres
are mostly thinner than 12 Ilm (Fig. F28). The adhesion between matrix and filler
is excellent because of the very good wettability of dry fillers by polyester resin.
Therefore, the failure of polymer concrete modifications containing hollow glass

spheres as a lightweight filler is attributed to the low thickness of the outer layer
and the lower filler strength associated with this.
Fig. F28. Fracture surface of polymer concrete with hollow glass spheres
3.5 Examination of Crack Propagation Behaviour

Under Fatigue Loading
Owing to the presence of an initial notch with a radius of 50 flm, the da/dN-M
curves (Fig. F29) of the polymer concretes start with a reduction in the crack
propagation speed until the threshold value M o is reached. This is the initiation
point of a sharp crack [II]. Above Mo, an increase in crack propagation speed
starts [12] .
The results for the various modified polymer concretes show a decrease in crack
resistance with increasing content of lightweight filler. On the one hand, this is
demonstrated by the decreased value of Mo, which characterizes the beginning of
stable crack propagation, and, on the other hand, by an increase of the constant m,
which characterizes the increase in the speed of stable crack propagation (Ta-
ble F6).
The region in which the crack propagation speed has a constant slope is de-
scribed by the Paris-Erdogan equation (FI2):
da/dN = C(M)m (FI2)
The constants in (FI2) determined for the different polymer concrete mixtures
are listed in Table F6.
10°r---r=========~-r==~======~===='-~====~
~.% I~itial MixIIn I
-r
'Light' 100 voI.%• : 'Glass Spheres'l00
··················.······iii·· ........... .
10-3
10"
~• •
10-6 • •
10.7
10-6
10 ~Ku
I 29
j
~Ko. 45~Ko
j
66 100
~K (MPamm l12)
Fig. F29. dafdN-AK curves of different polymer concretes
Table F6. Fracture mechanics values of dafdN-AK curves of different polymer concretes
Material Constants of the Paris- AKo Klc

Erdogan equation (MPammln) (MPammln)
C m
Initial mixture 3.6xlO-6 9 49 66

Series 1, 7,5xIO-B 16 32 45
100 % 'hollow glass spheres'
Series 3, 4.3xIO- 16 27 18 29
100% 'light'
The initial mixture shows the highest resistance against stable crack initiation
and propagation, The reason is that the quartzitic fillers have a higher particle
strength than do the lightweight fillers.
4 Summary
These investigations of damage behaviour have shown that the strength of polymer
concrete mixtures is considerably influenced by the strength of the largest filler
particles. For the manufacture of a lightweight concrete, i.e, with a bulk density
lower than 2 kg/dm3 , with a strength comparable to that of the initial mixture, only
the low-particle-size fraction should be substituted. The initiation of damage can

be correlated with the specific flexural strength.
By the method of multiple impact loading of a polymer concrete plate, the de-
termination of the damping index Ar is possible. However, this method is not sen-
sitive enough for the detection of clear differences between the various polymer
concrete mixtures. The logarithmic decrement, determined by means of modal
analysis, represents a more sensitive measure of the damping of mechanical vibra-
tions. These results show that the investigation of strength behaviour after different
amounts of impact pre-damage is of great interest.
Fracture mechanics investigations using mode I impact loading have shown that
only expanded-clay-modified polymer concrete (series 2) has a higher crack resis-
tance against unstable crack propagation, owing to the morphological differences
between the lightweight fillers used. In comparison with the initial mixture, with a
crack resistance of KId = 74 MPa mmll2, the resistance against unstable crack ini-
tiation is reduced to a smaller extent than in the cases of the 'hollow glass spheres'
and 'light' series.
The fracture mechanics characterization of polymer concrete under fatigue
loading was possible by means of da/dN-M( curves. Owing to the fact that it has
the highest particle strength, the initial mixture shows the highest resistance against
stable crack initiation and propagation, in comparison with the 'hollow glass
sphere' and the 'light' mixtures.
Acknowledgements
The authors would like to thank ACO Severin Ahlmann GmbH & Co. KG, Rends-
burg, Germany and the Ministry of Culture and Education of Sachsen-Anhalt,
Germany for fmancial support.
References
1. Kane J. F. (1991): Polymer concrete machine tool components. In: Proceedings of ICPIC
'91, San Francisco (USA)
2. Rikeou J. T. (1996): Polymer concrete design and specifications. In: Proceedings of the
International ICPIC Workshop on Polymers in Concrete for Central Europe, Bled, Slovenia,
September 10-12
3. Brandt A. M. (1995): Cement-Based Composites. E&FN Spon, London
4. Richtlinie fOr die Priifung und Giiteiiberwachung von Erzeugnissen aus Polyesterharzbeton.
(1980)
l-integral concept. In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Se-
ries, No. 17. Institut fOr Mechanik, Berlin, Chemnitz: 48-72
6. DIN 51290 (1991): Testing of Polymer Concretes (Reaction Resin Concretes) for Mechani-
cal Engineering Purposes
7. Wehner H., Hesse W., Grellmann W., Hildebrandt T. (1996): Untersuchungen zum Defor-
mations- und Bruchverhalten von modifizierten Polymerbetonen. In: Tagungsband Poly-
merwerkstoffe '96, Merseburg, Germany, September 18-20: 294-297
In: Fracture Mechanics, Micromechanics and Coupled Fields (FMC) Series, No.3. Institut
fur Mechanik, Berlin, Chemnitz: 102-111
mechanics and Coupled Fields (FMC) Series, No.3. Institut fur Mechanik, Berlin, Chem-
nitz: 142-151
ceedings of the 1st Conference on Mechanics, Prague, Czech, June 29-July 3, Vol. 5: 129-
133
II. Schott G. (1997): Werkstoffermiidung - Ermiidungsfestigkeit. Deutscher Verlag fur Grund-
stoffindustrie, Stuttgart
12. Anderson T. L. (1995): Fracture Mechanics: Fundamentals and Application. CRC Press,
Boca Raton
F.4 Fracture Mechanics Testing of Modified
Epoxy Resins with Mini-Compact Tension
(CT) Specimens
H. Walter, C. Bierogel, W. Grellmann and M. Fedtke,
Merseburg, and B. Michel, Berlin
1 Introduction
Epoxy resins are used in a wide range of applications. At room temperature

crosslinked epoxies typically exhibit a high modulus and a nearly elastic stress-
strain behaviour, but they have poor resistance to fracture. Complex material behav-
iour can often be observed owing to different mechanical and thermal properties
(for example, in a sandwich construction), which can impair the overall mechani-
cal-thermal reliability. In these systems, local defects such as cracks can appear as
a result of inhomogeneities of the composite material, internal stresses and high
thermal and mechanical loading gradients [1-3]. Consequently, the methods of
fracture mechanics have an increasing significance regarding the assessment of the
failure behaviour. The aim is an optimization of the toughness of epoxies under
extreme conditions. Concerning the dimensioning, geometry-independent parame-
ters have to be used in order to guarantee optimum reliability and operational
safety in industrial applications (pp. 3-26). The usual compact tension specimens
require a large amount of material and are often unsuitable for experimental inves-
tigations. In the future, the application of miniature specimens for the determina-
tion of fracture mechanics parameters should be made possible so that an efficient
toughness evaluation of newly developed materials can be performed [4].
In the present study, the influence of chemical-structure parameters and test
conditions (test temperature, crosshead speed) on the crack resistance behaviour
and fracture mechanics parameters is examined. The investigation of the fracture
behaviour of miniature specimens is based on the crack resistance analysis of stan-
dard specimens, which show stable and unstable crack propagation behaviour
under quasi-static loading conditions. Furthermore, the practical application of
fracture mechanics parameters to structural-integrity assessments is strongly re-
stricted by their limited transferability from specimens to components. Therefore,
an extensive investigation was conducted to check out the technical limits on
miniaturizing the geometry, the validity of specimen size criteria, and the thickness
independence of the fracture mechanics parameters in the presence of stress con-
centrations. The results of experiments with standard specimens and with minia-
ture specimens were compared, and the qualitative coherence that existed could be
explained. In addition, the fracture properties of commercial epoxy resins filled
with rigid glass particles were investigated.

520 H. Walter, C. Bierogel, W. Grellmann, M. Fedtke, B. Michel
2 Experimental
2.1 Materials
The test material chosen for all of the investigations was the epoxy resin Epilox
A1701 from Leuna Harze GmbH, Germany, a commercially available DGBEA
resin cured with a specially formulated anhydride hardener. The resin was used
both in the unfilled state and reinforced with silica particles (Table F7). The glass
particles were surface-treated. The flexibilizer M670 was used to improve the
resin-filler adhesion and the resin toughness. After mixing thoroughly with hard-
ener, the mixture was poured into a Teflon or silicone mould. The curing was
carried out in accordance with the manufacturer's instructions; this means with a
cure time of 3 h at 80°C followed by 10 h at 130°C. This procedure yielded a
material with a glass transition temperature of 152°C. All materials were condi-
tioned for 1 day in a standard atmosphere.
Table F7. Particulate fillers used in the experiments
Filler Manufacturer Kind of filler Diameter
Teco SiL44L C-E Minerals Glass powder 9-13 f.UIl

Spheriglass 5000 Potters Ballotini Glass ball 3.5-7 11m
Spheriglass 2024 Potters Ballotini Glass ball 100-200 11m
The fracture mechanics values of the cured resins were measured with miniature
compact tension specimens. The specimen dimensions were length L = 25.4 mm,
width W= 20.4 mm and thickness B = 1-10 mm. The ratio of the initial crack
length to the specimen width a/W was 0.55. In addition, the specimens were
notched with a razor blade.
2.2 Test Methods
The evaluation of the material behaviour was carried out by means of both con-
ventional mechanical tests and fracture mechanics methods. For the tests with CT
specimens, an Instron universal testing machine was used.- The load-line displace-
ment was measured with the help of a solid-state laser scanner system. The system
was developed specially for polymeric materials and demonstrates various advan-
tages compared with the conventional mechanical clip gauge system. This laser
double scanner is favoured for experimental fracture mechanics examinations and
works in the transmission mode with two parallel laser beams. With these two
beams, the load-line displacement and crack-mouth-opening displacement can be
determined by measurements without contact with the specimen. This is necessary
Fracture Mechanics Testing of Epoxy Resins 521
because mechanical extensometers influence the load-displacement behaviour of

polymer materials. If the conditions of the measurement are perfect, a resolution of
0.5 J.lm should be possible. The edges of the CT specimen must be of a high quality
because the accuracy of the measurements depends on the roughness of the outline.
This loss of accuracy is caused by the rotation and translation of the CT specimen
while the positions of the laser beams remain fixed (pp. 365-384).
In the present study, the toughness behaviour was estimated by means of two
fracture mechanics parameters: the stress intensity factor K and the J-integral (pp.
51-70, [5,6]). The stress field around a sharp crack in a linear elastic material can
be described by the stress intensity factor Kb which is given by
Fmax
KI = ,;;; f{a/W) , (F13)
B"W
where Fmax is the maximum load, f{a/W) is a geometrical function, B is the thick-
ness and W is the width.
The J values can be calculated from [7]
J = J (1- (0.751/-1)aa) (F14)

o W-a'
where
JO= 1/U (FI5)
B(W -a)
and
1/ = 2 + 0.522(1- a/W). (F16)
U is the area under the load versus load-line displacement curve.
The influence of the specimen thickness, in a range of 1-10 rom, and of the test
conditions (loading rate from 0.1 to 10 mmlmin and temperature from 23°C to
160°C) on the fracture mechanics behaviour was investigated. The effect of
specimen thickness was examined at room temperature with a loading rate of
1 mmlmin; the influence of the loading rate was also studied at room temperature
using CT specimens.
In addition, the materials were characterized by scanning electron microscopy,
used to observe the fracture surfaces.
3.1 Comparison of Standard and Miniature Specimens
Unfilled, unmodified epoxy resins are brittle and exhibit unstable crack propaga-
tion. The estimation of fracture mechanics parameters by the use of miniature
specimens is based on the analysis of the crack resistance behaviour against stable
522 H. Walter, C. Bieriigel, W. Grellmann, M. Fedtke, B. Michel
o Standard Specimens
100..------1
• Miniature Specimens
80
20
20 40 60 80 100 120
Load-Line Displacement v L (10. mm)
3
Fig. F30. Comparison of load versus load-line displacement curves of standard and miniature
specimens
and unstable crack propagation that is performed with standard specimens under
quasi-static loading. The aim of this study was the miniaturization of the speci-
mens and the verification of the validity of the size criteria. A further aim was to
examine the thickness independence of the fracture mechanics parameters.
The load versus load-line displacement curves of standard specimens and
miniature specimens are compared in Fig. F30. The K values were calculated from
(F13). As expected, the maximum load reached by the standard specimens is
greater. On the other hand, the load-line displacement and the stress intensity fac-
tor of the miniature specimens are smaller (Table FS).
Table F8. Comparison ofvaIues for standard and miniature specimens
CT specimens Fmax (N) Vmar (10-3 mm)
Miniature 69.8 92 21.1

Standard 89.9 105 23.1
3.2 Effect of Specimen Thickness
An important role in the assessment of the crack resistance is played by the trans-
ferability of laboratory results to components. Toughness values of polymers de-
pend on many factors, for example on the specimen thickness. Figure F31 shows
the K vs. B and J vs. B curves for specimen thicknesses ranging from I to 10 mm
at a crosshead speed of I mm1min at room temperature.
The values of J and K exhibit the same trend, which means that this unmodified
epoxy resin shows an energy- and load-determined fracture process. With in-
creasing specimen thickness, thickness-independent fracture mechanics values of
the material are certainly reached for specimen thicknesses B ~ 4 mm for the un-
modified epoxy resin tested.
60 0.5
A J
•
L
0.4
50 K
.g- 0.3
E E
E
'"
Q.
~
40 6
+ ~
• 0.2.E
~
£
0.1 .,
':t(.£ 30 .,·u
~ 0.0
20
10
2 3 4 5 6 7 8 9 10
Thickness B (mm)
Fig. F31. Values of K andJas a function of specimen thickness
The experimental estimation of the minimum specimen thickness Bmin is very

expensive owing to the great number of specimens which is necessary. Therefore,
the derivation of general size criteria concerning the various fracture mechanics
concepts, without requiring any knowledge of the specific thickness dependence of
the fracture mechanics parameters, is required. The empirical size criterion (F17)
is valid and is independent of the loading conditions and the nature of the failure
of the material, i.e. unstable or stable. By considering all experimental data points
for p vs. K Ic and p vs. KId that are known at present, an experimental dependence
of Emin can be determined (Fig. F32):
2
[ K 1c )
Emin =p-- (F17)
CT y
Generally, the proportionality constant P depends on the material and on the

loading conditions, but it is often considered to be constant. For the calculation of
the constant p, the experimentally determined thickness dependence was used. For
524 H. Walter, C. Bierligel, W. Grellmann, M. Fedtke, B. Michel
epoxy resins, P values of 15 and 26.2 have been reported in the literature [S]. A
double-logarithmic plot of P as a function of the stress intensity factor K enables a
description according to (FIS):
K Ic = KoP -n k . (FIS)
Ko and nk are material-independent parameters. An estimation of BDlin becomes
possible from the combination of (F 17) and (F IS):
. =(KIC )2(~)1 / nk (FI9)

Bmm a K '
y Ic
where~k 7 ~3 ..13..and ~ ';;' 7&8MPammll2.
Specimen Thickness
100 ~---------------+~ 0 1 mm
0 2mm
• 4mm
•... Smm
.,. 8mm
10 ~-----d~--------~
10mm
0
o. 11·':-o----~~~~~~1:-LO-O-----'---'-..:::.........<..-.........~1...J
OOO
K,. (MPa mm 112)

Fig. F32. Requirements on specimen geometry for the determination of the critical fracture
toughness parameter K,< with miniature specimens
For the J-integral, a similar empirical relationship between E and J values is

valid [9]. The constant Edecreases with increasing toughness, independently of the
standard or testing protocol used. For the epoxy resins examined Evalues from 950
to 1200 were determined [8]. The experimentally obtained fracture mechanics
parameters will be geometry-independent if they fulfil the size criterion of a mini-
mum specimen thickness EDlin. On the basis of the assessment of the proportional-
ity constants Pand E as well as the experimental determination, a specimen thick-
ness of 4 mm was chosen for the subsequent examinations. Thereby, the fracture
mechanics parameters could be used to express the influence of the material
structure, loading rate and test temperature on the toughness when miniature com-
pact tension (eT) specimens were used.
3.3 Effect of Test Conditions
3.3.1 Loading Rate
Figure F33 shows the fracture toughness determined with 4 nun thick specimens as
a function of the loading rate. The toughness of these epoxies decreases with in-
creasing loading rate. Local stress peaks in the specimen, which can be relieved
through sliding processes and conformational changes at a slow loading rate, are
one cause of this behaviour. These processes cannot take place completely with
increasing loading rate, i.e. the stress peaks only relax a little and the fracture
toughness simultaneously decreases [1,10].
60 Standard Specimens Miniature Specimens 0.5

0 K
0 J
•
~
K
J
rr
50
~ 0.4
~E
E 40 0.3 E
aI
a.. E
..
,,'_---r -----------
""-
~ , ~
-A- - - - -~ ....,
~ 30 ... ~ 0.2
0- __ ~
-0- ---
20 --------0 0.1
--0
10 0.0
0.1 1 10
Crosshead Speed (mm/min)
Fig. F33. Effect of loading rate on fracture mechanics parameters
3.3.2 Temperature
For the purposes of applications, it is especially important to determine limits of

use for epoxies; this requires a description of the mechanical properties as a func-
tion of temperature. The measured parameters are very sensitive to structural
changes induced by temperature (Fig. F34). Up to a temperature of 100°C the
toughness behaviour changes little, but if the temperature is increased further, the
fracture toughness increases. The maximum load reaches its highest value at 140°C,
just below the glass transition temperature Tg. The temperature Tg is defmed as the
temperature at the maximum of the mechanical loss factor tan 0. On the one hand,
this increase in fracture toughness at high temperatures can be attributed to an in-
Fig. F34. Effect oftemperature on fracture mechanics parameters
creased molecular mobility of network chains [11]. On the other hand, the tem-
perature-induced decrease in the yield stress of the polymer leads to larger plastic
zones in front of the crack tip, connected with crack blunting [1].
3.4 Modification of Epoxies
3.4.1 Influence of Fillers
The toughness improvement of polymers can be achieved by an enhancement of

the energy dissipation capability. Several authors have reviewed the fracture be-
haviour of unfilled and filled polymers. In general, the fracture toughness increases
with the addition of rigid particles as a modifier, at least for low volume fractions
of filler. A number of investigations have shown that the fracture toughness in-
creases with the inclusion of such rigid particles; deformation and fracture mecha-
nisms involving crazing processes and localized shear yielding have been reported
for crosslinked epoxies. Numerous mechanisms such as cavitation, micro-cracks,
filler cracks and the process of 'crack pinning' have been observed [1,11,12].
A propagating crack front will be temporarily pinned if it encounters particles
or inhomogeneities. An increase in load raises the degree of curvature between
pinning points, resulting in both a new fracture surface and further crack propaga-
tion. These processes lead to absorption of energy and therefore to an improve-
ment in toughness (Fig. F35).
• •••
Crack Breakaway
Crack Pinning
• •••
Rigid Particles
IC r a c k G row t h D ire c t ion
Fig. F35. 'Crack pinning' process
An increase in the fracture toughness occurs with the addition of fillers (Fig.
F36). All particles here were treated with a coupling agent to improve the adhesion
between resin and particles. Here, Teco SiL 44 L showed the best resistance
against crack propagation in terms of high values of fracture toughness. Besides
chemical linkage, this kind of filler shows pronounced physical adhesion effects as
a result of the particle geometry, leading to an increase in fracture toughness (Fig.
F37).
70
& Teco SiL 44 L (Glass Powder)
••
60 S 5000 (Glass 8all)
S 2024 (Glass 8all)
50
.r
-E
E40
co
a..
~
~30
'£
20
10
0 10 20 30 40 50 60 70
Filler Content (wt.%)
Fig. F36. Fracture toughness versus filler content
528 H. Walter, C. Bierogel, W. Grellmann, M. Fedtke, B. Michel
Fig. F37. SEM micrographs showing fracture surfaces of filled epoxies: Teco SiL 44 L, di-
ameter 9-13 11m (a); Spheriglass 5000, diameter 3.5-7 11m (b)
3.4.2 Influence of Flexibilizer and Filler
The influence of the chemical structure of the epoxy resin was investigated. By the
modification of the DGEBA resin with Epilox M670, flexible segments were built
in the network, and so the properties of the resin were strongly influenced. The
segments are responsible for an improvement in fracture toughness through net-
work expansion. With the addition of this flexibilizer, the glass transition tempera-
ture decreases from 150°C to 110°C because the intermolecular mobility of long-
70.---------------------------------,
• K (with Modifier)
•
60
• K (without Modifier)
50
5!~
•......
•.... ••
•....... ...
E
E 40
IV ~
a..
:E
;l30
Flexibilizer : 15 wt.% M670
20 Filler : Teco SiL 44 L
10
0 20 40 60
Filler Content (wt.%)
Fig. F38. Fracture toughness of modified epoxies versus filler content

chain molecules is improved. However, an increase in fracture toughness at the

room temperature cannot be observed with the addition of up to 20 wt.% of flexi-
bilizer. In comparison with other flexibilizers, the Epilox M670-modified epoxy
resin allows a high filler content owing to the low viscosity. With the addition of
fillers, a clear increase in toughness in comparison with unmodified epoxy was
achieved (Fig. F38). This increase in fracture toughness is achieved because of the
chemical activity of the flexibilizer and because of the filler in the epoxy resin
matrix. Because of the occurrence of stress relief, the flexibilizer increases the
ductility of the resin matrix, and the adhesion between the matrix and the filler
particles is improved simultaneously.
4 Conclusions
The aim of these investigations was the development of miniature compact tension
specimens. These miniature specimens allowed the evaluation of the fracture
toughness and the assessment of the deformation behaviour of epoxy. The values
of K and of the J-integral show the same behaviour as a function of specimen
thickness and loading rate. The dependences of the stress intensity factor K and of
the J values on specimen thickness, determined using miniature specimens, lead to
geometrical constants & and p that can be introduced into the general relations p =
f(K) and & = f(J). This means that the experimentally determined fracture mechan-
ics values fulfil the size criterion; therefore, they are independent of loading con-
ditions and of the nature of material failure. With increasing temperature, the
fracture toughness increases and reaches its maximum near the glass transition
temperature. With addition of fillers, an increase in toughness was observed, but
the combination of flexibilizer and filler leads to the greatest toughness increase.
The hybrid test methods used are very suitable for determining the specific ma-
terial behaviour of epoxy resin with miniature test specimens.
References
1. Ellis B. (1993): Chemistry and Technology of Epoxy Resins. Chapman & Hall, London
2. Kinjo N., Ogata M., Hishi K. (1988): Epoxy molding compounds as encapsulation materials
for microelectronics devices advance. Polym. Sci. 88: 22-122
3. Sommer E., Olaf J. M. (1994): Methods for determining mechanical properties of microsys-
terns. MaterialprUfung - Mater. Test. 36 (4): 124-127
4. Hinckley J. A. (1986): Small compact tension specimens for polymer toughness screening. J.
Appl. Polym. Sci. 32: 5653-5655
5. Anderson T. L. (1995): Fracture Mechanics. Fundamentals and Application. CRC Press,
Boca Raton
6. Kinloch A. J., Young R. J. (1983): Fracture Behaviour of Polymers. Applied Science Pub-
lishers
530 H. WaIter, C. Bierilgel, W. Grellmann, M. Fedtke, B. Michel
8. Grellmann W., Seidler S., Biertigel c., Straube E. (1990): Fracture mechanics toughness
characterization of epoxy resins under quasi-static load. In: Proceedings of 'Verstlirkte
Plaste'90', Dresden, Germany, October 16-17: 2.8/1-2.8/13
9. Grellmann W., Lach R. (1997): Toughness and relaxation behaviour ofpoly(methyl methac-
rylate), polystryrene and polycarbonate. Appl. Macromol. Chern. Phys. 253: 27-49
10. Scherzer T. (1993): FTIR-rheooptische Untersuchungen der molekularen Deformations-
mechanismen in Epoxidharzen. PhD thesis, University of Halle-Wittenberg, Germany
II. Schrtider N. (1998): Strukturbildung und Bruchverhalten von Multiphasen-Epoxidharzen
und Hybridkompositen modifiziert mit Oligomethacrylat-Flussigkautschuk. PhD thesis,
University of Freiburg, Germany
12. Htiffin F. (1995): Bruchmechanische Untersuchungen zur Wirkungsweise von Oligo- und
Polyetherestem mit Poly-(oxytetramethylen)-Segmenten als Z!ihmodifIkatoren von anhy-
dridgehlirteten Epoxidharzen. PhD thesis, University of Freiburg, Germany
G
Examples and Limits

of Application
G.1 Modelling of the Mechanical Behaviour
of Non-Linear Viscoelastic Materials
under a Multi-Dimensional State of Stress
E. Schmachtenberg, M. Wanders and N. M. Yazici, Essen
1 Abstract
The mechanical behaviour of non-linear viscoelastic materials depends on the

time, the temperature, the loading rate and the magnitude of the load. The aim of
the work described in this chapter was an investigation of the effect of a biaxial
state of stress on the characteristic functions, such as the secant modulus versus
strain. A model is presented that allows the simulation of non-linear viscoelastic
materials under a multi-dimensional state of stress.
Therefore, a suitable test specimen and testing machine were developed. Be-
cause the specimen is produced by injection moulding, both isotropic and aniso-
tropic specimens can be produced. The testing machine enables strain- and force-
controlled experiments.
Uniaxial quasi-static strain-controlled experiments on test specimens made from
polyamide 6 (PA 6) and from a styrene-butadiene-styrene (SBS) block copolymer
demonstrate the effect of the process parameters on the mechanical anisotropy. As
far as the isotropic PA 6 is concerned, approaches based on energy absorption and
equivalent stresses show the possible ways to transform the secant modulus from a
uniaxial to a biaxial state of stress in the case of quasi-static loads.
Dynamic experiments demonstrate that the load history must be taken into ac-
count when considering multi-dimensional states of stress. The approach men-
tioned above cannot achieve this. Therefore, a three-dimensional model has been
developed to simulate multi-dimensional load histories.
This three-dimensional deformation model is based upon the uniaxial deforma-
tion model and is constructed from a parallel arrangement of a certain number of
basic elements. Each basic element consists of an elastic Hookean element and a
damper system (3D damper) to describe the viscous properties. The arrangement is
calibrated by means of isothermal strain-controlled tensile tests at different tem-
peratures. Along with suitable calculation algorithms (relaxation algorithm and
retardation algorithm), this model offers the ability to simulate any multi-
dimensional load history caused by direct stresses. The model allows the simula-
tion of both quasi-static and dynamic loading at different temperatures.
By comparing the experimentally determined static and dynamic stress-strain
curves with simulated curves, the suitability of the model developed here is dem-
onstrated.
Relaxation effects are overpredicted by the modeLin comparison with experi-
mental results. Better calibration should lead to an improved correspondence.

534 E. Schmachtenberg, M. Wanders, N. M. Yazici
The aim of further investigations should be the expansion of this model to ther-
mal stresses and, later, implementation in fInite-element programs.
2 State of the Art
Every real component is subjected to a multi-dimensional state of stress. The load

on the material can be caused by mechanical, thermal and chemical stresses. The
characteristic data of the material, e.g. the modulus of elasticity and the admissible
stresses or strains, that are used to design components, however, are normally
evaluated by means of uniaxial experiments.
Investigations by Schmachtenberg [1] on a biaxially stressed tube show the ne-
cessity of a transformation of the characteristic data, as shown in Fig. G 1. The
state of stress is considered by means of a parameter M, evaluated from the energy
absorption at yield.
35
30
,.....,
N
25
~
~
0" 20
<II
<II
Q)
::
til
15
-.;;
~
10
Biaxial Stress State:
5
v=O.4 i,."=100 %'h S=23°C
0
0 2 4 6 8 10 12
Axial Strain &x (%)
Fig. Gl. Transfonnation of characteristic data for PE-lID tube stressed biaxially
Bardenheier [2] explains - concerning the stress at break - that this approach
has no general validity for plastics. Instead, a stress hypothesis is suggested on the
basis of shear-strain energy density (von Mises criterion). Therefore, the analysis
in [1] has to be reconsidered. Sarabi [3] dealt thoroughly with the long-time be-
Modelling of the Mechanical Behaviour 535
haviour of polymer materials under uniaxial and biaxial stress. The aim of this
investigation was the evaluation of a long-time admissible value, up to which a
component can be stressed without any damage. However, there is no adaptation
of the characteristic fimctions to the state of stress.
The aim of this chapter is to work out a rule to transfonn uniaxial measured char-
acteristic fimctions to a biaxial state of stress and to present a model to simulate
the mechanical behaviour under a multi-dimensional state of stress.
A suitable test specimen had to be developed that would enable the creation of
any biaxial state of stress and, following the CAMPUS standard, it had to be pro-
duced by injection moulding. By this means, it would be possible to produce both
isotropic and anisotropic test specimens. The testing machine that had to be devel-
oped for the experimental investigations needed to enable strain-controlled tensile
tests.
To compare different states of stress, uniaxial and biaxial experiments needed
to be carried out for different materials. A suitable rule for transforming the char-
acteristic parameters could then be developed, and a model for the simulation of a
multi-dimensional state of stress could be constructed.
3 Test Specimen and Testing Machine
A study of existing test specimens and experimental techniques for the multi-
dimensional testing of plastics led to the choice of a crucifonn test specimen. This
fonn of specimen was optimized by means of the ftnite-element method [4]. Figure
G2 shows the ftnal geometry of the test specimen.
Injection gate Production of a uniaxial

test specimen in flow direction
(similarly for transverse direction)
Fig. G2. Geometry of the test specimen (injection moulded)

The test specimens were produced by injection moulding by means of a double

tool (Fig. G3). To enable wriaxial tensile tests, wriaxial specimens were cut out of
the biaxial test specimens. In this procedure, test components were taken both
parallel and perpendicular to the flow direction.
The testing device built for the experimental investigation is shown in Fig. G4.
This device enables both strain- and force-controlled tensile tests.
Nozzle Side Clamping Side
Fig. G3. Injection mould for production of test specimens
Fig. G4. Biaxial testing device
During the test, both the forces at the clamps and the strains in the centre of the
specimens were measured by force sensors and strain gauges. By means of scale
factors evaluated by the FEM, the measured forces could be converted into
Modelling oftbe Mecbanical Bebaviour 537
stresses in the centre of the test specimen. The scale factors depend on the state of
stress. Therefore, the scale factors had to be evaluated for each state of stress to be
examined.
4 Experimental Investigations
To examine the influence of multi-dimensional states of stress on the mechanical

behaviour of thermoplastics, first of all the influences of the multi-dimensional
state of stress and of the anisotropy caused by alignment of the molecules have to
be separated. This requires the production of isotropic (non-oriented) and aniso-
tropic (oriented) test specimens. To achieve this aim, two approaches were used:
1. Production of test specimens with a defined orientation by means of suitable
processing parameters (process strategy).
2. Use of materials with different viscosity, which means with different flow
properties (material strategy).
The materials listed in Table G 1 were examined in the experiments.
Table Gl. Materials examined
Type Remarks Symbol Name of product/manufacturer
SBS block Impact-modified polystyrene SBS Styrolux 637 DIBASF AG

copolymer
PA6 Standard polyamide 6 with PA6 Durethan B30SIBayer AG
good flow properties
SBS has a higher dependence of the viscosity on the shearing rate. In addition,
the viscosity level is, in many situations, much higher than that ofPA 6. Therefore,
the SBS tends more towards an oriented structure. The orientations can be influ-
enced by the processing parameters.
4.1 Quasi-Static Experiments
Figure G5 shows the anisotropy as a function of the processing parameters. Uni-

axial tests and equally biaxial tensile tests on SBS are shown. It can be clearly seen
that high processing temperatures (the mass temperature TM , mould temperature TT
and injection-speed Vi are given) lead to lower anisotropy, which is discernible in
the smaller separation of the stress curves longitudinal and transverse to the flow
direction of the filling process.
25.-------r-------,-------,-------r------,-------,
SBS
20
23'C
0.1 %/min -------1-------
~ Equally Biaxial
~ 15 +---------r-----~~~--~~~~~~--r_----~~~------,
e
III
~ 10 +-------~~~~~~~1_~~1---------L-------~--------,
en -0- Longitudinal 220/40/20
Transverse, 220/40/20
...... Longitudinal, 240/60/40
Transverse, 240/60/40
...
TM ('C) I Tw ('C) I VE (mmls)
O~-------r------~--------+-------~-------+------~
o 0.25 0.5 0.75 1.25 1.5

Strain (%)
Fig. GS. Results of uniaxial and equally biaxial tensile tests on SBS
Unlike the SBS, PA 6 shows isotropic behaviour (Fig. G6). The stress-strain
curves longitudinal and transverse to the flow direction hardly differ.
80.------,------,------,------,-------,------,-----.
PAS Equally Biaxial
70 23 'C, Dry (Evaluated from
0.1 %/min Uniaxial Curve)
270/80/60
60
IV 50 +-------4+------~------~------~~----~------_+------~
Q. T M ('C) I T w ('C) I VE (mm/s)
:E
-; 40 +----+-----f-----«I=--~
!II
II:!
~ 30+------+----~~~~~----~~~----f------~----~
20
O~----_+------r_----~----~------+_----_+----~
o 0.25 0.5 0.75 1.25 1.5 1.75

Strain (%)
Fig. G6. Results of uniaxial and equally biaxial tensile tests on PA 6

540 E. Schmachtenberg, M. Wanders, N. M. Yaziei
Therefore, dynamic experiments were perfonned, as described in the next sec-

tion; the results make it clear that it is necessary to consider the load history.
4.2 Dynamic Experiments
Figure 08 shows the results of an alternating-creep experiment using conditioned

PA 6 under a 15 MPa: 0 MPa state of stress. First, within one minute, an external
force of 1700 N was applied in the flow direction to the test specimen. Simultane-
ously, a force of 144.5 N was applied in the perpendicular direction, i.e. at right
angles to the flow direction (t), to create a 15 MPa: 0 MPa stress state. The forces
used were calculated with the help of the scale factors mentioned in Sect. 3. After
a holding time of 20 minutes (load step 1), the state of stress was changed to a 0
MPa: 15 MPa state and then held again for 20 min (load step 2). Then the forces
along both axes were reduced to zero within one minute and, finally, were held at
zero for 20 min.
1750 1.75
...-~
I
1500
1250
..... ~
_ ~
~
-a-Force.1 I--- 1.5
--Force. t I--- 1.25
1000
f7 I ---Strain. I
f---
Load Step 1 Load Step 2 --Strain. t
750 PA6.Cond. I--- 0.75
g 500
23°C
0.5 ~c
~ l'a.t--,
0 250 0.25 ]i
II..
en
o
0
-250
.-
.
~.25
-500
1\ ~ No Loading
~~ r--.., I.... I
~.5
-750 ~.75
~ooo ~
o 5 10 15 20 25 30 35 40 45 50 55 60 65
Time (min)
Fig. GS. Results ofaltemating-creep experiment on PA 6 (conditioned, 15:0 MPa)
The strain versus time curves shown in Fig. 08 clearly demonstrate the creep of
the material due to the loading. In the first load step there is a spontaneous strain in
the flow direction of 0.94 %, whereas there is a transverse strain of -0.47 %. In the
first load step the material creeps to 1.48 % strain in the flow direction and to
-0.76 % strain transverse to the flow direction. The strain transverse to the load
direction is determined by the contraction behaviour of the test specimen. The
second load step leads to a spontaneous strain of 0.82 % in the direction of the
higher force and to a strain of -0.13 % in the direction of the stress-free axis, i.e. in
On the assumption that the multi-dimensional state of stress has no influence on

the secant modulus, the stress-strain curve for an equally biaxial stress can be
calculated from the uniaxial curve by means of the extended Hooke's law:
I
a"=--a'. (GI)
I-v
The assumption of a constant Poisson's ratio of v = 0.4 leads to the calculated
curve shown in Fig. G6. The difference between the calculated and the measured
curve could be explained by the influence of the multi-dimensional state of stress
on the secant modulus, or it could indicate a large deviation from the assumed
Poisson's ratio.
Figure G7 shows a possible way to transfer the secant modulus from one state
of stress to another. The figure shows the secant modulus for the equally biaxial
tensile test and for the uniaxial test, obtained from the results in Fig. G6, as a
function ofthe cone reference stress (m = 1.3) in the centre of the test specimen.
8000
PA6
7000 r-" - -- 23 ·C. Dry
U~iaxial 0.1 %/min
T M (·C) IT w (·C) I Ve (mmts) .... 270/80/60
6000
--
N~ 5000
E
~~
~ 4000
( ~\ ----.."'-
-
Equally Biaxial I---
W 3000
2000 I
1000
o
o 5 10 15 20 25 30 35 40
Cone Reference Stress (m = 1.3) (MPa)
Fig. G7. Secant modulus as a function of cone reference stress (m = 1.3)
The graphical representation in Fig. G7 demonstrates that a transformation of a

characteristic function of the material, the secant modulus, is possible by using the
cone reference stress when the load is quasi-static [4].
This method does not lead, however, to the ability to deal with a multi-
dimensional state of stress for non-linear viscoelastic materials. A distinct property
of non-linear viscoelastic materials, the memory function of plastics, is not rele-
vant to the result of quasi-static tests and is not covered by the above transforma-
tion either.
the flow direction. In load step 2 the material creeps in the tensile direction to
1.57 % strain, and transverse to the load direction to -0.71 % strain. After the
forces are set to zero, the strain transverse to the flow direction is reduced to
0.2 %, and the strain in the flow direction to 0.06 %.
The results of this experiment show the influence of the load history on the ma-
terial properties. The strain curves in load step 2 here are affected by the previous
stress in load step 1.
-
1000 0.5
900 - 0.45
-<>- Force, I
800 ...... Force, t - 0.4
700 ~ __ Strain, I - 0.35
600
r- _ Strain, t - 0.3
500 ~ Load Slep 1 Load Slep 2 0.25
SBS, 23'C L
~ 400 0.2 ~
~ 300 0.15 c
0
u. 200 0.1 ~
100 0.05
~
0 0
-100 No Loading
-0.05
-200 -0.1
-300 \ -0.15
~ '-
-400 -0.2
o 5 10 15 20 25 30 35 40 45 50 55 60 65
Time (min)
Fig. G9. Results of alternating-creep experiment on SBS (7: 0 MPa)
1000 0.5
-
900 -<>- Force, I r - 0.45
800 ........ Force, t r - 0.4
700 r - 0.35
600
500
... -- Strain, I
r--
...... Strain, t f--
0.3
0.25
~ 400 -
Load Slep 1 Load Slep 2 SBS, 23'C f-- 0.2 ;{!.
~
e....
~ 300 0.15 c
u. 200
0 0.1 ~
100 0.05
0 0
-
-100 No Loading - -0.05
•
-200 -0.1
-300 ~ I -0.15
-400 -0.2
o 5 10 15 20 25 30 35 40 45 50 55 60 65
Time (min)
Fig. GIO. Results of alternating-creep experiment on SBS (7:0 MPa), stress first applied
transverse to flow direction
The influence of the load history is shown more clearly when an anisotropic
material, for instance SBS, is subjected to an alternating stress. Figure G9 shows
the curves of the measured loads and strains for the SBS material when the load
(7 MPa) is first applied in the flow direction and then alternated.
Figure G 10 shows the corresponding curves for the case in which the stress is
first applied transverse to the flow direction.
Comparing the points of Figs. G9 and G 10, it is clear that the sequence of the
stresses has an influence on the behaviour of the anisotropic material. This indi-
cates the influence of the multi-dimensional load history on the material behaviour.
Figure GIl shows the results of a biaxial alternating-relaxation experiment with
dry PA 6.
0.7 21
!'cooa. --Strain, I
0.6
0.5
- -- Strain, t
-0- Stress, I
~
~
18
15
--Stress, t
~
~
c:
0.4
0.3
PA6, Dry
23 ·C
I 12 iii
9
a.
e
Ul
~
(J)
Ul
~
0.2 6 U5
0.1 I - - Load Step 1 Load Step 2 Load Step 3 3
0.0
. 0
1P 2P 30 40 ~ ~ 60 70
-0.1 -3
Time (min)
Fig. GIl. Results of biaxial alternating-relaxation experiment on PA 6 (dry, 0.5 %)
First, the longitudinal strain, parallel to the flow direction (I), was increased
within one minute to 0.5 %, and the transverse (t) strain was held at 0 %. This
corresponds to a strain-controlled experiment. After 20 min holding time in load
step 1, the strain transverse to the flow direction was increased to 0.5 % and the
longitudinal strain was reduced simultaneously to 0 %.
Load step 2 followed (holding time 20 min) and, after all strains had been re-
duced to 0 %, another holding time of 20 min (load step 3) completed the experi-
ment.
It can be seen the previous history of the relaxation in load step 1 has a clear in-
fluence on the second relaxation step (load step 2). After the load change, there is
a slight stress increase in the flow direction (I) at the beginning of load step 2,
caused by the relaxation in load step 1. If there was no simultaneous strain increase
to 0.5 % transverse to the flow direction, relaxation of the compressive stresses
caused by the decrease in strain in the flow direction would be the result. Owing to
the strain increase transverse to the flow direction, these compressive stresses are
compensated by lateral contraction effects, resulting in the tensile stress shown.
The experiments described above prove that different load histories lead to dif-
ferent material behaviour. A transformation based on energetic assumptions about
the actual state is not sufficient for a multi-dimensional calculation of the non-
linear viscoelastic behaviour. By considering the test results, a model for the
analysis of multi-dimensional states of stress was developed.
5 Development of the Model
On the basis of the tests described above, the following requirements were placed
on the model that was to be constructed:
- Non-linear viscoelastic effects have to be reproduced.
- Anisotropies must be taken into account.
- The influence of a multi-dimensional state of stress must be taken into account.
- The model must be usable with any load history.
- It must be possible to calibrate the model by means of data taken from standard
test methods or from the CAMPUS database.
- It must be possible to extend the model to take account of thermal stresses.
- Implementation in fmite-element programs should be possible.
Figure G 12 shows the basic element of the three-dimensional deformation
model developed for the simulation of multi-dimensional states of stress. The basic
element consists of an elastic element and a damper system (three-dimensional)
coupled together to describe the viscous properties. The three-dimensional basic
element can be regarded as a three-dimensional Maxwell element [4].
The elastic properties of the three-dimensional basic element are described by
the elastic modulus Ek,i and the Poisson's ratio Vkle,i'
Figure G 13 shows the principle of the operational mode of the three-dimen-
sional damper. The flow speeds in the three space directions depend on the pres-
sure differences applied to the dampers. Starting from the pressure differences,
regarded as being synonymous with stress differences, the plastic strains in each
direction can be evaluated by means of flow lines. The flow line of an element
describes the flow stress as a function of the temperature, the applied stress differ-
ence and the plastic strain rate. It is assumed that there is volume conservation for
flow within the damping system. This implies that flow leads to a volume rear-
rangement within the damping system.
Three-Dimensional Damper
(Damper System)
Vp ,; = const
Hookean Element
E 1,1. V 2 10
.1 . V 310, 1
E 2,1 . V 12e,l . v:ne,1
E 3.1 • V 3",1.
1 V 23,,1
O"l.,i
Valid for the Three

Space Directions:
Sk.i = e ke,i + Skp,i
Fig. GI2. Three-dimensional basic element (three-dimensional Maxwell element) [4]
Fig. GI3. Three-dimensional damper [4]

Modelling ofthe Mechanical Behaviour 545
For a more precise representation of the material behaviour, it is necessary to

arrange a number of these three-dimensional basic elements in parallel (Fig. G14).
In this way, it is possible to represent more precisely the relaxation behaviour of
non-linear viscoelastic materials.
For the parallel arrangement shown in Fig. G14, it is necessary that the sum of
the single-element stresses in each direction is equal to the total external stress:
n
CTI = LCTl,i , (G2a)
i=l
n
L
CT2 = CT 2,; , (G2b)
;=1
n
CT3 = LCT3; • (G2c)
;=1 '
Simultaneously, it is necessary that the external strain Ck for every basic element
consists of an elastic strain Ckf!,i and a plastic strain Ckp,i:
Cl = cl e,; + clp,i' (G3a)
(G3b)
C3 = c3e,; + c3p,; . (G3c)
Fig. G14. Three-dimensional defonnation model [4]

The model parameters were calibrated by means of uniaxial, strain-controlled,

isothermal tensile tests at different temperatures.
In [4], rules for calculating relaxation and retardation processes by means of the
three-dimensional deformation model are set out. In Fig. GIS, a comparison be-
tween an experiment and a simulation performed by means of the three-
dimensional model is shown for an equally biaxial tensile test. Since retardation
has not been included yet in the program, the stress in the free 3-direction was held
manually at zero. The model predicts too strong a relaxation. This difference could
be due, on the one hand, to the fact that the model was calibrated with stress-strain
curves from the CAMPUS database, recorded in accordance with ISO 527 and not
with a constant strain rate. On the other hand, the moisture content, the crystallin-
ity and therefore the morphology of the PA 6 have all a strong influence on the
mechanical behaviour. A better calibration should lead to better results.
40
~ /0
-0-- Stress 1, 2, Three·Dimensional Model
/ /V
__ Stress 3, Three-Dimensional Model ..;'
35
.............. Measured 2, Transverse
- Measured 1, Longitudinal .'~' ..rr'
.. 'V
30
~ T = 23·C. : = 0.2 %/min I /. '
.' .'
V/
25 PA6,Dry J
Iii'
a.. 20
5
VI
VI
15 ~V
V
~
CiS
10
5 #
0
/ ~
o 0.2 0.4 0.6 0.8 1.2

Strain (%)
Fig. G1S. Comparison of simulation and experiment for equally biaxial tension
6 Discussion
To consider and make precise predictions for non-linear viscoelastic materials

under multi-dimensional states of stress, the influence of the load history on the
material behaviour should be looked at, as well as the memory effect of the mate-
rial.
The three-dimensional deformation model described here enables one to make

predictions for multi-dimensional states of stress taking account of the non-linear
viscoelastic behaviour of the material. It is possible to use the model to evaluate
the stresses in the material for any multi-dimensional external stresses, even under
non-isothermal conditions. The inclusion of retardation in the program for the
three-dimensional deformation model will lead to an improved result for externally
applied stresses and for combined stress-strain loads.
The model predicts too large a relaxation under multi-dimensional load. One
reason could be in the calibration of the model parameters. The stress-strain
curves used for fitting were evaluated according to ISO 527 with a constant
crosshead speed, for which, strictly speaking, a constant strain rate cannot be as-
sumed. Further possible causes are variations of the temperature, moisture and
morphology of the test specimens, as well as the choice of the constant k value
used for the time-temperature shift according to the Arrhenius equation.
The aim of further investigations should be the inclusion of thermal expansion
to evaluate thermal stresses in the algorithms of the model. A subsequent embed-
ding of the model into finite-element simulations would create new possibilities for
simulating the thermo-mechanical behaviour of plastic components.
7 Comments and Prospects
The proposal to include effects of thermal strain in the algorithms of the model and
then to use the model in finite-element simulations offers new possibilities for
predicting the behaviour of real components. By means of such a model, predic-
tion of thermo-mechanical effects under external loads and prediction of shrinkage
due to internal stresses are likely to be possible.
Acknowledgements
We thank the German Research Foundation (DFG) for the financial support of this
research project and we also thank BASF AG, Ludwigshafen, Germany and
BAYER AG, Leverkusen (Germany) for the supply of the test materials Styrolux
637 D and Durethan B30S.
References
1. Schmachtenberg E. (1985): Die mechanischen Eigenschaften nichtlinear viskoelastischer

Werkstoffe. PhD thesis, RWTH Aachen, Germany
2. Bardenheier R (1982): Mechanisches Versagen von Polymerwerkstoffen. Kunststoffe -
Fortschrittsberichte, Vol. 8. Carl Hanser, Munich, Vienna
3. Sarabi B. (1984): Das Anstrengungsverhalten von Polymerwerkstoffen infolge ein- und

zweiachsigen Kriechens, Ermittlung von Langzeitbemessungskennwerten. PhD thesis, Uni-
versity of Kassel, Germany
4. Wanders M. (1999): Beitrag zur Entwickiung eines Modells zur Beschreibung des mechani-
schen Verhaltens nichtlinear viskoelastischer Werkstoffe unter mehrachsiger Beanspruchung.
PhD thesis, University of Essen, Germany
G.2 Detergent Resistance of PP/GF Composites
w. Grellmann, S. Seidler and C. Bierogel, Merseburg,
and R. Bischoff, Berlin
1 Introduction
The increasing requirements for lightweight construction are leading to a wider

acceptance of alternative materials for purposes where metals have traditionally
been used. As a result of new technologies, these materials are being used for the
substitution of conventional materials in components and structures. Polymer ma-
terials are finding increasing application in components subject to thermal and
detergent loading. At present, thermoplastics based on mass-produced polymers
are favoured because of their well-known advantages of properties, processing and
price. The reservations about the use of these materials are due to a lack of knowl-
edge about the parameters relevant to structural use and the lack of reliable mor-
phology-property correlations.
As a result of an inadequate description of the properties, an understanding in-
sufficient to allow consideration of multi-axial loading, and the viscoelastic defor-
mation behaviour of these materials, failures of polymer materials occur in prac-
tice. The choice of materials and the dimensioning of components are made
difficult by incorrect or unavailable values of parameters, and misjudgements of
multi-axial loading states are common. In the case of simultaneous thermal loading
and exposure to detergents, for instance in the lye tubs of washing machines (Fig.
G16), special problems occur, such as a severe reduction in lifetime. Normally, a
simulation of the long-time behaviour is impossible, and therefore experimental
investigations of the loading and deformation behaviour under the actual condi-
tions of use are essential. A successful replacement of high-grade steel in washing
machine lye tubs will require an accurate knowledge of the long-time behaviour.
Generally, the conditions of high detergent resistance and thermal stability can be
satisfied only by a reinforced polymer material. The material chosen for investiga-
tion, polypropylene with 30 wt. % short glass fibres, shows sufficient strength and
toughness, but the long-time behaviour under conditions of thermal loading and
detergent exposure is largely unknown.
Preliminary investigations of the mechanical behaviour and processing and the
construction of injection-moulding tools were carried out by BSH Bosch Siemens
Household Appliances GmbH, Berlin. Basically, a linear elastic analysis of the
strength by means of the fmite-element method (FEM) and a study of the flow
behaviour were used for a theoretical prediction of the strength and dimensional
stability of various short-fibre-reinforced polypropylene materials. The require-
ments on the material, based on a lifetime of 1000 hours, were the following:

550 W. Grellmann, S. Seidler, C. Bierogel, R. Bischoff
• static loading under its own weight and the weight of the lye tub when filled
• various dynamic tests related to the washing and spinning processes
• exposure to detergent at temperatures up to 95 DC
• impact loading during transport of the washing machine.
The critical factors affecting long-term reliability are loading by detergents,
water absorption at high temperatures (95 DC) and the vapour pressure. As a result
of varying wettability at the interface between water and vapour, a transition range
exists where the conditions under which diffusion and degradation occur are vari-
able. The aim of the investigations described here was the development of rules for
the use of polypropylene under such conditions, and of criteria for the choice of
materials. With this aim in mind, exposure tests were carried out on various poly-
propylene materials for up to 1000 hours and tests were performed to characterize
the material behaviour. The main topics of the program were the following:
• development of a test procedure for PP/GF composites
• investigation of resistance to detergents and vapour for various PP/GF com-
posites
• checking the sensitivity of the test methods chosen.
Laundry Input Water Input
Transport Protection
Fig. G16. Schematic representation ofa lye tub for a top-loading washing machine
2 Experimental
2.1 Materials
Polypropylene materials of various colours, with and without stabilizers, were used
in the investigations (Table G2). The materials were supplied by Targor, Mainz,
Germany; BSL, Schkopau, Germany; Borealis AG, Linz, Austria; and Thermofil,
Detergent Resistance 551
New Lane, Hampshire, UK. The specimens were conditioned for up to 1000 hours
in water vapour and in a detergent solution of 1 % washing powder at 95 °e.
All specimens were conditioned for 14 days under constant environmental con-
ditions after they were removed from the vapour or detergent solution at 100, 600
and 1000 hours. The initial state and the states at different ages were investigated
continuously.
Table G2. PP/GF composites chosen
Name Matrix material Fibre length (jlm) Fibre content (wt. %) }Idratio
PPI PP homopolymer 280 31.1 21.2

PP2 PP homopolymer 280 29.5 20.1
PP3 PP homopolymer 360 26.1 27.3
PPCI PP copolymer 300 22.1 22.7
PPC2 PP copolymer 290 20.6 22.0
2.2 Test Methods
The evaluation of the behaviour and suitability for long-term use of the materials
was carried out by means of conventional mechanical tests and fracture mechanics
methods. Specimens prepared according to ISO 3167 [1,2] were used for the char-
acterization and measurement of strength, deformation and toughness (Table G3).
Table G3. Test methods chosen for characterization of materials
Test methods Parameters Test conditions
Mechanical tests
Tensile test Et, OM,.sa DIN EN ISO 527
Flexural test Er, CTtM, CTfJ.5 DIN EN ISO 178
Ball-indentation hardness test HK and HK function DIN EN ISO 2039
Conventional toughness tests notch sensitivity
Flexural impact test (Charpy) acu DIN EN ISO 179
Flexural notch impact test (Charpy) acN DIN EN ISO 179
Fracture mechanics test
Instrumented Charpy impact test JIlT, Old MPKICIT
Heat resistance
Vicat test method Tv DIN 53460
Hybrid tests DIN EN ISO 527
Acoustic-emission analysis OSE, SsE, amplitude v-r= 2 mm!min
Laser extensometry bl, 61, heterogeneity He v-r=5mm!min
In addition, the instrumented Charpy impact test, with a pendulum hammer

speed of 1.5 mis, was used to quantify the toughness behaviour related to the re-
sistance against unstable crack propagation. The load-deflection diagrams re-
corded were analysed by assessment of the CTOD and of the J-integral by the
method of Sumpter and Turner (pp. 71-86).
In addition to the above tests, hybrid methods were also used. These methods
were used to detennine the damage kinetics and to quantify the criteria to be ap-
plied to the strength and defonnation of the materials. Use of such methods is
necessary because there are many problems with the application of specimen val-
ues to polymer structures. The cause of these problems is the wide variation of
thickness in the elements used in practice and the fact that the states of orientation
and residual stress in specimens and practical structures are not comparable. A test
method that can be applied to detect damage is acoustic-emission analysis (pp.
385-402, [3,4]). This method detects the onset of failure in composites in relation
to the state of stress or strain, when the failure mechanisms are active. The pa-
rameter amplitude distribution, which can also be calculated, allows one to assess
changes in the damage process caused by the environmental factors. A detailed
analysis of the source mechanisms, such as debonding or fibre fracture, is very
rarely possible. In addition to the acoustic-emission analysis, laser extensometry
was perfonned under tensile loading (Table G3). The reflectors required were
printed by a silk-screen method, with a spacing of 8 mm; by the use often fringes,
an initial total gauge length of 80 mm was created. By the additional use of micro-
structural methods, conclusions were drawn about the defonnation-detennined
changes of material behaviour from assessments of the local defonnation behav-
iour and calculation of the heterogeneity of the specimen volume investigated (pp.
365-384, [5]).
3 Results
3.1 Mechanical Characterization of Materials
The results of the exposure tests for up to 1000 hours showed the same general
tendencies for the various PP/GF composites under tensile and flexural loading.
Only the results of the exposure to detergent solution are shown here, because this
led to poorer values, relative to the initial state, than did the exposure to vapour.
For all materials investigated, a rapid decrease in strength by up to 20 % within the
first 1000 hours was found (Fig. G 17). The cause of this behaviour is the effect of
the heat treatment at 95°C. The superior perfonnance of the material PP3 is a
consequence of the extraordinarily high lid ratio of the glass fibres used (Table
G2). The initial states of the two copolymers show, as expected, lower strength
values than do the PP homopolymers, but at above 100 hours of exposure an un-
explained increase in strength occurs for PPC 1. The changes in the modulus of
elasticity caused by the heat treatment are different for all materials, but are insig-
nificant (Fig. G 18). Even though the loss of elasticity after 1000 hours amounts to
only 2 or 4 %, it will be necessary to investigate longer exposure times because the
tendencies are nearly linear. With increasing exposure time, the tensile strain at
break of the PP/GF composites decreases rapidly.
"
110
Wash Lye
100
\
~ 90
-........ '- -..... / ·PP3
V
~
1 80
~
70
\
~
~ \.L ~
r-..:- ~ ·PPI
·PPCl
60 --... r-- ·PPC2

~ ·PP2
50
o 200 400 600 800 1000
. 1200 1400
Exposure Time (h)
Fig. G17. Tensile strength OM ofPP materials as a function of exposure time
1.04 ...,.----.,..---.---,....---.---....----.-----.
Wash Lye
·PPI
1.02 +-I--+---+--+~=l_-=~--_I_-___1
~
-g 1.00 ..fE=----+---+--_I_~y~--+--_I_=___1
~ ·PP2
J:I
"3 0.98
.t:i -+---+----+-...::::!Io.....,..-"':::O'k-~"'d,"'_"...c..._+_-__I
]o
Z 0.96 +----11----+--1---+--+-..::....=+-=:-=-::--1
·PP3
0.94 +--.--1-...,....+--.-+--r-4--.,..--I--.-......j.-......-l
o 200 400 600 800 1000 1200 1400
Exposure Time (h)
Fig. GIS. Change of normalized Young's modulus Et as a function of exposure time
3.2 Evaluation of Toughness by Means of Fracture Mechanics
The results of conventional characterization by the Charpy impact test using un-
notched specimens, shown in Fig. G 19, demonstrate a separation of the PP materi-
als into two levels of toughness. For these test conditions, the values acu are nearly
the same for PP3 and the copolymer PPC 1. In order to clarify the structural causes
of the toughness behaviour, a fracture mechanics evaluation, complemented by
microstructural investigations, was performed. The load-deflection diagrams
measured during the instrumented Charpy impact tests were evaluated by the pro-
cedure described in A2.2 (pp. 71-86), and the resulting fracture mechanics pa-
rameters were used for quantification of the toughness.
70
Wash Lye
60
~
----
b ,................. ·PPCl
~ /
\ ---- ·PP3
~ r---.
·PPI
20 r--- ~ ·PPC2
. ~ "PP2
10
o 200 400 600 800 1000 1200 1400
Exposure Time (h)
Fig. G19. Charpy impact strength acu as a function of exposure time
For the razor-blade-notched specimens, the parameters .AlT and the critical crack-
opening displacement were calculated. Because a one-sided description of the
deformation-determined crack-opening displacement can lead to an overestimate
of the toughness behaviour, the evaluation was based on the energy-determined J-
integral (Fig. G20). If the .AlT values shown in Fig. G20 are used for the evaluation
of toughness, it becomes clear that the energy absorption capacity decreases with
increasing exposure time. The reduction in toughness can reach up to 50 % relative to
the initial state. With the exception of the decrease in the JIlT values ofPPC1 above
600 hours of exposure, no significant changes of resistance against unstable crack
propagation occur above 100 hours for the PP/GF materials. So, a state of nearly
constant behaviour is reached for the materials.
In contrast to the results of the conventional test methods, there exist clear dif-
ferences between the materials PP3 and PPC 1 in the initial state. These intensified
differences are caused by the higher sensitivity of the fracture mechanics parame-
ters. The high level of toughness for PPC 1 is caused by the deformation-
determined material behaviour (Fig. G21) and the dependences of bid (not shown).
The load-deflection diagrams in Fig. G21 reflect, for the other materials, a linear
elastic behaviour. So, in the case of unnotched specimens, PP3 and PPCI give
comparable areas under the load-deflection curves. At 600 hours, for all materials
investigated, the upper limit of toughness is reached (Fig. G20). The dominant po-
30
-- --
Wash Lye
25
1\
\
20
-PPCl
~~ /
·PP3
10
/--- ·PPC2
.~ ·PP2
5
~ -PPI
o
o 200
. 400 . 600 SOO 1000 1200
. 1400
Exposure Time (h)
Fig. G20. Influence of detergents on JIlT
SOO~----~----~--------~----~----~--------~----~----~
Wash Lye
PP3
0.4 O.S 1.2 1.6

Deflection f (nun)
Fig. G21. Load-deflection diagrams for PP/GF composites
556 W. Grellmann, S. Seidler, C. Bieriigel, R. Bischoff
sition of PPC 1 is caused by the very efficient adhesion between fibres and matrix
as well as the successful copolymerization, whereas the same procedure gives rise
to no increase of toughness for PPC2.
3.3 Hybrid Experimental Test Methods
Simultaneously with tensile testing, acoustic-emission measurements were carried

out for all materials. The amplitude and energy of acoustic emission obtained show
a wide range of behaviour among the materials investigated. In comparison with
the initial state, the amplitudes and the emitted energy decrease slightly after 100
hours of exposure.. Above an exposure time of 600 hours, the intensity of acoustic
emission decreases rapidly (Fig. G22). The heat treatment during exposure causes
a reduction of the orientation and residual stress produced by processing. At 1000
hours the emission reaches a minimum and the amplitude maximum is shifted to
lower values because the equilibrium between damage and adaptive processes is
displaced. The decrease of the high signal amplitudes, which is expressed in a change
30
Lye
25
,-.. 20-lf---+--f---+--f----1 ,-..

e e..-
~
il~+-~~~~rl--1
llo+-~-r~~~~
llS
~ ~
AE-Amplitude (dB)
,-..
e..-
~
»
g 15 ~I
!
!l.,.
.,
~I
~ ~
AE-Amplitude (dB) AE-Amplitude (dB)
Fig. G22. Amplitudes of acoustic emission for PP 1 after different exposure times
of the distribution function, is a result of the increasing damping, debonding of

fibres and domination of pull-out processes. For PP3, this observation agrees with
the rapid decrease of the critical stress O"SE (Fig. G23) above 600 hours.
The effects described above are also shown in the heterogeneity calculated
from the local deformations (Fig. G24) (pp. 365-384). With the exception ofPP3,
all materials show an increase of the heterogeneity He at 100 hours, which is caused
80
Wash Lye
70
...- 60
§
g50
'"'"
~
CI)
1l 40
:E
u 30
20
0 200 400 600 800 1000 1200 1400
Exposure Time (h)
Fig. G23. Critical stress CTSE of acoustic emission as a function of exposure time
70,-----r----r----~--_,----~----r_--_,
Wash Lye
o 200 400 600 1000 1200 1400

Exposure Time (h)
Fig. G24. Heterogeneity He oflocal deformations for PP materials (pp. 365-384)

by deformation-determined rearrangement processes due to annealing. If the

specimen size and volume are kept nearly constant, the degradation of the orienta-
tion and residual stress caused by manufacturing can be observed only in local
deformation processes, which can be detected by laser extensometry.
After a time of 600 hours, a clear reduction in the heterogeneity occurs. This de-
crease is caused by a transition to a constant state of equilibrium in the specimens.
However, Fig. G24 shows very different levels of heterogeneity among the speci-
mens at an exposure time of 1000 hours. Above 600 h of exposure, a decrease of
heterogeneity occurs, showing that an approximate state of equilibrium is being
approached, but this state has not yet been reached for all PP/GF materials at an
exposure time of 1000 hours.
A similar statement, but with a lower local measurement resolution, can be
made on the basis of ball-indentation measurements on the specimen surface. The
determination of the ball-indentation hardness on the surface of dumb-bell speci-
mens shows differences between the parallel part and the shoulder that are caused
by the state of orientation. The hardness in the area of the shoulders is lower than
that in the parallel part of the specimen. Hardness curves measured at different
points over the specimen length of PPC2 at times of 100 and 1000 hours show
different levels for the various measuring points (Fig. G25). Relative to the initial
state an increase of hardness occurs at 100 hours. This effect, which is caused by
the annealing, is similar to that seen using laser extensometry. At 1000 hours the
differences between the parallel part and the shoulder decrease as a result of plas-
ticizing effects of the detergent and an increasing tendency towards a state of
equilibrium.
70
"'(\ 1\ \
1\ \
,,\\~\\\
r-I \. \. :.:
! I ~
--
I / / / "'" lOOOh
:J. ~ •.-'1
PPC2
L.fJ-fi / I
03'\' I
II
II II / II 1 \\\ \
If ~ ~\" - PPC2
30
o
. .
20 40 60 80 100
. 120
lOOh
140
Point of Measurement (mm)
Fig. G25. Ball-indentation hardness HK ofPPC2 as a function of measuring point and time
4 Conclusions
The aim of these investigations of the thermal and environmental resistance of
selected fibre-reinforced PP materials to a detergent at a temperature of 95°C was
the development of a complete test programme for such materials.
The assessment of the PP/GF composites was performed by means of me chani-
cal characterization, measurement of various fracture mechanics parameters and
hybrid methods of polymer diagnostics. For various times (or states) of exposure,
the parameters were compared with those of the initial state of the materials in
order to simulate the variation over the lifetime of the components. The final levels
of the parameters have been presented in graphs and tables in comparison with the
initial state of the PP/GF composites investigated.
The results show that comparable matrix materials have very different long-time
resistance to the detergent used, owing to differences in the geometry of the glass
fibres, in the stabilizers and applied colours, and in the coupling agents. The de-
crease of the properties, which is expressed in the loss of toughness and strength as
well as in the reduction of deformation capability, depends strongly on the expo-
sure time and is caused by thermo-oxidative degradation and the effects of surfac-
tants at a temperature of about 100°C. As a result, the adhesion between the ma-
trix and glass fibres is influenced especially strongly, and so the dominant damage
processes are fibre pull-out and micro-crack initiation at the ends of the fibres.
The best materials for the proposed purpose are the homopolymer PP3 and the
copolymer PPC 1, even though these materials show deficiencies concerning heat
resistance and notch sensitivity. All of the investigations carried out indicate that
not only the fmal values of the parameters but also their changes during the life-
time are of importance. As a result of these reflections, additional exposure tests
for times up to 2000 hours, with more specimens being taken at intermediate times,
have been arranged. The main purpose of this extensive test programme, is the
evaluation and interpretation of the dominant damage processes of interest.
The basis on which one might recommend a test for the suitability of PP/GF
composites is, logically, the sensitivity of the test methods and of the chosen pa-
rameters regarding the possibility of distinguishing materials and regarding the
response of the material to environmental factors. Here, the application requires
parameters characterizing elasticity, toughness, heat resistance, and strength and
deformation behaviour. On the one hand, conventional impact tests, hardness
measurements and flexural tests provide insufficient selection criteria. On the other
hand, acoustic-emission analysis and laser extensometry are too expensive for a
simple test of suitability or an inspection test, particularly because of the amount of
data produced. Therefore, the following examination methods for tests of the suit-
ability ofPP/GF materials are proposed:
• tensile test, with determination of Young's modulus E" strength and tensile
strain at break
• ball-indentation hardness
• instrumented Charpy impact test, with determination of critical crack-opening

displacement and J values
• heat resistance according to the Vicat test.
If there is a requirement for detailed information on the state of the material or
on the damage kinetics as well as on the relevant failure mechanisms, hybrid ex-
perimental methods should be used in connection with microstructural examina-
tion. The experimental results described here also show that to determine the opti-
mal toughness, strength and deformation reserve under conditions of combined
environmental and thermal influences, special investigations concerning compati-
bilizer effects, fibre content and stabilization are necessary.
References
1. ISO DIS 3167 (1993): Plastics - Multipurpose Test Specimen

2. Breuer H., Dupp G., Jantz R., Wiibken G., Tiba M.H., Tiillmann R. (1994): Campus vor
weltweiter Verbreitung, Stand der intemationalen Normung. Kunst. - Ger. Plast. 84:
890-894
3. Bohse J., Kroh G. (1992): Micromechanics and AE-analysis of failure process of thermo-
plastic composites. J. Mat. Sci. 27: 298-306
4. Ritter R., Schmitz B. (1996): Photothermal inspections of adhesion strength and detection of
delaminatioDS. In: Balageas D., Busse G., Carlomagno C. M. (Eds.) Quantitative InfraRed
Thermography. Proceedings of Eurotherm Seminar No. 50, Stuttgart, Germany, September
2-5: 197-202
5. Grellmann W., Bier6gel C., K6nig S. (1997): Evaluation of deformation behaviour of poly-
amide using laserextensometry. Polym. Test. 16: 225-240
G.3 Material Optimization
of Polypropylene-Short-Glass-Fibre
Composites
B. Langer, C. Bierogel and W. Grellmann, Merseburg,
J. Fiebig and G. Aumayr, Linz
1 Introduction
The adaptation of material characteristics to specific applications by using differ-

ent fillers and reinforcements and/or using composite materials is an international
trend in the development of the properties of polymers. A necessary condition for
specific requirements on the strength and toughness level of polymers to be satis-
fied is a knowledge of the deformation and fracture mechanisms [1,2].
Owing to the continuing improvement of the mechanical-property spectrum of
polymers and a better descriptiveness of the deformation and fracture behaviour of
these materials, polymers are being used as substitutes for metallic materials. Fur-
thermore, in order to produce lightweight structures, it is necessary to use poly-
mers with favourable mechanical characteristics [3,4].
Using as an example the consideration of polypropylene/glass fibre (PP/GF)
composites for use in the lye tubs of washing machines (Fig. G26), a process of
material optimization regarding the glass fibre content and the assessment of the
deformation and fracture behaviour after environmental and thermal loading is
described in this chapter.
Fig. G26. Lye tub for a washing machine made from a PP/GF composite

562 B. Langer, C. Bierogel, W. Grellmann, J. Fiebig, G. Aumayr
2 Experimental
Three different PP matrix materials were used in the composites investigated: one
commercial polymer (PP), one copolymer (PPC!) and one combination of two
different copolymers (PPC2). These were used with volume contents of glass fi-
bres between 0 and 0.26. The assessment of the effectiveness of the fibre rein-
forcement and of the deformation and fracture behaviour under conditions of envi-
ronmental and thermal loading was performed using mechanical parameters
obtained from conventional tensile tests and fracture mechanics parameters related
to resistance against unstable and stable crack propagation determined from in-
strumented Charpy impact tests (lCITs). The quantification of the toughness val-
ues was carried out on the basis of the load--deflection behaviour by using linear
elastic fracture mechanics (LEFM), elastic-plastic fracture mechanics (EPFM) and
the crack resistance (R-) curve (Table G4) (pp. 71-86). Additionally, the hybrid
test method of laser extensometry was used for the characterization of the local
deformation behaviour and the heterogeneity of the specimen (pp. 549-560, [5,6]).
Table G4. Fracture mechanics parameters determined from the ICIT
Fracture mechanics concept Parameters
LEFM Fracture toughness

KId
EPFM Critical J value

JId
Critical crack-opening displacement
bid
R-curve Tearing modulus
TJ
3 Results
The results of the tensile tests as a function of fibre content (Fig. G27) show that
the modulus of elasticity in tension and the tensile strength increase with increas-
ing fibre content, as expected. However, at higher fibre contents, the tensile
strength shows a smaller increase than does the modulus of elasticity.
The composites with a PP matrix reach the highest values of the modulus of
elasticity and tensile strength. The graphs of the tensile strain at break (Fig. G27c)
and of the stress--strain behaviour (Fig. G27d) make it clear that there is an abrupt
change in the deformation behaviour as a result of the fibre reinforcement. The in-
vestigations of the resistance against unstable crack propagation lead to the result
Material Optimization 563
16
r
120
~ ~
/
(;'
12 (;'
~ 90
::E ~
6
----
'0
~
8
6
60
4 30
0 0
0 0.1 0.2 0.3 0.4 • PP 0 0.1 0.2 0.3 0.4
<I> v
A PPCl
<I> v
50.0 • PPC2 100
~ ~
37.5 ~ 75
,-.,
??
(;' .~ PPC2/GF (<(Jv = 0.03)
';'25.0 ~ 50
'" b
PPC2
12.5 25
II -
0 0
0 0.1 0.2 0.3 0.4 0 10 20 30 40
<I> v E(%)
Fig. G27. Results of tensile tests
that in the composites examined here, as well as in other composites with a ther-
moplastic matrix [7-9], the strength and deformation behaviour can be influenced
in various ways by the fibre reinforcement (Fig. G28). The energy-determined
value J Id (Fig. G2Sc) shows, as does the stress-determined value KId (Fig. G28a),
an increasing trend up to a fibre content of CfJv = 0.13. With higher fibre contents,
the toughness behaviour of the material is characterized by a dominant deforma-
tion constraint (Fig. G2Sb), and the resistance against unstable crack growth is
reduced. Therefore, the optimal fibre content for the toughness behaviour is
CfJv = 0.13.
This behaviour, however, is not so clearly pronounced for the composites with a
PP matrix. The energy-determined crack toughness value "'-d of the composites
with a PP matrix does not increase above CfJv = O.OS, whereas the KId values de-
scribe a stress-determined increase of the crack toughness behaviour with increas-
ing fibre content. It is clear from these results that only a multi-parametrical de-
scription of the deformation and fracture behaviour allows statements about the
micro-mechanical mechanisms that are active (pp. 3-26).
The techniques of fracture mechanics provide a very extensive range of meth-
ods for material testing that makes - if the techniques are correctly utilized and
applied - structure-oriented statements about materials possible. For interpretation
of the mechanisms and phenomena of the crack process, an exclusive reference to
the tensile strength, to the modulus of elasticity obtained from the stress-strain
curve or to the stress intensity factor obtained from the load versus load-line dis-
placement curve, for example, is not sufficient.
564 B. Langer, C. BieriSgel, W. Grellmann, J. Fiebig, G. Aumayr
30
a
S
~ 20
~ 10 PP
J PPCI
• PPC2
0
30
~
i
~:~
200
~
'-' 100
.J ~
0
20
c
16
E
.E 12
6-
....
;!! 8
0
4
~
0 0.1 0.2 0.3 0.4
(jIv
Fig. G28. Resistance against unstable crack growth as a function of fibre content: Kid (a), ~d
(b), Jld (c)
Figure G29 shows, using micrographs of the fracture surface of the PP/GF
composites with 'Pv = 0.04, the differences in toughness level between the various
matrix materials.
For a structure-oriented assessment of the crack behaviour, it is necessary to
analyse the complex connection between load, crack size, and resistance against
unstable and stable crack propagation. Crack resistance (R-) curves were deter-
mined by using the stop block method and the mUltiple-specimen method to esti-
mate the resistance against stable crack propagation. Figure G30 shows that the
resulting R-curves of the PP/glass fibre composites are characterized by a delayed
onset of stable crack growth, which shifts to greater il.a values with increasing
fibre content.
PP
KId Bid JId

(MPamm ln) (mm) (N/mm)
121 0.09 4.07
Assessment of Fracture Surface:
Frequent Pull-Out, Scaly Matrix

Areas, no Fibre-Matrix Adhesion
PPCI
Kid BId JId

(MPammln) (mm) (N/mm)
146 0.17 8.53
Pull-Out, Small Deformed Matrix

Areas, Low Fibre-Matrix
Adhesion
PPC2
Kid BId JId1IT

(MPa mmln) (mm) (N/mm)
147 0.20 10.28
Seldom Pull-Out, Strongly

Deformed Matrix Areas, Fibre-
Matrix Adhesion
Fig. G29. Micrographs of fracture surfaces ofPP (a), PPCI (b) and PPC2 (c) with fPv = 0.04
Furthermore, it should be observed that the crack propagation behaviour is con-

siderably changed by the fibre reinforcement; this is expressed in the lower slopes
of the straight lines describing the crack propagation. Because the technical initia-
tion parameter ~.2 determined in accordance with the recommendations of the
ESIS P2-92 standard [10] does not represent the crack resistance behaviour ofthe
PP/glass fibre composites examined here, the tearing modulus To was determined
566 B. Langer, C. Bieriigel, W. Grellmann, J. Fiebig, G. Aumayr
in order to provide a unified description of the changes in the crack propagation

behaviour due to the reinforcement (Fig. G31). The resistance against stable crack
growth is drastically decreased by the reinforcement, whereas the maximum reach-
able amounts of crack growth increase continuously. This means that the matrix
materials examined here exhibit the most favourable properties concerning resis-
tance against stable crack growth.
I
,Pel + 0.26 GF S"'j .... 4S""J S""]
Specimen 2
~ j
120
..-..
~
'"(:)
.....
'-'
80
<0
- - PPC2
40 --0- 0.03
--0- 0.08
- 0 19
0
2 3 -0- 0.26
Ila (nun)
Fig. G30. R-curves and load-deflection diagrams ofPP/glass fibre composites obtained using
the stop block method
r-------------------------------------~8
6 6
o 6. exP.
max
4 4
2 2
•• •
0~------~----~-------+------~------4,0
o 0.1 0.2 0.3 0.4
qlv
Fig. G31. Tearing modulus To and maximum reachable amount of crack growth .iamaxexp as a
function of fibre content rp.,
•
Considering the crack-opening displacement velocity t5 dk as a function of the
amount of crack growth (Fig. G32), it can be assumed that the values for the com-
posites with 'Pv = 0.03 and 0.08 are nearly constant. However, for the pure matrix
material and for the composites with higher glass-fibre contents, a constant value
•
of t5 dk does not occur until a defined amount of stable crack growth has been
passed. When this amount of stable crack growth has been passed, as described in
pp. 3-26 and pp. 281-300, the energy necessary for the crack growth increases
(right part of figure) and the matrix material consumes this amount of energy for
significantly smaller amount of crack growth (left part of figure).
Investigations were carried out on PPC2 with 'Pv = 0.13 to assess the strength
and toughness behaviour after environmental and thermal loading because this
composite had been characterized as toughness-optimized. The results in Fig. G33
show that the modulus of elasticity and the tensile strength are changed only
slightly, unlike the fracture mechanics values after storage in detergent solution for
2000 hours at a temperature of 90°C, but the tensile strength exhibits a stronger
drop than the modulus of elasticity (Fig. G33a). The graph of the fracture me-
chanics parameters related to resistance against unstable crack propagation J1d and
bid in the Fig. G33b that two processes separated in time lead to a decrease in the
toughness properties as a function of the exposure time.
.... - ...
0.16 10.
'";;;' 0.12
PPC2
e 7.5
0.16
'ie 0.08
. - ~
.€
r-~ 5.0
0.04 2.5
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.03 0.06 0.09 0,1
0.12 Lla (mrn) adk (m/s)
x. . ...
~ DD
+~
rJl
:§, 0.08 x ••
. x
D
Fibre Content
.0.03 • 0.19
0.04
• 0.08 D 0.26
x 0.13
o~----------------~==~====~
o 2 3
Lla (mrn)
Fig. G32. Relation between crack-opening displacement velocity, amount of stable crack
growthandJ
568 B. Langer, C. Bierllgel, W. Grellmann, J . Fiebig, G. Aumayr
10.0 160
a
PPC2 + 0.13 GF
S-
..
,..... 7.5
c
120
-;;-
~ ~ 0 0 0 Po
6
~
'0
5.0 80 :l;
--
~
b
W
aM
2.5 40
12 b 120
'?E
~
...
...:;
9
6
--
aId
90
60
~
.,
<:>
.s
'-"
J ld
~
3 30
0 800 1600 2400
Exposure Time (h)
Fig. G33. Influence of envirorunental and thennal loading on modulus of elasticity and tensile
strength (a) and onJld and ~d (b)
0.5 ---------~a
0.4
_ .!'~lyt!l~r-r - -___ _
0.3 /' - /
:C' I Copolymer
0.2
0.1
o0 200 400 600 800 1000

Exposure Time (h)
0.5.------=----.:....:......--,
0.4
:r: 0.3
20
250
0.2 ....... 15 200
0.1
CIO 150
iij 5 100
50 t (s)
00 0 1
4 5 10 0
Reflectors
Fig. G34. Relations between heterogeneity and exposure time (a), local stress distribution (b, d)
and integral strain (c)
Exposure to detergent solution for 100 h leads to the first reduction in proper-
ties because of annealing effects. With continued exposure up to a time of 600 h,
first a stabilization occurs and then the toughness properties are reduced a second
time because of degradation effects. The parallel behaviour of .lrd and bid in Fig.
G33b shows that the reduction in toughness can be traced back to a decrease in the
defonnation capability.
The local defonnation behaviour, measured by means of laser extensometry,
was quantified by means of the heterogeneity calculated from the measurements
(Fig. G34) (pp. 549-560, [5,11]). An increase of the heterogeneity He (pp. 365-
384) in comparison with the initial state was noticed after an exposure time of 100
h; this is attributed to defonnation-detennined rearrangement processes as a result
of annealing.
A degradation of the orientation and residual stress induced during manufacture
occurred because of local defonnation processes, although the external shape did
not change and the volume remained nearly constant. After 600 h, the heterogene-
ity decreases because the local defonnation is approaching an equilibrium state.
4 Summary
The investigations aimed at optimizing the glass fibre content of the material
showed that the composites with the PPC 1 and PPC2 matrix materials and a fibre
content of f/Jv = 0.13 have the optimum resistance against stable crack growth. The
resistance of the composites against stable crack growth is considerably lower than
that of the pure matrix materials. The modulus of elasticity and the tensile strength,
detennined in tensile tests, increase steadily with increasing fibre content; the
materials with the PP matrix have values a little smaller than those of the materials
with the PPC 1 and PPC2 matrices.
The defonnation and fracture behaviour is influenced in varying ways by the
environmental and thennal loading. After the exposure to detergent solution, the
modulus of elasticity and tensile strength are slightly reduced. The fracture me-
chanics values decrease because of two chronologically separated processes,
which can be traced back to annealing and degradation effects. Laser extensome-
try, which characterizes the local defonnation behaviour, was used to assess the
heterogeneity of specimens. The laser extensometry investigations demonstrated,
as did the fracture mechanics investigations using the ICIT, that the behaviour of
the composite materials examined was strongly influenced by a change of the de-
fonnation behaviour.
Acknowledgements
The authors thank Bosch and Siemens Household Appliances GmbH, Berlin, in
particular Dr. R. Bischoff, for cooperation in the studies of environmental effects
570 B. Langer, C. Bieriigel, W. Grellmann, J. Fiebig, G. Aumayr
on the impact behaviour of glass-fibre-reinforced PP materials. Our thanks go too

to Dr. H. Kaminsky, Dipl.-Ing. W. Posch, Dipl.-Phys. J. Fiebig and Dipl.-Ing. G.
Aumayr for providing the materials and the successful cooperation.
References
1. Herakovich C. T. (1997): Mechanics of Fibrous Composites. Wiley-VCH, Chichester
2. Karger-Kocsis J. (1995): Microstructural aspects of fracture in polypropylene and its filled,
chopped fibre and fibre mat reinforced composites. In: Karger-Kocsis J. (Ed.) Polypropyl-
ene. Structure, Blends and Composites. Vol. 3, Composites. Chapman & Hall, London:
142-201
3. Cahn R. W., Haasen P., Kramer E. J. (1993): Structure and properties of composites. In:
Chou T.-W. (Ed.) Material Science and Technology - A Comprehensive Treatment. Wiley-
VCH, Chichester
4. Hashemi S., Kooghilani M. (1995): Fracture toughness of injection moulded glass fiber
reinforced polypropylene. Polym. Engng. Sci. 35: 1124-1132
5. Grellmann W., Bierogel C. (1998): Laser extensometry applied. MaterialprUfimg - Mater.
Test. 40, 11-12: 452-459
6. Langer B., BierOgel C., Grellmann W., Aumayr G., Fiebig J. (2000): Material optimization
of PP-short glass fibre composites. In: Proceedings of Polymeric Materials P'2000, Halle,
Germany, September 25-27: 462
7. Grellmann W., Seidler S., Lauke B. (1991): Application of the J-integral-concept for the
description of toughness properties of fibre reinforced polyethylene composites. Polym.
Compos. 12,5: 320-326
8. Grellmann W., Seidler S., (1992): J-integral analysis of fibre-reinforced injection-moulded
thermoplastics. J. Polym. Engng. 11, 1-2: 71-101
9. Langer B. (1998): Bruchmechanische Bewertung von Polyamidwerkstoffen. Logos, Berlin
10. ESIS P2-92 (1992): ESIS-Procedure for Determining the Fracture Behaviour of Materials
11. Grellmann W., Bierogel C., Konig S. (1997): Evaluation of deformation behaviour in poly-
amide using laserextensometry. Polym. Test. 16: 225-240
G
Examples and Limits

of Application
G.4 Influence of Exposure
on the Impact Behaviour of Glass-Fibre-
Reinforced Polymer Composites
H. Walter, C. Bierogel and W. Grellmann, Merseburg,
and B. Rufke, Schkopau
1 Introduction
Glass-fibre-reinforced composites based on thermosets show good mechanical
properties, such as high specific strength and stiffness, combined with an excellent
chemical resistance. An increasing use of such materials in a wide range of engi-
neering applications and structures, especially as lining materials for containers
and pipelines, can be observed. In recent years, many research programmes have
been studied the mechanical properties of composites. However, independently of
these investigations, reservations about the application of composite materials in
the chemical industry exist because exact criteria for lifetime, chemical resistance
and moisture ageing are not or are rarely available. Additionally, the evaluation of
the damage state of a composite, as well as of the functionality, reliability and
impact behaviour, is complicated. A successful failure analysis of structures made
from composites depends strongly on suitable non-destructive test and inspection
methods.
The presence of defects requires a basic knowledge of the residual characteris-
tics of the materials used. In the work described here, the critical impact behaviour
of glass-fIbre-reinforced composites as a function of test and application condi-
tions was investigated. Specimens and components subjected to various ageing
conditions such as exposure to detergent solution and thermal loading were inves-
tigated by means of an instrumented falling-weight tester. The damage areas of the
impacted laminates produced were assessed with modem non-destructive test
methods. In relation to this, the main aim of the present study was the enlargement
of knowledge of the processes of continuing damage and of the failure mecha-
nisms of laminates. Therefore, the impact behaviour of undamaged and damaged
glass-fIbre-reinforced plates and components was investigated as a function of
load and test conditions. Further, by means of the compression-after-impact (CAl)
test, new knowledge of the residual strength and fatigue life for undamaged and
damaged test specimens, taking account of chemical effects, is expected.
2 Experimental
2.1 Materials and SpeCimens
For the experimental investigations, undamaged glass-fIbre-reinforced composites

based on Derakane 470-36, from the Dow company, Rheinmunster, Germany,

572 H. Walter, C. Bierogel, W. Grellmann, B. Rutke
were used. This special polymer matrix is a modification of the epoxy novo lac.
According to the manufacturer, the most important properties are a high crosslink
density and a high chemical and thermal resistance. The production of fibre-
reinforced laminates was carried out according to DIN 16965/16966 Part 5, pipe
type E, by the Kurotec company, Stade, Germany.
For the comparison of different material states, test plates with an average
thickness of approximately 6.5 mm were also produced. The plates were coated on
one side with a special resinous gel-coat layer; this layer had no support function,
but was applied only for protection from aggressive liquids or gases (for use in
linings).
2.2 Test Methods
A conventional tensile test was used for an initial classification of the initial state
of the materials used, in particular to characterize the stiffness and the strength.
Besides a description of the strength behaviour, this fundamental mechanical test
provides qualitative details of the deformation behaviour under quasi-static load-
ing conditions. The investigations were performed on an Instron 4507 universal
testing machine. The test specimens were cut to the required dimensions of
150 x 20 x 6.5 mm3 . The determination of the mechanical properties was carried
out in accordance with the ISO 527-4 test procedure [1], using a crosshead speed
of 5 mmlmin and an initial gauge length of 60 mm. Five specimens were exam-
ined under standard conditions (T= 23°C and 50 % air humidity).
In order to study damage zones of different depths and areas, experiments were
performed by means of a Fractovis instrumented falling-weight tester, made by
Ceast, Italy. The CAl test, which combines a falling-weight test, ultrasonic dam-
age detection and a compression test, was based on the procedure BSS 7620 [2]
developed by Boeing. Impact events with different impact energies were achieved
with a dropping steel indenter with a diameter of20 mm. For the impact tests, the
specimens were clamped with a steel plate on the sides according to the Boeing
specification, and the impact damage was created on the surface opposite to the
gel-coat layer. Because evaluation is impossible if multiple impacts occur, a de-
vice was used to catch the impact hammer. The desired impact energies were
obtained by adjusting the striker mass (1.99 kg to 23.3 kg) and choosing the striker
velocity (1 mls to 4 mls). From the results of the preliminary examinations, the
following optimum energy levels were ascertained:
• damage in the specimen still detectable with ultrasonic methods (1.57 J)
• visible damage on only the top surface (6.30 J)
• visible damage on top and bottom surfaces (25.6 J)
• breakthrough (> 40 J).
The comprehensive characterization of the test materials was performed by
means of an HFUS 2000 high-resolution ultrasonic testing system (Co. Dr.-Ing.
Hillger, Braunschweig, Germany). With this technique, it is possible to inspect
specimens for production accuracy, delaminations and initial defects and also to
Influence of Exposure on the Impact Behaviour 573
analyse acoustic parameters such as the damping factor and ultrasonic velocity.
From ultrasonic A-scans at different positions of the specimen surface, the ampli-
tude spectrum and transition time within synthetic apertures were determined and
saved as ultrasonic C- and B-scans. For all specimens, the immersion technique
was used, where water served as the coupling medium and preflow line simulta-
neously. Three resonant transducers with different frequencies (5, 10 and 50 MHz)
were used in the investigations. Focusing occurred at the back surface of the
specimen (gel coat), and the echoes of flaws between the surfaces and echoes from
the back surface were analysed. By recording ultrasonic B-, C- and D-scans, the
areas of impact damage in the sample volume could be detected and quantified in
relation to the impact energy.
Specimens with impact damage, as well as undamaged laminates, were tested,
employing the compression-after-impact (CAl) test procedure. The determination
of the residual strength after the impact was performed using the Instron 4507
universal testing machine with a crosshead speed of 1 mm/min. For the generation
of various pre-damage states, exposure to media such as water, detergent solution
and phenol was used. These liquids were selected because it is well known that
they cause a decrease in mechanical properties. Furthermore, it can be assumed
that the damage mechanisms of bodies made from the resin considered here are
also influenced. As a result of these considerations, all samples were exposed to
these liquids in closed containers for a fixed period and temperature, as listed in
Table G5.
Table GS. Exposure conditions used
Media Immersion Immersion

time temperature
Water 3000 h 23°C

Phenol 1000 h 23°C
Water vapour 1000h 100°C
Detergent solution 1000h 95°C
Using the instrumented falling-weight impact test, load-time curves were recorded
and the energy under these curves was calculated. The characteristic impact be-
haviour of fibre-reinforced composites is shown in Fig. G35. The amount of en-
ergy absorbed is determined by the upper plateau of the energy-time curve after
the impact. Consequently, the relative energy absorption can be described by:
. A
Energy absorptIOn = ~ 100 % . (G4)
Amax
574 H. Walter, C. Bieriigel, W. Grellmann, B. Rufke
5.80 "T""--.,....--~--,.---r---....,... 6.30
4.64 +----I-+--+--\-\c--+----f-----+ 5.04
3.48 3.78
6:
~ 2.32 2.52 fil
~ Ji
1.16 1.26
0 0
0 0.8 1.6 2.4 3.2 4.0
Time (ms)
Fig. G3S. Typical impact-loading and energy profiles versus time
The typical types of damage of the material systems used, such as delamina-
tions and cracks, were inspected by ultrasonic testing techniques. The sound ve-
locity of the materials investigated was in the range of 2600-2800 mls and corre-
sponds to values in the literature [3,4]. The best results could be obtained with the
point-focused HI0MP15 (10 MHz) transducer. This transducer combines the
transmitter and receiver functions and produces sound of a high depth resolution
owing to its short pulses. In order to detect useful ultrasonic echoes for these ma-
terials, an amplification between 44 and 50 dB was necessary. Under these condi-
tions of frequency and amplification for the materials investigated, defects> 0.26
mm are detectable. Generally, all laminates show a high acoustic scattering and
acoustic absorption.
Figure G36 shows results of ultrasonic C-scans of impact-damaged specimens.
The C-scan can be understood as a projection of internal defects onto the speci-
men surface. For lower impact energies (A < 1.6 1), the damage area in the interior
of the laminates is not detectable with certainty. The visible damage areas were
evaluated as a function of impact energy by means of a colour image system.
The ultrasonic B-scan can be understood as a view of a section perpendicular to
the surface of the test specimen. When this technique is used, defects such as
delaminations are represented in a way that shows their depth in the material
(Fig. G37). In contrast to the C-scan, the ultrasonic B-scans were performed from
the impacted surface. The back-scattering echoes were shaded by the induced
delaminations. Because impact damage produces surface cracks in the protection
layer on the inner surface of the specimens, liquids or gases may intrude more
easily. With increasing impact energy, an increase of the damaged cross-section
occurs from the impact side to the back panel in a conical manner. Such critical
damage to the material leads to a reduction in the residual strength of components
and encourages instability. The delamination can also be demonstrated by means
of microstructural analysis. The fracture surfaces, when investigated by SEM,
show a low adhesion between the fibres and the polymer matrix (Fig. G38). In the
case of an impact load, the most important types of damage are fibre breakage,
debonding and matrix cracks. If the adhesion is poor, debonding and matrix fail-
ures are the critical damage mechanisms, and the low mechanical properties of
such laminates are explainable. A subsequent exposure to a medium reduces the
level of properties because diffusion processes damage the fibre-matrix interface.
x(mm)
Fig. G36. Ultrasonic C-scan of specimens with impact defects caused by different impact
energies
" (mm)
Fig. G37. B-scan of specimens with impact defects

576 H. Walter, C. Biertigel, W. Grellmann, B. Rufke
Fig. G38. SEM micrograph of a fibre-reinforced polymer specimen with impact damage
The smallest damage area detectable by ultrasonic methods was produced with
an impact energy of 1.6 J. At this energy level, 45 % of the energy input was ab-
sorbed. With higher impact energies, the energy absorption and the corresponding
area of induced damage increased. When the specimen was loaded with an impact
energy of 25 J, more than 80 % of the energy input was absorbed, and clear dam-
age on the top and bottom surfaces of the specimens could be seen (Fig. G39). The
specimens showed penetration behaviour if the impact energy was higher than
40 J.
In the first series of compression-after-impact (CAl) tests, the compression
strength of undamaged specimens was determined using a 100 kN testing ma-
chine. In the examination of the undamaged samples, the limit of the load cell of
the testing machine was sometimes reached without fracture of the plates. The
horizontal edges of the samples were clamped to avoid out-of-plane deformations,
whereas the vertical edges were simply supported. Nearly the half of the speci-
mens used showed local buckling in the centre of the plate during compressive
testing; only these results were useful for the evaluation of the residual strength.
The delaminations caused by impact loading reduce the compressive strength and
also the shear strength, and promote premature compressive failure. With in-
creasing impact energies up to 14.3 J, the compressive strength and Young's
modulus are nearly constant; the long-time effects of the subsequent exposure to
the various agents could not be estimated. This means that the influence of such
damage on the stiffuess is negligible. At higher impact levels >14.3 J, the compres-
sive strength was diminished by 15 % at 25.6 J, and at an impact energy of 46 J
the strength decreased by 40 % (Fig. G40).
In this study, samples were exposed to various liquids and water vapour for
fixed periods as listed in Table G5, and afterwards were tested under the same
conditions as used for the undamaged specimens. It could be seen that at impact
energies A ~ 25.6 J penetration occurred, if the specimens had been exposed to
detergent solution or water vapour (for specimens in the initial state, penetration
occurred only at A> 40 J). Generally, the impact damage areas (the delamination
around the impact) of the exposed specimens were smaller than the damage areas
of the unexposed specimens. At an impact load A > 15 J these exposed specimens
can absorb more than 90 % of the impact energy supplied (Fig. G41). Delamina-
tion in the fibre-matrix interface is the dominant mode of damage for these mate-
rials.
100 1000
---
_ .• __ ...4
80 ,• - ....
,
~.
800
/~
"
~
~
c
0 60 600 g~
,, .'
.~
'"
~ ~
0
'" .' E
«
~
,, ....0'"
40 400
~ 0
c
,~ '"
«e
,
~
20 200
~ ,
,,
o iii
0
o 5 10 15 20 25 30
Impact Energy (J)
Fig. G39. Relative energy absorption versus impact energy
140
~
. -
120 1-
~
"'"
6
100
\ I\.
-
~
~ t-.
c
g
CI)
.~'" 80
r--
e'"'"0.
E
-
0
u 60
40
o 10 20 30 40 50 60
Impact Energy (J)
Fig. G40. Compressive strength of specimens in the initial state

578 H. Walter, C. Bierfigel, W. Grellmann, B. Rutke
The materials exposed to phenol show an insignificant decrease in impact re-

sistance. At impact energies greater than 50 J, the energy absorption is comparable
to that of the materials exposed to detergent solution and water. The damage areas
of the phenol-exposed specimens are smaller than these of the unexposed speci-
mens, but larger than the damage areas of the specimens exposed to detergent
solution and water vapour.
lOO~----~----,-----~----~----~---,
Impact Energy (1)
Fig. G41. Energy absorption of specimens exposed to various agents
The Young's moduli, determined from the CAl tests, of the specimens exposed
to detergent solution and water vapour (not shown here) decrease on average by
600 MPa in comparison with the non-exposed materials. The moduli of impact-
damaged specimens which were exposed to phenol and water under the same
conditions are nearly constant. Figure G42 shows a comparison of the residual
compressive strength for the various exposure states. For the material in the 'ini-
tial state', it becomes clear that impact energies < 14 J do not influence the load
capacity and stiffness. If the energy overshoots this critical limit; a clear decrease
in compressive strength of up to 16 % occurs. In contrast, the stability of the
specimens investigated is negligibly influenced by smaller amounts of impact
damage. The specimens which were exposed to detergent solution and water va-
pour show a clear decrease in compressive strength at smaller impact energies
(Fig. G42). In comparison with the initial state, the exposed materials show a
reduction in stability. The test specimens stored in detergent solution have the
absolute minimum of residual strength. If higher storage temperatures are used,
there will occur a decrease in residual strength too. According to the suppliers'
information, crack initiation and a decrease in mechanical properties are observ-
able at temperatures above 50 °e. These effects are caused by hydrolytic degrada-
tion and an accelerated diffusion rate. The reinforced epoxies used here show just
a small decrease in the mechanical properties if they are exposed to phenolic solu-
tions, because saponification of the resin occurs. For further investigations of the
material systems considered here, higher storage temperatures and longer times
are necessary. In order to determine suitable limits for impact damage, residual
stability and strength and to estimate the lifetime of components, long-time inves-
tigations will be necessary.
140
_ .•. - Initial State
- 0 - Detergent
--.- Water Vapour
120
~
fa
5
l:I 100
en \
!!
"r;;
[
d 80
60~---------+----------~------~~------~
o 20 40 60 80
Impact Energy (1)
Fig. G42. Residual compressive strengths of specimens for different storage conditions
4 Conclusion
By means of the investigations described above, it was shown that microcracks are
produced in protective coatings even by small impact energies. These cracks rep-
resent direct points of attack and can lead to failures of components if environ-
mental and thermal loading occurs for long periods. Besides tests of stability and
stiffness, possible changes in the crack propagation behaviour should also be con-
sidered. In the literature, no information exists about the damage behaviour of
glass-fIbre-reinforced composites exposed to the liquids used here. In contrast to
the results of the investigations of [5] and [6], a reduction in the mechanical prop-
erties of specimens caused by storage in water (room temperature) for 4 months
could not be found. Obviously, longer observation times are necessary for such
tests. Altogether, the specimens exposed to the various media show a decrease in
compressive strength only at higher exposure temperatures and greater impact
energies.
580 H. Walter, C. Bieriigel, W. GreUmann, B. Rutke
The glass-fibre-reinforced composites investigated here are mainly used to

make chemical pipelines and containers. On the basis of the present results, an
initial statement about the critical impact failure of laminate components as a
function of test and application conditions, as well as about the long-time stability,
is possible.
In the chemical industry, glass-fibre-reinforced pipelines and containers are ex-
posed to random impact loads of various energies. Low loads (for example ham-
mer blows) produce microcracks in the coating applied for protection against
chemical attack, which are not visible from the outside. Under conditions of in-
dustrial production, these cracks can cause critical states, which can strongly in-
fluence the operational safety and service life. The detection, analysis and evalua-
tion of critical damage by means of non-destructive and optical test methods were
successful. As a result of this precise knowledge of the dominant failure mecha-
nisms, decisions on further use or repair of defective components, taking account of
economic considerations, are also possible.
Acknowledgements
The authors gratefully acknowledge the financial support of the Ministry of Cul-
ture and Education of Sachsen-Anhalt, Germany. We would like to thank Dow
Chemical AG, Buna Sow Olefinverbund GmbH, for its cooperation.
References
1. ISO 527-4 (1997): Plastics - Detennination of Tensile Properties - Part 4: Test Conditions
for Isotropic and Orthotropic Fibre-Reinforced Plastic Composites
2. Boeing Specification Support Standard BSS 7620 (1986): Advanced Composite Compres-
sion Tests
3. Hillger W., von Wachter F. K. (1991) : Impactverhalten von Verbundwerkstoffen unter dem
EinfluB lluBerer Spannungen. Mat.-wiss. Werkstofftech. 22: 217-224
4. Hillger W., Heinrich W. (1992): Rechnergestiitzte manuelle UltraschallprUfungen fUr die
CFK-Instandsetzung.ln: Tagungsband DGZFP-Jahrestagung 92, Fulda, 33 Part 1: 190-197
5. Pastuska G., Lehmann V., Trubiroha P. (1989): Alterungsuntersuchungen an GFK- und
PMMA-Platten durch schlagartige Beanspruchung. MaterialprUfung 31: 142-146
6. Mohd Ishak Z. A., Lim N. C. (1994): Effect of moisture absorption on tensile properties of
short glass fiber reinforced polybutylene-terephthalate. Polym. Engng. Sci. 34: 1645-1655
G.S Physical Ageing and Post-Crystallization
of Polypropylene
J. Fiebig and M. Gahleitner, Linz
1 Introduction
Among thermoplastic polymers, polypropylene (PP) is one of the most rapidly

developing product families owing to its property profile, the range of possible
modifications and its excellent environmental record [I]. Associated with this
development are a quantitative growth well above the average and a continuous
expansion of its application areas. Technological applications demand tailor-made
PP systems with an optimized adjustment of polymer design, processability and
properties in the final application. Here, it is often assumed that the properties of a
moulded component remain constant for its whole lifetime - even if the producers
and processors of PP are well aware that the properties of polymers change con-
tinuously over a very long period of time and even if the producers and processors
take account of this fact, e.g. by fixing the time between processing and standard-
ized testing.
For polypropylene, as a semicrystalline material, the changes with time of the
material properties ofPP can be roughly split into
I. chemical changes in the material
2. post-crystallization
3. physical ageing.
In contrast to other polyolefms such as polyethylene and olefin elastomers, the
chain-scission reaction mainly affects the high-molecular-weight content ofPP and
strongly reduces the average chain length; a fact which can also be used for prod-
uct design by controlled degradation of the melt with radical-forming agents such
as peroxides [2]. This is especially relevant as the high-molecular-weight fraction
strongly influences the mechanics of the material through its activity as an inherent
nucleant and as 'tie molecules' between different crystalline regions. The normal
consequence of chemical ageing is embrittlement, a strong loss in toughness.
The influences of chemical changes on the properties during use, however, can
be minimized by stabilizer systems adapted to the actual requirements of the con-
ditions of processing and application (heat, UV and y irradiation, and exposure to
various media).
2 Physical Ageing of Polypropylene
Even if a functional part produced from a polymer would normally be considered

sound after the processing steps, changes in the physical properties can be ob-
served for a long period of time. The modulus and toughness, in particular, un-

582 J. Fiebig, M. Gahleitner
dergo steady changes even after long times. Most studies dedicated to this phe-
nomenon have been carried out on amorphous materials in the past [3]. This be-
haviour is not even polymer-specific, as some very early studies concerning inor-
ganic glasses, amorphous metals and even sugar show. The reason for the observed
changes with time was found in the relaxation behaviour of these materials.
The properties of semicrystalline polymers are mainly determined by the crys-
talline morphology developed during the moulding process [4-7]. The main influ-
ence factors are here the polymer parameters that are fixed during synthesis, such
as average molecular mass, molecular mass distribution and chain regularity, and
the combination of processing conditions and crystallization dynamics [8]. Ac-
cordingly, the property changes after moulding have been explained as being due
to a slow optimization of the crystalline structure and have been termed 'post-
crystallization'. This may be justified for changes closely connected to the mould-
ing process, especially at elevated temperatures. Structural changes far below the
crystalline melting point (e.g. at ambient temperature for polypropylene) cannot be
explained in a physically reasonable manner this way. Here, the influence of the
amorphous (and possibly low-molecular-weight) fractions of the polymer, a pro-
portion of which is highly mobile, has to be taken into account. These fractions act
as links between the crystalline segments and influence strongly the mechanical
properties. This point of view is supported by the strong changes in properties and
ageing behaviour when the glass transition of polypropylene is crossed near 0 °e.
On the other hand, it might be expected that these changes would occur in a nar-
row temperature range around this glass transition, but that above this region the
amorphous fraction would no longer influence the temperature dependence, owing
to its high mobility. In contrast to this expectation, a continuous change of me-
chanical properties with increasing temperature is observed up to the crystalline
melting point, the reason for which can be seen in the strong interaction between
crystalline and amorphous regions.
Amorphous fractions close to crystalline segments, where chains may be par-
tially integrated into both elements, are strongly hindered in their mobility and
require a significantly higher activation energy for relaxation. The consequence is
a glass transition extended towards higher temperatures as compared with amor-
phous polymers [2] and, in the case of polypropylene, a continuous softening be-
tween Tg and Tm. As the glass transition is a phenomenon of thermodynamic equi-
librium, this also explains the dependence of the properties of a semicrystalline
polymer on multiple factors. These properties are influenced by time, loading speed
(or frequency) and temperature as well as - much more strongly than in the case of
amorphous materials - by the thermal pre-history of the sample (memory effect).
The situation is even more complex in the case of heterophasic (high-impact)
copolymers containing ethylene, which consist of a PP matrix with inclusions of
PE and EPR. Here, structural changes in the elastomer phase, with a glass transi-
tion between -60 and -30 oe, can occur even at lower temperatures. A summary of
the effects and temperature regions is given in Table G6.
Physical Ageing of Polypropylene 583
Table G6. Mobility regions for physical ageing of polypropylene (temperature indications refer
to homopolymers based on standard catalyst systems)
Region Changes
Glassy 0-4°C Density, mechanical properties

(Tg, Tp) a (slightly)
Amorphous 50-100°C Density, mechanical proper-

mobility (To) ties,
order in amorphous regions
(significantly)
Crystalline 50-100 °C Density, mechanical proper-
mobility (Ta) ties, crystallinity in the meso-
morphic phase
Recrystalliza- 145-160°C 162-167°C Melting point, crystallinity
tion (TR) (Tm) b (transition at -+ a2)
Melt relaxation 162-167°C Phase morphology in
(Tm) heterophasic systems,
superstructures
In random copolymers containing ethylene, Tg can be lowered to -10 °C.

In random copolymers containing ethylene and in terpolymers containing ethylene and
another a olefm, Tm can be lowered to 140°C
3 Consequences of Structural Changes

for Material Properties
Physical ageing of polypropylene can be monitored through the changes of its bulk
properties such as density, mechanical properties and crystallinity (determined by
DSC or X-ray scattering), or through relaxation phenomena (measured by
DMTA). If an altered organization in the amorphous phase the predominant effect,
this may well explain the increase in density, as well as - because of the improved
stress transfer between crystalline segments - the increase in stiffuess.
Changes caused by physical ageing continue on a logarithmic timescale for a
very long time [9] and can - in the case of a sufficiently sensitive measuring tech-
nique - still be detected after months and even years. In the case of the modulus
and toughness in particular, this behaviour can be easily shown (see Fig. G43).
An agreement to perform measurements after a 'standardized' period (e.g. 96 h
after moulding) is a useful compromise to compare properties of different materi-
als. The engineer using these data for design, however, must account for further
changes over the lifespan of the product. The increase in stiffuess observed during
ageing at the application temperature is always connected with embrittlement,
especially in the case of low crystallinity, for example in film processing with high
cooling rates in the solidification step (see Fig. G44).
Apart from mechanical properties, ageing also influences 'secondary' properties

such as transparency, light scattering and stress whitening [10], and even - ifindi-
rectly - the reaction to radical-forming influences (UV and y radiation, solvent
effects, ESCC).
1.2
!
!I •..
i
1.1
~~~~
~ 1.0 .CI
·C
~
•
C
oi
~ 0.9
C
103 10'
Ageing Time (It)
Fig. G43. Influence of the storage time on the relative flexural modulus of injection-moulded
parts made from two PP homopolymers (open squares, Daplen KS 10, standard
grade, EB (96 h) = 1300 MPa; filled circles, Daplen KS 44 N, nucleated grade, EB
(96 h) = 1900 MPa)
120 1.6
~i
• • • 1.2
~
S
clOO
• J
!] 80
".p
x
0.8 §
.::;
!
0.4 Po.
x x
)C
x
60 o
o 50 100 ISO
Time of Conditioning (It)
Fig. G44. Development of the secant modulus (filled circles) and the penetration energy
(crosses) of quenched PP films as a consequence of physical ageing (data from
Schael [11])
4 Effects of Annealing and Physical Ageing

on Polypropylene
Physical ageing is a consequence of non-equilibrium effects below the glass tran-

sition temperature of the chain segments concerned. In purely amorphous polymers
the glass transition occurs in a relatively narrow temperature range, below which
ageing, together with the changes in properties mentioned above, occurs. The
progress of ageing depends strongly on the difference between the ageing tem-
perature and Tg• By raising the sample temperature above Tg, all ageing effects can
be erased, provided the exposure time at the higher temperature is long enough for
complete relaxation.
This is no longer possible in the case of semicrystalline polymers. As a conse-
quence of the extension of the glass transition towards higher temperatures, the
additional order created by the ageing process is erased only in those amorphous
regions that become mobile at temperatures up to the annealing temperature. For
regions with higher mobility restrictions, ageing is just accelerated. A complete
erasure of ageing effects is only possible by melting and reshaping the sample.
Thus, memory effects can exert a significant influence on the real-life properties of
such systems.
4.1 Prior Studies in this Area
When one scans the technical literature, it soon becomes obvious that the ageing
effects of PP have been investigated rather rarely in comparison with their techni-
cal importance. This may be connected to the fact that such investigations are very
time-consuming if carried out properly.
The fundamental effects of ageing at room temperature were outlined in the first
studies [11-13] using mechanical, density and X-ray investigations. No changes in
the crystallinity (as seen in WAXS or DSC) were found, but a steadily increasing
density with an approximately log-linear time dependence and an approximation
in the density between the crystalline and amorphous regions were determined by
SAXS. All this points to an increased state of order in the amorphous regions as
the dominant effect of ageing. From gas sorption and diffusion experiments (Ar,
Ne, CH2Ch) [12,19], the scale of these structural changes could be assessed, also
placing the changes in the amorphous or mesomorphic phase.
If the storage temperature is increased in an annealing process, the room-
temperature processes are not simply accelerated but new processes appear, finally
including recrystallization effects near the melting point [14]. In a series of papers
by Struik [15-18], PP was investigated, among other semicrystalline thermoplas-
tics, in particular with regard to the technologically relevant process of ageing
under external loading. Two more recent papers [20,21] deal with heterophasic
systems; here, changes in the matrix phase are combined with a change in the
elastomer morphology at higher temperatures, leading to corresponding changes in
mechanical behaviour.
586 J. Fiebig, M. Gableitner
4.2 New Investigations on Ageing and Annealing of PP
All investigations in the present work were performed on a highly crystalline PP

homopolymer; the relevant parameters of this material are summarized in Table
G7. Mechanical properties (flexural modulus according to ISO 178, Charpy im-
pact strength of notched specimens according to ISO 179 1eA and heat deflection
temperature according to ISO 75-2) and the density were determined on injection-
moulded samples of dimensions 80 x lOx 4 mm3 . In parallel, compression-
moulded plates 2 mm thick were used for torsion pendulum and WAXS investiga-
tions. DSC was omitted, as preliminary studies had shown no effect of ageing, in
accordance with the literature [13].
The results of a determination of the parameters after 24 h at room temperature
were used as reference values for all measurements. In addition to a series aged at
room temperature with measurements after 24, 96, 336, 1563 and 6144 h (ap-
proximately logarithmic timescale), samples were annealed at 80, 110 and 140°C
for 24 and 168 h. This annealing was followed by room-temperature storage and
measurements, which commenced after 2 h and were continued in the same way as
the pure ageing experiments.
Table G7. Key parameters of the PP grade used for ageing experiments (mechanical data
determined 96 h after moulding, according to standard)
Type MFR 230 °C/2.16 kg Flexural modulus Impact strength (+23 0c)
(glIO min) (MPa) (kJ/m2)
PP-HOMO 8.0 2100 2.2
3200
-;0-
~2800
rg
-=
~
01
B2400
ii::
2000
0.905 0.910 0.915 0.920
Density (glcm')
Fig. G45. Relation between density and flexural modulus for the PP-HOMO material in storage
at +23 °C without prior annealing (plus signs) and after annealing (80 °C124 h, open
squares; 80 °C/168 h, filled squares; 110 °C/24 h, open diamonds; 110 °C/168 h, filled
diamonds; 140 °C124 h, open circles; 140 °C/168 h, filled circles)
In the case of a simple acceleration of the ageing process in the annealing pro-
cedures, a universal correlation between density and flexural modulus would be
obtained. This, however, is not the case (see Fig. G45), indicating changes in the
material structure.
An interesting variation is also seen in the relation between flexural modulus and
impact strength (Fig. G46). While for ageing at room temperature, the development
of increasing flexural modulus combined with embrittlement is well known from
the literature, each annealing step improved the toughness of the material, resulting
in a higher 'equilibrium level' after subsequent ageing.
4.5 ,...-----"'T""-----r------,
!
h'
3.5 +-----=-~,.__--__+----_l
Ii 2.5 -t--'..---,~---+---"'..,._--"'od__::---__I
.§
80 ac
lloac
I.S-t--.-__.,...-..__+~-r__-r-_+__.-..__""T""_I
2000 2400 2800 3200
Flexural Modulus (MFa)
Fig. G46. Relation between flexural modulus and impact strength for the PP·HOMO material
in storage at +23'C without prior annealing (plus signs) and after annealing
(80'CI24 h, open squares; 80'C/168b, filled squares; 11O'CI24b, open diamonds;
110 'C/168 b, filled diamonds; 140 'C124 h, open circles; 140 'C/168 h, filled circles)
The effects of annealing on the heat deflection temperature are extremely

strong; room-temperature ageing, however, has no significant effect here. Starting
at a value of the HDT of 125°C, obtained according to ISO 75-2, values up to
151 °C were reached after annealing at 140°C. Additionally, Fig. G47 shows a
dramatic change of the shape of the recorded deflection curves: with increasing
annealing time, a plateau in the temperature-dependent deflection is reached,
which ends in a second increase only after the value of the annealing temperature
is passed. These effects also illustrate the sensitivity of the HDT experiment, in its
present standardized form, to small changes in the material. A temperature-
dependent determination of the modulus, as carried out in the torsion pendulum
test, appears to be better suited for material characterization.
Here, the annealing results in a significant change of the relaxation processes in
the material (see Fig. G48). The loss angle curve (tan 0) shows sharper maxima at
o°C (Tg) and 60-100 °C (Ta), with the latter transition also shifted to higher
588 J. Fiebig, M. Gableitner
temperatures. The increase in the HDT becomes obvious from the dynamic-
modulus curve (0').
The WAXS studies carried out on the 2 mm plates also showed no change of
the overall crystallinity in the room-temperature ageing, but a measurable effect as
a result of annealing. From this it is clear that annealing at higher temperatures
involves a mixture of relaxation and recrystallization processes, which do not
occur at room temperature even after long times.
0.4
0.3
S
.§,
= 0.2
'B..
..
~
Cl
0.1
0
20 60 100 140 180
Temperature eC)
Fig. G47. Deflection curves obtained according to ISO 75-2 (loading 0.45 MPa) for the PP-
HOMO material
104 0.15
:
-... ....
-=e::,
"/'
103 - - - ......... -- ------- r---,r--
V
I 0.10
gj
"3
"0
~
0
102 /. ~
<> "J~
.~
S 101 1\ j 0.05
U
,~
~-
..-
10° o
-150 -100 -50 o 50 100 150 200
Temperature eC)
Fig. G48. Torsion pendulum results for an annealed sample stored at +23 °C for 24 h (full
lines) and for a sample annealed for 24 h at +140 °C (dashed lines) for the PP-
HOMO material
5 Summary and Conclusions
In the physical ageing of polypropylene without external loading, two effects can
be distinguished:
• changes in the amorphous and/or mesomorphic regions of the material at room
temperature and below or slightly above, which lead to an increase of density
and modulus combined with embrittlement
• relaxation and recrystallization processes at higher temperatures (observed in
our study at above 80°C), which influence positively the HDT and impact
strength along with the modulus and density.
Evidence in the literature can be found for both effects, even if the experimental
data do not appear to be sufficient for a complete explanation. The investigations
presented here for the case of a PP homopolymer support the above assertions, but
a second series of measurements on a heterophasic EP copolymer would supple-
ment the results so as to provide a better understanding of the more complex be-
haviour of such materials.
In the light of the currently available data, an acceleration of ageing by a tem-
perature increase to obtain long-term data wiIllead to irrelevant results. The limits
of such procedures have been pointed out before in a study performed at the UK
National Physical Laboratory [23] of ageing under an external load.
Acknowledgements
The work presented in this paper was partially conducted within the framework of
the BRITE-EURAM project BE 95-2067, 'Structure development during solidifi-
cation in the processing of crystalline polymers' (DECRYPO), with fmandal sup-
port by the European Commission. The authors are grateful to Dr. C. Paulik and
Ing. H. Wolfschwenger, Borealis AG, Linz, Austria for support and helpful discus-
sions.
References
1. NeiBi W., Ledwinka H. (1993): Polypropylene - the future has only just begun. Kunstst. -
Germ. Plast. 83: 228-235
2. Hammerschmid K., Gahleitner M. (1998): Controlled rheology PP. In: Karger-Kocsis J.
(Ed.) Polypropylene - An A to Z Reference, Chapman & Hall, London
3. Struik L. C. E. (1978): Physical Ageing in Amorphous Polymers and Other Materials. El-
sevier, Amsterdam
4. Friedrich K. (1979): The effect of morphology on the strength and fracture process of poly-
propylene. Kunstst. - Germ. Plast. 69: 796-801
5. Fujiyama M., Wakino T. (1991): Distribution of higher-order structures in injection-

moulded polypropylenes. J. Appl. Polym. Sci. 43: 57-81
6. Liooauer S., Eder G., Janeschitz-Kriegl H., Jerschow P., Geymayer W., Ingolic E. (1993):
On the kinetics of shear induced crystallization in polypropylene. Int. Polym. Proc. 8: 236-
244
7. Jarus D., Scheibelhofer A., Hiltner A., Baer E. (1996): The effect of'skin-core' morphology on
the heat deflection temperature of polypropylene. J. Appl. Polym. Sci. 60: 209-219
8. Gahleitner M., Wolfschwenger J., Bemreitner K., NeiBl W., Bachner C. (1996): Crystalli-
nity and mechanical properties of PP-homopolymers as influenced by molecular structure
and nucleation. J. Appl. Polym. Sci. 61: 649-57
9. Wolfschwenger J., Paulik C., Fiebig J., Gahleitner M. (1996): Nachkristallisation von Poly-
propylen: Konsequenzen filr die Mecbanik. In: Proceedings of Polymerwerkstoffe '96 Mer-
seburg, Germany, September 18-20: 77-80
10. Altendorfer F., Frank H. P. (1980): Investigations of stress whitening of isotactic polypro-
pylene due to deformation. Kunstst. - Germ. Plast. 70: 98-101
11. Schael G. W. (1966): A study of morphological and physical properties of polypropylene
films. J. Appl. Polym. Sci. 10: 901-905
12. Kapur S., Rogers C. E. (1972): Aging of quenched polypropylene. J. Polym. Sci. A-2 10:
2107-2124
13. Agarwal M. K., Schultz J. M. (1981): The physical ageing of isotactic polypropylene.
Polym. Engng. Sci. 21: 776-781
14. Guerra C., Petraccone Y., Corradini P., DeRosa C., Napolitano R., Oirozzi B., Giunchi G.
(1984): Crystalline order and melting behaviour of isotactic polypropylene (alpha form). J.
15. Struik L. C. E. (1987): Mechanical and physical aging of semicrystalline polymers. Polymer
28: 1521-1533
16. Stroik L. C. E. (1987): Physical aging in isotactic polypropylene. Polymer 28: 1534-1542
17. Struik L. C. E. (1989): Mechanical behaviour and physical aging of semi-crystalline poly-
mers. Polymer 30: 799-814
18. Struik L. C. E. (1989): Mechanical behaviour and physical ageing of semi-crystalline poly-
mers. Polymer 30: 815-830
19. Yittoria Y. (1988): Investigation of the ageing of isotactic polypropylene via transport prop-
erties. Polymer 29: 1118-1123
20. lbhadon A. O. (1996): Physical aging in isotactic polypropylene. J. Appl. Polym. Sci. 62:
1843-1846
21. Mirabella Jr F. M. (1994): Phase separation and the kinetics of phase coarsening in commer-
cial impact polypropylene copolymers. J. Polym. Sci., Polym. Phys. 32: 1205-1215
22. Blom H. P., Teh 1. W., Rudin A. (1995): iPPIHDPE-blends: interactions at lower HOPE
contents. J. Appl. Polym. Sci. 58: 995-1006
23. Tomlins P. E., Read B. E. (1996): Creep and physical ageing of polypropylene: a compari-
son of models. NPL Report CMMT(A)23, National Physical Laboratory, Teddington, UK
Subject Index
abrasion rate 447 coefficient of friction 408, 415

abrasion volume 488 compliance 42, 123
acoustic emission 22,236,385,551 compliance changing rate 109
acrylonitrile-butadiene-styrene (ABS) compressive strength 576
43,89,111,301,317,335 compressor blade 429
acrylonitrile-styrene-acrylate (ASA) constraint 124, 324
33 corrosion 362
adhesion energy 46 crack blunting 23,40,67,526
alternative methods 11, 105ff crack energy factor 482
- approximate methods 11, 105ff crack extension 64,68, 114, 126, 182
- hinge model 12, 133ff, 254 crack growth 8ff, 28ff, 53, 64, 79ff,
- normalization method 11, 42, 87ff, 105ff, 121ff, 141, 161,215,
121ff,138 224,257,282,302,378,425,479,
- stability assessment diagram 11, 495,512,564
141ff,187 crack initiation 40,67,82,177,189,
American Society for Testing and 219,479"
Materials (ASTM) see standards crack length correction 52, 81, 88, 125
Arrhenius plot 185 crack-opening-displacement rate
automatic R-curve procedure 138 (velocity) 6,8, 19,285,567
crack propagation 5, 31, 51ff, 79, 108,
ball-indentation hardness 339, 558 128, 141ff, 162, 177, 181ff, 219, 233,
biaxial testing 536 243ff, 262, 301ff, 318,423, 479ff,
biocorrosion 451 493ff, 507, 521, 554, 562
butadiene rubber (BR) 485 crack propagation energy 72, 106,
butyl-acrylate copolymer (BA) 243 170ff, 235, 318
breaking elongation 339 crack propagation speed 31, 516
brittle-to-tough transition (BTT) 222, crack resistance curve 5, 18,21,41,
246,290,307,326,329 44, 54ff, 82, 91ff, IlIff, 124ff, 133ff,
blunting line 40 177, 215ff, 25 Iff, 263ff, 284ff, 313,
blunting process 177 326,501,566
crack tip 4, 40, 62, 87, 121, 158, 161,
cast polyamide (PA) 216 177,182,197,217,281, 294ff, 320,
carbon black 482, 496 361,372,398,482,526
carbon fibre (CF) 408, 430, 447 crack-tip-opening angle (CTOA) 64
carbon-fibre epoxy 387 crack-tip-opening displacement
carbon-fibre-reinforced polymer (C(T)OD) 4, 28,53,64,84, 128, 181,
(CFRP) 360, 448 187,230,314,320,425,562
Charpy impact strength 32, 233, 239, craze formation 193,200,294
339,551,554,587 craze microstructure 205
592 Subject Index
crazes, structures of203, 345ff energy absorption capacity 10, 169ff,

crazing 20, 185, 193,268,348 210,219,243,285,500,526,554
creep experiment 540 energy-determined fracture toughness
criteria for fracture see instability 6,87,182,314,523,563
criterion energy dissipation 5,79,215,297,
critical crack-opening displacement 302,386
see crack-tip-opening displacement energy dissipation capacity
critical interparticle distance see (capability) see energy absorption
interparticle distance capacity
critical tearing energy 499 energy method 80, 213
croiding 295 energy release rate 28ff, 388, 483
crosshead speed 126,212 entanglement network 195
crystallinity 163,282,447 environmental scanning electron
crystallization 154 microscope (ESEM) 270, 451, 454
C(T)OD testing for fracture toughness environmental stress-cracking
52,370 resistance 281
epoxide 31, 519, 572
damage 295,310,322,385,510,571 equivalent-energy concept 6, 13
damping behaviour 510 essential-work method 44ff
defect-selective imaging 355 ethylene content 284
defectoscopy 22, 355, 385 ethylene-propylene copolymer (EPM)
deformation energy 73, 318 17, 243ff
deformation-determined material ethylene-propylene rubber (EPR) 17,
parameter 7, 18,51,54,63,87, 181, 257, 284ff
284,286,314,427,554,569 ethylene-vinyl acetate copolymer
8-To stability assessment diagram (EV AC) 243ff
141ff
density 162,271,447 failure mode 6, 39, 395
depth wear rate 408 fatigue 31, 493,515
detergent resistance 549 fibre content 115,372,409,563
differential scanning calorimetry fibre-matrix interface 575
(DSC) 99, 162,260 filler content 486ff, 527
distribution oflocal deformation 377 finite-element method (FEM) 64, 430
dynamic-mechanical analysis (DMA) flow resistance 458
Fuller distribution 505
302
full-scale test 96
dynamic-mechanical-thermal analysis
fracture frequency 183, 312
(DMTA) 257, 583
fracture mechanics concepts
dynamic stress intensity factor see
- crack-tip-opening displacement
stress intensity factor
(CTOD) 4, 28,53,64,84, 128, 181,
187,230,314,320,425,562 .
elastic-plastic fracture mechanics
- elastic-plastic fracture mechanics
(EPFM) 4, 39
(EPFM) 4, 28,181
elastomer materials 479, 493 - equivalent-energy concept 6, 13
Elyashevich hypothesis 154 - J-integraI4, 29
energy absorption 76,576 - JTJ 7,84,171,219,245,252
Subject Index 593
- linear elastic fracture mechanics 212,230,244,257,302,317,423,

(LEFM) 4, 30fI 507,551,562
fracture mirror length 177, 186, 323 instrumented falling-weight test
fracture modes 6, 39, 395 (lFWIT) 42, 244, 572
fracture surface 32, 173, 178, 186, instrumented micro-hardness test 196,
225, 238fI, 253, 263fI, 295fI, 311, 457,471
322,347, 514fI, 528, 565, 574 instrumented tensile-impact test (lTIn
fracture toughness 495
- dynamic 426,512,514, 564fI inter-fibre fracture, analysis 390
- static 29ff, 408, 521 interface debonding energy 400
friction 405fI interfacial strength 36
friction energy 479 interlaminar fracture energy 400
interparticle distance 17, 148,249,
geometrical size criterion 261,265, 287fI, 329
- crack-tip-opening displacement 77,
84 J-controlled fracture see energy-
- J-integral 9, 55, 77, 84, 90, 313 determined fracture toughness
- stress intensity factor 76, 523 J-integraI4, 40, 65, 75
glass fibre-polypropylene (GFIPP) J-integral evaluation method
laminates 388 - Begley and Landes 40, 495
glass-fibre-reinforced polymer - Merkle and Corten 75, 305, 320
composites 374, 399, 115,549,561, - Sumpter and Turner 52, 75, 156,
571 170ff, 182,235,247,264,283,
glass transition temperature 19, 183, 320,425
193,209,245,258,281,290,302, J-R curve see crack resistance curve
582fI J-TJ stability assessment diagram 8,
141ff, 183
heat deflection temperature 281, 587 joint surgery 441
heterogeneity function 372,557,568 JTJ parameter 7, 84
high-density polyethylene (PE-HD) JTJ values 171, 176,219,221,224,
95, 161,229,442, 252,285
high-impact modified polystyrene 31,
193,537 key curve 11, 136
high-voltage transmission electron key curve method 110,122
microscopy 201,270,308,345 Kobayashi method 12, 109
hybrid experimental test method 22,
366,555 laser extensometry
- laser double scanner 52, 370
in-situ experiments on crack blunting - laser multi-scanner 371, 380
67 - longitudinalllateral365fI, 556, 569
in-situ methods 22,67,270,345,456 lid elongation test 473
inertial load 76,513 lifetime 95, 365, 385, 429, 438, 459,
instability criterion 6, 142 475,493,549,558,571,581
instrumented Charpy impact test linear elastic fracture mechanics
(ICIT)30, 73, 80, 106, 156, 164, 182, (LEFM) 4, 30fI, 181, 398, 562
594 Subject Index
LMN technique see normalization non-linear fracture mechanics see

method elastic-plastic fracture mechanics
load-deflection diagram 17, 34, 60, non-linear viscoelastic material
72ff, 80, 89, 106, 109, 113, 124, behaviour 533
133ff, 170, 172,234,251,262,283, normalization method 11,42, 121ff,
305,424,513,555,566 138
load-determined fracture toughness 7, notch impact strength see Charpy
175,523 impact strength
load-displacement curve 62, 379, 522 notch opening 10 1
load-extension diagram 495ff notch-pipe test (NPT) 96
load-penetration diagram 472, 511
local deformation behaviour 365ff, orientation, influence on fracture 13,
552,562 261,365
lock-in thermography see ultrasound
lock-in thermography PA 66-HI 215
low-blow method 80, 215 PAlGF compound 115ff, 372ff
Paris-Erdogan equation 515
macrocraze 348 particle diameter 17,249,261,288,
main (a) relaxation 19 309,319,342
mechanical loss factor 20, 183,259, particle (size) distribution 309,319,
342
303
PBT/GF compound 374
medium-density polyethylene (MDPE)
PC/ABS blend 19, 89, 110, 147
41
PE blends 267
melting point 163,260,271,447,583 PEIPP blends 229
micro-hardness see instrumented peel test 47
micro-hardness test phenolic resin 419
micro-mechanical fracture model 398 physical ageing 581ff
micro-yielding 268 physical crack initiation
microbial corrosion 461 - 8. 6
modal analysis, experimental 433 - ~ 6,282
modelling of toughness 64, 239ff, 254 plane strain state 217, 297, 312, 380
moisture content 218 plane stress state 217, 297, 312, 380
molecular weight 21 ff, 95, 163, 194ff, plastic-hinge model 182, 230, 546
282,301,318 plastic zone 323, 375
morphology-failure relationship 335 plastic-zone size 282ff, 324, 372ff
morphology-property correlation 22 poly-bis-maleimide (PBI) 408
morphology-toughness correlation 3, polyamide (PA) 39, 90ff, 115, 193,
16,22,162,177,297,301 209ff, 243,372,414,533
multiple crazing 294,309,321,336 poly(butylene terephthalate) (PBT)
multiple-specimen R-curve method 374
(MSM) 41, 79, 105, 121, 129, 133, polycarbonate (PC) 19,36, 144, 147,
141, 169, 213ff, 254, 564 181ff, 199,202
polyetheretherketone (PEEK) 406ff
network density 498 polyetheretherketoneketone (PEEKK)
non-destructive testing 22, 355ff 406
polyethemitrile (PEN) 406
Subject Index 595
polyethylene (PE) 95, 161,212,229, secondary (~) relaxation 19, 183,210,

285,442, 583
polyethylene notch test (PENT) 95, shear yielding 20,281,336
100 short chain branches 102, 161
poly(ethylene terephthalate) (PET) 193 short-fibre composite 38, 115,374,
polyimide (PI) 31, 408 408,549,561,
polymer concrete 505 short-fibre-reinforced polymer 38,
poly(methyl methacrylate) (PMMA) 115, 373ff, 406, 549, 561
19,30, 35ff, 181ff, 198,441,445, sidegrooves 169ff
505 single-specimen R-curve method
polypropylene (PP) 17,45,51,148, (SSM) 42, 105, 121,213
153ff, 257, 281ff, 369, 376ff, 395ff, sliding wear 409
549, 561, 581ff small-angle X-ray scattering (SAX:S)
polystyrene (PS) 19,36,38, 181ff, 164,200
193,317 small-scale yielding 4
polytetrafluoroethylene (PTFE) 415 specimen configuration, effect of 62,
polyurethane (PUR) 452, 458ff, 466ff, 211,522
474 specimen configurations
poly(vinyl chloride) (PVC) 51, 125ff, - centre-cracked tension specimen
198, 379ff (CCT) 64
PPIEPR blends 257 - compact tension specimen (CT) 13,
PPIEPRlPE reactor blends 145,282, 31,51,212,508,519,
292 - deeply-edge-notched tension
PP/GF compound 376ff, 395ff, 549, specimen (DDENT) 44, 212
561 - double-cantilever beam (DCB) 46,
pre-craze domains 194 388,395
pull-out 385,556,565 -double-edge-notched bend
pull-through test 473 specimen (DENB) 211
- double-edge-notched tension
quality control 3
specimen (DENT) 64, 211, 496
R-curve see crack resistance curve - mini-compact tension specimen
razor blade notch 44, 72ff, 99, 125, 519
155,168,211,244,301,320,424, - single-edge-notched bend
520,554 specimen (SENB) 29, 40,51,72,
rebound resilience 487 88, 107, 124, 134, 155, 168, 182,
reflecting fringes (laser extensometry) 211,230,243,301,320,424,507
367ff - single-edge-notched tension
specimen (SENT) 64, 98, 212, 485,
relationship between K, 0 and J 14,
496
143,171,175
- tapered double-cantilever beam
research in fracture mechanics 3-26
specimen (TDCB) 46
rubber content 246ff, 261ff, 305, 323
specimen geometry 51, 72, 76, 98,
scanning force microscopy 273 155,182,230,243,301,320,387,
scanning laser vibrometry 356 424, 485, 496, 507ff, 520
Schindler method 12, 106 specimen length method 80, 213, 215
596 Subject Index
specimen thickness 17,53, 76ff, 211, - Standard Draft ESIS TC 4 8,39,

380,522 52ff, 82, 87ff, 125, 128ff, 169,284,
stability limits 147 379,495
stable crack initiation and propagation stop block method 80, 213
6,40,79,133,141,169,182,219, stress-determined fracture toughness
302,325,486,507 see load-determined fracture
stable crack length (fracture mirror toughness
length)74, 81,113,139, 177ff, 185ff, stress intensity factor 4
253,296,311, 322ff - dynamic 74,107, 171ff, 516
standards - static 28ff, 53, 324, 482, 521ff
- ASTM D 3433-99 47 stress relaxation 321
- ASTM E 399-90 74, 85 stress-strain curve 4,64, 116,339,
- ASTM E 813-81 8, 88ff, 106, 351, 369ff, 499, 533ff, 563
144,212,324 stretch zone 23, 40, 67, 186, 189,314
- ASTM E 813-89 8,88ff structural mode1158
- ASTM E 1152-87 42, 89 styrene-acrylonitrile (SAN) 36, 138,
- ASTM E 1737-96 8 196,198,301,318,335
- ASTM E 1820-96 54 styrene-butadiene rubber (SBR) 485,
- BSS 7620 572 496
- DIN 16965 572 styrene-butadiene-styrene (SBS) 533,
- DIN 16966 572 538,541
- DIN 17770 164 sulphur content 496ff
- DIN 51004 162 supermolecular structure 153
- DIN 53444 343 support distance method 80, 213
- DIN 53455 230, 343
- DIN 53457 230 tear and fatigue analyser (TF A) 493
- DIN 53516 485 tearing modulus
- DIN EN ISO 1110 214 - defmition 5,85, 142
- DVM Merkblatt 002-1987 88ff - values 21,59,61,138,174,176,
- ESIS P2-92 8, 51ff, 61, 88ff, 565 219ff, 263, 285, 290,328,566
- ISO 75-2 585 technical crack initiation value
-ISOI78551,585 - ~.2 59, 61,174,189,211,284,315
- ISO 179 71ff, 230, 243, 301, 320, - JO.2 21,59,61,92, 112, 114, 129,
551,585, 138,171,174,176,189,217ff,
- ISO 527 162, 546ff, 551, 572 221,263, 284ff, 290, 293, 315,
- ISO 1133 162 327,502
- ISOIDIS 3167 229,301,365,551 temperature dependence oftoughness
- ISOIDIS 13478 96 12ff,60ff, 144, 183ff, 222, 245ff,
- ISOIDIS 13479-94 96 305ff,526
- ISOIDIS 13586,2-2000 30 tensile strain at break 486,497, 563
- ISOIDIS 15850-99 30 tensile stress 486, 497, 553, 563, 568
- ISO/TC 138/SC 4 N 144 101 thermoplastic urethane (TPU) 147,
- procedure for determining the 553,568
crack resistance behaviour using tie molecule density 165ff
the instrumented impact test time-dependent fracture mechanics
71-86 (values) 8, 19, 129,286,567
Subject Index 597
time to fracture 76, 101,513 Vicat heat transition temperature 339,

TPU/ABS blend 19, 147 551
tracheal prosthesis 451 voice prosthesis 451ff, 461ff, 471ff
transition from dominant unstable to void formation 162, 195
dominant stable crack propagation void growth 174
79,88,142 von Mises yield criterion 9, 64, 534
tribological properties 405ff, 419ff,
479ff wear 405ff, 479ff
Williams-Landel-Ferry equation 187
ultra-high-molecular weight work of deformation 5, 107
polyethylene (UHMWPE) 442
ultrasound lock-in thermography 357 X-ray diffraction 156,259
uniaxial test specimen 535
uniform elongation without necking yield stress 74, 126, 163, 188,219,
107 221,259,304, 339ff
unstable crack initiation 6, 72, 133, Young's modulus 74, 126, 163, 184,
149,301,325,551 198,219, 221ff, 232, 304, 339ff, 398,
447,474,539,553,563,568,584
Author Index
Aumayr, G. 561 Koch, T. 257

Kotek, J. 153
Baldrian, J. 153
Kotter, I. 51,429
Beerbaum, H. 161
Krietsch, T. 385
Berghaus, A. 451
Krohn, N. 355
Bethge, I. 243,451,471
KrUger, P. 301, 317
Bierogel, C. 365,471,519,549
Kucera, J. 95
561,571 '
Bischoff, R. 549 Lach, R. 141, 181,301,317,493
Bohse, J. 229, 385 Langer, B. 209, 561
Brandt, I. 441 Lebek, W.267
Busse, G. 355 Ludwig, H.-C. 335
Casar, T. 479 Maier, M. 429
Che, M. 51, 121 Michel, B. 519
Michler, G. H. 193,267,335
von Diest, K. 429
Moginger, B. 335
Dillenz, A. 355
Myalski, J. 419
Fedtke, M. 519
Neitzel, M. 429
Fiebig, J. 561, 581
Nezbedova, E. 95
Forster, S. 27
Niebergall, U. 229
Friedrich, K. 405
Futing, M. 451 Raab, M. 153
Ramsteiner, F. 27
Gahleitner, M. 581
Reincke, K. 243, 493
Godehardt, R. 267
Reinicke, P. 405
Grellmann, W. 3, 51, 71, 87,121,
Rufke, B. 571
133, 141, 153, 161, 181,209,229,
243,257,281,301,317,365,419, Salajka, Z. 95
429,451,471,479,493,505,519, Schmachtenberg, E. 533
549,561,571 Schuster, W. 27
Sebova, I. 461
Haberland, E.-I. 451, 461, 471
Seidler, S. 51, 71, 87,105,121,161,
Han, Y. 317
2~9,229,243,257,281,549
Hein, W. 441
Sterner, R. 133, 429
Heinrich, G. 479, 493
Stiefel, A. 461
Hesse, W. 71
St6Bel, R. 355
Hildebrandt, T. 505
Sturm, H. 229
Hoffmann, J. 405
Hyla, I. 419 Walter, H. 519, 571
Wanders, H. 533
Jung, K. 51, 257
Wehner, H. 505
Yazici, N. M. 533

Engineering Materials

Uploaded by

Copyright:

Available Formats

Engineering Materials

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Engineering Materials

Uploaded by

Copyright:

Available Formats

Engineering Materials

Springer-Verlag Berlin Heidelberg GmbH

Deformation and Fracture

With 447 Illustrations and 51 Tables

Prof. Dr. Sabine Seidler

Library of Congress Cataloging-in-Publication Data

© Springer-Verlag Berlin Heidelberg 2001

Typesetting: Cameraready from editors

The application of fracture mechanics to polymers and composites allows the

Merseburg, Vienna Wolfgang Grellmann

October 2000 Sabine Seidler

List of Authors.............................................................................. XIII

Terminology............... ................................................................... xxv

A1 State of the Art and Development Trends

A 1.1 New Developments in Toughness Evaluation of Polymers

A 1.2 Concepts of Fracture Mechanics for Polymers

A 2.1 Influence of Specimen Geometry and Loading Conditions

A 2.2 Procedure for Determining the Crack Resistance Behaviour

A 2.3 Possibilities and Limits of Standards and Drafts

A 2.4 The Relationship Between the Fracture Behaviour

A 3.1 Application of Single-Specimen Testing Methods

A 3.2 Application of Normalization Method for Determining

A 3.3 Calculation of J-R Curves Based on Load-Deflection Diagrams

A 3.4 An Alternative Method Based on J-TJ and ~T6 Stability

B 1.1 Supermolecular Structure and Mechanical Behaviour

B 1.2 Correlation Between Structure and Toughness Behaviour

B 1.3 Toughness and Relaxation Behaviour ofPMMA, PS and PC

B 1.4 Crazing in Amorphous Polymers - Formation of Fibrillated Crazes

B 1.5 Influence of Temperature and Moisture on Toughness Behaviour

B 2.1 Relationship Between Fracture Behaviour and Morphology

B 2.2 Influence of Modifier Content and Temperature

B 2.3 Morphology and Toughness ofPP/EPR Blends

B 2.4 Morphology and Micro-Mechanics

B 3.1 Toughness Optimization of Multi-Phase Polymer Materials

B 3.2 Crack Toughness Behaviour of ABS Materials

B 3.3 Fracture Mechanics Characterization of ABS Materials

B 3.4 Brittle Fracture of ABS

C Hybrid Methods of Polymer Testing

C.I Defect-Selective Imaging

C.2 Determination of Local Deformation Behaviour of Polymers

C.3 Damage Analysis of Composite Materials

D Technological Test Methods

0.1 Polymer-Based Composites for Friction and Wear Applications

0.2 Modification of Polymers by Means of Amorphous Carbon

0.3 Mechanical Vibration Behaviour of a Compressor Blade

E Biocompatible Materials and Medical Prostheses

E.l Polymer Materials in Joint Surgery

E.2 Material Parameters and ESEM Characterization

E.3 Microbial Corrosion of Pharyngo-Tracheal Shunt Valves

E.4 Deformation Behaviour of Voice Prostheses

F.l Crack Initiation, Wear and Molecular Structure

F.2 Investigation of Crack Propagation Behaviour

F.3 Characterization of Deformation Behaviour

F.4 Fracture Mechanics Testing of Modified Epoxy Resins

G Examples and Limits of Application

G.l Modelling of the Mechanical Behaviour

G.2 Detergent Resistance ofPP/GF Composites

G.3 Material Optimization of Polypropylene-Short-Glass-Fibre

G.4 Influence of Exposure on the Impact Behaviour

10000 .:; ,**~!'l!!!"'l':'~~~~~~~~~!j1