Mechanical Properties of Poly (Ether-Etherketone)
Mechanical Properties of Poly (Ether-Etherketone)
Mechanical Properties of Poly (Ether-Etherketone)
];
P E E K is a thermoplastic material and as such it can be
converted into a range of component shapes and sizes by
the full spectrum of fabrication technologies (extrusion,
injection moulding, etc.). It is also possible to compound
formulations based on P E E K and the incorporation of
short glass and carbon fibres is now a well-established
method for preparing injection moulding materials. These
compounds are generally referred to as micro-composites.
Fibre concentrations up to at least 40% w/w are being
evaluated.
A further area of technology where P E E K is used
relates to its pioneering role in continuous fibre reinforced
thermoplastic composites. It is a polyaromatic matrix and
therefore these materials have become known as aromatic
polymer composites (APC) being based on continuous
0032-3861/85/0913854)9503.00
1985 Butterworth & Co. (Publishers)Ltd.
1385
i0 a
'E
Z
I0 7
800C
2OoC f
150C
{0
i0 si0_~
I0
I I I
Stroin (%)
STIFFNESS
Creep
Engineering design based on a stiffness criterion will
require an insight into a modulus-time-temperature
function. The general deformation behaviour of thermoplastics is known to be non-linear viscoelastic 5. Consequently, there can be no unique value of a modulus at any
specific time under load or at a specific temperature;
instead a modulus will be dependent on the magnitude of
applied stress (or strain). The degree of non-linear viscoelasticity varies between different thermoolastics, but with
PEEK the relationship between stress and strain appears
to follow a simple pattern.
Figure 1 illustrates 100s isochronous stress-strain 6
plots for P E E K at temperatures in the range 20-180C.
The curves are virtually parallel lines on these double
logarithmic plots and are straight up to strain levels of
~1%. This simple behaviour suggests an ease in
predicting intermediate temperature behaviour. The
'glass transition' temperature for PEEK is 143C 1
(although it depends on test rate) consequently there is a
large shift in deformational characteristics between 80
and 150C.
The time dependence of modulus is investigated in
tensile creep tests 6, where experimental results have been
collected in the temperature range 20-180C and for long
times under load (~ 1 year). Detailed families of creep
curves for specific conditions are presented in the manufacturer's booklet 7. The purpose here is to consider the
interacting effects of stress (strain), time and temperature
on modulus to examine overall trends.
A tensile modulus can be defined at a specific constant
stress or at a specified constant strain. For both
definitions of modulus the influence of time and temperature dependence can then be explored. Experimental
creep data are generated at a specific test temperature, and
usually at several levels of applied stress. The data can be
stored in a computer data-bank which then enables
detailed interpolations and crossplots 8 to be conducted.
Consequently, it is possible to express several creep
functions at various test temperatures as if the short-term
stiffness of the material was similar. These functions are
illustrated in Figure 2 where a common 10 s strain of 0.5%
is used to explore the effect of temperature on the creep
characteristic. For the creep curves at higher temperature
the applied stresses are smaller because stiffness falls with
temperature. With reference to Figure 2, it is then
1386
Figure !
1.5
2.4 MN m-2
1.0
03
N m-2
0.5
20"C
i01
I
I02
I
I03
I
104
I
105
I
I0 s
iOT
Time (seconds)
E
b
__e
2-
~20C
~80C
1200C
"
I0
I
iOI
I02
5C,sico
IOs
I04
105
Time ( seconds )
I0 e
I07
Peek
min
-80-40
40 80 120 160200240
I ruth
'r
i-
'E
-4
Figure 4
time
versus
temperature and
1.0
Q9
~Aromatic
polymer
0.8
07
0.6
0.5
0.4
0.5
Q2
0.1
O0
-90
Unreinforced p~ek
--tO
-50
-~K)
I
t
I
I
-I0
I0
30
Disc angle (degrees)
50
"tO
I
90
Anisotropy
It is encouraging to identify some simplifying trends in
the deformational characteristics of PEEK. Introduction
of fibre reinforcement, whether long or short, carbon or
glass, adds a further complication; namely, that of
anisotropy. Perhaps the clearest and simplest manifestation of anisotropy is illustrated in a disc test t 0. A 115 mm
diam. disc is subjected to small strain three line bending
and its stiffness is measured at 10c~intervals relative to the
direction of principal flow or some other reference
direction. Figure 5 illustrates a normalized disc stiffness
versus direction for four different materials. Normalized
disc stiffness is expressed as the ratio of a specific stiffness
for any of the four specimens relative to the maximum
observed stiffness (32.9 GN m -2) for the APC material
reinforced with continuous carbon fibre in a cross plied
lay-up ('mat').
The injection moulded single edge-gated PEEK discs
show different levels of anisotropy. Virtually no
anisotropy is exhibited for the unreinforced PEEK, such
that the anisotropy factor would be 1 (anisotropy factor is
the ratio of maximum to minimum stiffness). The 30~
w/w short carbon fibre reinforced PEEK compound has
an anisotropy factor of 1.7, whilst the 30/0 short glass fibre
reinforced PEEK compound has an anisotropy factor of
1.4. The higher stiffness and anisotropy of the short
carbon-fibre reinforced PEEK will be partly due to higher
fibre volume fraction. Other factors such as the fibre
orientation in the skin and core of the moulding as a result
of processing may also be important.
The anisotropy of the APC material is naturally
governed by the characteristics of the carbon fibre
orientation. As can be seen from Figure 5, the deformational characteristics for this type of material are more
involved. APC materials can be prepared in various ways
the most common being the laying-up of the impregnated
tapes. Prediction of stiffness performance for these
materials can be complex and although outside the scope
of this paper, it is nevertheless possible to comment on
approach.
The stiffness of different lay-ups of continuous carbon
fibre composites can be predicted from the pseudoelastic* constants of simple consolidated uniaxial sheet or
tape. This is the building block of the more complicated
structures. When fibre symmetry is assumed (i.e. isotropy
in the plane normal to the fibres), five independent
pseudo-elastic constants are required for the sheet 1~
Tensile testing of thin sheets along and normal to the fibre
direction enables three of these to be determined, torsion
about the fibre axis gives a fourth 12. Figure 6 illustrates
five pseudo-elastic constants:
E 1l--modulus in the fibre direction
Ez2--modulus normal to the fibre direction
vx2 and v2~--lateral contraction ratios
G~2--shear modulus in the plane
1387
0.38
z,21
GIz (GN m-z)
0.04
4.0
I
Ell
= -o- I
El
~f2
I
I
= -E2
EI
tJl3
E:5
2f
"
E2
J
E22
0" 2
E2
1/21
= --E I
E2
3
//23
=-~3
E2
1)12
v21
'8O
70
~ 6o
o
:~
50
~ 40
~ 3oi
20 I
I0
TOUGNESS
Preamble
An objective measure of toughness is difficult to obtain
because of geometry dependence. Therefore, a simple
expression of absorbed impact energy per unit ligament
area from a Charpy or Izod test is very likely to mislead.
1388
ol
-I00
Ductile-brittle transition
I
-80
t
t
[
-60 -40 -20
I
I
I
I
0
20 40 60
Temperature (=C)
I
80
I
I00
[
I
120 140
/w CF
LU
/w CF
4
~/"
=,
20
-[00
I
-80
I
-60
~aflon-energy
~
I
-40
I
-20
I
0
30% w/w GF
20
40
60
80
I00
120
140
Temperature (*C)
45
40
Energy absorbed
(Joules)
//
45
4O
//
55
35-
5O
50
Normalized to 3 mm
total energy
25
25
20
2O
15
15
Normalized to 3 mm
initiation ener~_
I0
"1"
5
0
-80
I0
5
Initiation
I
-60
I
-40
-20
I
I
I
0
20
40
Temperature (*C)
I
60
I
80
IOO
with that for the glass fibre system. This may be an effect of
the fibre volume fraction because it is higher for the
carbon composite, but might also be related to the flow
anisotropy.
The component of energy required to initiate a crack is
small compared with the total fracture energy,
particularly at temperatures above -20C, for both
compounds. Consequently, crack propagation would be
expected to be the dominant energy absorbing process. At
temperatures below -20C, crack propagation energy
decreases relative to the energy required to initiate a
crack. With reference to Figure 7 for the unreinforced
material, this is co-incident with the PEEK becoming
brittle. Therefore, it is implicit for the fibre-reinforced
PEEK that the crack initiation process is fibre dominated,
whilst propagation is matrix dominated. Consequently,
when the matrix is tough, crack propagation is the energy
absorbing process; but when the matrix is brittle, then
crack initiation becomes the energy absorbing process.
The impact behaviour for the continuous carbon fibre
PEEK composites exhibit different failure/fracture
characteristics. Figure 9 illustrates impact energy versus
temperature plots for quasi-isotropic lay-ups which are
2 mm thick. (Data normalized to 3 mm thickness are also
included for comparison with the injection moulded
discs.) Again, initial failure is through cracking followed
by subsequent propagation. But observation of the
fractured specimens shows that the overall area of impact
damage is restricted to that in and around the impact area,
i.e. long crack propagation is restricted by the lay-up. The
energy absorbed in initiating and propagating failure is
greater for APC than for the microcomposites. This
results in a considerable increase in the total energy
absorbed. By contrast to the microcomposites, the total
energy absorbed by APC does not fall markedly between
- 20 and - 50C. This suggests that the fibres dominate
the propagation process for APC as well as initiation.
Intrinsic toughness
Instrumented falling weight and other impact techniques enable comparative toughness to be measured.
However results from these techniques are all geometry
dependent and this limits their usefulness in both
providing understanding and enabling design for
toughness. Consequently it is helpful to use linear elastic
fracture mechanics methods to measure intrinsic
toughness 17. Two toughness parameters are usually of
interest:
Kc~--a plane strain value for the critical stress field
intensity factor.
Gc~--a plane strain value for the critical strain energy
release rate.
A relevant value of yield strength is necessary to complete
the toughness picture, and the merits of these three
parameters are discussed elsewhere 18. It is not our
intention to present a comprehensive fracture toughness
picture for PEEK, but rather to present a set of intrinsic
toughness values and to comment on the importance of
establishing experimental conditions that lead to
interpretable data.
While it is relatively easy to obtain 'fracture toughness
parameters' these will only be of use if they are indeed
intrinsic properties. To ensure that this is the case, certain
criteria must be imposed to ensure that the measured
1389
(1)
bmi. = 2 5 ( Kc~'~2
" \,z,
-->
(2)
Wmin
Wmm_--5(gc'
32
k% /
Usually if condition (1) is not met then the measured Kc
(or Go) will be larger than the plane strain value. However
if condition (2) is not met then the measured value will be
too small. Hence if neither condition is met the fracture
toughness results will be difficult to interpret.
The fracture toughness measurements were made on
notched beam specimens tested in three line flexure, the
experimental techniques being described elsewhere / 1. The
specimen dimensions were 70 x 11 x 3 mm, and they were
supported on a 50mm span, with notches of nominal
10~m tip radius machined into their surface. A range of
notch depths were used.
Fracture toughness results for unreinforced PEEK are
shown in Figure 10, using two test conditions, namely
23C with a test speed of 100ram rain- ~ and - 40C with
Static strength
Strength is considered in terms of its magnitude and the
mode of failure; time under load and environmental
temperature are also prime factors. In general terms,
unreinforced PEEK exhibits a yielding or shear banding
process whilst fibre-reinforced composites are dominated
by the fibre and show crack-like fractures.
150
K c = 4.5 MN m-3/2
Specimen dimensions
w = .3 mm
b = 12 mm
I00
/./.
A
Z
v
/'S"
v/
/
i
Figure 10
1390
_4olc
,# , /
50
/ i
A
I
0
3
I m/S
0.5
1.0
YJ"0" mm
I10
I00 -
*.~-.~
9o
80
'E
Z
v
~"~-'~"-,,~
o)
O3
5C
7O
6O
5O
50
150C
2O
I0
I I IIIIlll
qO I
Figure 12
I IIIlll]
I I Illllll
103
i0z
I I llllltl
I I l llllll
104
I05
Time (s)
I I ltllll
I I Itllll
107
I06
108
80
"?0
'E 60
'wcF
~ 5o
y. 40 -
~0
w/w GF
0 and 90 unreinforced peek
20
I0
I
i01
Figure 13
I
i0 ,?.
I
103
I
t
104
105
Time (seconds)
I
i06
I
i0?
1(38
zero- tension
IO0
90
80
7O
Unnotched spelmen
15 6O
Z
5O
~
4O
50
20
190
I0
170 E 150z
IO
iso-
I10-
[02
[03
(Cycles)
I04
105
90>-
705O
I
I
I
-I00 -80 -60
Figure i I
Yield stress
-120
I
I
I
I
-40 -20
0
20
Temperature (C)
versus
I
40
I
60
I
80
I00
1391
Dynamic strength
Increasing the test severity enables an even more
critical view of the strength characteristics of PEEK and
its composites. Dynamic loading and the introduction of
stress concentrations are known to precipitate crack-like
fractures in many plastics 24. Figure 14 illustrates zerotension fatigue plots for unreinforced PEEK for notched
(250#m tip radius) and unnotched specimens. Neither
curve relates to classical brittle fractures even though the
presence of the notch reduces the strength. In the
experiments conducted on unhatched specimens the
degree of ductility extends up to gross yielding
particularly at the higher stress levels; for the notched
specimens, a much lower level of ductility is observed.
The relative fatigue performance of PEEK and its
injection moulding compounds is shown in Figure 15 for
90 specimens. Again, carbon-fibre-reinforced PEEK is
stronger than the glass-fibre-reinforced PEEK. However,
from the remarks about the fibre alignment and
anisotropy it is apparent that the choice of 90 specimens
cut from single edge-gated discs favours the carbon fibre
material.
Perhaps of greater significance is the comparison of
strength between static and dynamic loading. To this end,
'e23C,
II0
I00
CONCLUSIONS
Successful engineering application involves making
certain choices correctly:
1.
2.
3.
90
""~'~1L30%
w/w Glass
I10
I00
7o I
fE 90
60
-~ 8o
50
Q' 70
3.
O~
40
SC
Figure 15
6O
90 Specimens
I
I0
I0 z
103
Cycles to fracture
104
105
5O
I0e
1392
Dynamic strength
(square waveform 0,5 Hz)
Figure 16
I0
102
103
104
Time under load (seconds)
I05
I0e
ioo
,~ 8o
Z
Dynamic loading
0.5 Hz square waveform
6o
Static loading
ACKNOWLEDGEMENTS
,~ 4o
20
i
O01 , , ,,,,,,I
I0 2
i t,l,I
io 3
i iiiiiii
iI
iiii1[
I IIIII1[
i illttl[
104
I05
I06
Time under load (seconds)
I IIIltl
I0?
IO8
1.0
J=
APC-uniaxial CF
~0,75
"6
'~ o,5o
P'?(o,9o
(90% specimens)
4
5
6
7
8
0,25
'(,9
0.0
I0 o
I01
102
103
IO4
IO5
I0e
I07
9
10
Number of cycles
11
12
13
14
15
16
17
18
19
20
21
22
23
24
POLYMER,
(Conference
issue)
1393