See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/248365041

Toughening of ceramic earthenwares by quartz inclusions: An ancient

art revisited

Article in Acta Metallurgica et Materialia · August 1995

DOI: 10.1016/0956-7151(95)00006-H

CITATIONS READS

88 182

3 authors, including:

George Vekinis Y. Maniatis

National Center for Scientific Research Demokritos National Center for Scientific Research Demokritos
99 PUBLICATIONS 1,830 CITATIONS 177 PUBLICATIONS 3,887 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Y. Maniatis on 22 October 2017.

The user has requested enhancement of the downloaded file.

 Acta metall, mater. Vol.43, No. 8, pp. 2959-2965, 1995
ElsevierScienceLtd
~ Pergamon 0956-7151(95)00006-2 Copyright © 1995.Acta MetallurgicaInc.
Printed in Great Britain.All rights reserved
0956-7151/95$9.50+ 0.00

TOUGHENING OF CERAMIC EARTHENWARES BY

QUARTZ INCLUSIONS: AN ANCIENT ART REVISITED
V. KILIKOGLOU,G. VEKINISand Y. MANIATIS
Institute of Materials Science, NCSR "Demokritos", 15310Athens, Greece

(Received 3 August 1994; in revised form 16 December 1994)

Abstract--Since ancient times randomly distributed quartz is routinely included in ceramic earthenwares
as an aid to drying. Although it has been noticed that this practice also provides the ware with higher
capability for flaw tolerance so that a crack does not become readily unstable in the event of overloading
during service, no systematic study of the effect or a satisfactory model has been reported in the open
literature. This work reports the results of such an investigation on the influenceof grain size and volume
fraction of quartz on the elastic modulus, strength and fracture energy dissipation of such composites
based on fired clays. In addition, we have studied the fracture micromechanismsby dynamic in situ optical
microscopy and scanning electron microscopy and introduced a simple model to explain the results, based
on the local effects of the stress field of the inclusions.

INTRODUCTION significantly" the properties of dried clay from

Fayoum in Egypt [3], while small additions of saw-
Intentional tempering of clay--the mixing-in of other dust ( < 2 wt%) improve the compressive strength of
materials--was common practice in antiquity, some non-calcereous clays from Nigeria [4]. More
especially when large vessels were to be made. Potters relevant is a single study on archaeological pottery [5]
had many reasons to temper; the most important which considers enhancement of crack propagation
were reduction of shrinkage during drying and firing, resistance but the results are mainly used to explain
promotion of uniform drying of vessels and improve- pottery function.
ment of thermal shock resistance during firing [1]. In this paper we present the results of a systematic
However, tempering enhanced the mechanical prop- study of the dependence of the elastic and mechanical
erties of the final products and, although the under- properties of such composites on size and volume
lying details were probably not understood, the fraction of quartz and in particular the conditions
practice was widespread which underlined its inherent under which fracture energy is maximised. Specifi-
usefulness. For example, potters were aware of the cally, we have studied the effect of quartz as temper,
fact that tempering prevented ceramic vessels failing on the fracture strength and toughness (energy dissi-
catastrophically and therefore they were able to pation) of quartz-tempered fire clays, through a series
repair pots with long cracks [2]. In modern times too of experiments on experimental briquettes. In
the use of tempering as a drying agent is well known addition, elastic modulus, density, shrinkage and
in the heavy clay industry. Sand or a "lighter" type unfired composite properties were also determined.
of clay (highly caicarous), is routinely mixed with the
base clay, up to about 20% vol. fraction. This import- EXPERIMENTAL
ant technological advantage of tempering is con-
Processing
sidered highly relevant to present technological needs
and is therefore at the core of this investigation. The experimental briquettes were manufactured
Nevertheless, even though this practice is wide- using a calcerous clay containing 15% CaO from the
spread, its impact on the mechanical properties, as far area of Pikermi in Attica, Greece, normally used by
as we can ascertain, has never been properly investi- the brick industry. Such calcareous clays are known
gated. The effect of various non-plastic additives on to display excellent postfiring properties [6]. The clay
some physical and mechanical properties of fired was fractionated in the laboratory and a fraction with
clays (earthenwares) have been reported in the litera- grain size < 30/zm was received and used as the
ture mainly in the form of a few semi-quantitative matrix material for all specimens. It was mixed with
observations. It is generally accepted that all such naturally weathered sand (99.5% SiO2) of four differ-
additives affect the performance of clays and ceramics ent average grain sizes: d = 750, 400, 250 and 100 pm
in a similar way that they affect the properties of in volume fractions of Vr= 5, 10, 20 and 40% for
cements: sand with grain size 500-1000/~m and each grain size. The 750/~m-40% combination was
30% vol. fraction has been reported to "increase not included because the fired briquettes were too

2959
2960 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ

weak to test accurately. In addition, a number of Preliminary investigations showed that all
briquettes were made from the matrix clay alone materials containing more than about 10% quartz
containing no additional quartz. In this way a total grains failed in a stable manner. Therefore, in these
of 16 different powder mixes with all combinations of cases the total energy G could be calculated as the
variables were prepared. Water was added to the total area under the load-deflection curve. For spec-
mixes to provide plasticity and briquettes of size imens that contained 10% or less quartz grains and
100 × 60 x 10mm were produced by uniaxial com- which displayed unstable fracture, their fracture
paction at a pressure of approx. 100 kPa. They were toughness K~c and Young's modulus were measured
air-dried for over 72 h at ambient temperature and independently, from which Glc was then calculated
humidity and fired at 950°C in air with a heating rate using equation (1).
of 200°C/h and a soaking time of 1 h. This firing K~c was measured in 4-point bending of straight-
schedule was dictated by previous work [6] which edge-notched-beam (SENB) specimens using the
showed that fine calcareous clays exhibit a homo- analysis of Srawley [7]
geneous and reproducible cellular microstructure at
firing temperatures between 850 and 1050°C as a 3 P f ( S a - S2)x/~tF(~ )
(2)
result of the development of an extensive glass
filament network which is formed around the voids where F(~) = 1 . 1 2 2 - 1.121~ + 3.74~ 2 + 3.873~ 3 -
created at an earlier stage by the liberation of CO2 19.05~ 4 + 22.55~ 5 and ~ = t/dand the dimensions S~,
from CaCO3. After firing, the large surfaces of the $2, d, b and t are shown in Fig. 1.
briquettes were made parallel to approx. + 10 #m by The load at fracture, Pc, is determined at the
surface grinding with a 1200 carborundum wheel and maximum point of the load~leflection curve for the
cut by a diamond wheel to produce test specimens of specimens. However, a notch introduced by diamond
size 10 x 10 × 60 mm. cutting does not offer an atomically sharp crack as
assumed in the analysis. This exaggerates the value of
Characterisation K~c as shown by the results of Munz et al. [8]. A
conservative value was therefore obtained by dividing
Performance of structural materials such as heavy the result by a factor of 1.25.
clays is critically dependent on their mechanical In order to avoid loading misalignments, the spec-
properties. Whether these clays are used in construc- imens were prepared as follows: after diamond sec-
tion or in manufacturing of vessels, the properties
tioning from identical fired briquettes the rectangular
that are decisive in determining the material's efficacy beam specimens were ground parallel and polished on
are its strength and, most importantly, its capacity for
a 1200 carborundum polishing wheel. Their edges
absorbing energy during fracture, measured by the were slightly rounded to avoid stress concentrations
fracture energy G. during loading and the notch was diamond machined
The fracture energy G can be defined as the using a 0.15 mm thick wafering blade on the centre of
total energy required to produce a unit area of
a surface normal to the original pressing direction. In
fracture surface. In the case of the homogeneous, all cases the specimen dimensions were kept close to
truly brittle materials, which exhibit unstable crack
10 × 10 x 60 mm. This size was chosen to minimise
propagation, it can be determined by measuring the shear stresses during loading while ensuring that a
"critical strain energy release rate", i.e. the toughness statistically adequate number of quartz grains were
G1¢, given by included in the eventual fracture surface. The notch
depth was kept close to 2 mm to avoid edge stress
Gtc=KUE (1) effects during testing.
The testing configuration is shown in Fig. 1. A
where K~c is the fracture toughness of the material as prepared notched rectangular beam was loaded in a
measured using standard fracture mechanics methods self-aligning 4-point bending jig on an INSTRON
and E is its Young's modulus. For such materials 1195 100 kN testing frame at a constant loading rate
energy is expended in breaking atomic bonds and G1¢ of 50#xm/min and the load was monitored as a
can be considered to be the intrinsic fracture energy
7. In the case of composites Glc can be considered as P/2 $2 P/2
the fracture energy of the matrix.
By contrast, where the presence of various micro-
structures causes stable crack propagation, G cannot
be calculated using equation (1). In such cases the
total fracture energy is given by 7 -I-energy dissipa-
tion Gd. Gd is the energy dissipated during fracture
and appears on the load-displacement curve as the
region of stable crack propagation, and is best P/2 S1 P~2
measured by the total area under the stable part Fig. 1. Schematic diagram of the SENB fracture toughness
curve. testing method in 4-point bending.
 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ 2961

function of deflection. The outer loading points were The micromechanics of fracture were investigated
spaced at $1 = 40.8 mm and the inner loading points by dynamic in situ microscopy. A video camera and
had a span $2 = 20 mm. The estimated measurement colour monitor were used to observe and record
accuracy was +2.5/zm (,,,1%) for deflection the fracture process at magnifications up to 100
measurements and +2.5 N ( ~ 1%) for load measure- times. This, in conjunction with a fast video printer
ments. The experimental conditions used, especially provides the possibility of monitoring the progress
the specimen size, rigidity of frame and strain rate are of fracture and allows correlation with the load-
known to influence significantly the fracture deflection curve in real time. In addition, scanning
behaviour of toughened brittle materials: small spec- electron microscopy (SEM) was used to study the
imens require very low loading rates which would fracture surfaces and in particular the quartz-matrix
enable accurate load-deflection monitoring during interfaces.
crack propagation.
The calculation of Gt¢ from equation (1) and the RESULTS AND DISCUSSION
analysis of fracture energy dissipation requires
knowledge of the mechanical and elastic properties of Five typical load--displacement curves (SENB) are
the material under identical loading configuration. shown schematically in Fig. 2 for specimens with
Thus Young's modulus E was measured from the different Vf but same quartz inclusion size d
same load~leflection curve obtained from the bend- ( ~ 4 0 0 # m ) . The shapes for other quartz inclusion
ing experiments. This was carried out by measuring sizes are similar. As Vf increases the apparent post-
the gradient of the curve at the stage just prior to fracture behaviour of the material changes: for
fracture initiation and calculating E using standard Vf ~< 10% we have typical brittle ceramic behaviour
beam deflection theory [10]. Independently, for with sharp load drop at the fracture point but no
confirmation, E was also measured in compression apparent energy dissipation after fracture initiation
(for specifically prepared rectangular parallel sided was observed under the loading conditions used. On
specimens) and by the ultra-sonic time-of-flight the other hand, for values of Vf > 10% the material
method with comparable results. displays pronounced crack stability giving fracture
Finally, transverse rupture strength (TRS) was energy dissipation as evidenced by the tail in the
measured in order to obtain information on the curve after fracture initiation. In this case the crack
influence of quartz and relate the strength of was observed to arrest soon after initial fracture and
the specimens to their flaw size and fracture energy. to advance stably to the opposite edge of the speci-
The tests were carried out on unnotched but identi- men with increasing deflection. This was particularly
cally prepared and sized specimens as used for pronounced for Vf = 40%. SEM observations indi-
the SENB tests, using 3-point bending to minimise cated that crack arrest occurred at quartz-matrix
possible misalignment effects. The calculation of TRS interfaces. It was also found that fracture behaviour
was done using the standard bending mechanics was independent of the quartz grain size. The
analysis observed value of Vf for stable-unstable transition
(10--20%) is probably related to the stiffness of our
3PfSl testing equipment. Nevertheless, the trend is clear: the
TRS = (3)
2bd 2 amount of energy dissipation during fracture
increases significantly with volume fraction.
where the parameters are defined in Fig. 1. The effect of quartz on Young's modulus is unex-
All other loading and testing parameters were pected. Figure 3 shows that the Young's modulus
identical to those used in the SENB tests. TRS was of the composites decrease significantly with increas-
taken as the average of at least seven experiments. ing Vf for all quartz sizes even though quartz has a
The variability of measurements in tempered speci- higher modulus than fireclay (Eq,ar~=80GPa,
mens was found to be significantly lower than in Ec~,y= 22 GPa) and so, theoretically, the modulus of
untempered ones, probably due to the beneficial
presence of the equi-sized quartz inclusions. 25
I --e--- d=lOOpm
--II-- d=250pm
20
+d=4OOl.zm
0%.
d---X--
=750Hrmie
5%

A 10%
o 0/A4 0.
o i i i J i i i

5 10 15 20 25 30 35 40 45
Displacement Volume fraction, %
Fig. 2. Typical load-displacement curves for a series of Fig. 3. Variation of Young's modulus with volume fraction
specimens with a range of Vf for d = 400 #m. and size d of inclusions.
2962 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ

45
40 11111¸

35.
~. 3 0 . Opm !
25.
¢,~ 2 0 -
't5-
10.
5-

0 5 10 15 20 25 30 35 40 45
Volume fraction, %

Fig. 5. Variation of transverse rupture strength (TRS) with

volume fraction and size d of inclusions.
Fig. 4. SEM micrograph of the material. Debonded inter-
faces between SiO2 inclusions and the matrix are clearly
visible. equation implies that the modulus should be indepen-
dent of the pore size (the quartz grain size in our case)
the composites should increase [11-14]. A clue to this as indeed appears from our results.
behaviour is presented by the SEM micrograph of The TRS, as expected, decreases significantly with
Fig. 4 which shows a debonded quartz grain in the increasing Vf but, in addition, appears to decrease
clay matrix. As this was observed in most of the with increasing inclusion size d as well (Fig. 5). The
specimens examined it appears that the inclusions decrease is particularly pronounced for Vr=40%
adhere very poorly to the matrix and, in many cases, where TRS is up to 75% lower. This trend is also
actually debonded from it. Such debonding between obvious in the curves of Fig. 2 where materials
the grains and the matrix is probably the result of the displaying unstable fracture generally have much
large expansion accompanying quartz transformation higher strength than those displaying stable fracture
which takes place at about 580°C and which exerts a behaviour.
tensile stress on the surrounding matrix prior to full The total fracture energy G as a function of Vf is
densification. On cooling an equivalent contraction shown in Fig. 6. It includes contributions from both
occurs and, as the fired matrix is unable to follow the and Gd. As no toughening was observed for Vr = 5
contraction, the net result is debonding and separ- and 10% we only have contribution from ~, whereas
ation of the two phases, although the separation is for higher Vf there is substantial energy dissipation
reduced a little by the larger thermal contraction of Gd. For such a composite, ? depends mainly on the
the matrix (~ = 9 x 10-6°C -I) than that of the quartz microstructural nature of the matrix: intrinsic
(g = 1.5 × 10-6°C-'). Direct confirmation of this has strength of crystal planes (for transgranular fracture)
been found by SEM observations, as shown in Fig. 4, or the strength of the interfaces (for intergranular
and by densitometry which showed that total poros- fracture) [17]. Thus, ~ for these composites would be
ity increases with increasing quartz size and volume expected to be equal to the matrix toughness G~¢ and
fraction [15]. Such debonding has been observed in be constant. Our results show this: G~¢ was found to
many industrial fireclay products with quartz as the be approximately constant in all cases, being in the
drying agent. range of 25-35J/m 2 which is reasonable for the
The results suggest that, as far as elastic behaviour matrix material. Close examination of the results in
is concerned, the material behaves like one containing Fig. 6 for Vf = 5 and 10%, reveals that, in fact, Gic
an equivalent population of pores of effective diam- decreases slightly, a trend which is reproducible and
eter equal to that of the quartz inclusions. The elastic has been observed for higher values of Vf as well [15].
behaviour of such materials has been analysed by This is probably the result of significantly increased
MacKenzie [16], who showed that the Young's porosity as shown in Fig. 7 ( V f = 4 0 % and
modulus of a material with a Poisson's ratio about
0.3, containing a population of closed pores of con- 120
stant diameter, can be described by an equation of the
form
80
E = Eo(1 - 1.9Vf + 0.9V 2) (4) 60

where E 0 is the modulus of the uncracked solid and

Vr is the volume fraction of pores. The relation is ~ 20 ~ ~
shown superimposed on the results in Fig. 3. It is
evident that the results are consistent with this 0 5 10 ~5 20 25 30 35 40
equation especially if one takes into account the Volume fraction, %
experimental error of such measurements and the Fig. 6. Total fracture energy as a function of volume
approximate nature of equation (4). In addition, this fraction and size d of inclusions.
 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ 2963

Fig. 7. Micrograph of a specimen with Vf=40% and

d = 400 #m showing the extensive network of cracks in the
matrix.

d = 400 #m). However, the total energy of fracture

(intrinsic toughness plus energy dissipation) increases
dramatically for Vf> 10%. The influence of the
quartz inclusions is thus significant: the total energy
dissipated during fracture increases nearly 4-fold at a
volume fraction of about 20%. After the maximum,
the total energy appears to decrease slightly, probably
due to the extensive damage observed and higher
porosity present at Vf = 40%.
The micromechanisms of toughening were studied
by dynamic in situ microscopy aided by subsequent
SEM of the fracture process. Figure 8 (Vr = 40% and
d = 250 #m) presents a sequence of micrographs of a
crack propagation during loading. A number of
events can be discerned: the crack propagates by
matrix cracking or along a debonded interface and,
in the case of low volume fractions of quartz, the
crack traverses large areas of matrix material before
meeting an inclusion. For large Vf, the crack "jumps"
easily from one inclusion to an adjoining one. These
fracture processes result in crack deflection and crack
bifurcation which interact to produce "bridges"
spanning the crack surfaces, a number of which are
visible in Fig. 8. In these cases, material, cracked Fig. 8. Sequence of micrographs from an in situ experiment
showing crack deflection and bifurcation as a result of
around its boundary but wedged or pinched into both crack inclusion interaction. Vf= 40%, d = 250 pm.
crack faces, does work against friction (sometimes
involving localised crushing) as it is pulled free. The
action of such energy dissipation mechanisms are turn is balanced by a hydrostatic tensile stress on the
known to result in R-curve behaviour (increase in matrix material around them. This results in matrix
toughness) which is more pronounced for higher microdamage and tensile residual stresses in the
populations of inclusions. It is significant that such immediate vicinity of the inclusions: the "damaged
energy dissipation mechanisms have been determined zone". On firing, this damaged zone is highly suscep-
recently to be responsible for R-curve behaviour in tible to cracking and, in conjunction with matrix-in-
monolithic alumina [9]. clusion debonding, defines an "effective flaw radius",
The energy dissipation observed in our experiments re~, over which the local stress falls to the fracture
can be modelled by considering what happens to the strength af, of the dried clay. An example of such
material during preparation. We start by considering damage in the vicinity of an inclusion is shown in
the effect of drying on the matrix integrity and Fig. 9. We can get a rough estimate of reefas follows:
generation of residual stresses and microdamage in we assume (reasonably, e.g. [10]) that the magnitude
the material: before drying the two phases are in of the stress a around a rigid spherical inclusion
intimate contact with one another. During drying, under hydrostatic pressure falls off as approx. 1/r 2,
however, the matrix contracts and exerts a hydro- from ao at the interface to a at distance r from the
static compressive stress on the inclusions which in inclusion. The fracture strength ar is given by
AM 43/8--F
2964 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ

Table 2. Calculated values of N, deer,Acfr/A and d~aWspecimens with

inclusion size 400/am. Values for other inclusion sizes are similar
Damage zone Effectiveflaw
Number of radius r~ radius, rna.
Vr (%) particles(N) (/am) A,et/A (/am)
5 40 562 0.40 424
10 80 547 0.75 518
20 159 516 1.33 1872
40 318 447 1.99 6972

inclusions will not offer any further toughening but

will only serve to decrease the strength.
The variation of T R S with microstructure is con-
Fig. 9. SEM micrograph of an inclusion-matrix interface. sistent with the previous model. Using the basic linear
Damaged matrix in the vicinity of a quartz grain is clearly elastic fracture mechanics equation for the fracture of
visible. brittle materials, Klc = atsx/na, where a is the half-
diameter of the largest enclosed flaw, and taking
trts ~ TRS/1.5 [11] and K~c and E from the present
Ee(1 - Vr) where E is the Young's modulus of the results, we find that the calculated flaw radius raw
dried matrix, e the matrix shrinkage (strain) occurring (given in Table 2) is very close to ten as long as
during drying and Vt is the volume fraction of the Vf~< 10%. The value of rnaw for Vr~> 20% is much
inclusions whose presence reduce the effective shrink- larger probably due to the large overlap of the
age by a factor of (1 - Vf). The extent of the damage individual stress fields.
zone ten will then be given by the radius over which F r o m our results, it appears that a volume fraction
tr falls to the fracture stress of the dried matrix a t. To of quartz inclusions around 20% is an optimum level
calculate this, we carried out a separate set of for toughness as G is maximised at the expense of a
measurements on dried untempered specimens and reduction in T R S of about 40%. These results are
determined the Young's modulus, the linear shrink- valid for the materials prepared in this work: hand-
age and the fracture stress. The results are shown in pressure fabrication without pressure compaction etc.
Table I. Nevertheless, they reflect a definite trend on the
Using these results we calculate that a o ~ mechanical properties of such tempered earthen-
7 0 ( 1 - Vt). Matrix damage due to a stress a thus wares. The results may be different if the specimens
becomes negligible when tr equals the fracture were to be prepared by uni-axial or even isostatic
strength, at = 8.4 MPa at r = reg from which the total compaction or extrusion as used in modern heavy-
damaged area can be calculated. The values of Vf and clay industries. This is being investigated at present.
d define the number of inclusions N, intersecting the
fracture surface from which the total damaged area CONCLUSIONS
can be calculated. The values for N, re~ and the ratio
A,~/A (where A is the actual fracture surface) for The incorporation of a random dispersion of
Vr = 5, 10, 20 and 40% and for a nominal inclusion quartz inclusions in earthenwares (tempering)
size (400#m) are tabulated in Table 2. The table appears to be an effective way of increasing the
shows that, at volume fractions up to 10%, the toughness of the brittle matrix offering substantial
effective damage area is a fraction of the total fracture flaw tolerance. The mechanism of toughening is
area providing limited crack deflection and energy frictional energy dissipation as a result of crack
dissipation. At volume fractions greater than about deflection and bifurcation. This is aided by weak
10% the individual stress fields of the inclusions begin interfaces and localised matrix damage introduced
to interact strongly with the crack producing exten- during drying and firing. As a result, Young's modu-
sive microcrack networks (Fig. 9), encouraging crack lus and transverse rupture strength decreases but
deflection and bifurcation which increases energy toughness increases 4-fold compared to untempered
dissipation. At Vt> 20% the individual stress field material. In the system studied here there appears to
zones cover the whole of the specimen cross-sectional be an optimum volume fraction of inclusions which
area and thus the toughening effect at these levels of is independent of inclusion size around 20% above
Vf is constant. Adding more than about 20% which no further increase in toughness is found. The
results can be explained by considering the extent of
microdamage introduced in the matrix during drying
Table 1. Properties of dried matrix material (average of three
measurements) and firing due to the presence of the quartz
Young's modulus, E (GPa) 3.37 _+0.4 inclusions.
Transverse rupture strength, TRS (MPa) 12.6 _+1.5
Estimated tensile strength, o'f (MPa)t 8.4 _+1.5 Acknowledgements--We would like to express our gratitude
Ultimate compressivestress, UCS (MPa) 12.9 +_1.5 to Professor M. F. Ashby for his incisive and helpful
Linear shrinkage:~, e (%) 2 + 0.2 comments and suggestions. We would also like to thank
tCalculated as = TRS/1.5. 3~Untildry. Professor G. Tsamasfyros of the National Technical
 KILIKOGLOU et al.: TOUGHENING OF EARTHENWARES BY QUARTZ 2965

University of Athens for permission to use his laboratory 8. D. Munz, R. T. Bubsey and J. L. Shannon Jr, J. Am.
equipment. Ceram. Soc. Notes 63, 300 (1980).
9. G. Vekinis, M. F. Ashby and P. W. R. Beaumont, Acta
metall, mater. 39, 2583 (1991).
REFERENCES 10. S. P. Timoshenko and J. N. Goodier, in Theory of
Elasticity, 3rd edn. McGraw-Hill, New York (1970).
1. I. V. Pinto, M. B. Schiffer, S. Smith and J. M. Skibo, 11. R. W. Hertzberg, in Deformation and Fracture Mech-
Archeomaterials 1, 119 (1987). anics of Engineering Materials, 3rd edn. Wiley, New
2. K. Brannigan, in Aegean Metalwork o f the Early and York (1989).
Middle Bronze Age. Oxford University Press, Oxford 12. Z. Hashin and S. Strikman, J. appl. Phys. 33, 3125
(1974). (1962).
3. B. A. Sabrah and E. A. M. Ebied, Am. Ceram. Soc. Bull. 13. J. D. Eshelby, Proc. R. Soc. Lond. A241, 376 (1957).
65, 759 (1985). 14. M. F. Ashby, Acta metall, mater. 41, 1313 (1993).
4. D. A. Okongwu, Am. Ceram. Soc. Bull. 67, 1409 15. V. Kilikoglou, G. Vekinis and Y. Maniatis, to be
(1988). submitted to Archaeometry.
5. G. Bronitsky and R. Hamer, Am. Antiq. 51, 89 16. J. K. MacKenzie, Proc. Phys. Soc. (Lond.)!]163, 2
(1986). (1950).
6. Y. Maniatis and M. S. Tite, J. Arch. Sci. 8, 59 (1981). 17. B. Lawn, in Fracture o f Brittle Solids, 2nd edn.
7. J. E. Srawley, Int. J. Fract. Mech. 12, 475 (1976). Cambridge University Press, Cambridge (1993).

View publication stats