Journal of the European Ceramic Society 23 (2003) 3089–3096

[Link]/locate/jeurceramsoc

Young’s modulus measurements of magnesia–spinel composites

using load–deflection curves, sonic modulus, strain gauges and
Rayleigh waves
Cemail Aksela,*, Frank L. Rileyb
a
Department of Materials Science and Engineering, Anadolu University, Iki Eylül Campus, Eskisehir 26470, Turkey
b
Department of Materials, School of Process, Environmental and Materials Engineering, University of Leeds, Leeds LS2 9JT, UK

Received 21 September 2002; received in revised form 16 January 2003; accepted 1 February 2003

Abstract
The extent of interlinking of the microcracking and a decrease in strength and modulus values were determined to be a function
of both spinel particle size and volume fraction to allow calculation of thermal shock parameter, R000 . Measurements of Young’s
modulus were carried out both at room temperature and after thermal shock testing by using the load–deflection curves (defined as
‘‘mechanical’’ modulus) and by the sonic modulus technique. The values obtained from these methods were significantly different
for quenched/unquenched samples. To understand the basis for these differences, strain gauge and Rayleigh wave methods were
also used to determine Young’s modulus of the composites. Modulus values obtained from these methods confirmed the differences
measured, and provided a guide to the values to be used in calculating thermal shock parameters. The mechanical modulus tech-
nique was considered the most meaningful indicator of Young’s modulus for a situation in which large mechanical strains were to
be applied to the materials during thermal shock.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Mechanical modulus; MgAl2O4; MgO; Rayleigh waves; Sonic modulus; Strain gauge; Thermal shock

1. Introduction effect of porosity is expected to be small. Minimum

changes are observed when the solid phase is con-
The elastic properties of ceramics determine mechan- tinuous. The largest changes occur when the pore phase
ical behaviour and are dependent on crystal structure is continuous where only point contact exists between
and bonding. Young’s modulus decreases as tempera- solid grains.2 When the pores are randomly distributed,
ture is increased, due to lattice expansion.1 The effects the effect of porosity on strength and elastic moduli is
of porosity on the physical properties of ceramics have smaller.2
also been observed in many studies on sintering and The reduction in Young’s modulus cannot be
designed in development programme to establish opti- explained only by the porosity. The decrease in moduli
mum forming and firing procedures for commercial is attributed to the formation of spontaneous micro-
production. However, changes in processing and com- cracking.3 In terms of elastic strain on a stress–strain
position variables usually produce changes in crystal curve, the introduction of either pores or pre-existing
size, pore size and shape. It is therefore difficult to eval- cracks leads to a reduction in Young’s modulus and to
uate the effect of porosity alone. The solid phase is an equivalent increase in elastic strain for a given stress.
continuous with isolated pores. Porosity does not affect The increase in elastic strain on the basis of the crack-
the expansion coefficient in isotropic polycrystalline opening displacement is caused by the presence of radial
bodies. In bodies composed of anisotropic crystals the and/or annular cracks associated with the pores.4 The
values of Young’s modulus decrease with increasing
* Corresponding author. Tel.: +90-222-335-0580x6355; fax: +90-
degree of microcracking, depending on the nature of
222-323-9501. crack interaction. Heavily microcracked materials exhi-
E-mail address: caksel@[Link] (C. Aksel). bit a Young’s modulus substantially lower than non-
0955-2219/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0955-2219(03)00103-1
3090 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096

microcracked materials.5 For example, MgO has a effective elastic modulus of a cracked medium depends
compact microstructure with a small percentage of dis- on crack shape, density and crack preferred orientation.
continuities, which can initiate cracking. MgO–chromite The measured effective elastic properties of the damaged
bricks exhibit lower Young’s modulus as compared to material represent the effect of microcracks.23
MgO bricks.6 For example in the MgO–spinel system, In this work, fully dense materials have been obtained
Young’s modulus falls to a minimum with increasing and thus the influence of porosity on modulus has been
spinel content at 30–40% addition.1,7 However, neglected, but attention has been given to the relation-
Young’s modulus increases significantly with spinel ship between composition and microcracking, which
content > 40% (e.g. back to the MgO value). As the affects Young’s modulus significantly. The extent of
spinel becomes the continuous phase, Young’s modulus interlinking of the microcracking has been investigated
stayed approximately constant between 50 and 80%.1 as functions of spinel particle size and volume fraction,
Magnesium aluminate spinel (MgAl2O4) is an impor- and the reasons for this are examined. The effects of
tant constituent of MgO-based refractory materials. varying the amounts of spinel with different particle
Spinel particles are added in various proportions to sizes on both strength and Young’s modulus using vari-
MgO to improve thermal shock resistance.8ÿ10 Thermal ous methods were investigated to allow calculation of
shock only occurs during the heating and cooling peri- thermal shock parameter, R000 . The reliability of different
ods, when temperature changes can be large and rapid, methods determining the modulus values has been
leading to large thermal stresses.11 Fired MgO–spinel established on the basis of thermal shock data to allow
brick exhibits high resistance to thermo-mechanical detailed modelling of the thermal shock behaviour of
stress.12 The reason for these improvements13 is linked magnesia-spinel composite materials of compositions
to the large difference in thermal expansion coefficient and microstructures matching those of less pure refrac-
between MgO (13.610ÿ6  Cÿ1 from 25  C to 1000  C) tory materials.
and spinel (8.410ÿ6  Cÿ1 from 25  C to 1000  C).14,15
This difference leads to the development of large tensile
stresses, and eventually extensive microcracking, around 2. Experimental
spinel grains during cooling from fabrication tempera-
tures in excess of 1600  C.12,16,17 Due to the microcrack MgO powder of > 98.0% purity (‘light’: GPR, BDH,
development around the spinel grains, the radial cracks Poole, UK) was calcined at 1300  C for 2 h, to produce
produced readily link together.16,17 The improved resis- a powder with a mean particle size of 0.5 mm. A > 97%
tance to thermal shock observed in MgO–spinel com- dense materials were prepared by hot-pressing at
posites can therefore be attributed to the microcrack 1700  C and 20 MPa for 25 min. 12 and 24 mm spinel
networks developed around the spinel particles.18ÿ20 powders were obtained by crushing and sieving high
There seems to exist a critical volume fraction above density ( 98% purity) spinel ‘pebbles’ (Britmag 67,
which spontaneous microcracking will occur and this Redland Minerals, UK). The stoichiometry of spinel by
critical volume fraction is related to the critical grain weight% is given as follows: 66.3% Al2O3, 31.6% MgO
size. The reason for the dependence on volume fraction and 2.1% the total amounts of impurities. MgO con-
is associated with the residual tensile stresses in the taining spinel particles reached the theoretical density
composites. It was found that tensile stresses in the ( 99%), by hot-pressing at 1720  C and 20 MPa for 25
matrix will increase with increasing volume fraction of min. Samples for measuring Young’s modulus with flat,
an additive. Therefore, composites using particles with parallel ends were required. The ends of these samples
sizes less than the critical size at a specific volume frac- were cut parallel to each other using a diamond cutter
tion do not eliminate the formation of spontaneous (Struers Accutom-2), which allowed precise alignment
microcracking at higher volume fractions. Spontaneous of the samples and performed precision cutting.
microcracking leads to reductions in Young’s modulus.3 Young’s modulus values of MgO and MgO–spinel
In terms of microcracking, Young’s modulus values composites were measured, as a function of particle size
are important to determine thermal shock resistance. A and spinel addition at room temperature using various
non-uniform distribution of cracks of random or pre- techniques (Sections 2.1–2.4). The R000 parameter,
ferred orientation can also be used to advantage in expressing the ability of a material to resist crack pro-
improving thermal stress resistance.21 As the micro- pagation and further damage on thermal shocking, was
cracking increases, Young’s modulus decreases depend- calculated by the formula, {R000 =E/[ 2f .(1ÿ)]},25ÿ27
ing on the size of the cracks.22 As significant amount of where E is Young’s modulus,  is strength, and  is
microcracks appears after thermal shock because of Poisson’s ratio. Thermal-shock tests were then made by
thermally induced stresses, permanent loss of stiffness holding the rectangular specimens for at least 20 min to
and strength occur.23 It is considered that this beha- allow for temperature equilibration in a vertical tube
viour of strength, and Young’s modulus in composites furnace maintained at 1000  C and dropping them into
was due to the growth and combination of cracks.24 The a container of silicone oil at room temperature, which
 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096 3091

was being stirred with a magnetic stirrer (at room tem- resulting mechanical vibrations of the specimen and
perature). After cleaning with acetone, samples were transform them into an electrical signal that could be
dried in an oven at 110  C before breaking in three- displayed on the screen of an oscilloscope used to detect
point bend. Young’s moduli of composites and pure fundamental and harmonic vibration frequencies.29
MgO were measured after quenching from 200 to Longitudinal vibration occurs by a sinusoidal force
1000  C, at 200  C intervals, using load–deflection applied to one of the parallel ends. The stress wave
curves (mechanical modulus-Em) and the sonic modulus propagates down the specimen and is reflected at the
(Es) techniques, as a function of quench temperature. other end. The frequency of vibration of the specimen
Thermal shock results were then evaluated on the basis was recorded that resulted in a maximum displacement,
of the calculated R000 parameter. The CamScan 4 SEM having a well-defined peak on the indicator.30 The
equipped with an EDX system for elemental analysis standard equation 29,30 for the sonic modulus (ES) is:
was also used to examine the size, shape, texture of the ES ¼ 4:f 2 :l2 : ð2Þ
particles, the presence and position of second phases in
the polished surface of the multiphase materials. where: f=frequency of longitudinal mode of vibration
(velocity of sound/wavelength), l=length of the specimen,
2.1. Modulus from stress–strain data and =density of the specimen. Specimens having
either very small or very large ratios of length to thickness
Modulus measurements were made at room tempera- may be difficult to excite in the fundamental modes of
ture and after thermal shock testing in three-point bend, vibration. To apply this method, the ratio must be
using bars of dimension of 2632.5 mm3. Five speci- between 5 and 20.30 This method is non-destructive, and
mens were normally used to obtain each mean value, specimens used for this method can subsequently be
with a support roller span of 20 mm and a cross-head used for other tests.
speed of 0.2 mm minÿ1. Mechanical modulus (Em)
values were calculated by drawing a tangent to the 2.3. Strain gauges
steepest initial straight-line portion of the load–deflec-
tion curve, where the stiffness of the machine was also In order to compare and confirm the results obtained
considered, using Eq. (1)28 from the mechanical and sonic modulus methods, spe-
Em ¼ L 3 m= 4WD 3
ÿ

ð1Þ cimens were also loaded with a strain gauge in the cen-
tral zone between the inner loading rollers on the tensile
where L=support span, W=width of bar, D=depth of faces, using a support roller span of 20 mm, and a cross-
bar, and m=slope of the tangent of the initial straight- head speed of 0.2 mm minÿ1. Measurements were made
line portion of the load–deflection curve. using the four-point bend test.31 The equations devel-
oped32,33 for the measurement of strain (), resistance
2.2. Sonic modulus (R), and output voltage data (d ) are:

A sonic modulus method was also used to measure  ¼ 100½ðR ÿ rÞ=ðkSG :Rފ ð3Þ
Young’s modulus. This is a procedure for measuring the 
resonance frequency, based on the vibration of a rec- R ¼ ½rðd =GÞ þ 300Š=½Vo ÿ d =GŠ ð4Þ
tangularly shaped specimen at room temperature.29 The
test required the samples to be electrically conductive, where:
and samples were painted on both their ends with silver r=(resistance of amplifier plus nominal resistance of
paint, and a thin stripe along their length. The sample strain gauge at 21  C(120
))
was clamped at its middle point by two knife edge sup- d =Output voltage (plus 5.106 to make positive the
ports on the testing apparatus, electrically earthed, and all data)
placed between two electrodes with a voltage of  200 G=Amplifier gain setting/108
V. The electrodes were brought very close to the speci- Vo=Half bridge voltage across Wheatstone Bridge
men ends (usually < 0.2 mm) to form parallel-plate (2.5 V)
capacitors with the flat ends of the specimen. In addi- kSG=Strain gauge constant (2.14).
tion, a small periodic voltage of the order of a few mV
was fed to one of the electrodes, the driver electrode, 2.4. Rayleigh waves
which set the sample into vibration caused by the oscil-
lating electrostatic field. The parallel-plate capacitor on To measure the Rayleigh wave velocity in the line-
the other end sensed the oscillation by means of chan- focus acoustic microscope a ZnO transducer is used to
ging charges due to changing capacitance caused by the launch an acoustic wave along a sapphire lens. Water
vibration. Test specimens were vibrated in a broad fre- acts as a couplant to transmit the acoustic energy from
quency range: the electrode was used to sense the the lens to specimen. At the sapphire/water interface
3092 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096

waves are refracted towards the sample. Two waves are the coarser the spinel particle size. Spinel content, but
of interest: one travels straight down to the specimen predominantly spinel particle size, significantly affected
and is reflected back along the axis of the lens. The sec- the strength. There was a significantly greater decrease
ond wave is refracted at the Rayleigh angle, R, typically in strength for 24 mm compared to 12 mm spinel parti-
about 30 . This wave excites a leaky Rayleigh wave in cles, and the 24 mm spinel particles were associated with
the surface of the specimen, which in turn radiates a formation of longer cracks than the 12 mm particles
wave back towards the lens. The net signal received at (Figs. 1 and 2). The total extent of cracking also
the transducer is the phase-sensitive sum of these two increased with increasing concentration of spinel
waves. As the specimen is scanned up towards the lens (Fig. 3).
(z-scan) the voltage output, V(z), shows oscillations, the The Young’s modulus of the pure hot-pressed MgO
period of which,
z, is related to the velocity of the at room temperature was in good agreement with lit-
Rayleigh wave, vR, by the following equation:34,35 erature values of  290 GPa36 ( 25 mm grain size), and
258 GPa38 ( 7 mm grain size). The Young’s modulus of
Vw spinel composites generally decreased with increasing
VR ¼  ð5Þ
1 ÿ ð1 ÿ Vw =2fDzÞ2
1=2 spinel content (Fig. 4). There was little difference
between the 12 and 24 mm spinel composites. The
decrease in mechanical Young’s modulus was most sig-
where vw is the velocity of sound in water (  1500 m sÿ1 nificant, for up to 10% additions. At 10%, Young’s
at room temperature) and f is the frequency of the modulus was approximately 50% of the value for pure
sound generated. In the microscope used in this work, MgO. Increasing the amount of spinel from 10 to 30%,
f=225 MHz. The shear velocity, vs is:

vs ¼ vR = 1:14418 ÿ 0:25771 þ 0:126612

ÿ

ð6Þ

where vR is the Rayleigh wave velocity and  Poisson’s

ratio. Given the Rayleigh wave velocity, Poisson’s ratio
(: 0.29) and density (: 3.52 Mg mÿ3); elastic modulus
E was calculated from the equation as follow:
E ¼ 2Gð1 þ Þ ¼ 2v2s ð1 þ Þ ð7Þ

3. Results and discussion

3.1. Strength and Young’s modulus

Fig. 1. SEM micrograph of a 20% 12 mm spinel composite (dark grey:
MgO, light grey: spinel).
As shown in Table 1, the room temperature strength
of the hot-pressed pure MgO was in good agreement
with the literature values of  225 and 230 MPa36,37 for
 32 and 25 mm MgO grain sizes, respectively. The
addition of up to 10% of the spinel powder to MgO
decreased the strength significantly ( 75%). Further
additions caused no further loss (though they may have
been a slight increase at 30% addition, but the scatter of
data was larger). The decrease in strength was greater

Table 1
Bend strength as a function of spinel content and particle size

Spinel (wt.%) s (MPa)

0 5 10 20 30

12 mm spinel 233 7 14217 9424 62 28 10322

24 mm spinel 233 7 9114 4817 50 10 7023 Fig. 2. SEM micrograph of a 10% 24 mm spinel composite (dark grey:
MgO, light grey: spinel).
 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096 3093

had a relatively small effect. The loss of modulus can be sonic modulus values followed a similar pattern to the
explained in terms of radial crack propagation and ulti- mechanical modulus. However, sonic modulus values
mately interlinking (Figs. 1–3). It appears that crack for spinel composites, showed only small changes with
interlinking is completed at about 10% addition of spi- increasing spinel content, compared to pure MgO. Fig. 4
nel, undermining the mechanical integrity of the mat- showed that when the spinel content of composites
erial, and changes little after this point. In contrast, increased, both values decreased, but the mechanical
Young’s modulus values determined by the sonic modulus showed a more marked loss of Young’s mod-
method were approximately twice the mechanical mod- ulus than the sonic modulus.
ulus values for spinel amounts 510%, and showed a
smaller decrease with increasing spinel content. 3.2. R000 parameter
The modulus values determined using strain gauges
for pure MgO were in reasonable agreement with the Because of the importance of the strength and
sonic modulus and mechanical modulus values. There is Young’s modulus (E/ 2) ratios for calculation of ther-
thus consistency, within the limits of experimental mal shock parameter (R000 ), it was of interest to investi-
errors, between these data. On the other hand, Young’s gate how strength and Young’s modulus varied for each
modulus for pure MgO calculated using Rayleigh waves composition. The R000 parameter25ÿ27 gives information
was less than the calculated from the stress–strain curve about available energy at fracture for resisting to crack
(7%), and the value measured using the sonic modulus propagation, and high values indicate an improvement
technique (12%). in thermal shock resistance.
Young’s modulus values calculated using Rayleigh The variations of R000 parameter for MgO and spinel
waves were close to values measured using the sonic composites are shown in Fig. 5. On the basis of mechan-
modulus technique, and the values obtained using strain ical modulus (Em) and sonic modulus (Es) data, 24 mm
gauges were also close to the values from the stress– spinel composites attained by far the highest R000 para-
strain measurements. For this reason, mechanical mod- meter values. For the 24 mm spinel, the maximum R000
ulus and sonic modulus methods were used for both value was at 10% addition; the 12 mm spinel B had a
unquenched and quenched samples, containing spinel maximum at  20%. The maximum R000 parameters for
powders, because significantly different values were the 12 and 24 mm spinel composites were approximately
obtained from each method, the origin of which needed 7–12 times (using Em) and 11–21 times (using Es) larger,
to be investigated.
Mechanical Young’s moduli of composites decreased
with additions of spinel, for each particle size (Fig. 1).
SEM observations showed that (in Figs. 1–3) coarse
particles were associated with longer crack formation,
compared to fine particles. The crack lengths also
increased with increasing concentration of spinel.
The decrease in mechanical modulus was significantly
greater, the higher the spinel content. The changes in

Fig. 4. Young’s modulus of MgO and MgO–spinel composites as a

Fig. 3. SEM micrograph of a 30% 24 mm spinel composite (dark grey: function of spinel content and particle size: mechanical modulus (Em),
MgO, light grey: spinel). sonic modulus (Es), strain gauges (Sg), and Rayleigh waves (Rw).
3094 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096

respectively, than that of pure MgO (Fig. 5). These increasing shock temperature up to  575  C. After
trends indicate that strength values were initially more quenching between 575 and 625  C, the Young’s mod-
sensitive to composition than Young’s modulus, but as ulus of MgO fell to about 30% of the initial value,
modulus values decreased slightly with increasing addi- because of the nucleation of microcracks and their pro-
tions, strength remained approximately constant (Fig. 4 pagation. Above 625  C, Young’s modulus decreased
and Table 1). It is seen that the effect of incorporating more slowly with increasing temperature (Fig. 6), pre-
spinel into magnesia is to cause a sharp decrease in sumably because the number of cracks increased only
strength, through microcrack generation. The strength slightly with increase in the thermal shock temperature.
and Young’s modulus values decreased with spinel In contrast, spinel composites tested showed a stable
additions, and the influence of the spinel was observed trend (Fig. 6). The cracks initiated by thermal shock
to be more marked the larger the spinel particle size. seemed to propagate only a short distance and were
Strengths of the coarser spinel composites were in gen- then arrested rather than that the cracks had a sudden
eral much more sensitive to composition than was extension, giving catastrophic fracture. For this reason,
Young’s modulus, noting the wide spread of particle there was a gradual decrease in the Young’s modulus of
sizes. R000 parameter shows that the smaller the spinel composites.
particle size, the lower the resistance to crack propaga- Fig. 6 shows that the mechanical Young’s modulus of
tion (Fig. 5). Materials containing 12 mm spinel had pure MgO decreased abruptly in the temperature range
short length cracks originating from the spinel particles; 575–625  C. Above this temperature, mechanical mod-
the 24 mm spinel particles were the origin of longer ulus decreased gradually with increasing shock tem-
cracks, and more crack initiation sites (Figs. 1–3). On perature. In contrast, there was in general a slight
the basis of the calculated R000 values, coarse spinel par- decrease in sonic modulus values of MgO with increas-
ticles are generally more beneficial than the fine parti- ing shock temperature. Even when the crack length
cles, for which a very much larger volume appears to be increased and more cracks were introduced by thermal
required to achieve the same improvement. As shown in shocking, the sonic technique was unable to detect the
Fig. 5, the thermal shock resistance of 10% 24 mm spinel microcracks. A similar, but more stable trend, was also
composites was predicted to be  2 times greater than observed for the spinel composites by using both meth-
those obtained from 20% 12 mm spinel composites. It ods: there was a marked decline in mechanical modulus
was therefore expected that resistance to crack propa- values whereas sonic modulus values remained almost
gation in general would be greater with materials con- at the same level, with increasing quench temperature.
taining the coarsest spinel particle size. On the basis of the prediction from the R000 parameter
values; sonic modulus data seemed to illustrate the
3.3. Thermal shock testing highest resistance to crack propagation, for 10% 24 mm
spinel composites, in comparison with the values
The Young’s moduli of pure MgO and spinel compo- obtained from mechanical modulus (Fig. 5). However,
sites, were also measured after shock from temperatures thermal shock results showed that sonic modulus values
at 200  C intervals up to 1000  C (Fig. 6). For each of MgO were still higher than that of spinel composites.
material the Young’s modulus decreased gradually with On the other hand, mechanical modulus values of spinel
composites was much higher than pure MgO at the
highest quench temperature used, even more cracks

Fig. 5. R000 parameter as a function of spinel content and particle size: Fig. 6. Mechanical modulus from load–deflection curve (Em) and
mechanical modulus (Em) and sonic modulus (Es). sonic modulus (Es), as a function of quench temperature.
 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096 3095

were introduced by thermal shocking (Fig. 6). R000 para- thermal shocking (in MgO for example) the sound wave
meter values obtained from mechanical modulus does not really detect the presence of the cracks. This is
method are thus in good agreement with the thermal because the crack is insufficiently wide to prevent brid-
shock test data. For this reason, mechanical modulus, ging by the crack front. In the case of very severely
Em, was found to be the most reliable indicator for shocked materials a decrease in the sonic modulus
evaluating Young’s modulus data rather than using values starts to be seen (Fig. 6), and the greater crack
sonic modulus technique for those materials. width, or density now appears to have an influence on
These results suggest that this behaviour of Young’s the sound wave. The key criteria may be whether the
modulus was because of the formation and interlinking cracks are sufficiently wide, or have propagated to form
of microcracks, which is most marked for MgO, in a continuous link amongst the spinel particles. The
comparison with the spinel composites. Spinel compo- above observations explain that the sonic modulus
sites were not as badly cracked, as in the MgO materi- technique is not a sensitive detector of certain types of
als, and a high concentration of larger cracks was not flaw generated in these materials, and gives consistently
occurred at the critical quench temperature of MgO. A higher values than the mechanical modulus technique. It
gradual decrease in the Young’s modulus occurred was therefore decided that the evaluation of thermal
because of the high levels of pre-formed thermal expan- shock parameters should be more reliable on mechan-
sion mismatch microcracks already present in the struc- ical Young’s modulus measurements, calculated from
ture. With increase in microcrack density after thermal load deflection curves. This also applied to quenched
shock, crack interaction effects became more important. samples, for which even lower mechanical modulus
The high thermal shock resistance was attributed to the values were obtained.
interlinking of high level of pre-existing microcracking
occurred in the spinel materials.
4. Conclusions
3.4. Comparison of Young’s modulus measurements
It is suggested that Rayleigh waves and the sonic
Young’s modulus values obtained for spinel compo- modulus technique cannot detect fine microcracks, and
sites using strain gauges were very close (e.g. with an it cannot therefore give reliable information about
 5% deviation) to the values from the stress–strain cracked materials. Even when more cracks were intro-
mechanical measurements, and the values obtained duced by quenching from above the critical quench
using Rayleigh waves were quite close (e.g. in the range temperature for MgO, the sonic technique could not
of  10%) to the values measured using the sonic mod- provide a sensitive detector of microcracking. Further-
ulus technique (Fig. 4). Mechanical modulus and sonic more, the load applied during the sonic modulus mea-
modulus methods used in this work gave significantly surements to specimens was insignificant, and further
different values. It is suggested that when a material has extended microcracking would not have been caused.
a higher expansion coefficient than the second phase The mechanical modulus technique (using the load–
particles it contains, then during cooling, the matrix will deflection curves) was considered the most meaningful
clamp down onto the second phase. It is clear that in the indicator of Young’s modulus in the evaluation of ther-
case of the MgO–spinel composites, this results in the mal shock parameters for a situation in which large
production of very narrow radial microcracks (Figs. 1– mechanical strains were to be applied to materials by
3), because of the hoop stress developed around the thermal shock. This is also supported by the R000 para-
spinel particles. These materials then provide a con- meter values and confirmed by the thermal shock test
tinuous mechanical path through the MgO matrix for data. In addition, Young’s modulus values of spinel
the sound pulse to follow. In contrast, wider micro- composites in general decreased with increasing spinel
cracks occurring in more severely shocked materials content. Modulus appeared to be more sensitive to spi-
disrupt the sound pulse, and must then be no direct nel content than to spinel particle size.
solid path: a lower sonic modulus value is seen, because
of a decrease in the velocity of the sound wave, and
hence resonance frequency. However, there was in Acknowledgements
general a slight decrease in sonic modulus values with
even increasing shock temperature (Fig. 6). The Redland Minerals (UK) is thanked for supplies of
mechanical modulus would clearly be influenced by the materials. The contributions of Dr. P.D. Warren, Prof.
matrix microcracking to a much greater extent. Interpreta- B. Rand and the late Prof. R.W. Davidge are gratefully
tion of the sonic modus data is not straightforward. It acknowledged. Special thanks go to Dr. C. Lawrence
appears that in materials which contain relatively small for his help with Rayleigh Wave method. The authors
amounts of local microcracking arising from the thermal also wish to thank Dr. Roger Morrell, whose invaluable
expansion mismatch (the composites), or from milder expertise in this field was greatly appreciated. Dr. P.
3096 C. Aksel, F.L. Riley / Journal of the European Ceramic Society 23 (2003) 3089–3096

Bartha, Dr. S. Plint, and M.W. Roberts are also 17. Aksel, C., Davidge, R. W., Knott, P. and Riley, F. L., Mechan-
thanked for helpful discussions. This work was in part ical properties of magnesia–magnesium aluminate spinel compo-
sites. In III. Ceramic Congress Proceedings Book, Engineering
carried out as part of a PhD programme in the Depart-
Ceramics, Vol. 2, Istanbul, Turkey, 1996, pp. 172–179.
ment of Materials of the University of Leeds, and the 18. Aksel, C., Rand, B., Riley, F. L. and Warren, P. D., Mechanical
authors thank the Technical Staff of the Department for properties of magnesia–spinel composites. J. Eur. Ceram. Soc.,
their assistance. Financial support was provided in part 2002, 22, 745–754.
by the Council of Higher Education in Turkey. 19. Aksel, C. and Warren, P. D., Thermal shock parameters [R, R000
and R0000 ] of magnesia-spinel composites. J. Eur. Ceram. Soc.,
2003, 23, 301–308.
References 20. Aksel, C., Davidge, R. W., Warren, P. D. and Riley, F. L.,
Investigation of thermal shock resistance in model magnesia–spi-
1. Cooper, S. C. and Hodson, P. T. A., Magnesia–magnesium alu- nel refractory materials. In IV. Ceramic Congress, Proceedings
minate spinel as a refractory. Trans. J. Brit. Ceram. Soc., 1982, Book, Part 1, Eskisehir, Turkey, 1998, pp. 193–199.
81, 121–128. 21. Kingery, W. D., Factors affecting thermal stress resistance of
2. Coble, R. L. and Kingery, W. D., Effect of porosity on physical ceramic materials. J. Am. Ceram. Soc., 1955, 38, 3–15.
properties of sintered alumina. J. Am. Ceram. Soc., 1956, 39, 22. Hasselman, D. P. H. and Singh, J. P., Analysis of thermal stress
377–385. resistance of microcracked brittle ceramics. Am. Ceram. Soc.
3. Chou, Y. S. and Green, D. J., Silicon carbide platelet/alumina Bull., 1979, 58, 856–860.
composites: II, mechanical properties. J. Am. Ceram. Soc., 1993, 23. Rokhlin, S. I., Chu, Y. C. and Baaklini, G. Y., Assessment of
76, 1452–1458. damage in ceramics and ceramic matrix composites using ultra-
4. Krstic, V. D. and Ericson, W. H., A model for the porosity sonic techniques. Journal of Engineering for Gas Turbines and
dependence of Young’s modulus in brittle solids based on crack Power, 1995, 117, 417–423.
opening displacement. J. Mater. Sci., 1987, 22, 2881–2886. 24. Matsuuhita, K., Kuratani, S., Okamoto, T. and Shimada, M.,
5. Hasselman, D. P. H., Thermal stress resistance parameters for Young’s modulus and internal friction in alumina subjected to
brittle refractory ceramics: a compendium. Am. Ceram. Soc. thermal shock. J. Mater. Sci. Lett., 1984, 3, 345–348.
Bull., 1970, 49, 1033–1037. 25. Hasselman, D. P. H., Elastic energy at fracture and surface
6. Szmeja, B. T. and Wala, T. B., Thermal shock acoustic emission energy as design criteria for thermal shock. J. Am. Ceram. Soc.,
and microstructure of refractories II. Basic refractories. Ceram. 1963, 46, 535–540.
Int., 1994, 20, 367–377. 26. Hasselman, D. P. H., Unified theory of thermal shock fracture
7. Aksel, C., Thermal Shock Behaviour and Mechanical Properties of initiation and crack propagation in brittle ceramics. J. Am.
Magnesia–Spinel Composites. PhD Thesis, Department of Mate- Ceram. Soc., 1969, 52, 600–604.
rials Engineering, University of Leeds, Leeds, UK, 1998. 27. Sack, R. A., Extension of Griffith’s theory of rupture to three
8. Bartha, P., Magnesia spinel bricks—properties, production and dimensions. Proc. Physical Soc., London, 1946, 58, 729–736.
use. In Proc. Int. Symp. Refractories, Refractory Raw Materials 28. ASTM D790M-86, Standard Test Methods for Flexural Proper-
and High Performance Refractory Products, ed. X. Zhong et al. ties of Unreinforced and Reinforced Plastics and Electrical Insu-
Pergamon, Hangzhou, 1989, pp. 661–674. lating Materials, Annual Book of ASTM Standards, Vol. 08.01.
9. Klischat, H. J. and Bartha, P., Further development of magnesia ASTM, 1988, pp. 290–298.
spinel bricks with their own specific properties for lining the 29. ASTM C885-881, Standard Test Method for Young’s Modulus of
transition and sintering zones of rotary cement kilns. World Refractory Shapes by Sonic Resonance, Annual Book of ASTM
Cement, 1992, 52–58. Standards, Vol. 17. ASTM, 1982, pp. 964–970.
10. Kimura, M., Yasuda, Y. and Nishio, H., Development of mag- 30. ASTM C747-774, Standard Test Method for Moduli of elasticity
nesia spinel bricks for rotary cement kilns in Japan. In Proc. 26th and fundamental frequencies of carbon and graphite materials by
Int. Col. Ref., Interceram Special Issue, 1984, Vol. 33, Aachen, sonic resonance, Annual Book of ASTM Standards, Vol. 17.
Germany, 1983 pp. 344–376. ASTM, 1982, pp. 764–774.
11. Geisler, T. A., Finite element analysis of thermal stresses in 31. ASTM C1161-90, Standard Test Method for Flexural Strength of
cement kiln brick. In UNITECR ’89, Anaheim, USA, 1989, pp. Advanced Ceramics at Ambient Temperature, Annual Book of
1317–1327. ASTM Standards, Vol. 15.01. ASTM, 1991, pp. 327–333.
12. Soady, J. S., and Plint, S., A quantitative thermal shock 32. Strain Gauges Catalogue. TML Pam E-101I, Tokyo Sokki Ken-
approach to the development of magnesia-spinel refractories for kyujo Co., Ltd., 8-2, Minami-ohi 6-chome, Shinagawa-Ku,
the cement kiln. In UNITECR ’91, Aachen, Germany, 1991, pp. Tokyo 140-8560, Japan.
443–449. 33. Marshall, C. W. and Maringer, R. E., Dimensional Instability—
13. Tabbert, W. and Klischat, H. J., Magnesia spinel bricks for the An Introduction. Pergamon Press, Battelle, Columbus Labora-
cement industry. In Proceedings Beijing China Symposium, China, tories, Columbus, OH, USA, 1977.
1992, pp. 424–430. 34. Rayleigh, L., On waves propagated along the plane surface of an
14. Shackelford, J. F., Alexander, W. and Park, J. S., (eds), CRC elastic solid. Proc. Lond. Math. Soc., 1885, 17, 4–11.
Materials Science and Engineering Handbook. CRC Press, Boca 35. Briggs, G. A. D., Acoustic Microscopy. Oxford University Press,
Raton, FL, 1994. Oxford, 1992.
15. Burnett, S. J., Properties of Refractory Materials. UKAEA 36. Davidge, R. W., Mechanical Behaviour of Ceramics. Cambridge
Research Group Report, Harwell, 1969. University Press, Cambridge, 1979.
16. Aksel, C., Davidge, R. W., Warren, P. D. and Riley, F. L., 37. Rice, R. W., Strength and fracture of hot-pressed MgO. Proc.
Mechanical properties of model magnesia–spinel composite Brit. Ceram. Soc., 1972, 20, 329–363.
materials. In Euro Ceramics V, Part 3, Extended Abstracts of the 38. Llorca, J. and Ogawa, T., Crack wake effects on MgO fracture
5th Conference and Exhibition of the European Ceramic Society, resistance. In Fracture Mechanics of Ceramics, Vol. 9, eds. R. C.,
Key Engineering Materials, Vols 132–136, Versailles, France, Bradt, D. P. H., Hasselman, D., Munz, M. Sakai, V. Ya., Shev-
1997, pp. 1774–1777. chenko. 1992, pp. 305–317.