MIATERIALS ENE S ENGINEERING ELSEVIER Materials Science and Engineering A278(2000)187-19 www.elsevier.com/locate/msea The mechanical behaviour of glass and glass-ceramic matrix composites K.-L. Choy a, * P. Duplock a, P.S. Rogers a, J. Churchman-Davies b, M.T. Pirzada Department of Materials, Imperial College of Science, Technology and Medicine, Prince Consort Road, London SW 2BP, UK b 5la Charlton Road, Wantage, Oxon, OX 12 8H. UK Received 10 November 1998: received in revised form 10 August 1999 Abstract The mechanical behaviour of several glass and glass-ceramic composites was studied, with particular interest in axial and off-axis properties. The embrittlement of a cross-ply composite of carbon fibre reinforced borosilicate glass was attributed to the formation of a fibre /matrix interfacial reaction layer during processing. The tensile properties for cross-ply BMAS(BaO, Mgo Al,O3, Sio,) glass-ceramic composites reinforced with silicon carbide fibres exhibited higher matrix cracking stresses in the 90 and 0 plies than the carbon fibre reinforced borosilicate glass composite. These were attributed to the presence of residual stresses that caused the glass-ceramic matrix in compression. Failure strengths in excess of 300 MPa were found for these glass-ceramic composites. Off-axis properties for cross-ply laminates were found to strongly depend on the volume fraction of fibres Composites with lower fibre volume fractions exhibited higher interlaminar strengths than those with higher fibre volume fractions. There also appeared to be a link between inhomogeneous microstructures and a larger variation in mechanical properties o 2000 Elsevier Science S.A. All rights reserved. Keywords: Glass; Glass-ceramics: Matrix composites; Mechanical properties 1. Introduction of faws must be combined with an increase in the resistance to crack growth and a decrease in the sensi- Advanced materials for structural applications at tivity of the strength to the size of those flaws that are high temperatures have been largely restricted to ad- present. These criteria have led to the development of vanced metallic alloys. In high temperature applica- ceramic composites, where two phases are combined tions, a sophisticated breed of Ni-, Co- and Fe-base with weak inter-linking bonds for crack deflection. Fi alloys has evolved both in terms of microstructural and bres of many different shapes, sizes and geometries can structural design. Nevertheless, the high temperature be embedded in a ceramic matrix. This paper is con- applications of such alloys are still restricted by the cerned with long fibre composites that contain rela melting points of the metals. For continued develop- tively large volume fractions of continuous aligned ment, ceramics offer one of the few avenues to a fibres in unidirectional composites or laminae, the lam significant increase in the service temperatures. Their inae being stacked together and consolidated to multi low density, low chemical reactivity and high hardness directional laminates. The technology is most advanced offer additional potential for extending performance for glass and glass-ceramic matrix systems, such as limits beyond those offered by metallic materials [1]. borosilicate lithium-alumina-silicate (Las),magne The reliability of structural ceramics in engineering sium-alumina-silicate(MAS)and barium osumilite applications is a major problem. Improving the proper- (BMAS). A slurry impregnation route is employed to ties of these brittle ceramic materials is achieved by flaw incorporate the fibres into the unconsolidated matrix, control and toughening [2]. A reduction in the incidence followed by hot pressing for consolidation. Most devel- opment work has been carried out using multifilament author. Tel /fax: +44-171-5946750 tary tows of small diameter(7-15 um) SiC or C fibres kchoy@ic ac uk(K.-L Choy) [3] 0921-5093/00/S- see front matter c 2000 Elsevier Science S.A. All rights reserved. PI:s0921-5093099)00572-9
Materials Science and Engineering A278 (2000) 187–194 The mechanical behaviour of glass and glass–ceramic matrix composites K.-L. Choy a,*, P. Duplock a , P.S. Rogers a , J. Churchman-Davies b , M.T. Pirzada a a Department of Materials, Imperial College of Science, Technology and Medicine, Prince Consort Road, London SW7 2BP, UK b 51a Charlton Road, Wantage, Oxon, OX 12 8HJ, UK Received 10 November 1998; received in revised form 10 August 1999 Abstract The mechanical behaviour of several glass and glass–ceramic composites was studied, with particular interest in axial and off-axis properties. The embrittlement of a cross-ply composite of carbon fibre reinforced borosilicate glass was attributed to the formation of a fibre/matrix interfacial reaction layer during processing. The tensile properties for cross-ply BMAS (BaO, MgO, A12O3, SiO2) glass-ceramic composites reinforced with silicon carbide fibres exhibited higher matrix cracking stresses in the 90 and 0° plies than the carbon fibre reinforced borosilicate glass composite. These were attributed to the presence of residual stresses that caused the glass–ceramic matrix in compression. Failure strengths in excess of 300 MPa were found for these glass–ceramic composites. Off-axis properties for cross-ply laminates were found to strongly depend on the volume fraction of fibres. Composites with lower fibre volume fractions exhibited higher interlaminar strengths than those with higher fibre volume fractions. There also appeared to be a link between inhomogeneous microstructures and a larger variation in mechanical properties. © 2000 Elsevier Science S.A. All rights reserved. Keywords: Glass; Glass–ceramics; Matrix composites; Mechanical properties www.elsevier.com/locate/msea 1. Introduction Advanced materials for structural applications at high temperatures have been largely restricted to advanced metallic alloys. In high temperature applications, a sophisticated breed of Ni-, Co- and Fe-base alloys has evolved both in terms of microstructural and structural design. Nevertheless, the high temperature applications of such alloys are still restricted by the melting points of the metals. For continued development, ceramics offer one of the few avenues to a significant increase in the service temperatures. Their low density, low chemical reactivity and high hardness offer additional potential for extending performance limits beyond those offered by metallic materials [1]. The reliability of structural ceramics in engineering applications is a major problem. Improving the properties of these brittle ceramic materials is achieved by flaw control and toughening [2]. A reduction in the incidence of flaws must be combined with an increase in the resistance to crack growth and a decrease in the sensitivity of the strength to the size of those flaws that are present. These criteria have led to the development of ceramic composites, where two phases are combined with weak inter-linking bonds for crack deflection. Fibres of many different shapes, sizes and geometries can be embedded in a ceramic matrix. This paper is concerned with long fibre composites that contain relatively large volume fractions of continuous aligned fibres in unidirectional composites or laminae, the laminae being stacked together and consolidated to multidirectional laminates. The technology is most advanced for glass and glass–ceramic matrix systems, such as borosilicate lithium–alumina–silicate (LAS), magnesium–alumina–silicate (MAS) and barium osumillite (BMAS). A slurry impregnation route is employed to incorporate the fibres into the unconsolidated matrix, followed by hot pressing for consolidation. Most development work has been carried out using multifilamentary tows of small diameter (7–15 mm) SiC or C fibres [3]. * Corresponding author. Tel./fax: +44-171-5946750. E-mail address: k.choy@ic.ac.uk (K.-L. Choy) 0921-5093/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved. PII: S0921-5093(99)00572-9
K.L. Choy et al./ Materials Science and Engineering 4278 (2000)187-194 bide 'Nicalon'and'Tyranno' fibres were used to rein Properties of fibres [1] force glass and glass-ceramic matrices. The properties Fibre Tensile strength of these fibres are given in table 1 (GPa (GPa Carbon‘HM Silicon carbide 000 2. 2. Composites Silicon carbide glass-ceramic matrix composites were investigated for their mechanical roperties. These composites along with fibre/matrix compositions, fibre diameters, lamina thickness, Hot pressed glass and glass-ceramic composites are overall stacking sequenced and plate geometries are he highest strength ceramic matrix composites(CMCs) summarised in Table 2. The glass matrix composites available. The properties of a CMc depend on vari- both had a matrix of borosilicate glass though the ables that include, fibre/matrix strength and moduli, embedded fibres were carbon and silicon carbide in a fibre/matrix bond strength, fibre volume fraction and cross-ply and unidirectional lay-up respectively. The fibre diameter. The thermal coefficient of expansion two glass composites both had a BMAS matrix with between the fibre and matrix can affect the composite silicon carbide fibres in a cross-ply stacking sequence strength, as well as the porosity within the matrix For these two later composites only the volume Accurate mechanical data is essential for designers con- fraction of fibres varied. Samples C and d were templating the use of such ceramic composites in struc- termed "fibre-rich' and 'fibre-poor'respectively. These tural applicatio materials represented deviations from the usual fibre This paper reports on an investigation conducted to distribution produced in such composites determine the mechanical properties of a range of glass and glass-ceramic matrix composites. The tensile prop- erties. stress-strain behaviour. interlaminar shear and 23. Fabrication flexural properties have been investigated; the results of which are considered in the light of the microstructures all the composites were prepared by slurry present. Thus far, failure mechanisms evaluated by impregnation and hot pressing technique. The process other researchers have mainly considered axial proper- for mposite starts by y winding ng slurry ties of a unidirectional long fibre composite, which are impregnated tow onto the mandrel to form stacked in a cross-ply arrangement(0/90). Furthermore, monolayer tapes. The slurry consists of water, a resin his paper will also briefly examine some of the off-axis binder(Pva with 40% PC)[4] and glass powder properties of the composites. The anisotropic nature of These tapes are cut up to make plies which are then these composites with respect to interlaminar/transverse stacked and densified to form the final composite in a strengths is also discussed hot pressing operation. For the borosilicate glass matrix, the composite was hot-pressed at temperature of about 1000oC. In the case of the 2. Experimenta glass-ceramic matrices, to achieve 100% barium osumilite, the composites were ' ceramed or matrices 2.1. Fibres crystallised at temperatures of 1200-13000C. This process prevented the formation of undesirable celsian Carbon'Hercules Magnamite'(HM) and silicon car- and hexcelsian phases in the matrix [4] Table 2 Composites provided for mechanical testing Sample Fibre Matrix Fibre diameter Lamina thickness Stacking Plate geometry Carbon Hercules Borosilicate 275 100×100×2.2 Magnamite Silicon carbide "Nicalon Borosilicate 100×100×6.3 C· fibre Silicon carbide arium osumilite l50×150×2.4 Tyranno D'fbre.rich Silicon carbide 210×210×1.9 (BMAS)
188 K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 Table 1 Properties of fibres [1] Fibre Young’s modulus Tensile strength (GPa) (GPa) Carbon ‘HM’ 2.7 350 Silicon carbide 2.4 190 ‘Nicalon’ Silicon carbide 200 2.7 ‘Tyranno’ bide ‘Nicalon’ and ‘Tyranno’ fibres were used to reinforce glass and glass–ceramic matrices. The properties of these fibres are given in Table 1. 2.2. Composites Two glass and two glass–ceramic matrix composites were investigated for their mechanical properties. These composites along with fibre/matrix compositions, fibre diameters, lamina thickness, overall stacking sequenced and plate geometries are summarised in Table 2. The glass matrix composites both had a matrix of borosilicate glass though the embedded fibres were carbon and silicon carbide in a cross-ply and unidirectional lay-up respectively. The two glass composites both had a BMAS matrix with silicon carbide fibres in a cross-ply stacking sequence. For these two later composites only the volume fraction of fibres varied. Samples C and D were termed ‘fibre-rich’ and ‘fibre-poor’ respectively. These materials represented deviations from the usual fibre distribution produced in such composites. 2.3. Fabrication All the composites were prepared by slurry impregnation and hot pressing technique. The process for making the composite starts by winding slurry impregnated tow onto the mandrel to form monolayer tapes. The slurry consists of water, a resin binder (PVA with 40% PC) [4] and glass powder. These tapes are cut up to make plies which are then stacked and densified to form the final composite in a hot pressing operation. For the borosilicate glass matrix, the composite was hot-pressed at a temperature of about 1000°C. In the case of the glass–ceramic matrices, to achieve 100% barium osumilite, the composites were ‘ceramed’ or matrices crystallised at temperatures of 1200–1300°C. This process prevented the formation of undesirable celsian and hexcelsian phases in the matrix [4]. Hot pressed glass and glass–ceramic composites are the highest strength ceramic matrix composites (CMCs) available. The properties of a CMC depend on variables that include, fibre/matrix strength and moduli, fibre/matrix bond strength, fibre volume fraction and fibre diameter. The thermal coefficient of expansion between the fibre and matrix can affect the composite strength, as well as the porosity within the matrix. Accurate mechanical data is essential for designers contemplating the use of such ceramic composites in structural applications. This paper reports on an investigation conducted to determine the mechanical properties of a range of glass and glass–ceramic matrix composites. The tensile properties, stress–strain behaviour, interlaminar shear and flexural properties have been investigated; the results of which are considered in the light of the microstructures present. Thus far, failure mechanisms evaluated by other researchers have mainly considered axial properties of a unidirectional long fibre composite, which are stacked in a cross-ply arrangement (0/90). Furthermore, this paper will also briefly examine some of the off-axis properties of the composites. The anisotropic nature of these composites with respect to interlaminar/transverse strengths is also discussed. 2. Experimental 2.1. Fibres Carbon ‘Hercules Magnamite’ (HM) and silicon carTable 2 Composites provided for mechanical testing Sample Fibre Matrix Fibre diameter Plate geometry Lamina thickness Stacking (mm) (mm) sequence (mm) A 7 100 Carbon ‘Hercules Borosilicate 275 (0/90)2s ×100×2.2 Magnamite’ B Silicon carbide ‘Nicalon’ Borosilicate 7–10 – (0) 100×100×6.3 Silicon carbide 8 300 (0/90) C ‘fibre-poor’ Barium osumilite 2s 150×150×2.4 ‘Tyranno’ (BMAS) D ‘fibre-rich’ Silicon carbide Barium osumilite 210 8 160 (0/90)3s ×210×1.9 ‘Tyranno’ (BMAS)
K.L. Choy et al./ Materials Science and Engineering 4278 (2000)187-194 this type of end tab was not a suitable representative of Specimen dimensions as recommended by ASTM, BSI and Rae the composite. Subsequently, aluminium end tabs were d(mm) b(mm) /(mm) investigated. Using sample D, a specimen of 25 mm by 100 mm was cut(as recommended by ASTM). Tapered ASTM aluminium end tabs were produced(about 4 mm thick), d+1 with a tab length of approximately 30 mm, leaving a RAE Conclusion sample gauge length of 40 mm. To prevent slippage d+10 form the testing machine, 'Heavy Duty Araldite' was Minimum S for interlaminar shear determination testing rig. used to bond the aluminium tabs. The results proved positive, failure of the specimen occurred in the gauge 2. 4. Image analysis section, without any end tab slippage. Alur minium e tabs were implemented for the final testing The fabricated composites were sectioned, ground ASTM [8], and Royal Aerospace Establishment [91 and polished for image analysis. Both the volume frac- tandards were studied with regards to determining tion of fibres and porosity of the as-fabricated com suitable tensile test dimensions. A large specimen width posites were determined using image analysis coupled was chosen to avoid edge effects and cut edges were with optical microscopy. The errors inherent in this ground down to avoid frictional effects. Testing was conducted on an instron model 4206 at a strain rate of include representativeness of the sample, quality of 2x 10-4 analysis method are well established [5, 6] and sources sample preparation, operator bias and instrument er rors. The main errors found to be incurred during these 2. 6. Interlaminar shear properties particular tests were setting of detector thresholds for feature discrimination and the inhomogeneous mi One important deficiency which affects the applica tion of most fibrous ceramic composites is interlaminar rostructure of the composite materials. At the lowest shear strength(ILS). Laminated and 2-d prefom-rein magnification, such features were readily distinguish ble but one had to compromise between the available forced CMCs are more susceptible to failure in the time for such measurements with respect to the desired matrix-rich interlaminar regions. Such interlaminar fail- ure, or delamination may lead to loss of stiffness and accuracy. possible structural failure. An understanding of failure The outcome of the analysis yielded quantitative mechanisms leading to ILS is therefore crucial information on the volume fraction of fibres V. The The short beam shear test method is found to be data were averaged and presented with their 95% confi- simple and relatively inexpensive for determining shear dence limits. Furthermore, the porosity was also esti- properties. The shear stress distribution along the thick mated but was limited to being determined in the 0 ness of the specimen is a parabolic function, which plies where fibres are observed end on(performed at reaches a maximum at the neutral axis and zero at the higher magnification). This analysis was not performed upper/lower surfaces. However, due to non-constant on the plies where the fibres were transversely aligned bending moments along the shear plane, the maximum due to inaccuracies introduced by fibre tearing mecha shear stress does not occur on the neutral axis [10. A nisms during the sample preparation. Thus, the values more appropriate testing procedure would make use of for porosity are restricted to the bulk of the 0 plies and the thin-walled tube torsion test, thought to exhibit hence, do not represent the matrix rich regions existing pure shear [11]. However, the short beam shear test was between the laminae chosen due to its overriding simplicity. Short beam shear testing in three-point bending re- quires a span between the outer loading points which is 2.5. Tensile testing small enough to induce the shear failure and prevent any tensile failure in the test specimen. The maximum Problems with implementing tensile testing for CMCs shear stress is given as t=3/4(P/bd) and the maxi- are well documented [l], and attempts have been made mum tensile/compressive stresses occurring at the sur o perfect techniques, including elaborate specimen faces for rectangular bar is given by, preparation involving cast integrated epoxy end tabs c=3/2(PS/b]2). Thus, the ratio of tensile(or com 7]. It has been recommended that procedures used for pressive) stress to shear stress yields a parameter, lymer matrix composites be followed [], as well as namely 2S/d. As a rough working rule for laminated established standards [8, 9]. Glass reinforced plastic and CMCs, tensile or compressive failure will occur at aluminium tabs were both investigated as potential end S/d> 20 and shear at S/d<5 [12]. Suitable specimen tabs dimensions were chosen in accordance to ASTM [13], reinforced plastic end tabs was observed. Therefore, Establishment [15] as shown in Table val Aerospace Failure of the specimens in the region of the glass British Standards Institution [14] and R
K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 189 Table 3 Specimen dimensions as recommended by ASTM, BSI and RAE Source d (mm) b (mm) l (mm) S (mm) ASTM 6 – 4 – d d BSI 3 10 20–25 5d+1 RAE 2 5d 5d+10 5d Fixed 7 10 5d+10 da Conclusion a Minimum S for interlaminar shear determination testing rig. this type of end tab was not a suitable representative of the composite. Subsequently, aluminium end tabs were investigated. Using sample D, a specimen of 25 mm by 100 mm was cut (as recommended by ASTM). Tapered aluminium end tabs were produced (about 4 mm thick), with a tab length of approximately 30 mm, leaving a sample gauge length of 40 mm. To prevent slippage form the testing machine, ‘Heavy Duty Araldite’ was used to bond the aluminium tabs. The results proved positive, failure of the specimen occurred in the gauge section, without any end tab slippage. Aluminium end tabs were implemented for the final testing. ASTM [8], and Royal Aerospace Establishment [9] standards were studied with regards to determining suitable tensile test dimensions. A large specimen width was chosen to avoid edge effects and cut edges were ground down to avoid frictional effects. Testing was conducted on an Instron model 4206 at a strain rate of 2×10−4 s−1 . 2.6. Interlaminar shear properties One important deficiency which affects the application of most fibrous ceramic composites is interlaminar shear strength (ILS). Laminated and 2-d prefom-reinforced CMCs are more susceptible to failure in the matrix-rich interlaminar regions. Such interlaminar failure, or delamination may lead to loss of stiffness and possible structural failure. An understanding of failure mechanisms leading to ILS is therefore crucial. The short beam shear test method is found to be very simple and relatively inexpensive for determining shear properties. The shear stress distribution along the thickness of the specimen is a parabolic function, which reaches a maximum at the neutral axis and zero at the upper/lower surfaces. However, due to non-constant bending moments along the shear plane, the maximum shear stress does not occur on the neutral axis [10]. A more appropriate testing procedure would make use of the thin-walled tube torsion test, thought to exhibit pure shear [11]. However, the short beam shear test was chosen due to its overriding simplicity. Short beam shear testing in three-point bending requires a span between the outer loading points which is small enough to induce the shear failure and prevent any tensile failure in the test specimen. The maximum shear stress is given as tc=3/4{P/bd} and the maximum tensile/compressive stresses occurring at the surfaces for a rectangular bar is given by, sc=3/2{PS/bd2 }. Thus, the ratio of tensile (or compressive) stress to shear stress yields a parameter, namely 2S/d. As a rough working rule for laminated CMCs, tensile or compressive failure will occur at S/d\20 and shear at S/dB5 [12]. Suitable specimen dimensions were chosen in accordance to ASTM [13], British Standards Institution [14] and Royal Aerospace Establishment [15] as shown in Table 3. 2.4. Image analysis The fabricated composites were sectioned, ground and polished for image analysis. Both the volume fraction of fibres and porosity of the as-fabricated composites were determined using image analysis coupled with optical microscopy. The errors inherent in this analysis method are well established [5,6] and sources include representativeness of the sample, quality of sample preparation, operator bias and instrument errors. The main errors found to be incurred during these particular tests were setting of detector thresholds for feature discrimination and the inhomogeneous microstructure of the composite materials. At the lowest magnification, such features were readily distinguishable but one had to compromise between the available time for such measurements with respect to the desired accuracy. The outcome of the analysis yielded quantitative information on the volume fraction of fibres Vf . The data were averaged and presented with their 95% confi- dence limits. Furthermore, the porosity was also estimated but was limited to being determined in the 0° plies where fibres are observed end on (performed at higher magnification). This analysis was not performed on the plies where the fibres were transversely aligned due to inaccuracies introduced by fibre tearing mechanisms during the sample preparation. Thus, the values for porosity are restricted to the bulk of the 0° plies and hence, do not represent the matrix rich regions existing between the laminae. 2.5. Tensile testing Problems with implementing tensile testing for CMCs are well documented [1], and attempts have been made to perfect techniques, including elaborate specimen preparation involving cast integrated epoxy end tabs [7]. It has been recommended that procedures used for polymer matrix composites be followed [1], as well as established standards [8,9]. Glass reinforced plastic and aluminium tabs were both investigated as potential end tabs. Failure of the specimens in the region of the glass reinforced plastic end tabs was observed. Therefore
K.L. Choy et al./ Materials Science and Engineering 4278 (2000)187-194 钟:计“家 espii Fig. 1 SEM micrographs for composite samples A, B, C, and D (Table 2). Magnification x 150 for all micrographs (a) Sample A,(b) Sample 2.7. Flexural properties 3.2. Tensile properties Flexural failure is expected to occur for a CMC Measurements of the strain width and thickness of beam, when 2S/d> cu/te Using previous values, a the specimen were taken at several points, and the suitable span was determined. Three point bending was minimum value of cross-sectional area was used in onducted using the same procedure as ILs testing on calculations. Stress values and Youngs moduli were samples A,C and D and by application of the equa- calculated using: o=P/bd, and E=(AP/An(/bd) ion, E=Sm/4bd, the results were obtained where m is the initial linear portion of the load deflection curve 3. Results and discussion Table 4 Summary of microstructural observation Fig. I shows the scanning electron micrographs of Sami Stacking Comments the composites: (a) Sample A and(b) Sample B(c) sequence Sample C and (d) Sample D. All micrographs shown are at magnification x 150. a summary of the mi- A (0/90) Uneven fibre distribution in plies/ ma- crostructural observations for the glass/glass-ceramic trix rich regions between plies with matrix composites is presented in Table 3. Table 4 Porosity on a fine and large scale ob- summarises the microstructural observations for the glass/glass-ceramic matrix composites. (0/90) Fibre-poor n fibre distribution in plies/matrix rich regions between 3. 1. Image analysis plies with significant porosity Fibre-rich, tightly packed fibres in plies with some bundles/no matrix The fibre volume fraction and porosity for com- posites are summarised in table 5
190 K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 Fig. 1. SEM micrographs for composite samples A, B, C, and D (Table 2). Magnification ×150 for all micrographs. (a) Sample A, (b) Sample B, (c) Sample C, (d) Sample D. 2.7. Flexural properties Flexural failure is expected to occur for a CMC beam, when 2S/d\scu/tc. Using previous values, a suitable span was determined. Three point bending was conducted using the same procedure as ILS testing on samples A, C and D and by application of the equation, Ec=S3 m/4bd3 , the results were obtained where m is the initial linear portion of the load deflection curve. 3. Results and discussion Fig. 1 shows the scanning electron micrographs of the composites: (a) Sample A and (b) Sample B (c) Sample C and (d) Sample D. All micrographs shown are at magnification ×150. A summary of the microstructural observations for the glass/glass–ceramic matrix composites is presented in Table 3. Table 4 summarises the microstructural observations for the glass/glass–ceramic matrix composites. 3.1. Image analysis The fibre volume fraction and porosity for composites are summarised in Table 5. 3.2. Tensile properties Measurements of the strain, width and thickness of the specimen were taken at several points, and the minimum value of cross-sectional area was used in calculations. Stress values and Young’s modulli were calculated using: s=P/bd, and E=(DP/Dl)(l/bd), Table 4 Summary of microstructural observation Sample Comments Stacking sequence A Uneven fibre distribution in plies (0/90) /ma- 2s trix rich regions between plies with significant porosity. B Porosity on a fine and large scale ob- (0) served. C ‘Fibre-poor’, uneven fibre distribution (0/90)2s in plies/matrix rich regions between plies with significant porosity. D ‘Fibre-rich’, tightly packed fibres in (0/90)3s plies with some bundles/no matrix rich regions
K.L. Choy et al./ Materials Science and Engineering 4278 (2000)187-194 Table 5 Fibre volume fraction and pore Sample Fibre volume fraction, V(%) Porosity 300 ( magnification×100 250 35+1 5.55+0.92 200 The number of fields for fibre volume fraction and porosity were 115+0.3 45+4 0.9+0.85 where a is tensile strength, I is gauge length of exten- someter, P is load on sample and d is sample thickness The tensile results obtained are tabulated in table 6 0.2 0.6 and a typical stress-strain curve is shown in Fig. 2. STRAIN The stress-strain curves for tensile testing exhibit wo proportional limits, corresponding to initiation of Otu - transverse cracking in 90 plies transverse cracking in the 90 plies and matrix microc Omu=matrix microcracking in 0 plies stresses in the 0 plies. A lower transverse stress is expected for sample A, Fig. 2. Stress-strain curve observed of sample DProportional limits which has a matrix as opposed to samples C and for otu, and omu are shown by arrows. D whose BMAs matrix is a glass-ceramic having higher strengths. Sample C exhibits small variation in results whereas d has a large variation that makes the composite failure stress, cu. The microcracking comparison between the two glass-ceramic sample behaviour of the (0/90)3s, CAS composite with 45% difficult. A similar study conducted on a CAS matrix silicon carbide fibres [16], was found to be the highest posite reinforced with Nicalon fibres in a(0/90) literature value for cross-ply composites, with matrix figuration (V=45%)observed non-linearity for microcracking in the 0o plies and final composite failure stress-strain curves at 55 mPa which is much lower at 175 and 284 MPa respectively, indicating a much broader stress range between the two deformation than the observed value for our glass-ceramic mechanisms than for our composites. Narrow stress composites. ranges betw een microcracking The next deviation on the stress-strain curve, mark- hence, high stresses for the second proportional limit, ing the onset of matrix microcracking in the 0 plies is can be expected if the coefficients of thermal expansion again low. This was expected for the borosilicate sam-(CTE) of the matrix is lower than that of the fibre ple. In light of image analysis work conducted, the resulting in the compression of the matrix and subse- lower value of 210MPa for sample C(cf. with D)is quent suppression of matrix microcracking expected. We would expect composite containing only example, the CTE of silicon carbide(Tyranno)is 4.8x 33+ 2% of fibres (cf. with sample D which has Vr 10-6 K-I and the borosilicate glass matrix is 3.5 x 45+ 6%)to have a lower microcracking stress as dis- 10-6 K[3]. However, little is known about the cussed earlier where the Averston Cooper and Kelley residual stress distribution in this particular composite, (ACK) model predicted tresses for crack but this explanation could well explain the observed ing with increasing fibre volume fraction. The value of phenomena 261+8 MPa achieved for sample D is remarkably high The failure of the end tabs meant tensile strength for matrix microcracking stress and was very close to results for sample d only were obtained. As indicated Table 6 Measured tensile properties of samples A to D Gt (mPa Lu(%0) σmu(MPa) (%) (MPa) E(GPa) 0.06 his sample was unsuitably thick for tensile testing 97+1 0.08+0.04 0.04 l01+13 88±27 0.08+003 26l+8 0.38+0.3 290.320 100+1
K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 191 Table 5 Fibre volume fraction and porosity for compositea Fibre volume fraction, Vf Sample (%) Porosity (%) (magnification ×100) (magnification ×500) A 0.85 3591 90.03 B 5.55 4692 90.92 C 3392 1.1590.31 D 0.9 4594 90.85 a The number of fields for fibre volume fraction and porosity were 10 and 20, respectively. Fig. 2. Stress–strain curve observed of sample D. Proportional limits for stu, and smu are shown by arrows. where s is tensile strength, l is gauge length of extensometer, P is load on sample and d is sample thickness. The tensile results obtained are tabulated in Table 6 and a typical stress–strain curve is shown in Fig. 2. The stress–strain curves for tensile testing exhibit two proportional limits, corresponding to initiation of transverse cracking in the 90° plies and matrix microcracking at higher stresses in the 0o plies. A lower transverse cracking stress is expected for sample A, which has a glass/matrix as opposed to samples C and D whose BMAS matrix is a glass–ceramic having higher strengths. Sample C exhibits small variation in results whereas D has a large variation that makes comparison between the two glass–ceramic samples difficult. A similar study conducted on a CAS matrix composite reinforced with Nicalon fibres in a (0/90)3S configuration (Vf=45%) observed non-linearity for stress–strain curves at 55 MPa which is much lower than the observed value for our glass–ceramic composites. The next deviation on the stress–strain curve, marking the onset of matrix microcracking in the 0° plies is again low. This was expected for the borosilicate sample. In light of image analysis work conducted, the lower value of 210MPa for sample C (cf. with D) is expected. We would expect composite containing only 3392% of fibres (cf. with sample D which has Vf= 4596%) to have a lower microcracking stress as discussed earlier where the Averston Cooper and Kelley (ACK) model predicted increases in stresses for cracking with increasing fibre volume fraction. The value of 26198 MPa achieved for sample D is remarkably high for matrix microcracking stress and was very close to the composite failure stress, scu. The microcracking behaviour of the (0/90)3s, CAS composite with 45% silicon carbide fibres [16], was found to be the highest literature value for cross-ply composites, with matrix microcracking in the 0° plies and final composite failure at 175 and 284 MPa respectively, indicating a much broader stress range between the two deformation mechanisms than for our composites. Narrow stress ranges between microcracking and final failure, and hence, high stresses for the second proportional limit, can be expected if the coefficients of thermal expansion (CTE) of the matrix is lower than that of the fibre, resulting in the compression of the matrix and subsequent suppression of matrix microcracking [3]. For example, the CTE of silicon carbide (Tyranno) is 4.8× 10−6 K−1 and the borosilicate glass matrix is 3.5× 10−6 K−1 [3]. However, little is known about the residual stress distribution in this particular composite, but this explanation could well explain the observed phenomena. The failure of the end tabs meant tensile strength results for sample D only were obtained. As indicated, Table 6 Measured tensile properties of samples A to D stu (MPa) otu (%) smu (MPa) omu (%) scu (MPa) Ec (GPa) A 50.6 0.06 152 0.2 – 69 B This sample was unsuitably thick for tensile testing C 9791 0.0890.04 210 0.044 – 101913 D 0.08 88927 261 90.03 98 0.3890.3 290, 320 10091
K.L. Choy et al./ Materials Science and Engineering 4278 (2000)187-194 the values appear to be particularly high if compared evidence of a reaction layer could be observed, though with similar glass-ceramic systems. For our matrix a more in depth investigation possibly including TEM rich, sample C, we expect a lower failure stress. This is would be beneficial in helping to explain the observed due to it being a composite with an equivalent volume fracture behaviour fraction of fibres and where these carbon theoretically have the same strength as the silicon carbide as shown 3.3. Interlaminar shear properties in Table 1. In the case of sample A, we would expect a value of aeu comparable to C. This would indicate a The results obtained at a cross-head rate of 1.3 mm strength well over 200 MPa Discrepancies between the min are summarised in Table 7. A typical load-dis- predicted results and the final outcome may well be due placement graph obtained during testing is shown in to interface degradation during processing, oxidation of Fig 3 carbon fibres at elevated temperatures and the brittle Interlaminar shear strength(ILS)tests conducted at a reaction layer formed between fibres and borosilicate span to depth ratio of 7 produced pure shear deforma matrix may well have produced what appears to be tion in all samples, making test results valid for calcula reduced mechanical properties. tion Damage tended to concentrate in the matrix Young's modulus values are in good agreement with regions only, and for cross-ply samples this meant similar glass-ceramic composites with a cross-ply lay shearing between lamina. The formation of a crack up[17] though sample C would be expected to have a corresponded to the rounded peak on the load maxim lower modulus value due to is lower volume fraction After shear crack formation the load drop is not The most interesting value is that obtained for sample catastrophic and a diminished load is maintained dur A(69 GPa)which is extremely low, especially when ing further deformation, corresponding to a substantial ROM predictions using data obtained in Table I where energy absorption Carbon HM fibres have a modulus almost 50% A similar study on a Las glass-ceramic reinforced greater than for silicon carbide, are considered. These with silicon carbide [18] determined that this load bear bility of the composite after shearing can be interfacial reactions must have taken place during pro- attributed to the combined effect of sliding resistance of cessing, resulting in fibre degradation and embrittle ment of the composite. Some initial SEM beams produced after failure characterisation of the interface was attempted but, no A study of the micrographs of these samples pro- vided an insight into the variation in results between Table 7 samples C and D, i.e. absence of matrix between the The interlaminar shear strength results plies which would result in reduced ILS strength. Mi crographs of C also indicated significant porosity be- ample TaPP(MPa) tween the plies in these matrix-rich regions. If these A 13.3+0.2 defects, that can reduce strengths of composites, could B 24.6+0.5 be reduced by altering the processing conditions, it may 33.9±43 be possible to achieve even higher ILs values that D 15.1+0.7 would results in improved off-axis properties for these composites The values obtained and the small scatter in results 400 were very encouraging and contrary to the data repro- 35 ducibility problems encountered, as stated in the intro- duction, by previous workers. ILS strengths for a 250 es cross-ply silicon carbide (Vr=35%) reinforced CAS found to lie in the range 35-45 Sample C having a similar volume fraction of fibres was 6 9 found to have a comparable iLs value of 33.9 +4.3 MPa indicating good agreement in our results with 100 published data. Sample C exhibited the highest value for all four samples. The other two cross-ply laminates (A and D)exhibited similar shear strength values, though the reasons for their diminished results (cf with C)can be attributed to different reasons. Failure, as stated, occurred in the matrix rich regions of the Fig. 3. Load-displacement curve for short beam shear test(Sample matrix meant shear failure was observed on application
192 K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 the values appear to be particularly high if compared with similar glass–ceramic systems. For our matrix rich, sample C, we expect a lower failure stress. This is due to it being a composite with an equivalent volume fraction of fibres and where these carbon theoretically have the same strength as the silicon carbide as shown in Table 1. In the case of sample A, we would expect a value of scu comparable to C. This would indicate a strength well over 200 MPa. Discrepancies between the predicted results and the final outcome may well be due to interface degradation during processing, oxidation of carbon fibres at elevated temperatures and the brittle reaction layer formed between fibres and borosilicate matrix may well have produced what appears to be reduced mechanical properties. Young’s modulus values are in good agreement with similar glass–ceramic composites with a cross-ply layup [17], though sample C would be expected to have a lower modulus value due to is lower volume fraction. The most interesting value is that obtained for sample A (69 GPa) which is extremely low, especially when ROM predictions using data obtained in Table 1 where ‘Carbon HM’ fibres have a modulus almost 50% greater than for silicon carbide, are considered. These results enforce the idea that significant fibre/matrix interfacial reactions must have taken place during processing, resulting in fibre degradation and embrittlement of the composite. Some initial SEM characterisation of the interface was attempted but, no evidence of a reaction layer could be observed, though a more in depth investigation possibly including TEM would be beneficial in helping to explain the observed fracture behaviour. 3.3. Interlaminar shear properties The results obtained at a cross-head rate of 1.3 mm min−1 are summarised in Table 7. A typical load–displacement graph obtained during testing is shown in Fig. 3. Interlaminar shear strength (ILS) tests conducted at a span to depth ratio of 7 produced pure shear deformation in all samples, making test results valid for calculation. Damage tended to concentrate in the matrix regions only, and for cross-ply samples this meant shearing between lamina. The formation of a crack corresponded to the rounded peak on the load maxim. After shear crack formation the load drop is not catastrophic and a diminished load is maintained during further deformation, corresponding to a substantial energy absorption. A similar study on a LAS glass–ceramic reinforced with silicon carbide [18] determined that this load bearing capability of the composite after shearing can be attributed to the combined effect of sliding resistance of the shear cracks and load bearing from the two half beams produced after failure. A study of the micrographs of these samples provided an insight into the variation in results between samples C and D, i.e. absence of matrix between the plies which would result in reduced ILS strength. Micrographs of C also indicated significant porosity between the plies in these matrix-rich regions. If these defects, that can reduce strengths of composites, could be reduced by altering the processing conditions, it may be possible to achieve even higher ILS values that would results in improved off-axis properties for these composites. The values obtained and the small scatter in results were very encouraging and contrary to the data reproducibility problems encountered, as stated in the introduction, by previous workers. ILS strengths for a cross-ply silicon carbide (Vf=35%) reinforced CAS system were found to lie in the range 35–45 MPa [19]. Sample C having a similar volume fraction of fibres was found to have a comparable ILS value of 33.994.3 MPa indicating good agreement in our results with published data. Sample C exhibited the highest value for all four samples. The other two cross-ply laminates (A and D) exhibited similar shear strength values, though the reasons for their diminished results (cf. with C) can be attributed to different reasons. Failure, as stated, occurred in the matrix rich regions of the composite. Samples A with a lower strength borosilicate matrix meant shear failure was observed on application Table 7 The interlaminar shear strength results Sample tc app (MPa) A 13.390.2 B 24.690.5 C 33.994.3 D 15.190.7 Fig. 3. Load–displacement curve for short beam shear test (Sample A)
K.L. Choy et al./ Materials Science and Engineering 4278 (2000 )187-194 4. Conclusions 450 00 Micrographs and image analysis work showed the four samples(A, B, C and D) exhibited a certain degree of inhomogeneity within the plies, where fibre bundling could be contrasted with less tightly packed areas 4009 Cross-ply laminates, and in particular Sample A and C exhibited matrix rich regions in between plies and sig- nificant porosity due to poor consolidation of the ma- 150 trix. Image analysis work determined fibre volume fractions and indicated porosity levels Sample A exhibited low transverse ou and matrix microcracking om due to its borosilicate matrix. Evi- 0 dence of low composite failure stress and Youngs modulus for this sample indicated the possibility of DEFLECTION (mm fibre/matrix reactions during processing resulting in embrittleness. Both glass-ceramic samples exhibited Fig. 4. Load-deflection curve for flexural testing(Sample C) and it as suggested tha the suppression of microcracking in the matrix was due Table 8 to mismatches in the coefficients of thermal expansions Flexural properties of some glass-ceramic composites between fibre/matrix, resulting in the matrix being un- Reference Matrix Fibre Lay-up vr%) acP EcP der compression. Final composite strengths were only obtained for 90)3s34±5530115 sample D, which was found to be in excess of 300 MPa, CAS Nicalon (0/90)s 19 320 and considerably higher than reported literature values on similar glass-ceramic systems. Only ILs could be performed on sample B(due to its unsuitable thick- ness). The short beam shear test was found to be successful in terms of data reproducibility, for the of a lower load. Sample B should necessarily exhibit a determination of the interlaminar shear properties of higher strength than A, even though they have the same these composites. Failure in matrix rich regions was matrix material due to its unidirectional lay-up. The observed and it was found that a layer of matrix results obtained of 24.6 MPa +0.5 must be dependent between plies, present in lower-fibre volume fraction on the significant porosity observed on both micro- composites improved this off-axis property of these graphs and during image analysis work materials. Flexural testing was only found to be suc cessful, when a correct span to depth ratio was chosen. 3. 4. Flexural properties Thus illustrating the poor off-axis composite properties; indicating the promising potential for this silicon car Samples A, C, and D were tested, only sampleC bide reinforced BMAS system, in terms of mechanical failed in a flexural manner required for calculations of properties apparent strength(oupp) and modulus(ecpP). For Sam This study appears to be the first to investigate the ple C, aapp was found to be 678+ 22 MPa and Eapp axial and off-axis properties of glass-ceramic com was 142+9 GPa. A load-displacement curve produced posites with varying volume fractions of fibres. While during testing is shown in Fig. 4 tensile strengths are dictated by the failure strength of Mixed mode failure, with multiple shearing for sam- the fibres, and increasing volume fraction of fibres ples A and d meant that apparent strengths and mod- results in an increase in strength, the opposite appears uli could not be obtained. This highlights the extremely to be the case for interlaminar shear strengths. This off poor off-axis properties for these composites, resulting axis property appears to be dependent on matrix rich in their anisotropic behaviour. Sample C failed in the regions between plies and increased ILS strengths are manner required for calculations. Strengths and moduli favoured by lower-fibre-volume fraction composite for similar glass-ceramic composites are presented in These phenomena must be thoroughly considered dur- Table 8. The value of acpp of 678+ 22 MPa is ex- ing composite design and criteria for the degree of remely high in comparison with those obtained from acceptable anisotropy should be taken into account the literature and indicates that the possibilities fo A homogeneous sample is of vital importance. Inhe these silicon carbide reinforced BMAS composites are mogeneous microstructures lead to large standard devi very promising in terms of mechanical properties. ation in results making design selection very difficult for
K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 193 Fig. 4. Load–deflection curve for flexural testing (Sample C). 4. Conclusions Micrographs and image analysis work showed the four samples (A, B, C and D) exhibited a certain degree of inhomogeneity within the plies, where fibre bundling could be contrasted with less tightly packed areas. Cross-ply laminates, and in particular Sample A and C exhibited matrix rich regions in between plies and significant porosity due to poor consolidation of the matrix. Image analysis work determined fibre volume fractions and indicated porosity levels. Sample A exhibited low transverse stu and matrix microcracking smu due to its borosilicate matrix. Evidence of low composite failure stress and Young’s modulus for this sample indicated the possibility of fibre/matrix reactions during processing resulting in embrittleness. Both glass–ceramic samples exhibited high values for stu and smu, and it was suggested that the suppression of microcracking in the matrix was due to mismatches in the coefficients of thermal expansions between fibre/matrix, resulting in the matrix being under compression. Final composite strengths were only obtained for sample D, which was found to be in excess of 300 MPa, and considerably higher than reported literature values on similar glass–ceramic systems. Only ILS could be performed on sample B (due to its unsuitable thickness). The short beam shear test was found to be successful in terms of data reproducibility, for the determination of the interlaminar shear properties of these composites. Failure in matrix rich regions was observed and it was found that a layer of matrix between plies, present in lower-fibre volume fraction composites improved this off-axis property of these materials. Flexural testing was only found to be successful, when a correct span to depth ratio was chosen. Thus illustrating the poor off-axis composite properties; indicating the promising potential for this silicon carbide reinforced BMAS system, in terms of mechanical properties. This study appears to be the first to investigate the axial and off-axis properties of glass–ceramic composites with varying volume fractions of fibres. While tensile strengths are dictated by the failure strength of the fibres, and increasing volume fraction of fibres results in an increase in strength, the opposite appears to be the case for interlaminar shear strengths. This off axis property appears to be dependent on matrix rich regions between plies and increased ILS strengths are favoured by lower-fibre-volume fraction composite. These phenomena must be thoroughly considered during composite design and criteria for the degree of acceptable anisotropy should be taken into account. A homogeneous sample is of vital importance. Inhomogeneous microstructures lead to large standard deviation in results making design selection very difficult for Table 8 Flexural properties of some glass–ceramic composites Reference Fibre Lay-up Matrix Vf (%) scu app Ec app 20 (0 CAS Nicalon /90)3s 3495 115 530 (0/90) 17 320 – CAS Nicalon 4s 19 of a lower load. Sample B should necessarily exhibit a higher strength than A, even though they have the same matrix material due to its unidirectional lay-up. The results obtained of 24.6 MPa90.5 must be dependent on the significant porosity observed on both micrographs and during image analysis work. 3.4. Flexural properties Samples A, C, and D were tested, only sample C failed in a flexural manner required for calculations of apparent strength (su app) and modulus (Ec app). For Sample C, su app was found to be 678922 MPa and Ec app was 14299 GPa. A load-displacement curve produced during testing is shown in Fig. 4. Mixed mode failure, with multiple shearing for samples A and D meant that apparent strengths and moduli could not be obtained. This highlights the extremely poor off-axis properties for these composites, resulting in their anisotropic behaviour. Sample C failed in the manner required for calculations. Strengths and moduli for similar glass–ceramic composites are presented in Table 8. The value of scu app of 678922 MPa is extremely high in comparison with those obtained from the literature and indicates that the possibilities for these silicon carbide reinforced BMAS composites are very promising in terms of mechanical properties
K.L. Choy et al./ Materials Science and Engineering 4278 (2000 )187-194 engineers. This is especially the case in composites with 9 P.T. Curtis, Method of Test for the Tensile Stre ulus of multidirectional Fibre Reinforced Aerospace Establishment, 1988, pp. 27-29 [0KT. Keward, Fiber Sci. Technol. 5(1972)85. [l C.C. Chiao, R L. Moore, T.T. Chiao, Measurement of shear References properties of fiber composites. Part 1: Evaluation of Test Meth- ds8(3)(1977)16l-169 U D.C. Larsen, S.L. Stuchly, in: K.S. Mazdiyasni (Ed. ) The Me. 12Rw. Davidge, J.J. R. Davies, Ceramic Matrix Fibre Composites chanical Evaluation of Ceramic Fibre Composites, Fibre rein. Mechanical Testing and Performance. McGraw-Hill. New York forced Ceramic Composites: Materials, Processing and [3 Anon, Standard Test Method for Apparent Interlaminar Shear 2AJ. Evans, J. Am. Ceram Soc. 73(1990)187 Strength of Parallel Fiber Composites by Short Beam Method 3]D.C. Phillips, in: R. Warren (Ed ) Long Fibre Reinforced D2344-84. ASTM Standards and Literature References for Com- Ceramics, Ceramic Matrix Composites, Blackie and Son, Glas- posite Materials, 2nd edn, American Society for Testing and gow,1992,pp.167-198 Materials, Philadelphia, PA, 1990, pp. 15-17. 4N. Mason, Protective Glazes for Ceramic Matrix Composites, [14 Anonymous, Short Span Three Point Bend Testing, Advanced Long Project, Department of Materials Science, Imperial Col- Technical Ceramics, British Standards Institution, 1991. pp 8-10CEN/TC184 5G.F. Vander Voort, Image Analysis, Materials Characterisation [15] P.T. Curtis, Method of Test for Interlaminar Shear Strength of [ol]. Metals Handbook, 9th edn, American Society for Metals, aterials Park, OH, 1986, pp. 309-316 pp. 10-ll Report 88,012 [6G.F. Vander Voort, Quantitative Microscopy, Metallography [16 L P. Zawada, L M. Butkus, G.A. Hartman, J. Am. Ceram Soc. Principles and Practice, McGraw-Hill, New York, 1984, p. 410 74(1991)2851 [7S.w. Wang, A. Parvizi-Majidi, J. Mat. Sci. 27(1992)5483. [17 D.S. Beyerle, S M. Spearing, A.G. Evans, J. Am. Ceram Soc. 75 [8 Anonymous, Standard Test Method for Tensile Properties of (1992)3321 Fiber-resin Composite Materials D3039-76, 2nd edn, American [(18S. Jansson, F.A. Jeckie, Acta. Metall. Mater. 40(1992) Society for Testing and Materials, Philadelphia, PA, 1990, pp 9 [19N.JJ. Fang, T.-w. Chou, J. Am. Ceram. Soc. 76(10)(
194 K.-L. Choy et al. / Materials Science and Engineering A278 (2000) 187–194 engineers. This is especially the case in composites with low fibre volume fractions. References [1] D.C. Larsen, S.L. Stuchly, in: K.S. Mazdiyasni (Ed.), The Mechanical Evaluation of Ceramic Fibre Composites, Fibre Reinforced Ceramic Composites: Materials, Processing and Technology, Noyes, Park Ridge, NJ, 1990, p. 182. [2] A.J. Evans, J. Am. Ceram. Soc. 73 (1990) 187. [3] D.C. Phillips, in: R. Warren (Ed.), Long Fibre Reinforced Ceramics, Ceramic Matrix Composites, Blackie and Son, Glasgow, 1992, pp. 167–198. [4] N. Mason, Protective Glazes for Ceramic Matrix Composites, Long Project, Department of Materials Science, Imperial College, 1994. [5] G.F. Vander Voort, Image Analysis, Materials Characterisation [101], Metals Handbook, 9th edn, American Society for Metals, Materials Park, OH, 1986, pp. 309–316. [6] G.F. Vander Voort, Quantitative Microscopy, Metallography. Principles and Practice, McGraw-Hill, New York, 1984, p. 410. [7] S.W. Wang, A. Parvizi-Majidi, J. Mat. Sci. 27 (1992) 5483. [8] Anonymous, Standard Test Method for Tensile Properties of Fiber-resin Composite Materials D3039-76, 2nd edn, American Society for Testing and Materials, Philadelphia, PA, 1990, pp. 26–30. [9] P.T. Curtis, Method of Test for the Tensile Strength and modulus of Multidirectional Fibre Reinforced Plastics, Royal Aerospace Establishment, 1988, pp. 27–29 Technical report 88,012. [10] K.T. Keward, Fiber Sci. Technol. 5 (1972) 85. [11] C.C. Chiao, R.L. Moore, T.T. Chiao, Measurement of shear properties of fiber composites. Part 1: Evaluation of Test Methods 8 (3) (1977) 161–169. [12] R.W. Davidge, J.J.R. Davies, Ceramic Matrix Fibre Composites: Mechanical Testing and Performance, McGraw-Hill, New York, pp. 249–266. [13] Anon, Standard Test Method for Apparent Interlaminar Shear Strength of Parallel Fiber Composites by Short Beam Method, D2344-84, ASTM Standards and Literature References for Composite Materials, 2nd edn, American Society for Testing and Materials, Philadelphia, PA, 1990, pp. 15–17. [14] Anonymous, Short Span Three Point Bend Testing, Advanced Technical Ceramics, British Standards Institution, 1991, pp. 8–10 CEN/TC184. [15] P.T. Curtis, Method of Test for Interlaminar Shear Strength of Fibre Reinforced Plastics, Royal Aerospace Establishment, 1988, pp. 10–11 Report 88,012. [16] L.P. Zawada, L.M. Butkus, G.A. Hartman, J. Am. Ceram. Soc. 74 (1991) 2851. [17] D.S. Beyerle, S.M. Spearing, A.G. Evans, J. Am. Ceram. Soc. 75 (1992) 3321. [18] S. Jansson, F.A. Jeckie, Acta. Metall. Mater. 40 (1992) 2967. [19] N.J.J. Fang, T.-W. Chou, J. Am. Ceram. Soc. 76 (10) (1993) 2539.