《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_Fracture behaviour of mullite fibre reinforced–mullite

Availableonlineatwww.sciencedirect.com DIRECTO COMPOSITES SCIENCE AND TECHNOLOGY Composites Science and Technology 65(2005)325-333 fracture behaviour of mullite fibre reinforced-mullite matrix composites under quasi-static and ballistic impact loading A.R. Boccaccini a,, S. Atiq a, D.N. Boccaccini b, I. Dlouhy, C. Kaya d Department of Materials, Imperial College London, South Kensington Campus, Prince Consort Road, SW7 2BP UK b Department of Materials and Environmental Engineering, University of Modena and Reggio Emilia, 41100 Modena, Italy Institute of Physics of Materials, d wolfson Centre for Materials Processing and Mechanical Engineering, Brunel UniVersity, Uxbridge UB8 3PH, UK Received 1 February 2004: received in revised form 31 July 2004: accepted 21 August 2004 vailable online 23 September 2004 Abstract The fracture behaviour and damage development in mullite fibre reinforced-mullite matrix composites have been investigated using chevron notch technique and ballistic impact tests. Fracture toughness(Kic)values in the range of 1.8-3.3 MPa m were determined using the chevron notched specimen technique. A large variability of Kle data due to the complex(heterogeneous)com- posite microstructure was found. Extensive fibre pull-out occurred during failure which was due to a favourable matrix/fibre inter- facial bond given by NdPOa coating of the fibres. The materials response under ballistic impact loads was studied using a gas gun. The projectiles were glass balls of 10.15 mm in diameter and weighing 1. 4 g. The projectile velocity was in the range 77. 6-207.5 m/s. The remanent load carrying capability of composite samples after ballistic tests was measured to quantify ballistic impact induced microstructural damage. The composites retained some of their load bearing capacity even after penetration of the projectile, since structural damage caused by projectiles remained localised, preventing catastrophic failure. Penetration by the projectile occurred at impact energy of about 4 J for the conditions investigated. Understanding crack propagation and damage development under bal listic impact loads may open new opportunities for the use of these composites in lightweight armour applications. c 2004 Elsevier ltd. all rights reserved. Keywords: A Ceram composites: Mullite matrix; B. Toughness 1. Introduction taking into account also the development of reliable and cost-effective fabrication procedures [1-10]. The development of oxide fibre reinforced-oxide Mullite appears as an ideal oxide matrix for high tem matrix composites is a promising way of achieving light- perature applications due to its high fracture strength at weight, structural materials combining high-temperature elevated temperatures, good thermal shock resistance strength with improved fracture toughness, damage high chemical and thermal stability and high creep tolerance, thermal shock and oxidation resistance [1]. resistance [ll, 12]. In its monolithic form, however, mul- Significant research effort has been devoted to the opti- lite exhibits low fracture toughness [12]. The fracture misation of these ceramic matrix composite systems, toughness and flaw resistance of mullite can be increased by the introduction of high strength continuous ceramic Corresponding author. Tel. +44 207 5946731; fax: +44 20 fibres [4, 5, 8]. Toughening mechanisms like fibre debond E-mail address: a boccaccini@imperial ac uk(A R. Boccaccini). tion are activated in such a composite and contribute Permanent address: PCSIR Labs, P.O. Box 387, Quetta, Pakistan. to a non- linear stress-strain response and thus high -3538/S- see front matter 2004 Elsevier Ltd. All rights reserved. ech.2004.08002

Fracture behaviour of mullite fibre reinforced–mullite matrix composites under quasi-static and ballistic impact loading A.R. Boccaccini a,*, S. Atiq a,1, D.N. Boccaccini b , I. Dlouhy c , C. Kaya d a Department of Materials, Imperial College London, South Kensington Campus, Prince Consort Road, London SW7 2BP, UK b Department of Materials and Environmental Engineering, University of Modena and Reggio Emilia, 41100 Modena, Italy c Institute of Physics of Materials, Czech Academy of Sciences, CZ-61662 Brno, Czech Republic d Wolfson Centre for Materials Processing and Mechanical Engineering, Brunel University, Uxbridge UB8 3PH, UK Received 1 February 2004; received in revised form 31 July 2004; accepted 21 August 2004 Available online 23 September 2004 Abstract The fracture behaviour and damage development in mullite fibre reinforced–mullite matrix composites have been investigated using chevron notch technique and ballistic impact tests. Fracture toughness (KIc) values in the range of 1.8–3.3 MPa m1/2 were determined using the chevron notched specimen technique. A large variability of KIc data due to the complex (heterogeneous) com￾posite microstructure was found. Extensive fibre pull-out occurred during failure, which was due to a favourable matrix/fibre inter￾facial bond given by NdPO4 coating of the fibres. The materials response under ballistic impact loads was studied using a gas gun. The projectiles were glass balls of 10.15 mm in diameter and weighing 1.4 g. The projectile velocity was in the range 77.6–207.5 m/s. The remanent load carrying capability of composite samples after ballistic tests was measured to quantify ballistic impact induced microstructural damage. The composites retained some of their load bearing capacity even after penetration of the projectile, since structural damage caused by projectiles remained localised, preventing catastrophic failure. Penetration by the projectile occurred at impact energy of about 4 J for the conditions investigated. Understanding crack propagation and damage development under bal￾listic impact loads may open new opportunities for the use of these composites in lightweight armour applications.
2004 Elsevier Ltd. All rights reserved. Keywords: A. Ceramic matrix composites; Mullite matrix; B. Toughness 1. Introduction The development of oxide fibre reinforced–oxide matrix composites is a promising way of achieving light￾weight, structural materials combining high-temperature strength with improved fracture toughness, damage tolerance, thermal shock and oxidation resistance [1]. Significant research effort has been devoted to the opti￾misation of these ceramic matrix composite systems, taking into account also the development of reliable and cost-effective fabrication procedures [1–10]. Mullite appears as an ideal oxide matrix for high tem￾perature applications due to its high fracture strength at elevated temperatures, good thermal shock resistance, high chemical and thermal stability and high creep resistance [11,12]. In its monolithic form, however, mul￾lite exhibits low fracture toughness [12]. The fracture toughness and flaw resistance of mullite can be increased by the introduction of high strength continuous ceramic fibres [4,5,8]. Toughening mechanisms like fibre debond￾ing, fibre bridging and pull-out as well as crack deflec￾tion are activated in such a composite and contribute to a non-linear stress–strain response and thus high 0266-3538/$ - see front matter
2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.compscitech.2004.08.002 * Corresponding author. Tel.: +44 207 5946731; fax: +44 207 5843194. E-mail address: a.boccaccini@imperial.ac.uk (A.R. Boccaccini). 1 Permanent address: PCSIR Labs, P.O. Box 387, Quetta, Pakistan. www.elsevier.com/locate/compscitech COMPOSITES SCIENCE AND TECHNOLOGY Composites Science and Technology 65 (2005) 325–333

R Boccaccini et al./ Composites Science and Technology 65(2005 )325-333 energy dissipation before fracture [13]. Several mullite 2. Experimental matrix composites with oxide and non-oxide fibre rein- forcement have been developed in the last 20 years 2. 1. Material 4, 5, 8,9, 12, 14-19]. Most previous investigations have concentrated on the study of the fracture strength of The material investigated was a composite formed by the composites at room and elevated temperatures, with a mullite matrix containing homogeneously distributed less emphasis given to other properties such as thermal ultra-fine (70-350 nm) porosity reinforced with hock resistance or fracture behaviour under dynamic NdPOa-coated mullite woven fibre mats(Nextel 720) (e.g. ballistic impact) loads. Moreover, there has been The details of the material fabrication, which was done only limited work aimed at measuring the fracture by a combination of electrophoretic deposition, pressure toughness of fibre reinforced-mullite matrix composites filtration and pressureless sintering, have been published [5,19 elsewhere [26]. The typical microstructure of the com- In brittle matrices reinforced by ceramic fibres posite is shown in Fig. 1 fibre bridging and fibre pull-out mechanisms cause the toughening [13]. Both these mechanisms in- 2. 2. Fracture toughness determination crease, to some extent, behind the crack tip in the proc. ess zone wake [20, 21]. Thus, the crack growth resistance Samples in the form of rectangular test bars of nom- rises as the crack propagates and leaves the wake In inal dimensions 4.9x 1.9 x 45 mm were used for frac- these materials, it is difficult to define the intrinsic frac- ture toughness determination by means of the CN ture toughness(Klc)as a material parameter due to a ris- specimen technique. Chevron notches with angles of ing crack growth resistance curve [20]. It is not possible 90 were cut using a thin diamond wheel. A 3-point to characterise this group of materials by known meth- bending test(with span of 16 mm) at a constant cross- ods of R-curve determination either, because of major head speed of 0.01 mm/min was employed. Graphs of crack deflection along the fibre matrix interface and load versus time were recorded and the maximum force the resultant delamination instead of single was determined from each trace. The fracture toughness crack propagation. Nevertheless, an exact method of value was calculated from the maximum load (Fmax)and fracture behaviour quantification is needed, mainly for the corresponding minimum value of the geometrical comparison purposes, if a further development of fibre calibration function(Ymin). The calculation of the func reinforced-brittle matrix composites is desired, includ tion y foran bend bars was based on the use of ing the assessment of their possible structural degrada- Bluhm's slice model [27]. Details of the procedure used tion in service. In the work of Ha and Chawla [19] it for calculating Kl have been given elsewhere [28]. The was shown that using the chevron notched (CN) speci- Cn depth ao. necessary for calculating the calibration men technique the fracture toughness of mullite matrix function Ymin, was measured from optical micrograph composites containing mullite fibres with different coat- of fractured specimens. The acoustic emission technique ings could be measured. In our previous investigations, (AE) was used during the test. Traces of cumulative the CN technique has been successfully used to measure number of counts(AE events) were obtained in the same fracture toughness in SiC fibre reinforced-glass matrix time scale as the load vs time plots. This technique al- composites which have been subjected to thermal shock lows for an accurate detection of the microcrack initia and thermal ageing [22-25] tion at the Cn, which occurs when a sharp increase in la he purpose of the present work was to analyse the the number of AE events is observed. Valid measure- acture behaviour of a novel mullite fibre/mullite ma- ments for computing Klc are those in which this increase trix composite, which has been recently developed [26]. of AE events coincides with the maximum load follow- Since some preliminary characterisation of the mechan- ing the linear part of the force versus time trace, as ical properties of this composite already exists, i.e. explained below. The crack propagation in different high-temperature fracture strength and thermal cycling samples was observed by low magnification optical resistance [26]. the focus of the present study was on microscopy crack propagation behaviour under quasi-static condi- tions and on the composite macroscopic response under 2.3. Ballistic impact test and evaluation ballistic impact loading. The suitability of the CN tech nique for determination of the fracture toughness of this Ballistic impact tests were carried out by impacting fibre reinforced-mullite matrix composite was also composite tiles of dimensions 75 mm x 75 mm x 2.65 assessed. The materials response under impact loads mm with projectiles of varying velocities using a labo- was studied for low-velocity ballistic conditions using ratory gas gun. The projectiles were glass balls meas a gas gun. The remanent load carrying capability of uring 10.15 mm in diameter and weighing 1. 4 g. The composite samples after the ballistic test was measured projectile velocity range during this investigation was to quantify ballistic impact induced damage 77.6-207.5 m/s. The samples were mechanically

energy dissipation before fracture [13]. Several mullite matrix composites with oxide and non-oxide fibre rein￾forcement have been developed in the last 20 years [4,5,8,9,12,14–19]. Most previous investigations have concentrated on the study of the fracture strength of the composites at room and elevated temperatures, with less emphasis given to other properties such as thermal shock resistance or fracture behaviour under dynamic (e.g. ballistic impact) loads. Moreover, there has been only limited work aimed at measuring the fracture toughness of fibre reinforced–mullite matrix composites [15,19]. In brittle matrices reinforced by ceramic fibres, elastic fibre bridging and fibre pull-out mechanisms mainly cause the toughening [13]. Both these mechanisms in￾crease, to some extent, behind the crack tip in the proc￾ess zone wake [20,21]. Thus, the crack growth resistance rises as the crack propagates and leaves the wake. In these materials, it is difficult to define the intrinsic frac￾ture toughness (KIc) as a material parameter due to a ris￾ing crack growth resistance curve [20]. It is not possible to characterise this group of materials by known meth￾ods of R-curve determination either, because of major crack deflection along the fibre matrix interface and the resultant delamination process instead of single crack propagation. Nevertheless, an exact method of fracture behaviour quantification is needed, mainly for comparison purposes, if a further development of fibre reinforced–brittle matrix composites is desired, includ￾ing the assessment of their possible structural degrada￾tion in service. In the work of Ha and Chawla [19], it was shown that using the chevron notched (CN) speci￾men technique the fracture toughness of mullite matrix composites containing mullite fibres with different coat￾ings could be measured. In our previous investigations, the CN technique has been successfully used to measure fracture toughness in SiC fibre reinforced–glass matrix composites which have been subjected to thermal shock and thermal ageing [22–25]. The purpose of the present work was to analyse the fracture behaviour of a novel mullite fibre/mullite ma￾trix composite, which has been recently developed [26]. Since some preliminary characterisation of the mechan￾ical properties of this composite already exists, i.e. high-temperature fracture strength and thermal cycling resistance [26], the focus of the present study was on crack propagation behaviour under quasi-static condi￾tions and on the composite macroscopic response under ballistic impact loading. The suitability of the CN tech￾nique for determination of the fracture toughness of this fibre reinforced–mullite matrix composite was also assessed. The materials response under impact loads was studied for low-velocity ballistic conditions using a gas gun. The remanent load carrying capability of composite samples after the ballistic test was measured to quantify ballistic impact induced damage. 2. Experimental 2.1. Material The material investigated was a composite formed by a mullite matrix containing homogeneously distributed ultra-fine (70–350 nm) porosity reinforced with NdPO4-coated mullite woven fibre mats (NextelTM 720). The details of the material fabrication, which was done by a combination of electrophoretic deposition, pressure filtration and pressureless sintering, have been published elsewhere [26]. The typical microstructure of the com￾posite is shown in Fig. 1. 2.2. Fracture toughness determination Samples in the form of rectangular test bars of nom￾inal dimensions 4.9 · 1.9 · 45 mm3 were used for frac￾ture toughness determination by means of the CN specimen technique. Chevron notches with angles of 90 were cut using a thin diamond wheel. A 3-point bending test (with span of 16 mm) at a constant cross￾head speed of 0.01 mm/min was employed. Graphs of load versus time were recorded and the maximum force was determined from each trace. The fracture toughness value was calculated from the maximum load (Fmax) and the corresponding minimum value of the geometrical calibration function ðY
minÞ. The calculation of the func￾tion Y
min for CN bend bars was based on the use of Bluhm
s slice model [27]. Details of the procedure used for calculating KIc have been given elsewhere [28]. The CN depth a0, necessary for calculating the calibration function Y
min, was measured from optical micrographs of fractured specimens. The acoustic emission technique (AE) was used during the test. Traces of cumulative number of counts (AE events) were obtained in the same time scale as the load vs. time plots. This technique al￾lows for an accurate detection of the microcrack initia￾tion at the CN, which occurs when a sharp increase in the number of AE events is observed. Valid measure￾ments for computing KIc are those in which this increase of AE events coincides with the maximum load follow￾ing the linear part of the force versus time trace, as explained below. The crack propagation in different samples was observed by low magnification optical microscopy. 2.3. Ballistic impact test and evaluation Ballistic impact tests were carried out by impacting composite tiles of dimensions 75 mm · 75 mm · 2.65 mm with projectiles of varying velocities using a labo￾ratory gas gun. The projectiles were glass balls meas￾uring 10.15 mm in diameter and weighing 1.4 g. The projectile velocity range during this investigation was 77.6–207.5 m/s. The samples were mechanically 326 A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333

A R Boccaccini et al. / Composites Science and Technology 65(2005)325-333 克K Fig. I. SEM micrograph showing the typical microstructure of the mullite fibre reinforced mullite matrix composite investigated [26] clamped to a steel sample holder without a backing Subsequently, samples for 4-point flexure strength plate. The gun employed for this study was a single test measuring 75 mm in length, 32 mm in width and stage laboratory gas gun capable of firing spherical 2.65 mm in height were carefully cut from the impacted up to 400 m/s. A detailed description of the facility impression left by the ballistic impact was placed projectiles with diameters up to 13 mm at velocities tiles. The samples were cut in such a way that th used is given by McQuillan [29]and a schematic dia- the centre of the sample. At least five samples for each gram of the gas gun is shown in Fig. 2. The gun uses testing condition were considered. As-received and im- compressed nitrogen gas to fire the projectile and is pacted samples were tested on a universal testing ma operated by a bursting-diaphragm firing mechanism. chine using a 4-point flexure fixture with 30-mm inner The compressed gas is transferred from the cylinder and 60-mm outer spans. Each specimen was placed in to the gas reservoir (on one end of the barrel), which the fixture such that its impacted side was under tension is joined to a breech adaptor. A suitable diaphragm is and the point of impact was located in the centre of the placed between the barrel and the breech adaptor. inner spans. Tests were conducted at a speed of I mm/ The pressure in the reservoir causes the diaphragm min using a 100 kN load cell. On the basis of data ob- to rupture, shooting the projectile through the barrel. tained during the 4point bending tests, Youngs modu The velocity of the projectile is controlled by the lus was calculated using the following relation The macroscopic damage of the samples after E=o impacts was recorded using a digital camera(Olympus D-510) where Breech Bursting Taper tube Target device Muzzle Fig. 2. Schematic diagram showing the gas gun used for the ballistic impact tests [291

clamped to a steel sample holder without a backing plate. The gun employed for this study was a single stage laboratory gas gun capable of firing spherical projectiles with diameters up to 13 mm at velocities up to 400 m/s. A detailed description of the facility used is given by McQuillan [29] and a schematic dia￾gram of the gas gun is shown in Fig. 2. The gun uses compressed nitrogen gas to fire the projectile and is operated by a bursting-diaphragm firing mechanism. The compressed gas is transferred from the cylinder to the gas reservoir (on one end of the barrel), which is joined to a breech adaptor. A suitable diaphragm is placed between the barrel and the breech adaptor. The pressure in the reservoir causes the diaphragm to rupture, shooting the projectile through the barrel. The velocity of the projectile is controlled by the pressure, which is required to burst the diaphragm. The macroscopic damage of the samples after impacts was recorded using a digital camera (Olympus D-510). Subsequently, samples for 4-point flexure strength test measuring 75 mm in length, 32 mm in width and 2.65 mm in height were carefully cut from the impacted tiles. The samples were cut in such a way that the impression left by the ballistic impact was placed in the centre of the sample. At least five samples for each testing condition were considered. As-received and im￾pacted samples were tested on a universal testing ma￾chine using a 4-point flexure fixture with 30-mm inner and 60-mm outer spans. Each specimen was placed in the fixture such that its impacted side was under tension and the point of impact was located in the centre of the inner spans. Tests were conducted at a speed of 1 mm/ min using a 100 kN load cell. On the basis of data ob￾tained during the 4-point bending tests, Young
s modu￾lus was calculated using the following relation: E ¼ Fl2 ol1 16Jyo ; where, To auxiliary gas reservoir Target Bursting diaphragm Breech Barrel Muzzle Timing device Taper tube Sabot containing projectile Fig. 2. Schematic diagram showing the gas gun used for the ballistic impact tests [29]. Fig. 1. SEM micrograph showing the typical microstructure of the mullite fibre reinforced mullite matrix composite investigated [26]. A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333 327

AR Boccaccini et al. Composites Science and Technology 65(2005 )325-333 J Table I shows the primary data of fracture toughness determination using the CN specimen technique on as- with: b, width of sample; h, height of sample: h1=15 received specimens. The fracture toughness values deter l。=30mm;F, ion at load mined in this study by the cn specimen technique are F, measured using transducers characterised by high scattering of data, as shown in From the load-deflection curves the load for fracture Table l and discussed further below initiation was recorded. a digital camera (Olympus Fig 3 shows typical load-time curves obtained dur- D-510)was used to document the macroscopic deforma- ing CN 3-point bending tests. In Fig 3(a), this curve is tion of the sample during 4-point flexure strength test supplemented by records of cumulative number of AE Fracture surfaces of selected samples were observed by events generated by the two-channel ae kit(chI and scanning electron microscopy (SEM) ch2). Further typical examples of load-time curves ob- served in this investigation are plotted in Fig. 3(b)for specimens ckl, ck5 and ck6. The load-time records ex- 3. Results and discussion hibit typical pop-in effects that can be observed when the maximum load is reached. The pop-ins can be as- 3. 1. Fracture toughness and crack propagation signed to the stopping of the major crack propagation at the fibre/matrix interface followed by interface The mullite matrix composite containi debonding and delamination or even local change of oated woven mullite fibre mats(Nextelm 720)exhibits fracture mode followed by further crack development relatively high fracture strength(235 and 224 MPa at This is seen underneath the CN, as indicated in the room temperature and at 1300C, respectively), as re- optical image of a sample after testing shown in ported elsewhere [26]. The damage tolerant behaviour Fig 4. Delamination in the plane shown in Fig. 4 of the composite, which was ascribed to extensive fibre observed as consequence of the free fracture surface pull-out occurring during failure, has been also con- oriented perpendicularly to the cn plane very close firmed in the previous study [26] to the notch root Table I Primary data for the fracture toughness determination using chevron notched specimen technique in mullite fibre reinforced mullite matrix composites Specimen Specimen width Specimen height Fracture force Chevron notch Calibration function Fracture tough Critical crack (MP 1.74 46.I 171 8.01 5.01 1.85 8.73 Irface 1x105 5x10 20040060080010001200 200400600800100012001400 time s] ne[ s Fig. 3. (a) Examples of load-time curves obtained from 3-point bending test on chevron notched specimens. Typical curves of cumulative number of AE events generated by two channels(chl, ch2 )are also shown. The arrows(a, b)indicate different fracture stages, as discussed in the text. (b) Typical load-time curves obtained from 3-point bending test on different chevron notched specimens(ckl, ck5, ck6), exhibiting different fracture behaviour due to crack propagation occurring in different regions of the composite

J ¼ bh3 12 ; with: b, width of sample; h, height of sample; l1 = 15 mm; lo = 30 mm; F, max. load and yo, deflection at load F, measured using transducers. From the load–deflection curves the load for fracture initiation was recorded. A digital camera (Olympus D-510) was used to document the macroscopic deforma￾tion of the sample during 4-point flexure strength test. Fracture surfaces of selected samples were observed by scanning electron microscopy (SEM). 3. Results and discussion 3.1. Fracture toughness and crack propagation The mullite matrix composite containing NdPO4- coated woven mullite fibre mats (NextelTM 720) exhibits relatively high fracture strength (235 and 224 MPa at room temperature and at 1300 C, respectively), as re￾ported elsewhere [26]. The damage tolerant behaviour of the composite, which was ascribed to extensive fibre pull-out occurring during failure, has been also con- firmed in the previous study [26]. Table 1 shows the primary data of fracture toughness determination using the CN specimen technique on as￾received specimens. The fracture toughness values deter￾mined in this study by the CN specimen technique are characterised by high scattering of data, as shown in Table 1 and discussed further below. Fig. 3 shows typical load–time curves obtained dur￾ing CN 3-point bending tests. In Fig. 3(a), this curve is supplemented by records of cumulative number of AE events generated by the two-channel AE kit (ch1 and ch2). Further typical examples of load–time curves ob￾served in this investigation are plotted in Fig. 3(b) for specimens ck1, ck5 and ck6. The load–time records ex￾hibit typical pop-in effects that can be observed when the maximum load is reached. The pop-ins can be as￾signed to the stopping of the major crack propagation at the fibre/matrix interface followed by interface debonding and delamination or even local change of fracture mode followed by further crack development. This is seen underneath the CN, as indicated in the optical image of a sample after testing shown in Fig. 4. Delamination in the plane shown in Fig. 4 is observed as consequence of the free fracture surface oriented perpendicularly to the CN plane very close to the notch root. Table 1 Primary data for the fracture toughness determination using chevron notched specimen technique in mullite fibre reinforced mullite matrix composites Specimen Specimen width b (mm) Specimen height h (mm) Fracture force Fmax (N) Chevron notch depth a (mm) Calibration function Y
min ð–Þ Fracture toughness KIc (MPa m1/2) Critical crack lengtha acr (mm) ck1 1.74 4.86 29.8 2.39 13.53 3.27 2.77 ck2 1.81 4.92 18.1 2.33 12.62 1.80 2.73 ck3 1.91 4.99 46.1 1.71 8.01 2.74 2.11 ck4 1.96 4.85 36 1.97 10.38 2.74 2.39 ck5 2.04 4.96 28 2.09 10.79 2.09 2.52 ck6 1.95 5.01 34 1.85 8.73 2.15 2.27 a Depth from specimen surface. 0 200 400 600 800 1000 1200 0 10 20 30 40 load [ N ] 5x104 1x105 AE events [ counts ] ch 1 ch 2 0 (a) (b) b a 0 200 400 600 800 1000 1200 1400 time[ s ] 0 10 20 30 40 load [ N ] ck_6 ck_1 ck_5 time [ s] Fig. 3. (a) Examples of load–time curves obtained from 3-point bending test on chevron notched specimens. Typical curves of cumulative number of AE events generated by two channels (ch1,ch2) are also shown. The arrows (a,b) indicate different fracture stages, as discussed in the text. (b) Typical load–time curves obtained from 3-point bending test on different chevron notched specimens (ck1, ck5, ck6), exhibiting different fracture behaviour due to crack propagation occurring in different regions of the composite. 328 A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333

A R Boccaccini et al. / Composites Science and Technology 65(2005)325-333 linearity for specimen ck6. This particular behaviour for each sample is connected with the different micro- structural phases at the Cn tip, i.e. at the moment of crack initiation and/or at the critical crack length, acr, (when the crack reaches the length corresponding to the transition from stable to unstable propagation). In addition, the intersection of the matrix/fibre interface with the major crack plane results in very high tendency for delamination. In the case that the crack tip is still lo- ated in the Cn area(triangle), the driving force and CN 1 mm geometry inhibit the delamination development, this being an advantage of the cn ge Fig4.Optical macrograph of a chevron notched specimen after the SiC fibre reinforced-glass matrix composites in a previ test, showing microcracks in direction perpendicular to the maj ous investigation [24]. Once the major crack tip has the opportunity to intersect the fibre/ matrix interface after departure from the Cn triangle, very extensive delani- Based on the analysis of load-time curves, calculation nation takes place and the major crack plane is thus of critical crack length and analysis of fracture morphol- no longer kept ogy, the fracture behaviour of the composites can be The results of AE analysis support well the previous nterpreted. Fig. 5 shows an optical macrograph of the explanation. In cases where"pop-in"events are ob- pecimens fracture plane. The Cn tip and the direction served on the load-time(and load-deflection)curves,a he crack development are marked by arrows. The rapid increase of AE events is detected, as registered ack length, acr, corresponding to the minimum value by the curve of cumulative number of AE events against on the calibration function, Ymin, as reported in time. This can be seen in Fig 3(a) on both AE curves, Table 1, corresponds to the transition of crack propaga- chl and ch2, generated by two-channel E events ion from controlled to uncontrolled regimes. From the registration analysis of crack initiation and propagation in different When a tensile stress is applied to a composite, sev- samples, it follows that, due to the complex microstruc ral fracture phenomena can occur in the material ture of the material, different possible regions exist with including matrix microcracking, fibre-matrix debonding the sample where crack initiation at the Cn tip may and delamination along the matrix/fibre interface [13]. occur. This leads to significant variability in the material The aE technique used here indicated that individual response and thus to the high scatter of fracture tough- fracture events in the present composites started at ess values found, as reported in Table 1. Fractographic about 1/2 to 2/3 of the maximum load(arrow"a"in observations support this explanation, as elaborated Fig 3(a)). Microstructural changes responsible for this below AE events should be matrix microcracking and partial Support of the above follows also from Fig. 3(b), i.e. local fibre/matrix interfacial decohesion. Deflection very different fracture behaviour may be expected for from linearity of the load-time trace was typical for this each specimen. It is possible to identify the crack devel- stage of CN specimen loading, as well as at loads close opment at the CN tip for example by "pop-ins "present to the maximum load (arrow "bin Fig. 3(a)). This on traces of specimens ckl and ck5, or deflection from was associated with a strong increase in the cumulative number of AE events. Although not conclusively proved in this study, the pronounced increase in AE events is thought to be due to the major crack propagation through the mullite matrix as well as due to fibre/ matrix debonding and fibre fracture. An increase in the number of AE events observed at the end of the linear part of the load-deflection trace indicates therefore that the actual crack has developed at the Cn tip. In thes se cases measurements of Kle can be taken to be valid, 1 mm the unstable fracture (at maximum force) occurs from a propagating crack perpendicular to the fibre axis A typical fracture surface of a CN sample is shown in Fig. 5. Optical macrograph of the fracture plane of a chevron notched st specimen, showing the chevron notch tip and crack Fig. 6. Extensive fibre pull-out, typical of this kind of direction as marked by the arrow. The calculated value of the critical composites [26]is observed. However, due to the previ- crack length aer is also shown. A section of the fracture surface on the ously discussed effects of heterogeneous microstructure right side of the image is not focused because of fracture relief of the composite on Klc values, it is not possible to

Based on the analysis of load–time curves, calculation of critical crack length and analysis of fracture morphol￾ogy, the fracture behaviour of the composites can be interpreted. Fig. 5 shows an optical macrograph of the specimens fracture plane. The CN tip and the direction of the crack development are marked by arrows. The crack length, acr, corresponding to the minimum value on the calibration function, Y
min, as reported in Table 1, corresponds to the transition of crack propaga￾tion from controlled to uncontrolled regimes. From the analysis of crack initiation and propagation in different samples, it follows that, due to the complex microstruc￾ture of the material, different possible regions exist with￾in the sample where crack initiation at the CN tip may occur. This leads to significant variability in the material response and thus to the high scatter of fracture tough￾ness values found, as reported in Table 1. Fractographic observations support this explanation, as elaborated below. Support of the above follows also from Fig. 3(b), i.e. very different fracture behaviour may be expected for each specimen. It is possible to identify the crack devel￾opment at the CN tip, for example by ‘‘pop-ins’’ present on traces of specimens ck1 and ck5, or deflection from linearity for specimen ck6. This particular behaviour for each sample is connected with the different micro￾structural phases at the CN tip, i.e. at the moment of crack initiation and/or at the critical crack length, acr, (when the crack reaches the length corresponding to the transition from stable to unstable propagation). In addition, the intersection of the matrix/fibre interface with the major crack plane results in very high tendency for delamination. In the case that the crack tip is still lo￾cated in the CN area (triangle), the driving force and CN geometry inhibit the delamination development, this being an advantage of the CN geometry as proven for SiC fibre reinforced–glass matrix composites in a previ￾ous investigation [24]. Once the major crack tip has the opportunity to intersect the fibre/matrix interface after departure from the CN triangle, very extensive delami￾nation takes place and the major crack plane is thus no longer kept. The results of AE analysis support well the previous explanation. In cases where ‘‘pop-in’’ events are ob￾served on the load–time (and load–deflection) curves, a rapid increase of AE events is detected, as registered by the curve of cumulative number of AE events against time. This can be seen in Fig. 3(a) on both AE curves, ch1 and ch2, generated by two-channel AE events registration. When a tensile stress is applied to a composite, sev￾eral fracture phenomena can occur in the material, including matrix microcracking, fibre–matrix debonding and delamination along the matrix/fibre interface [13]. The AE technique used here indicated that individual fracture events in the present composites started at about 1/2 to 2/3 of the maximum load (arrow ‘‘a’’ in Fig. 3(a)). Microstructural changes responsible for this AE events should be matrix microcracking and partial local fibre/matrix interfacial decohesion. Deflection from linearity of the load–time trace was typical for this stage of CN specimen loading, as well as at loads close to the maximum load (arrow ‘‘b’’ in Fig. 3(a)). This was associated with a strong increase in the cumulative number of AE events. Although not conclusively proved in this study, the pronounced increase in AE events is thought to be due to the major crack propagation through the mullite matrix as well as due to fibre/matrix debonding and fibre fracture. An increase in the number of AE events observed at the end of the linear part of the load–deflection trace indicates therefore that the actual crack has developed at the CN tip. In these cases, the measurements of KIc can be taken to be valid, since the unstable fracture (at maximum force) occurs from a propagating crack perpendicular to the fibre axis. A typical fracture surface of a CN sample is shown in Fig. 6. Extensive fibre pull-out, typical of this kind of composites [26] is observed. However, due to the previ￾ously discussed effects of heterogeneous microstructure of the composite on KIc values, it is not possible to Fig. 4. Optical macrograph of a chevron notched specimen after the test, showing microcracks in direction perpendicular to the major crack. Fig. 5. Optical macrograph of the fracture plane of a chevron notched test specimen, showing the chevron notch tip and crack propagation direction as marked by the arrow. The calculated value of the critical crack length acr is also shown. A section of the fracture surface on the right side of the image is not focused because of fracture relief. A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333 329

R Boccaccini et al./ Composites Science and Technology 65(2005 )325-333 2 cm Fig. 6. Typical fracture surface of the composite investigated showing quantitatively correlate pull-out behaviour with the neasured Kl dat 3. 2. Ballistic impact resistance and damage Mullite fibre reinforced-mullite matrix composites, projectile impacts, demonstrated a 2 cm typical composite behaviour with the sample remaining in one piece despite some substantial localised damage, especially for high impact energy. Fig. 7(a) shows the macroscopic damage caused by a projectile having im- after ballistic test at: (a) low impact energy (4.2 J)and (b) high impa pact energy of 4.2 J wherein a considerable number of energy(30.1 J) fibres pulling-out on the rear face of the sample can be seen. Owing to the relatively low impact energy, the sample was not penetrated during this impact. Impact ing at higher velocities however led to penetration of the projectile through the samples causing failure as for example shown in Fig. 7(b) for a sample impacted with a projectile of 30.1 J energy. The sample stayed in one piece due to the presence of the fibres. Due to 20.15 he complex nature of the ceramic composite micro- structure(see Fig. 1), post-impact microscopic examina tion of the impacted surface did not reveal much structural detail. In the first instance however it seemed that the structural damage was highly localised around 005 In order to assess the effects of ballistic impact energy on microstructural damage and on structural integrity of the composite, the damaged samples were subjected Displacement(mm) placement curves for as-received and impact damaged Fig 8 Load-displacement curves for: (a)as-received and samples(impact energy 10. 35 J). As mentioned above, under investigation does not 心 the impression left by the ballistic impact was in the cen tre of the sample. As expected, the composite under investigation does not fail catastrophically, even after having been substantially damaged by the impact of of the projectile). This behaviour is in agreement with projectiles. Instead, the material retains its load bearing literature reports [30] on continuous fibre reinforced capacity after the commencement of failure(penetration glass-ceramic matrix composites. Fig 9 documents the

quantitatively correlate pull-out behaviour with the measured KIc data. 3.2. Ballistic impact resistance and damage Mullite fibre reinforced–mullite matrix composites, when subjected to projectile impacts, demonstrated a typical composite behaviour with the sample remaining in one piece despite some substantial localised damage, especially for high impact energy. Fig. 7(a) shows the macroscopic damage caused by a projectile having im￾pact energy of 4.2 J wherein a considerable number of fibres pulling-out on the rear face of the sample can be seen. Owing to the relatively low impact energy, the sample was not penetrated during this impact. Impact￾ing at higher velocities however led to penetration of the projectile through the samples causing failure as for example shown in Fig. 7(b) for a sample impacted with a projectile of 30.1 J energy. The sample stayed in one piece due to the presence of the fibres. Due to the complex nature of the ceramic composite micro￾structure (see Fig. 1), post-impact microscopic examina￾tion of the impacted surface did not reveal much structural detail. In the first instance however it seemed that the structural damage was highly localised around the point of impact. In order to assess the effects of ballistic impact energy on microstructural damage and on structural integrity of the composite, the damaged samples were subjected to 4-point flexural strength test. Fig. 8 shows load–dis￾placement curves for as-received and impact damaged samples (impact energy 10.35 J). As mentioned above, the impression left by the ballistic impact was in the cen￾tre of the sample. As expected, the composite under investigation does not fail catastrophically, even after having been substantially damaged by the impact of projectiles. Instead, the material retains its load bearing capacity after the commencement of failure (penetration of the projectile). This behaviour is in agreement with literature reports [30] on continuous fibre reinforced glass–ceramic matrix composites. Fig. 9 documents the Fig. 6. Typical fracture surface of the composite investigated showing extensive fibre pull-out. Fig. 7. Macrographs showing the macroscopic composite damage after ballistic test at: (a) low impact energy (4.2 J) and (b) high impact energy (30.1 J). Fig. 8. Load–displacement curves for: (a) as-received and (b) impact damaged samples in 4-point flexural strength tests. The composite under investigation does not fail catastrophically, even after having been substantially damaged by the impact of projectiles. 330 A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333

A R Boccaccini et al. / Composites Science and Technology 65(2005)325-333 331 Fig. 9. Macrograph showing the high level of deformation achieved during the 4-point flexural strength test by a sample that had been impacted at an energy of 10.35 J. The sample did not break into two fragments, demonstrating a true composite"pseudo-plastic behaviour high level of deformation achieved dur iring 4-point flex- ral strength test in a sample that had been impacted at an energy of 10.35 J. The sample did not break into two fragments, demonstrating a true composite, ""pseu do-plastic"behaviour. a plot of relative Youngs modulus as a function of impact energy is presented in Fig. 10(a). The elastic modulus of the as-received material ther lot (71.6 GPa), but comparable to that of similar oxide/ oxide composites as reported in the literature [10, 15]. 30 Decrease in elastic modulus after ballistic impact is ob- served up to the point of penetration of the projectile$20 (4.2 J impact energy) where structural damage imum Samples impacted with higher energy projectiles show an increase in Youngs modulus, indicating less structural damage. This accordance with the Impact Energy (J) existing understanding that structural damage under ballistic impact increases to a point where impact en- ergy is just sufficient to cause penetration as observed 2 Iso in polymer and glass-ceramic matrix composit 140 The continuous decrease of Youngs modulus with increasing impact energy of projectile(below 4.2 J) which is related to the cumulative development of 80 microstructural damage in the sample, may be ana lysed by considering a model linking elastic constant and microcracking density. Assuming that micro- 20 cracking in the matrix is the dominant damage mech anism, the approach proposed by Budiansky and 20 OConnell [32] for the elastic modulus of a cracked Impact Energy (J) body could be appropriate, which introduces a dam age parameter based on area and perimeter of uni Fig. 10. (a) Elastic modulus and(b) fracture load of ballistic impacted mullite fibre reinforced-mullite matrix composites as a function of formly distributed cracks. However the damage impact energy The values shown are averages of five measurements ntroduced in the present composites under increasing and the relative error was in all cases <10%

high level of deformation achieved during 4-point flexu￾ral strength test in a sample that had been impacted at an energy of 10.35 J. The sample did not break into two fragments, demonstrating a true composite, ‘‘pseu￾do-plastic’’ behaviour. A plot of relative Young
s modulus as a function of impact energy is presented in Fig. 10(a). The elastic modulus of the as-received material is rather low (71.6 GPa), but comparable to that of similar oxide/ oxide composites as reported in the literature [10,15]. Decrease in elastic modulus after ballistic impact is ob￾served up to the point of penetration of the projectile (4.2 J impact energy) where structural damage is max￾imum. Samples impacted with higher energy projectiles show an increase in Young
s modulus, indicating less structural damage. This is in accordance with the existing understanding that structural damage under ballistic impact increases to a point where impact en￾ergy is just sufficient to cause penetration as observed also in polymer and glass–ceramic matrix composites [30,31]. The continuous decrease of Young
s modulus with increasing impact energy of projectile (below 4.2 J), which is related to the cumulative development of microstructural damage in the sample, may be ana￾lysed by considering a model linking elastic constants and microcracking density. Assuming that micro￾cracking in the matrix is the dominant damage mech￾anism, the approach proposed by Budiansky and O
Connell [32] for the elastic modulus of a cracked body could be appropriate, which introduces a dam￾age parameter based on area and perimeter of uni￾formly distributed cracks. However the damage introduced in the present composites under increasing Fig. 9. Macrograph showing the high level of deformation achieved during the 4-point flexural strength test by a sample that had been impacted at an energy of 10.35 J. The sample did not break into two fragments, demonstrating a true composite ‘‘pseudo-plastic’’ behaviour. 0 10 20 30 40 50 60 70 80 0 5 10 15 20 25 30 35 Impact Energy (J) Young's Modulus (GPa) (a) 0 20 40 60 80 100 120 140 160 180 0 10 20 30 4 Impact Energy (J) Failure Commencement Load (N) (b) 0 Fig. 10. (a) Elastic modulus and (b) fracture load of ballistic impacted mullite fibre reinforced–mullite matrix composites as a function of impact energy. The values shown are averages of five measurements and the relative error was in all cases <10%. A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333 331

AR Boccaccini et al. Composites Science and Technology 65(2005)325-333 impact energy may involve more complex mechanisms Republic. S. Atiq acknowledges the Government of than purely matrix microcracking, including fibre/ma- Pakistan for a fellowship trix interface debond Thus a predictive model for the ballistic impact behaviour of the present composites must take into References consideration the complex microstructure of the com- posites and the effect of interfaces. The formulation [1 Marshall DB, Davis JB. Ceramics for future power generation of such a model is beyond the scope of the present technology: fiber reinforced oxide composites. Curr Opin Solid State Mater Sci 2001: 5: 283-9. [2] Holmquist M, Lundber R, Sudre O, Razzell AG, Molliex L Considering the failure commencement load as an Benoit J, et al. Alumina/alumina composite with a porous indicator of residual strength of the composite, it was ronia interphase, processing, properties ad component testing. found that this reaches a minimum value when the im- J Eur Ceram Soc 2000: 20: 599-606 pact energy is just below that required for penetration 3] Chawla KK, Coffin C, Xu ZR. Interface eng n oxide of the projectile(Fig. 10(b)). Further increase in ballistic fibre/oxide matrix composites. Int Mater Rev 2000: 45: 165-8 4 Peters PWM, Daniels B, Clements F, Vogel WD. Mechanical impact energy after penetration of the sample by the characterisation of mullite based ceramic matrix composites at projectile results in increase of the load carrying capabil st temperatures up to 1200C. J Eur Ceram Soc 2000: 20: 531-5 ity of the composite, due to less microstructural damage [5]Kanka B. Schneider J. Aluminosilicate fiber/mullite matrix being introduced in the sample. This behaviour is in gen composites with favourable high ature properties. J Eur Ceram Soc2000:20:619-93 eral agreement with the literature on ballistic resistance [6] Kramb VA, John R, Zawada LP Notched fracture behaviour of of composite materials [31]. an oxide/oxide ceramic matrix composite. J Am Ceram Soc [7 Lewis MH, Tye A, Butler E, Al-Dawery I. Development of 4. Conclusion terfaces in oxide matrix composites. Key Eng Mater 1999: 164- 5:351-6 The work has demonstrated that the cn technic [8 Schmuecker M, Schneider H, Chawla KK, Xu ZR, Ha J-s Thermal degradation of fibre coatings in mullite-fibre- reinforced can be used to assess fracture behaviour in mullite fibre mullite composites. J Am Ceram Soc 1997: 80: 2136-40 reinforced-mullite matrix composites of complex micro- [9] Ha J-s, Chawla kK, Engdahl EE. Effect of processing and fiber structure. The fracture toughness(Kl) values deter coating on fibre matrix interaction in mullite fibre-mullite matrix mined using the Cn technique were in the range of composites. Mater Sci Eng A 1993: 161: 303-8. 1.8-3.3 MPa m". The large scatter of fracture tough [10] Holmquist MG, Lange FF Processing and properties of a porous oxide matrix com reinforced with continuous oxide fibres. J ness values is typical for this kind of materials due to Am Ceram soc2003:86:173340. the complex(heterogeneous)composite microstructure [ll] Schneider H, Okada K, Pask JA, editors. Mullite and mullite The observed fibre pull-out effect is due to a favourable ceramics. Chichester: Wiley: 1994. matrix/ fibre interfacial bond given by NdPOA coating of [2 Somiya S, Davies RF, Pask JA, editors. Ceramics transactions. Mullite and mullite matrix osites. vol. 6. Westerville the fibres OH: American Ceramic Society: 1990 Mullite fibre reinforced-mullite matrix composites, [13] Chawla KK. Ceramic matrix composites. 2nd ed. Norwell(MA), when subjected to ballistic impact loading by firing Dordrecht. The Netherlands: Kluwer Academic Press: 2003 pherical projectiles, retain some of their load bearing [14] Chawla KK, Xu ZR, Ha J-s Processing, structure and properties capacity after penetration by the projectile. This is of mullite fiber mullite matrix composites. J Eur Ceram Soc due to the fact that structural damage caused by pro- [15] Yeheskel O, Balmer ML, Cranmer DC Interfacial chemistry of jectiles remains localised preventing catastrophic fail mullite-mullite composites. Ceram Eng Sci Proc 1988: 9(7- ure. For the conditions of the present tests, :687-94 penetration by the projectile occurs at impact energy [16] Boccaccini AR, MacLaren I,Lewis MH, Ponton CBElectroph of about 4J. which indicates that there is a limiting va- retic deposition infiltration of 2-D woven SiC fibre mats with lue for the material to be useful in ballistic armour of mullite composition. J Eur Ceram Soc 1997;17:154 applications when it is used on its own(without back- [17] Kaya C,Gu X, Al-Dawery I,Butler EG.Microstructural ing layers) development of woven mullite fibre-reinforced-mullite ceramic atrix composites by infiltration processing. Sci Technol Adv Mater2002:3:35-4 [18 She J, Mechnich P, Schneider H. Schmuecker M, Kanka B Effect Acknowledgements of cyclic infiltrations on microstructure and mechanical behaviour of porous mullite/mullite composites. Mater Sci Eng A The research was partially funded from Grants of 2002:325:19-24 NATO(No. PST CLG.977558)and Royal Society 19 Ha JS, Chawla KK. Mechanical behaviour of mullite reinforced with mullite fibres. Mater Sci Eng A 1995: A203: 1271-6 (London, UK). Part of the experimental work was [20] Akatsu T, Yasuda E, Sakai M Fracture toughness and work of financially supported by Grant No. A2041003 of the fracture of toughened brittle materials in chevron notch geometry. Grant Agency of the Academy of Sciences, Czech Fract Mech Ceram 1996: 11: 245

impact energy may involve more complex mechanisms than purely matrix microcracking, including fibre/ma￾trix interface debonding and localised fibre fracture. Thus a predictive model for the ballistic impact behaviour of the present composites must take into consideration the complex microstructure of the com￾posites and the effect of interfaces. The formulation of such a model is beyond the scope of the present experimental study. Considering the failure commencement load as an indicator of residual strength of the composite, it was found that this reaches a minimum value when the im￾pact energy is just below that required for penetration of the projectile (Fig. 10(b)). Further increase in ballistic impact energy after penetration of the sample by the projectile results in increase of the load carrying capabil￾ity of the composite, due to less microstructural damage being introduced in the sample. This behaviour is in gen￾eral agreement with the literature on ballistic resistance of composite materials [31]. 4. Conclusion The work has demonstrated that the CN technique can be used to assess fracture behaviour in mullite fibre reinforced–mullite matrix composites of complex micro￾structure. The fracture toughness (KIc) values deter￾mined using the CN technique were in the range of 1.8–3.3 MPa m1/2. The large scatter of fracture tough￾ness values is typical for this kind of materials due to the complex (heterogeneous) composite microstructure. The observed fibre pull-out effect is due to a favourable matrix/fibre interfacial bond given by NdPO4 coating of the fibres. Mullite fibre reinforced–mullite matrix composites, when subjected to ballistic impact loading by firing spherical projectiles, retain some of their load bearing capacity after penetration by the projectile. This is due to the fact that structural damage caused by pro￾jectiles remains localised preventing catastrophic fail￾ure. For the conditions of the present tests, penetration by the projectile occurs at impact energy of about 4 J, which indicates that there is a limiting va￾lue for the material to be useful in ballistic armour applications when it is used on its own (without back￾ing layers). Acknowledgements The research was partially funded from Grants of NATO (No. PST.CLG.977558) and Royal Society (London, UK). Part of the experimental work was financially supported by Grant No. A2041003 of the Grant Agency of the Academy of Sciences, Czech Republic. S. Atiq acknowledges the Government of Pakistan for a fellowship. References [1] Marshall DB, Davis JB. Ceramics for future power generation technology: fiber reinforced oxide composites. Curr Opin Solid State Mater Sci 2001;5:283–9. [2] Holmquist M, Lundber R, Sudre O, Razzell AG, Molliex L, Benoit J, et al. Alumina/alumina composite with a porous zirconia interphase, processing, properties ad component testing. J Eur Ceram Soc 2000;20:599–606. [3] Chawla KK, Coffin C, Xu ZR. Interface engineering in oxide fibre/oxide matrix composites. Int Mater Rev 2000;45:165–89. [4] Peters PWM, Daniels B, Clements F, Vogel WD. Mechanical characterisation of mullite based ceramic matrix composites at test temperatures up to 1200 C. J Eur Ceram Soc 2000;20:531–5. [5] Kanka B, Schneider J. Aluminosilicate fiber/mullite matrix composites with favourable high temperature properties. J Eur Ceram Soc 2000;20:619–93. [6] Kramb VA, John R, Zawada LP. Notched fracture behaviour of an oxide/oxide ceramic matrix composite. J Am Ceram Soc 1999;82:3087–96. [7] Lewis MH, Tye A, Butler E, Al-Dawery I. Development of interfaces in oxide matrix composites. Key Eng Mater 1999;164– 165:351–6. [8] Schmuecker M, Schneider H, Chawla KK, Xu ZR, Ha J-S. Thermal degradation of fibre coatings in mullite–fibre-reinforced mullite composites. J Am Ceram Soc 1997;80:2136–40. [9] Ha J-S, Chawla KK, Engdahl EE. Effect of processing and fiber coating on fibre matrix interaction in mullite fibre–mullite matrix composites. Mater Sci Eng A 1993;161:303–8. [10] Holmquist MG, Lange FF. Processing and properties of a porous oxide matrix composite reinforced with continuous oxide fibres. J Am Ceram Soc 2003;86:1733–40. [11] Schneider H, Okada K, Pask JA, editors. Mullite and mullite ceramics. Chichester: Wiley; 1994. [12] Somiya S, Davies RF, Pask JA, editors. Ceramics transactions. Mullite and mullite matrix composites, vol. 6. Westerville, OH: American Ceramic Society; 1990. [13] Chawla KK. Ceramic matrix composites. 2nd ed. Norwell (MA), Dordrecht, The Netherlands: Kluwer Academic Press; 2003. [14] Chawla KK, Xu ZR, Ha J-S. Processing, structure and properties of mullite fiber mullite matrix composites. J Eur Ceram Soc 1996;16:293–9. [15] Yeheskel O, Balmer ML, Cranmer DC. Interfacial chemistry of mullite–mullite composites. Ceram Eng Sci Proc 1988;9(7– ):687–94. [16] Boccaccini AR, MacLaren I, Lewis MH, Ponton CB. Electroph￾oretic deposition infiltration of 2-D woven SiC fibre mats with mixed sols of mullite composition. J Eur Ceram Soc 1997;17:1545–50. [17] Kaya C, Gu X, Al-Dawery I, Butler EG. Microstructural development of woven mullite fibre-reinforced–mullite ceramic matrix composites by infiltration processing. Sci Technol Adv Mater 2002;3:35–44. [18] She J, Mechnich P, Schneider H, Schmuecker M, Kanka B. Effect of cyclic infiltrations on microstructure and mechanical behaviour of porous mullite/mullite composites. Mater Sci Eng A 2002;325:19–24. [19] Ha JS, Chawla KK. Mechanical behaviour of mullite composites reinforced with mullite fibres. Mater Sci Eng A 1995;A203:1271–6. [20] Akatsu T, Yasuda E, Sakai M. Fracture toughness and work of fracture of toughened brittle materials in chevron notch geometry. Fract Mech Ceram 1996;11:245. 332 A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333

A R Boccaccini et al. / Composites Science and Technology 65(2005)325-333 21 Evans AG, Zok Fw. Review. The physics and mechanics of [26 Kaya C, Butler EG, Selcuk A, Boccaccini AR, Lewis MH Mullite fibre-reinforced brittle matrix composites. J Mat Sci 1994: 29 Nextel- 720)fibre reinforced mullite matrix composites exhibit. 3857-96 favourable thermomechanical properties. J Eur Ceram So AR, Janczak-Rusch J, Dlouhy I. Thermal aging 002222333-4 a SiC-fiber reinforced glass matrix com In non- 27 Shang-Xian W. Fracture toughness determiantion of bearing steel tmosphere. Mater Chem Phys 1998: 53: 155-64 hevron-notched three point bending specimen. Eng Fract [23] Ken H, Dlouhy I. Determining the fracture Mech1984:19:221-32. istance of fibre reinforced glass matrix composites by means of [28] Holzmann M. man J. valka l. the use of chevron the chevron-notched flexural technique. Mater Sci Eng A en for fracture toughness determination. Metal 2001:308:11 Mater I994:32:3-13 4 Dlouhy I, boccaccio toughness []McQuillan F. PhD Thesis, Imperial College, University of ites reinforced by continuous Scripta Mater 2001: 44: 531-7 30 Rawlings RD. Glass-ceramic matrix composites. Composites 25] Dlouhy I, Reinisch M, Boccaccini AR. Fracture toughness and 1994:5:372-9 ork of fracture of Sic-fibre reinforced glass matrix composite 31]Matthews FL, Rawlings RD. Composite materials: engineering n: Bradt RC. Munz d. Sakai M. Shevchenko v. white K nd science. Woodhead Publishing Limited: 1999 editors. Fracture mechanics of ceramics. vol. 13. New York 32]Budiansky B, OConnell R. Elastic moduli of a cracked solid. Int J Kluwer Acacemic/Plenum Publishers: 2002. p. 203-l Solid Struct 1976: 12: 81-97

[21] Evans AG, Zok FW. Review. The physics and mechanics of fibre-reinforced brittle matrix composites. J Mat Sci 1994; 29: 3857–96. [22] Boccaccini AR, Janczak-Rusch J, Dlouhy´ I. Thermal aging behavior of a SiC-fiber reinforced glass matrix composite in non￾oxidizing atmosphere. Mater Chem Phys 1998;53:155–64. [23] Boccaccini AR, Kern H, Dlouhy I. Determining the fracture resistance of fibre reinforced glass matrix composites by means of the chevron-notched flexural technique. Mater Sci Eng A 2001;308:111–7. [24] Dlouhy I, Boccaccini AR. Reliability of the chevron notch technique for fracture toughness determination in glass compos￾ites reinforced by continuous fibres. Scripta Mater 2001;44:531–7. [25] Dlouhy´ I, Reinisch M, Boccaccini AR. Fracture toughness and work of fracture of SiC-fibre reinforced glass matrix composite. In: Bradt RC, Munz D, Sakai M, Shevchenko V, White K, editors. Fracture mechanics of ceramics, vol. 13. New York: Kluwer Acacemic/Plenum Publishers; 2002. p. 203–12. [26] Kaya C, Butler EG, Selcuk A, Boccaccini AR, Lewis MH. Mullite (NextelTM 720) fibre reinforced mullite matrix composites exhibit￾ing favourable thermomechanical properties. J Eur Ceram Soc 2002;22:2333–42. [27] Shang-Xian W. Fracture toughness determiantion of bearing steel using chevron-notched three point bending specimen. Eng Fract Mech 1984;19:221–32. [28] Dlouhy I, Holzmann M, Man J, Valka L. The use of chevron nothed specimen for fracture toughness determination. Metal Mater 1994;32:3–13. [29] McQuillan F. PhD Thesis, Imperial College, University of London; 1992. [30] Rawlings RD. Glass–ceramic matrix composites. Composites 1994;5:372–9. [31] Matthews FL, Rawlings RD. Composite materials: engineering and science. Woodhead Publishing Limited; 1999. [32] Budiansky B, O
Connell R. Elastic moduli of a cracked solid. Int J Solid Struct 1976;12:81–97. A.R. Boccaccini et al. / Composites Science and Technology 65 (2005) 325–333 333