《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_Effect of thermomechanical loads on microstructural damage and on the resulting thermomechanical behaviour of silicon carbide fibre-reinforced glass matrix composites

MATERAL CHARACIERAAION ELSEVIER Materials Characterization 54(2005)75-83 Effect of thermomechanical loads on microstructural damage and on the resulting thermomechanical behaviour of silicon carbide fibre-reinforced glass matrix composites A.R. Boccaccinia,*. A.M. Torre. C.R. oldanib D.N. Boccaccinic Department of Materials, Imperial College London, Prince Consort Rd, London SW7 2BP UK DEpartamento de Materiales, Universidad Nacional de Cordoba, argentina Dipartimento di Ingegneria dei Materiali e dell'Ambiente, Universita di Modena e, Reggio Emilia, Modena, itah Received 23 July 2004; received in revised form 26 October 2004: accepted 5 November 2004 Abstract The development of microstructural damage in SiC fibre(Nicalon )reinforced glass matrix composites subjected to different mechanical and thermal loads was investigated by assessing the change of the thermal expansion coefficient and the resistance to impact loads of the composites. The thermal expansion coefficient, measured by dilatometry after thermal shock tests from 650C to room temperature, was found to be 'insensitive to microstructural damage such as matrix microcracking, matrix softening or matrix/fibre interface degradation. The impact resistance of the composites, measured by a pendulum-type apparatus, was high even after subjecting the samples to different thermomechanical loads, h as repetitive thermal shocks from 650C for up to 20 cycles. However, the samples showed an appreciable degradation of their impact resistance after thermal aging in air at 700C for 100 h. This was shown to be due to extensive porosity formation due to softening of the glass matrix and due to oxidation of the carbon-rich fibre/matrix interfaces C 2004 Elsevier Inc. All rights reserved Keywords: Glass matrix composites; Thermal expansion; Microstructural damage; Impact behaviour; Thermal shock; Thermal aging 1. Introduction Silicate glasses possess a combination of attractive properties as: high hardness, high resistance to Corresponding author. Tel. +44 20 7594 6731; fax: +44 20 che low density, high wear resistance, 75843194. insulating electrical properties and optical transpar E-mail address. aboccaccini@imperial ac uk ency. The main disadvantages that glasses present for (A.R. Boccaccini) technical applications are their extreme fragility, their 1044-5803/S- see front matter e 2004 Elsevier Inc. All rights reserved doi:10.1016 j. matcha.2004.11.0

Effect of thermomechanical loads on microstructural damage and on the resulting thermomechanical behaviour of silicon carbide fibre-reinforced glass matrix composites A.R. Boccaccinia,*, A.M. Torreb , C.R. Oldanib , D.N. Boccaccinic a Department of Materials, Imperial College London, Prince Consort Rd., London SW7 2BP, UK b Departamento de Materiales, Universidad Nacional de Co´rdoba, Argentina c Dipartimento di Ingegneria dei Materiali e dell’Ambiente, Universita` di Modena e, Reggio Emilia, Modena, Italy Received 23 July 2004; received in revised form 26 October 2004; accepted 5 November 2004 Abstract The development of microstructural damage in SiC fibre (NicalonR) reinforced glass matrix composites subjected to different mechanical and thermal loads was investigated by assessing the change of the thermal expansion coefficient and the resistance to impact loads of the composites. The thermal expansion coefficient, measured by dilatometry after thermal shock tests from 650 8C to room temperature, was found to be dinsensitiveT to microstructural damage such as matrix microcracking, matrix softening or matrix/fibre interface degradation. The impact resistance of the composites, measured by a pendulum-type apparatus, was high even after subjecting the samples to different thermomechanical loads, such as repetitive thermal shocks from 650 8C for up to 20 cycles. However, the samples showed an appreciable degradation of their impact resistance after thermal aging in air at 700 8C for 100 h. This was shown to be due to extensive porosity formation due to softening of the glass matrix and due to oxidation of the carbon-rich fibre/matrix interfaces. D 2004 Elsevier Inc. All rights reserved. Keywords: Glass matrix composites; Thermal expansion; Microstructural damage; Impact behaviour; Thermal shock; Thermal aging 1. Introduction Silicate glasses possess a combination of attractive properties such as: high hardness, high resistance to chemical attack, low density, high wear resistance, insulating electrical properties and optical transpar￾ency. The main disadvantages that glasses present for technical applications are their extreme fragility, their 1044-5803/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.matchar.2004.11.001 * Corresponding author. Tel.: +44 20 7594 6731; fax: +44 20 7584 3194. E-mail address:a.boccaccini@imperial.ac.uk (A.R. Boccaccini). Materials Characterization 54 (2005) 75 – 83

L.R. Boccaccini et al. Materials Characterization 54(2005)75-8 low resistance to thermal shock and their low ical loads resulting in microstructural damage. Meas- mechanical strength, which makes them unsuitable urements of the thermal expansion coefficient were as structural materials. The introduction of reinforce- utilised for the first time to assess the sensitivity of ments in the form of fibres in glass matrices, forming this parameter to microstructural changes, as a composites, can change these characteristics of possible method for detection of microstructural glasses making it possible to develop materials damage in its early stages of developI tolerant to microstructural defects or cracks. with evaluation of the mechanical behaviour of the high mechanical resistance and a very good resistance composite material after thermomechanical loads to thermal shock [1-4]. The composite materials thus was carried out through impact tests. Scanning obtained constitute advanced lightweight structural electron microscopy (SEM) was used to verify the materials, which can be utilised at high temperatures occurrence of energy-dissipation mechanisms, such as and under mechanical loads in oxidant atmospheres fibre pull-out, during composite fracture Therefore, these materials may find applications in energy-conversion systems, aerospace vehicles and structures, as well as in the building industry 2. Experimental fireproof components or thermal protection elements that must work under mechanical stresses [3-5]. This 2. 1. Material favourable flaw-tolerant behaviour of the composites is not only due to the high modulus of elasticity and The material investigated was a commercially mechanical resistance of the fibres but also due to th available unidirectional SiC-Nicalon&(NL 202) presence of a thin carbon-rich fibre/matrix interfacial fibre-reinforced borosilicate (duraN)glass-matrix layer(20-30 nm thickness). This layer, produced composite fabricated by Schott-Glaswerke (Mainz, during the high-temperature fabrication step, origi- Germany)[14. The composite was prepared by the nates energy-dissipation mechanisms during fracture, ethod. Details of this technique can be such as crack deflection and fibre"pull-out, provid read in previous publications [15, 16]. Nominal ing the appropriate matrix/fibre bonding strength properties of the matrix, fibre and composite are leading to favourable"pseudo-ductile"fracture behav iven in Table 1 [17, 18]. The material was received in iour [3, 5]. To achieve the satisfactory use of these the form of prismatic test bars of nominal dimensions materials in the applications above mentioned and to 4.5X3.8X100 mm. The density of the composites was extend their application to other technical areas, it is 2.4 g cm and the fibre volume fraction.4. This necessary to have a broad knowledge of their material is being considered for applications in the thermomechanical behaviour. This explains the grow- glass and non-ferrous metallurgy industries, for the ing quantity of studies published in the specialised fabrication of tools for handling of hot glassware and literature in the last decade about the behaviour of metal parts, respectively [14] glass-matrix composite materials when submitted to thermal gradients and combined thermomechanical stresses [6-13]. There is, however, further need in Table developing techniques to characterise microstructural Properties of composite constituents and of the composite [15-17] damage in composites after thermomechanical dam Property latrix DURAN Fibre sic age loads is required to assess the expecte (borosilicate Nicalon(40 voL% lifetime of components in service. The objective of (NL202) fibre content) this study is to investigate the influence of mechanical Density (g cm") 2.23 and thermal loads on the development of micro- Young's structural damage and in the resulting thermomechan- modulus(GPa) ical behaviour of silicon carbide(Nicalon)fibre- 0.22 0.200.2 Thermal expansion 3. 25x10 3×10-63.10×10-6 reinforced glass matrix composite materials. Thermal oefficient(K-) shock, thermal cycling, thermal aging and mechanical Tensile strength 60 600700 re-stressing were studied as typical thermomechan (MPa)

low resistance to thermal shock and their low mechanical strength, which makes them unsuitable as structural materials. The introduction of reinforce￾ments in the form of fibres in glass matrices, forming composites, can change these characteristics of glasses making it possible to develop materials tolerant to microstructural defects or cracks, with high mechanical resistance and a very good resistance to thermal shock [1–4]. The composite materials thus obtained constitute advanced lightweight structural materials, which can be utilised at high temperatures and under mechanical loads in oxidant atmospheres. Therefore, these materials may find applications in energy-conversion systems, aerospace vehicles and structures, as well as in the building industry as fireproof components or thermal protection elements that must work under mechanical stresses [3–5]. This favourable flaw-tolerant behaviour of the composites is not only due to the high modulus of elasticity and mechanical resistance of the fibres, but also due to the presence of a thin carbon-rich fibre/matrix interfacial layer (20–30 nm thickness). This layer, produced during the high-temperature fabrication step, origi￾nates energy-dissipation mechanisms during fracture, such as crack deflection and fibre bpull-outQ, provid￾ing the appropriate matrix/fibre bonding strength leading to favourable bpseudo-ductileQ fracture behav￾iour [3,5]. To achieve the satisfactory use of these materials in the applications above mentioned and to extend their application to other technical areas, it is necessary to have a broad knowledge of their thermomechanical behaviour. This explains the grow￾ing quantity of studies published in the specialised literature in the last decade about the behaviour of glass-matrix composite materials when submitted to thermal gradients and combined thermomechanical stresses [6–13]. There is, however, further need in developing techniques to characterise microstructural damage in composites after thermomechanical dam￾age loads. This is required to assess the expected lifetime of components in service. The objective of this study is to investigate the influence of mechanical and thermal loads on the development of micro￾structural damage and in the resulting thermomechan￾ical behaviour of silicon carbide (NicalonR) fibre￾reinforced glass matrix composite materials. Thermal shock, thermal cycling, thermal aging and mechanical pre-stressing were studied as typical thermomechan￾ical loads resulting in microstructural damage. Meas￾urements of the thermal expansion coefficient were utilised for the first time to assess the sensitivity of this parameter to microstructural changes, as a possible method for detection of microstructural damage in its early stages of development. The evaluation of the mechanical behaviour of the composite material after thermomechanical loads was carried out through impact tests. Scanning electron microscopy (SEM) was used to verify the occurrence of energy-dissipation mechanisms, such as fibre pull-out, during composite fracture. 2. Experimental 2.1. Material The material investigated was a commercially available unidirectional SiC-NicalonR (NL 202) fibre-reinforced borosilicate (DURANR) glass–matrix composite fabricated by Schott-Glaswerke (Mainz, Germany) [14]. The composite was prepared by the sol–gel slurry method. Details of this technique can be read in previous publications [15,16]. Nominal properties of the matrix, fibre and composite are given in Table 1 [17,18]. The material was received in the form of prismatic test bars of nominal dimensions 4.5
3.8
100 mm. The density of the composites was 2.4 g cm3 and the fibre volume fraction ~0.4. This material is being considered for applications in the glass and non-ferrous metallurgy industries, for the fabrication of tools for handling of hot glassware and metal parts, respectively [14]. Table 1 Properties of composite constituents and of the composite [15–17] Property Matrix DURANR (borosilicate glass) Fibre SiC NicalonR (NL202) Composite (40 vol.% fibre content) Density (g cm3 ) 2.23 2.55 2.4 Young’s modulus (GPa) 63 198 119 Poisson’s ratio 0.22 0.20 0.21 Thermal expansion coefficient (K1 ) 3.25
106 3
106 3.10
106 Tensile strength (MPa) 60 2750 600–700 76 A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 2.2. Thermal shock tests applied loads investigated were both below and above the fracture load, which was taken from results in the The thermal shock tests involved heating the literature on the same composite [12] amples in air in a muffle furnace at a pre-determined temperature(650C) for 10 min and then dropping 2.6. Microstructural characterisation them abruptly into an unstirred water bath maintained at room temperature(20C). The thermal gradient SEM of polished sections was conducted in an was chosen on the basis of previous studies [12, 18] in attempt to characterise the microstructural damage order to ensure that the microstructural damage induced in the specimens. Since the main energy induced in the samples could be attributed only to dissipation mechanism that occurs during fracture of the effect of thermal shock and not to other environ- these materials is the phenomenon of fibre pull-out mental influences such as, for example, fibre/matrix [10, 12], whose occurrence an nsion can be interface or SiC fibre oxidation. After each thermal appreciated visually, SEM observations of fracture shock test, the samples were dried in an oven at 100 surfaces were carried out C for I h. They were carefully inspected, visually, for the appearance of any macroscopic damage, such as 2.7. Impact test delamination, chipping or fibre protrusion. Samples in form of test bars were impacted with an 2.3. Thermal cycling energy of 4 J. This value of energy was adopted on the basis of impact test data obtained in previous For the thermal cycling tests, the samples were studies [19]. a pendulous apparatus with a mass and alternated quickly between high temperature(700C) arm especially designed to deliver the necessary and room temperature for different number of cycles energy to ensure the same operative conditions was (up to 1000 cycles). A computer program controlled used. A custom-made data acquisition system was the movement of the sample holder. Details of the adapted to the impact tester for observing if signifi equipment used for this test can be found elsewhere cant variations of energy consumed by the samples [10]. The time at high temperature was set at 15 min, occurred during impact. More detailed information of with 5 min being allowed for the sample to reach the these experiments is given elsewhere [19] target temperature of 700C, while the time at room temperature was fixed at 5 min. Thus, a complete 2.8. Measurement of the thermal expansion coefficient thermal cycle lasted 20 min. After a determined number of cycles, the samples were inspected care The thermal expansion coefficient (ax) of compo- fully for evidence of any macroscopic damage, such ite samples before and after thermomechanical as a change of the surface colour, delamination, loading was determined using a high temperature protrusion of fibres, spalling and/or chipping dilatometer (NETZSCH GmbH TMA 402). The samples were cut with a diamond saw to an 2. 4. Thermal aging approximate length of 23 mm. Before carrying out the experiments, the dimensions of the samples were The thermal aging experiments involved heating measured using a vernier with a precision of 0.05 samples at 600 and 700C in a fumace in an air mm. The measurements were carried out in normal atmosphere for long periods of time (up to 100 h). atmosphere between room temperature(20C)and After the thermal exposure, the samples were cooled 750C with a heating rate of 5 K min. At least two down slowly inside the fumace measurements were performed for each specimen and the results were averaged. The values were calculated 2.5. Mechanical pre-stressing for temperature intervals: 20-300, 20-400 and 20- 00C. The accuracy of the measured values was Selected samples were subjected to mechanical Ax=+lx10K. A high reproducibility of results loading in a three-point bending configuration. The was confirmed

2.2. Thermal shock tests The thermal shock tests involved heating the samples in air in a muffle furnace at a pre-determined temperature (650 8C) for 10 min and then dropping them abruptly into an unstirred water bath maintained at room temperature (20 8C). The thermal gradient was chosen on the basis of previous studies [12,18] in order to ensure that the microstructural damage induced in the samples could be attributed only to the effect of thermal shock and not to other environ￾mental influences such as, for example, fibre/matrix interface or SiC fibre oxidation. After each thermal shock test, the samples were dried in an oven at 100 8C for 1 h. They were carefully inspected, visually, for the appearance of any macroscopic damage, such as delamination, chipping or fibre protrusion. 2.3. Thermal cycling For the thermal cycling tests, the samples were alternated quickly between high temperature (700 8C) and room temperature for different number of cycles (up to 1000 cycles). A computer program controlled the movement of the sample holder. Details of the equipment used for this test can be found elsewhere [10]. The time at high temperature was set at 15 min, with 5 min being allowed for the sample to reach the target temperature of 700 8C, while the time at room temperature was fixed at 5 min. Thus, a complete thermal cycle lasted 20 min. After a determined number of cycles, the samples were inspected care￾fully for evidence of any macroscopic damage, such as a change of the surface colour, delamination, protrusion of fibres, spalling and/or chipping. 2.4. Thermal aging The thermal aging experiments involved heating samples at 600 and 700 8C in a furnace in an air atmosphere for long periods of time (up to 100 h). After the thermal exposure, the samples were cooled down slowly inside the furnace. 2.5. Mechanical pre-stressing Selected samples were subjected to mechanical loading in a three-point bending configuration. The applied loads investigated were both below and above the fracture load, which was taken from results in the literature on the same composite [12]. 2.6. Microstructural characterisation SEM of polished sections was conducted in an attempt to characterise the microstructural damage induced in the specimens. Since the main energy￾dissipation mechanism that occurs during fracture of these materials is the phenomenon of fibre pull-out [10,12], whose occurrence and extension can be appreciated visually, SEM observations of fracture surfaces were carried out. 2.7. Impact test Samples in form of test bars were impacted with an energy of ~4 J. This value of energy was adopted on the basis of impact test data obtained in previous studies [19]. A pendulous apparatus with a mass and arm especially designed to deliver the necessary energy to ensure the same operative conditions was used. A custom-made data acquisition system was adapted to the impact tester for observing if signifi￾cant variations of energy consumed by the samples occurred during impact. More detailed information of these experiments is given elsewhere [19]. 2.8. Measurement of the thermal expansion coefficient The thermal expansion coefficient (a) of compo￾site samples before and after thermomechanical loading was determined using a high temperature dilatometer (NETZSCH GmbH TMA 402). The samples were cut with a diamond saw to an approximate length of 23 mm. Before carrying out the experiments, the dimensions of the samples were measured using a vernier with a precision of 0.05 mm. The measurements were carried out in normal atmosphere between room temperature (20 8C) and 750 8C with a heating rate of 5 K min1 . At least two measurements were performed for each specimen and the results were averaged. The values were calculated for temperature intervals: 20–300, 20–400 and 20– 500 8C. The accuracy of the measured values was Da=F1
107 K1 . A high reproducibility of results was confirmed. A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83 77

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 3. Results and discussion Negligible differences exist between the values measured for the different samples, except for the 3.I. Microstructure of the investigated material specimen subjected 20 times to thermal shock (AT=630C), indicating that this sample may have A SEM micrograph showing the microstructure of suffered a higher degree of microstructural damage a polished section of an as received sample is shown than the rest. The experimental values of the thermal in Fig. 1. It can be seen that the matrix material is expansion coefficient obtained for the temperature dense, without residual porosity, and the composite ranges 20-400 and 20-500C were very similar to exhibits a fairly regular fibre distribution wn In Fig Fig 3 shows the thermal expansion curve obtained 3.2. Evaluation of the thermal expansion behaviour expansion that characterises the glass transition Considering the values of thermal expansion temperature (Ty) of the glass matrix (525C for efficient pertaining to the duran)glass matrix dURAN glass)[17] is not apparent due to the and to the silicon carbide fibres(am and dominant influence of the fibres respectively ), and taking into account that the fibre ig. 4 shows the thermal expansion curves of both volume fraction is -0.4, the theoretical value of the the first and second measurements performed on the thermal expansion coefficient of the composite can be same sample which had been subjected to thermal calculated by a simple rule of mixtures: sho oc= aflf + amM measurement a“wave" between300and400°Ccan be observed. this is characteristic of materials that where Pr and Vm are the fibre and matrix volume possess internal stresses. These stresses are generated fractions, respectively. With the values of ar and om as result of the progressive number of thermal shocks given in Table 1, the thermal expansion coefficient [ 18]. Concerning the second measurement on the obtained from Eq. (1)is: ac-3. 19x10-6K- same sample, a different value of thermal expansion Fig. 2 shows the experimental values of the coefficient was obtained, which is very close to that thermal expansion coefficient measured in the temper- calculated by Eq. (1). Moreover, no"wave"is ature range of 20-300C for all samples investigated. observed in the curve, as shown in Fig. 4. This As can be this graph, and taking into suggests that during heating the sample up to 750C, account the average error of the measurements during the first measurement of the thermal expansion (+lx10K), the measured values are very close coefficient, an internal relaxation of stresses was to that calculated by the rule of mixtures(Eq(1). induced. Previous studies on the same material [12 have shown that thermal shocks of At=570C do not cause microstructural damage in this type of compo site, at least up to 21 cycles. However, thermal shocks with△T≥690° C produce serious microscopic degra- dation of the material in the form of microcracks delamination and oxidation of the carbon-rich inter face and of Nicalon SiC fibres. In the same previous study [12], microstructural damage was confirmed on samples subjected to thermal shock for 20 cycles with a thermal gradient equal to that utilised in this work (AT=630C). In that study, a forced-vibration technique and flexural strength tests were used to 50m detect microstructural damage. The damage detected was in the form of small microcracks In the Fig. 1 SEM micrograph of the polished section of the composit glass ial investigated. A regular distribution of Sic fibres in the matrix, without major delamination or mass changes matrix is observed and absence of porosit samples. This precludes the

3. Results and discussion 3.1. Microstructure of the investigated material A SEM micrograph showing the microstructure of a polished section of an as received sample is shown in Fig. 1. It can be seen that the matrix material is dense, without residual porosity, and the composite exhibits a fairly regular fibre distribution. 3.2. Evaluation of the thermal expansion behaviour Considering the values of thermal expansion coefficient pertaining to the (DURANR) glass matrix and to the silicon carbide fibres (am and af, respectively), and taking into account that the fibre volume fraction is ~0.4, the theoretical value of the thermal expansion coefficient of the composite can be calculated by a simple rule of mixtures: aC ¼ afVf þ amVm ð1Þ where Vf and Vm are the fibre and matrix volume fractions, respectively. With the values of af and am given in Table 1, the thermal expansion coefficient obtained from Eq. (1) is: aC=3.19
106 K1 . Fig. 2 shows the experimental values of the thermal expansion coefficient measured in the temper￾ature range of 20–300 8C for all samples investigated. As can be seen in this graph, and taking into account the average error of the measurements (F1
107 K1 ), the measured values are very close to that calculated by the rule of mixtures (Eq. (1)). Negligible differences exist between the values measured for the different samples, except for the specimen subjected 20 times to thermal shock (DT=630 8C), indicating that this sample may have suffered a higher degree of microstructural damage than the rest. The experimental values of the thermal expansion coefficient obtained for the temperature ranges 20–400 and 20–500 8C were very similar to those shown in Fig. 2. Fig. 3 shows the thermal expansion curve obtained for the as-received sample. The quick increase of expansion that characterises the glass transition temperature (Tg) of the glass matrix (525 8C for DURANR glass) [17] is not apparent due to the dominant influence of the fibres. Fig. 4 shows the thermal expansion curves of both the first and second measurements performed on the same sample which had been subjected to thermal shock for 20 cycles (DT=630 8C). In the first measurement a bwaveQ between 300 and 400 8C can be observed. This is characteristic of materials that possess internal stresses. These stresses are generated as result of the progressive number of thermal shocks [18]. Concerning the second measurement on the same sample, a different value of thermal expansion coefficient was obtained, which is very close to that calculated by Eq. (1). Moreover, no bwaveQ is observed in the curve, as shown in Fig. 4. This suggests that during heating the sample up to 750 8C, during the first measurement of the thermal expansion coefficient, an internal relaxation of stresses was induced. Previous studies on the same material [12] have shown that thermal shocks of DT=570 8C do not cause microstructural damage in this type of compo￾site, at least up to 21 cycles. However, thermal shocks with DTz690 8C produce serious microscopic degra￾dation of the material in the form of microcracks, delamination and oxidation of the carbon-rich inter￾face and of NicalonR SiC fibres. In the same previous study [12], microstructural damage was confirmed on samples subjected to thermal shock for 20 cycles with a thermal gradient equal to that utilised in this work (DT=630 8C). In that study, a forced-vibration technique and flexural strength tests were used to detect microstructural damage. The damage detected was in the form of small microcracks in the glass matrix, without major delamination or mass changes of the samples. This precludes the possibility of a Fig. 1. SEM micrograph of the polished section of the composite material investigated. A regular distribution of SiC fibres in the glass matrix is observed and absence of porosity. 78 A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 a), b), c)Samples in the as-received condition. d), e) Samples thermally aged for 100 hs at 600 and 700 C, respectively 20-300°C ample mechanically pre-stressed above the fracture load j, k) Samples mechanically pre-stressed above the fracture load subjected to thermal shock. 4 3,5 a) b) c) d) e) f g) h) i) 1) k) Fig. 2. values obtained for the thermal expansion coefficient in the range 20-300C for all composites investigated (as-received and after thermomechanical loading, as indicated in the legend): (O) first measurement, (O)second measurement, (O) third measurement. The error average in the measurements was±1×10-7K-1 significant oxidation of the interface at those inter-wave"in the thermal expansion curve recorded mediate temperatures. Matrix microcracking occurs during the first measurement of ac(Fig. 4) because in a quenching test, the sample surface is Regarding the samples thermally cycled in air,a brought to the temperature of the cooling medium previous study [10] has shown that microstructural rapidly, whereas the interior of the samples remains at damage can be mainly attributed to partial oxidation high temperature. This temperature gradient creates of the interfaces and to softening of the glass matrix internal stresses, which are tensile at the surface and during the exposure to a high-temperature oxidising compressive in the interior. With an increasing environment. In the previous study [10, thermal number of quenching cycles, the tensile stresses at cycling in air from 700C to room temperature the surfaces can reach a critical value sufficient to resulted in the generation of microstructural damage microcracks in the matrix. In this investigation, which was detected after 77 cycles by the simulta- nulated residual stresses are detected by the neous decrease in Youngs modulus and increase 2 200300 Temperature(C) Fig. 3. Curve representing the thermal expansion behaviour of the composite material investigated in the initial state(as-received). Heating rate=s K min

significant oxidation of the interface at those inter￾mediate temperatures. Matrix microcracking occurs because in a quenching test, the sample surface is brought to the temperature of the cooling medium rapidly, whereas the interior of the samples remains at high temperature. This temperature gradient creates internal stresses, which are tensile at the surface and compressive in the interior. With an increasing number of quenching cycles, the tensile stresses at the surfaces can reach a critical value sufficient to create microcracks in the matrix. In this investigation, the accumulated residual stresses are detected by the bwaveQ in the thermal expansion curve recorded during the first measurement of ac (Fig. 4). Regarding the samples thermally cycled in air, a previous study [10] has shown that microstructural damage can be mainly attributed to partial oxidation of the interfaces and to softening of the glass matrix during the exposure to a high-temperature oxidising environment. In the previous study [10], thermal cycling in air from 700 8C to room temperature resulted in the generation of microstructural damage, which was detected after 77 cycles by the simulta￾neous decrease in Young’s modulus and increase in Fig. 2. Values obtained for the thermal expansion coefficient in the range 20–300 8C for all composites investigated (as-received and after thermomechanical loading, as indicated in the legend): (o) first measurement, (n) second measurement, ( R ) third measurement. The error average in the measurements was F1
107 K1 . Fig. 3. Curve representing the thermal expansion behaviour of the composite material investigated in the initial state (as-received). Heating rate=5 K min1 . A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83 79

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 5 Second measurement Temperature(C) Fig 4. Curves representing the thermal expansion of a composite material sample submitted to thermal shock 20 times(47-630C):(a)first neasurement,(b) second meast nt. Heating rate=s K min the intemal friction of the composite [10]. In the The sample subjected to thermal aging (100 h, 700 present study, two samples were thermally cycled with C)presents the most severe damage. This sample did I and 17 cycles, expecting to find microstructural amage in its early stage of development. However, no significant variations of the thermal expansion coefficients of the samples that could be attributed to microstructural damage were found(see Fig. 3) 3.3. Evaluation of the mechanical behaviour by mpact strength tes Fig. 5(a, b) shows images of the fracture surface of as-received sample broken by impact test(impact energy=4 J). Evidence of extensive fibre pull-out can be observed, which is indicative of the favourable 1mm quasi-ductile" fracture behaviour characteristic of this composite material. The samples subjected to thermal shock presented fracture surfaces very similar to those of the as received samples, with evidence of significant fibre pull-out, as shown in Fig. 6. In this investigation, no significant variations in the impact strength of the samples were found even after 21 thermal shocks with △7=630-650°C. Assuming that fibre pull-out depends fundamentally on the properties of the fibre/matrix interface and because thermal shock induces damage mainly in the form of matrix micro- cracking [12], the"quasi-ductile"mechanical behav iour induced by the fibre pull-out phenomenon is no altered under the thermal shock conditions inves tigated. Therefore, the composite material retains its Fig. 5. SEM micrographs showing the fracture surface of an as- received sample submitted to impact test (energy 4 J)at (a) low and high resistance to impact loads, even after having b) high magnification. The typical fibre pull-out effect, character been subjected to thermal shocks of AT=630 istic of these composite materials, is observed

the internal friction of the composite [10]. In the present study, two samples were thermally cycled with 1 and 17 cycles, expecting to find microstructural damage in its early stage of development. However, no significant variations of the thermal expansion coefficients of the samples that could be attributed to microstructural damage were found (see Fig. 3). 3.3. Evaluation of the mechanical behaviour by impact strength tests Fig. 5(a,b) shows images of the fracture surface of an as-received sample broken by impact test (impact energy=4 J). Evidence of extensive fibre pull-out can be observed, which is indicative of the favourable bquasi-ductileQ fracture behaviour characteristic of this composite material. The samples subjected to thermal shock presented fracture surfaces very similar to those of the as￾received samples, with evidence of significant fibre pull-out, as shown in Fig. 6. In this investigation, no significant variations in the impact strength of the samples were found even after 21 thermal shocks with DT=630–650 8C. Assuming that fibre pull-out depends fundamentally on the properties of the fibre/matrix interface and because thermal shock induces damage mainly in the form of matrix micro￾cracking [12], the bquasi-ductileQ mechanical behav￾iour induced by the fibre pull-out phenomenon is not altered under the thermal shock conditions inves￾tigated. Therefore, the composite material retains its high resistance to impact loads, even after having been subjected to thermal shocks of DT=630 8C. The sample subjected to thermal aging (100 h, 700 8C) presents the most severe damage. This sample did Fig. 4. Curves representing the thermal expansion of a composite material sample submitted to thermal shock 20 times (DT=630 8C): (a) first measurement, (b) second measurement. Heating rate=5 K min1 . Fig. 5. SEM micrographs showing the fracture surface of an as￾received sample submitted to impact test (energy 4 J) at (a) low and (b) high magnification. The typical fibre pull-out effect, character￾istic of these composite materials, is observed. 80 A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 of fibres and softening of the glass matrix with consequent microstructural rearrangement [10]. This form of microstructural damage has been observed and discussed in other fibre-reinforced glass and glass-ceramic matrix composites aged at intermediate temperatures(in the range 400-800C)for different times(from a few hours to over 500 h)[7, 20 Fig. 8 confirms the same form of damage in the present composi observed by SEM on th sample thermally aged for 100 h at 700C. Despite 50um this extensive microstructural damage, the thermal expansion coefficient measured did not show a large difference with that of the as-received sample(see Fig. 6. SEM micrograph showing the fracture surface of a sample Fig. 3). This verifies theoretical results in the literature that had been subjected to thermal shock 20 times(A7-630"C), predicting that the thermal expansion coefficient of after impact test(4 J). Extensive fibre pull-out, similar to that bserved in the sample in the as-received state(Fig. 5), is observed materials is not affected by porosity [21] not exhibit a unique surface of fracture, and in a zone 4. Final remarks and conclusions that extends from the point of impact towards one of the sample edges, total debonding of the matrix (delamination) could be observed with only the fibre The thermal expansion coefficient(ac)was meas- bundles remaining on the fractured surface. Moreover, ured in Sic-Nicalon fibre reinforced glass matrix ome of these fibres have been fractured during the composite materials subjected to different mechanical impact test. Fig. 7 shows a macroscopic photograph of and thermal loads. The objective was to evaluate the the sample thermally aged(100 h, 700oC), which possibility to use this coefficient to monitor changes illustrates the morphology of fracture of the specimen in the material microstructure(microstructural dam- The degradation of the mechanical behaviour of this age) which occurred as result of the thermomechanical material as a result of the prolonged exposition to high mainly to the degradation of the interface, debonding Fig. 7. Macrograph showing the sample submitted to themal aging Fig. 8. SEM micrograph showing porosity induced in the glass (100 h, 700C), after impact test. Extensive macroscopic damage is matrix after thermal aging for 100 h at 700C, due to glass viscous observed and fracture by delamination along the fibres has occurred. deformation

not exhibit a unique surface of fracture, and in a zone that extends from the point of impact towards one of the sample edges, total debonding of the matrix (delamination) could be observed with only the fibre bundles remaining on the fractured surface. Moreover, some of these fibres have been fractured during the impact test. Fig. 7 shows a macroscopic photograph of the sample thermally aged (100 h, 700 8C), which illustrates the morphology of fracture of the specimen. The degradation of the mechanical behaviour of this material as a result of the prolonged exposition to high temperature in oxidant environments is attributed mainly to the degradation of the interface, debonding of fibres and softening of the glass matrix with consequent microstructural rearrangement [10]. This form of microstructural damage has been observed and discussed in other fibre-reinforced glass and glass-ceramic matrix composites aged at intermediate temperatures (in the range 400–800 8C) for different times (from a few hours to over 500 h) [7,20]. Fig. 8 confirms the same form of damage in the present composites, as observed by SEM on the sample thermally aged for 100 h at 700 8C. Despite this extensive microstructural damage, the thermal expansion coefficient measured did not show a large difference with that of the as-received sample (see Fig. 3). This verifies theoretical results in the literature predicting that the thermal expansion coefficient of materials is not affected by porosity [21]. 4. Final remarks and conclusions The thermal expansion coefficient (ac) was meas￾ured in SiC-NicalonR fibre reinforced glass matrix composite materials subjected to different mechanical and thermal loads. The objective was to evaluate the possibility to use this coefficient to monitor changes in the material microstructure (microstructural dam￾age) which occurred as result of the themomechanical Fig. 7. Macrograph showing the sample submitted to thermal aging (100 h, 700 8C), after impact test. Extensive macroscopic damage is observed and fracture by delamination along the fibres has occurred. Fig. 8. SEM micrograph showing porosity induced in the glass matrix after thermal aging for 100 h at 700 8C, due to glass viscous deformation. Fig. 6. SEM micrograph showing the fracture surface of a sample that had been subjected to thermal shock 20 times (DT=630 8C), after impact test (4 J). Extensive fibre pull-out, similar to that observed in the sample in the as-received state (Fig. 5), is observed. A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83 81

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 loads applied to the material. No significant changes of the carbon-rich matrix/fibre interface. Because the of the thermal expansion coefficient were found for embrittling effect is not significant in this composite the majority of the samples observed. This demon- material, it is possible to conclude that matrix soften strates the dimensional stability of the composite even ing is the limiting factor for applications involving after having been subjected to diverse forms of high temperatures (-700C) Interestingly, the sam mechanical and thermal loads. However, a significant ples thermally cycled using the conditions of this increase in the value of the thermal expansion study had a higher resistance to impact than the as- coefficient (approximately 18%)was observed for received samples. This permits one to conclude that a the sample submitted a 20 cycles to thermal shock thermal treatment("annealing)in the post-production (47-630C) This is probably due to internal residual stage(for example 5 h at 700C)may improve the stresses produced by the successive thermal shocks. mechanical behaviour of the composite material under The fact that no significant changes in the thermal study expansion coefficients were found, for the majority of the samples investigated, does not necessarily indicate that the samples are completely"free"of micro- Acknowledgements structural damage. This is particularly evident in the sample thermally aged at 700C for 100 h, which The measurements of the thermal expansion despite having suffered extensive microstructural coefficient were performed in the Institute of Materi damage(Fig. 8), exhibited a similar ae to that of the als Technology of the Technical University of as-received material. The most important conclusion ILmenau, Germany, when one of the authors(AMT) of this part of the study is, therefore, that thermal was a visiting student. The impact tests were carried expansion coefficient measurements are not suff out in the installations of the CIMM. Cordoba ciently"" to detect microstructural damage Argentina. The authors wish to thank the assistance in these materials. Only when internal changes in the of Dr. V. Winkler, Dr. D. Raab and all technical staff material were very large, did the coefficient show a at the Institute of Materials Technology, Technical gnificant variation. It is clear then that other non- University of Ilmenau, and at the CIMM, that made destructive techniques should be employed to detect possible this work. microstructural damage in its early development in these composite materials. In particular, measurement of the intemal friction by the forced-vibration reso- References nance method seems to be the most suitable technique, as has been shown in previous studies [10, 12, 18 [1] Chawla KK. Ceramic matrix composites. London: Chapman Impact tests were also carried out on the composite and Hall: 1993 materials after samples were submitted to thermome- 22 Boccaccini AR. Glass and glass-ceramic matrix composite chanical loads. The material in the as-received materials. A review. J Ceram Soc Jpn 2001: 109(7): S99-109. condition showed a quasi-ductile"behaviour typical 3] Prewo KM. The development of fibre reinforced glasses and glass-ceramics. In: Tressler RE, Messing G, Pantano C of these composites with a fracture surface exhibiting Newnham R, editors. Tailoring multiphase and composite extensive fibre pull-out. This very favourable behav- ceramics. Mat Sci Res. voL. 20 York: Plenum press iour was observed also in samples subjected to 1986.p.529-47 thermal shock, thermal cycling and mechanical pre 4 Chawla KK. The high-tem application of ceramic stresses. This demonstrates the great capacity of the sites.J Miner Mater Metal Soc (JOM)1995 47(12):19-21 composite material investigated to bear this type of 5 Rawlings RD. Glass-ceramic matrix composites. Composit load. However, similar results were not found in 1994;25:372-9 amples submitted to thermal aging at 700C. In these 6 Phillips DC, Park N, Lee RJ. The impact behaviour of high experiments, a marked weakening of the material was erformance, ceramic matrix fibre composites. Comp Sci observed. This can be mainly attributed to matrix Technol1990;37:249-65 softening(due to viscous flow of the glass at high [7] Bleay SM, Harris B, Scott VD, Cooke RG, Habib FA. Mechanical and structural characterization of nicalon fibre. temperature)and to oxidation of the fibre surface and reinforced borosilicate glass. J Mater Sci 1996: 31: 5933-40

loads applied to the material. No significant changes of the thermal expansion coefficient were found for the majority of the samples observed. This demon￾strates the dimensional stability of the composite even after having been subjected to diverse forms of mechanical and thermal loads. However, a significant increase in the value of the thermal expansion coefficient (approximately 18%) was observed for the sample submitted a 20 cycles to thermal shock (DT=630 8C). This is probably due to internal residual stresses produced by the successive thermal shocks. The fact that no significant changes in the thermal expansion coefficients were found, for the majority of the samples investigated, does not necessarily indicate that the samples are completely bfreeQ of micro￾structural damage. This is particularly evident in the sample thermally aged at 700 8C for 100 h, which despite having suffered extensive microstructural damage (Fig. 8), exhibited a similar ac to that of the as-received material. The most important conclusion of this part of the study is, therefore, that thermal expansion coefficient measurements are not suffi￾ciently bsensitiveQ to detect microstructural damage in these materials. Only when internal changes in the material were very large, did the coefficient show a significant variation. It is clear then that other non￾destructive techniques should be employed to detect microstructural damage in its early development in these composite materials. In particular, measurement of the internal friction by the forced-vibration reso￾nance method seems to be the most suitable technique, as has been shown in previous studies [10,12,18]. Impact tests were also carried out on the composite materials after samples were submitted to thermome￾chanical loads. The material in the as-received condition showed a bquasi-ductileQ behaviour typical of these composites with a fracture surface exhibiting extensive fibre pull-out. This very favourable behav￾iour was observed also in samples subjected to thermal shock, thermal cycling and mechanical pre￾stresses. This demonstrates the great capacity of the composite material investigated to bear this type of load. However, similar results were not found in samples submitted to thermal aging at 700 8C. In these experiments, a marked weakening of the material was observed. This can be mainly attributed to matrix softening (due to viscous flow of the glass at high temperature) and to oxidation of the fibre surface and of the carbon-rich matrix/fibre interface. Because the embrittling effect is not significant in this composite material, it is possible to conclude that matrix soften￾ing is the limiting factor for applications involving high temperatures (~700 8C). Interestingly, the sam￾ples thermally cycled using the conditions of this study had a higher resistance to impact than the as￾received samples. This permits one to conclude that a thermal treatment (bannealingQ) in the post-production stage (for example 5 h at 700 8C) may improve the mechanical behaviour of the composite material under study. Acknowledgements The measurements of the thermal expansion coefficient were performed in the Institute of Materi￾als Technology of the Technical University of Ilmenau, Germany, when one of the authors (AMT) was a visiting student. The impact tests were carried out in the installations of the CIMM, Cordoba, Argentina. The authors wish to thank the assistance of Dr. V. Winkler, Dr. D. Raab and all technical staff at the Institute of Materials Technology, Technical University of Ilmenau, and at the CIMM, that made possible this work. References [1] Chawla KK. Ceramic matrix composites. London7 Chapman and Hall; 1993. [2] Boccaccini AR. Glass and glass-ceramic matrix composite materials. A review. J Ceram Soc Jpn 2001;109(7):S99–109. [3] Prewo KM. The development of fibre reinforced glasses and glass-ceramics. In: Tressler RE, Messing G, Pantano C, Newnham R, editors. Tailoring multiphase and composite ceramics. Mat Sci Res, vol. 20. New York7 Plenum Press, 1986. p. 529 – 47. [4] Chawla KK. The high-temperature application of ceramic matrix composites. J Miner Mater Metal Soc (JOM) 1995; 47(12):19 – 21. [5] Rawlings RD. Glass-ceramic matrix composites. Composites 1994;25:372 – 9. [6] Phillips DC, Park N, Lee RJ. The impact behaviour of high performance, ceramic matrix fibre composites. Comp Sci Technol 1990;37:249 – 65. [7] Bleay SM, Harris B, Scott VD, Cooke RG, Habib FA. Mechanical and structural characterization of nicalon fibre￾reinforced borosilicate glass. J Mater Sci 1996;31:5933 – 40. 82 A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83

A.R. Boccaccini et al. Materials Characterization 54(2005)75-83 [8]Blisset M, Smith PA, Yeomans JA. Thermal shock behaviour (15 Pannhorst W, Spallek M, Bruckner R, Hegeler H, Reich C, of unidirectional silicon carbide fibre reinforced calciu Grathwohl G, et al. Fibre-reinforced glasses and glass aluminosilicate. J Mater Sci 1997: 32: 317-25 ceramics fabricated by a novel process. Ceram Eng Sci Proc N. Kishi T. Thermal shock resistance of SiC fibre-reinforced borosilicate glass and lithium alumino [16] Hegeler H, Brickner R. Fibre reinforced glasses. J Mater Sci licate matrix composites. J Mater Sci 1993: 28: 735-41 1989;24:1191-201 [10] Boccaccini AR, Strutt A, Vecchio KS, Mendoza D, Chawla [17] Klug T, Fleischer V, Bruckner R. Thermal KK. Ponton CB. et al. Behaviour of nicalon-fibre reinforced behaviour of fibre-reinforced duran glass. Glastech Ber glass matrix composites under thermal cycling conditions. 1993:66:201-6. Compos, Part A Appl Sci Manuf 1998; 29(11): 1343-52 [18 Boccaccini AR, Janczak-Rusch J, Pearce DH, Kern H. [11] Metcalfe BL, Donald Iw, Bradley DJ. Preparation and Assessment of damage induced by thermal shock in Sic. properties of a SiC fibre-reinforced glass-ceramic matrix fibre-reinforced borosilicate glass composites. Compos Sci composite. Br Ceram Trans 1993: 92: 13-20 Technol1999;59:105-12. [12] Boccaccini AR, Pearce DH, Janczak J, Beier W, Ponton CB. [19 Boccaccini AR. Habilitation Thesis, Technical University of Investigation of the cyclic thermal shock behaviour of mena(Germany) 2000. fibre reinforced glass matrix composites using a non-destruc 120 Plucknett KP, Cain RL, Lewis MH. Interface degradation in tive forced resonance technique. Mater Sci Technol 1997; 13 CAS/Nicalon during elevated temperature aging. Mater Res Symp Proc1995:365:421-6 [13] Sutherland S, Plucknett KP, Lewis MH. High mechanical and [21]Ondracek G. Microstructure-thermomechanical-property thermal stability of silicate matrix composites. Compos Eng relations of two-phase and porous materials. Mater Chem Phys 995:5:1367-78 1986:15:281-31 [14 Bier W. Tough and strong. Schott-Inf 1995: 73: 3-6

[8] Blisset MJ, Smith PA, Yeomans JA. Thermal shock behaviour of unidirectional silicon carbide fibre reinforced calcium aluminosilicate. J Mater Sci 1997;32:317 – 25. [9] Kagawa Y, Kurosawa N, Kishi T. Thermal shock resistance of SiC fibre-reinforced borosilicate glass and lithium aluminosi￾licate matrix composites. J Mater Sci 1993;28:735 – 41. [10] Boccaccini AR, Strutt A, Vecchio KS, Mendoza D, Chawla KK , Ponton CB, et al. Behaviour of nicalon-fibre reinforced glass matrix composites under thermal cycling conditions. Compos, Part A Appl Sci Manuf 1998;29(11):1343 – 52. [11] Metcalfe BL, Donald IW, Bradley DJ. Preparation and properties of a SiC fibre-reinforced glass-ceramic matrix composite. Br Ceram Trans 1993;92:13 – 20. [12] Boccaccini AR, Pearce DH, Janczak J, Beier W, Ponton CB. Investigation of the cyclic thermal shock behaviour of fibre reinforced glass matrix composites using a non-destruc￾tive forced resonance technique. Mater Sci Technol 1997;13: 852 – 9. [13] Sutherland S, Plucknett KP, Lewis MH. High mechanical and thermal stability of silicate matrix composites. Compos Eng 1995;5:1367 – 78. [14] Bier W. Tough and strong. Schott-Inf 1995;73:3 – 6. [15] Pannhorst W, Spallek M, Brqckner R, Hegeler H, Reich C, Grathwohl G, et al. Fibre-reinforced glasses and glass￾ceramics fabricated by a novel process. Ceram Eng Sci Proc 1990;11:947 – 63. [16] Hegeler H, Brqckner R. Fibre reinforced glasses. J Mater Sci 1989;24:1191 – 201. [17] Klug T, Fleischer V, Brqckner R. Thermal expansion behaviour of fibre-reinforced DURAN glass. Glastech Ber 1993;66:201 – 6. [18] Boccaccini AR, Janczak-Rusch J, Pearce DH, Kern H. Assessment of damage induced by thermal shock in SiC￾fibre-reinforced borosilicate glass composites. Compos Sci Technol 1999;59:105 – 12. [19] Boccaccini AR. Habilitation Thesis, Technical University of Ilmenau (Germany) 2000. [20] Plucknett KP, Cain RL, Lewis MH. Interface degradation in CAS/Nicalon during elevated temperature aging. Mater Res Symp Proc 1995;365:421 – 6. [21] Ondracek G. Microstructure-thermomechanical-property cor￾relations of two-phase and porous materials. Mater Chem Phys 1986;15:281 – 313. A.R. Boccaccini et al. / Materials Characterization 54 (2005) 75–83 83