COMPOSITES SCIENCE AND TECHNOLOGY ELSEVIER Composites Science and Technology 59(1999)801-811 Fibre strength parameters measured in situ for ceramic-matrix composites tested at elevated temperature in vacuum and in air lan J Davies Takashi Ishikawa, * Masaki Shibuya, Tetsuro Hirokawa rframe Division, National Aerospace Laboratory, 6-13-1 Ohsawa, Mitaka-Shi, Tokyo 181, Japan cOrporate Research and Development, Ube Industries Ltd, 1978-10 Kogushi, UbeShi, Yamaguchi 755, Japan Research and Development Department, Industrial Textile Division, Shikibo Ltd, 1500-5 Shibahara-Minami, Yokaichi-Shi, Shiga 527, Japan Received 3 July 1997; received in revised form 15 July 1998; accepted 7 January 1999 Abstract In situ fibre fracture characteristics have been investigated for Si-Ti-C-O fibres after tensile testing up to 1380"C in vacuum and in air. Specimens tested in air at 1 100 and 1200 C generally had flat fracture surfaces with less than 20% of fibres exhibiting fracture mirrors: this is attributed to oxygen ingress into the fibre bundles. Fibre strength characteristics normalised to a 10-3 m gauge length indicated that fibres tested in air at elevated temperature have significantly lower strengths and average Weibull parameter. m, compared to the room-temperature, 1200 and 1300 C/vacuum cases, and this is attributed to oxygen damage of the fibre toge- her with oxidation of the fibre/matrix interface. The fibre/matrix interface shear strength, t, was low for the room-temperature specimens and increased slightly with temperature when tested in vacuum, possibly as a result of a change in the thermal mismatch between fibres and matrix. Values of r for specimens tested at 1100 and 1200 C in air were an order of magnitude greater than those for room-temperature specimens, indicating a significant degree of oxidation damage at the fibre/ matrix interface to have occurred C 1999 Elsevier Science Ltd. All rights reserved 1. ntroduction shear strength, t, and fibre properties such as the radius r, and the Weibull strength parameters, So and m omposites(CMCs) that utilise con- Values for So and m may be obtained by measuring the tinuous fibres as reinforcement have many potential in situ strength of fibres and fitting a two-parameter high-temperature structural applications, particularly in Weibull curve [7] to the resulting data,i.e the area of space re-entry vehicles. Recent advances include production of 3-D woven composites based on F=1-c-(÷) the sic/Sic system that possess short-term tensile (1) strengths of nearly 400 MPa in vacuum at room tem- perature and 1200C with tensile strains to failure in where F is the cumulative failure probability of fibres at excess of 1%[1-3]. However, the mechanical properties a stress, S, and S, and m are empirical constants known of CMCs often degrade at elevated temperature in the as the Weibull strength parameters. Although the ex situ presence of oxygen which is known to attack the fibre/ strength of ceramic fibres is often well known, there is a matrix interface. Although studies have shown that relative dearth of data concerning in situ fibre strength sealing the surface of CMCs with a glass-based com- One method of estimating in situ fibre strength involves pound may allow similar mechanical properties to be the measurement of mirror radi-a feature often pre- achieved at elevated temperature in air and in vacuum sent on the fracture surface of ceramic fibres. Fig. I [4], improvement of the fibre/matrix interface is still a illustrates such a fracture mirror observed on the frac- ture surface of a Tyranno Si-Ti-C-O fibre It can be Recent work [5,6] has indicated that several impor seen that the fracture mirror is cent tant composite mechanical properties may be predicted to the initiating defect in the fibre and is surrounded by from a knowledge only of the fibre/matrix interface a region of multiple fracture planes. Past research has shown most fracture mirrors to 4 Corresponding author. initiate at flaws present on the surface of the fibre [8, 9 0266-3538/99/S- see front matter C 1999 Elsevier Science Ltd. All rights reserved. PlI:S0266-3538(99)00011-1
Fibre strength parameters measured in situ for ceramic-matrix composites tested at elevated temperature in vacuum and in air Ian J. Davies a , Takashi Ishikawa a,*, Masaki Shibuya b, Tetsuro Hirokawa c a Airframe Division, National Aerospace Laboratory, 6-13-1 Ohsawa, Mitaka-Shi, Tokyo 181, Japan bCorporate Research and Development, Ube Industries Ltd, 1978-10 Kogushi, Ube-Shi, Yamaguchi 755, Japan c Research and Development Department, Industrial Textile Division, Shikibo Ltd, 1500-5 Shibahara-Minami, Yokaichi-Shi, Shiga 527, Japan Received 3 July 1997; received in revised form 15 July 1998; accepted 7 January 1999 Abstract In situ ®bre fracture characteristics have been investigated for Si±Ti±C±O ®bres after tensile testing up to 1380C in vacuum and in air. Specimens tested in air at 1100 and 1200C generally had ¯at fracture surfaces with less than 20% of ®bres exhibiting fracture mirrors: this is attributed to oxygen ingress into the ®bre bundles. Fibre strength characteristics normalised to a 10ÿ3 m gauge length indicated that ®bres tested in air at elevated temperature have signi®cantly lower strengths and average Weibull parameter, m, compared to the room-temperature, 1200 and 1300C/vacuum cases, and this is attributed to oxygen damage of the ®bre together with oxidation of the ®bre/matrix interface. The ®bre/matrix interface shear strength, , was low for the room-temperature specimens and increased slightly with temperature when tested in vacuum, possibly as a result of a change in the thermal mismatch between ®bres and matrix. Values of for specimens tested at 1100 and 1200C in air were an order of magnitude greater than those for room-temperature specimens, indicating a signi®cant degree of oxidation damage at the ®bre/matrix interface to have occurred. # 1999 Elsevier Science Ltd. All rights reserved. 1. Introduction Ceramic-matrix composites (CMCs) that utilise continuous ®bres as reinforcement have many potential high-temperature structural applications, particularly in the area of space re-entry vehicles. Recent advances include production of 3-D woven composites based on the SiC/SiC system that possess short-term tensile strengths of nearly 400 MPa in vacuum at room temperature and 1200C with tensile strains to failure in excess of 1% [1±3]. However, the mechanical properties of CMCs often degrade at elevated temperature in the presence of oxygen which is known to attack the ®bre/ matrix interface. Although studies have shown that sealing the surface of CMCs with a glass-based compound may allow similar mechanical properties to be achieved at elevated temperature in air and in vacuum [4], improvement of the ®bre/matrix interface is still a major area of research. Recent work [5,6] has indicated that several important composite mechanical properties may be predicted from a knowledge only of the ®bre/matrix interface shear strength, , and ®bre properties such as the radius, r, and the Weibull strength parameters, So and m. Values for So and m may be obtained by measuring the in situ strength of ®bres and ®tting a two-parameter Weibull curve [7] to the resulting data, i.e. F 1 ÿ eÿ S So m 1 where F is the cumulative failure probability of ®bres at a stress, S, and So and m are empirical constants known as the Weibull strength parameters. Although the ex situ strength of ceramic ®bres is often well known, there is a relative dearth of data concerning in situ ®bre strength. One method of estimating in situ ®bre strength involves the measurement of mirror radiiÐa feature often present on the fracture surface of ceramic ®bres. Fig. 1 illustrates such a fracture mirror observed on the fracture surface of a Tyranno1 Si±Ti±C±O ®bre. It can be seen that the fracture mirror is a smooth region adjacent to the initiating defect in the ®bre and is surrounded by a region of multiple fracture planes. Past research has shown most fracture mirrors to initiate at ¯aws present on the surface of the ®bre [8,9] Composites Science and Technology 59 (1999) 801±811 0266-3538/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0266-3538(99)00011-1 * Corresponding author
I.. Davies et al. Composites Science and Technology 59(1999)801-811 which is vulnerable to degradation during fibre and approximations for So and m with the correct form of composite processing. That fracture initiation does not the relationship being [5] normally occur within the fibre bulk suggests the aver- ge flaw size and or flaw density to be significantly lar- F=l ger on the fibre surface compared to that within the fibre bulk. A common equation linking mirror radius, with S" and m being the uncorrected Weibull strength Im, to fibre strength, S is of the form parameters In order to obtain the relevant in situ fibre strength S (2) parameters, So and m, it is required to use correction factors that have been determined [5 to be of the form shown in Fig. 2. It may be observed from Fig. 2 that Eqs (1)and (3)gives similar results when m A 4 but values for So and m obtained from Eq (1)for the case of m< 4 will Am= bric respectively underestimate and overestimate actual valu where Am and Bm are empirical constants and Kic is the It should be noted that Eq 3)assumes no knowledge racture toughness of the fibre. a value of Kic a 1 of the specimen gauge length. However, the value of S MPa m/ 2 has been estimated for Nicalon"SiC-based obtained depends strongly on the specimen gauge length fibres [ 8] whilst values of 3. 5 [10]and 2.51 [11] have been with large specimens having reduced strengths com- suggested for Bm. Although values of Bm and Kic have not pared to smaller specimens. It is thus necessary when been determined for Tyranno Si-Ti-C-O fibres, they comparing S, for different data sets to normalise differ might be expected to be similar to those for Nicalon" ent gauge lengths to a standard gauge length, Lo, by fibres as the microstructure and chemistry of Tyranno" using the relationship [9] and Nicalon"fibres are alike in many respects. It has recently been shown that Eq. (1) provides only where S, is the predicted value of S, at the standard gauge length, L', and Lo is the gauge length for a spe cimen with Weibull strength parameters So and m It has been suggested that Lo=10-3 m is an appro priate standard gauge length for CMCs as this is the order of the fibre pull-out length. The reason why fibre th is significant for CMCs is that fo posites that fail as a result of multiple matrix cracking as is the case for most“good”CMCs, the gradual transfer of stress from matrix to fibre away from the sam2 4 um 1.1 (b) E0.8 0.7 0.6 um Weibull modulus . m Fig. 1. Scanning electron micrographs illustrating a typical fractu Fig. 2. Relationship between Weibull scale parameters(S, mr) mirror observed on the surface of Tyranno" Si-TiC-O fibres:(a) mined from fracture mirror data and underlying fibre strength general view, and(b) detailed view of fracture mirror. meters(So, m)[5
which is vulnerable to degradation during ®bre and composite processing. That fracture initiation does not normally occur within the ®bre bulk suggests the average ¯aw size and/or ¯aw density to be signi®cantly larger on the ®bre surface compared to that within the ®bre bulk. A common equation linking mirror radius, rm, to ®bre strength, S, is of the form: S Am rm p 2 with Am BmKIC where Am and Bm are empirical constants and KIC is the fracture toughness of the ®bre. A value of KIC 1 MPa m1/2 has been estimated for Nicalon1 SiC-based ®bres [8] whilst values of 3.5 [10] and 2.51 [11] have been suggested for Bm. Although values of Bm and KIC have not been determined for Tyranno1 Si±Ti±C±O ®bres, they might be expected to be similar to those for Nicalon1 ®bres as the microstructure and chemistry of Tyranno1 and Nicalon1 ®bres are alike in many respects. It has recently been shown that Eq. (1) provides only approximations for So and m with the correct form of the relationship being [5]: F 1 ÿ eÿ S S m 3 with S and m being the uncorrected Weibull strength parameters. In order to obtain the relevant in situ ®bre strength parameters, So and m, it is required to use correction factors that have been determined [5] to be of the form shown in Fig. 2. It may be observed from Fig. 2 that Eqs. (1) and (3) gives similar results when m 4 but values for So and m obtained from Eq. (1) for the case of m < 4 will respectively underestimate and overestimate actual values. It should be noted that Eq. (3) assumes no knowledge of the specimen gauge length. However, the value of S obtained depends strongly on the specimen gauge length with large specimens having reduced strengths compared to smaller specimens. It is thus necessary when comparing So for dierent data sets to normalise dierent gauge lengths to a standard gauge length, L0 o, by using the relationship [9]: S0 o So Lo L0 o 1 m 4 where S0 o is the predicted value of So at the standard gauge length, L0 o, and Lo is the gauge length for a specimen with Weibull strength parameters So and m. It has been suggested that L0 o 10ÿ3 m is an appropriate standard gauge length for CMCs as this is the order of the ®bre pull-out length. The reason why ®bre pull-out length is signi®cant for CMCs is that, for composites that fail as a result of multiple matrix cracking (as is the case for most ``good'' CMCs), the gradual transfer of stress from matrix to ®bre away from the Fig. 1. Scanning electron micrographs illustrating a typical fracture mirror observed on the surface of Tyranno1 Si±Ti±C±O ®bres: (a) general view, and (b) detailed view of fracture mirror. Fig. 2. Relationship between Weibull scale parameters (S, m) determined from fracture mirror data and underlying ®bre strength parameters So; m [5]. 802 I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811
I.. Davies et al. Composites Science and Technology 59(1999)801-811 crack front means that the fibres are effectively tested at reason, the present paper will investigate in situ fibre a gauge length, 8c, independent of the composite speci- properties for composites based on the SiC/SiC system men gauge length and deduced to have the form [6, 12]: after tensile testing up to 1380 C in vacuum and air h) rS (m) 2. Experimental procedure with The composite investigated in this report was on the siC/Sic system with Tyranno" Si-Ti-C-O (m)≈0716+m6f0rm≥1 fibres being utilised that had been surface-modi order to improve interface properties. Ex situ strength data for standard and surface-modified LoxM fibres [14] where() is the mean fibre pull-out length after com- has been presented in Table I whilst Auger depth pro- ure files at the fibre surface [15] are shown in Fig. 3. Normalising So values for in situ fibres would be Increases in So and m following surface modification xpected to give reasonably accurate values for S, at were presumably due to a reduction in the average Lo= 10-m as Sc= Lo g Lo. However, some caution should be used when Lo > L as is often the case for ex situ fibre strength where L, may be typically 25x10-3 Table 1 m. In such cases, researchers [13] have discovered Ex situ strength parameters for TyrannoSH-Ti-C-O fibres calculated from data in Ref (14 appreciable differences(up to 50%) when comparing values of So obtained from experimental gauge lengths Tyranno Si-Ti-C-0 So(GPa) Le (le=lo) and those predicted from Eq.(4)for LoxM fibre (GPa) gauge length of Lo after testing at a gauge length LoStandard 3.32(±0.01)10.84(±0.32)4.47(±0.13) (Le》Le) Surface-modified4.23(±0.01)12.54(±0.39)5.47(±0.17) Once values of So and m have been established, they may be used to estimate the ultimate tensile strength of a composite, SUTs, that failed through multiple matrix cracking using the following equation [5] (a) 2+1 m+2 m+2 where V is the fibre volume fraction in the direction of Comparison between experimental and predicted alues for suts has met with some success, though deviations of 20-30% have been seen in other systems [9]. From Eq.(6)it may be observed that composite strength is determined essentially by the fibre archi- 100 tecture, V, fibre properties, So, and m, and also the interfacial shear strength, T, which can be derived from (b) rearrangement of Eq (5)such that [6] ri(m)So T 4(h Thus, in situ observation of fibres makes possible the derivation of s nd t that composite properties. Although the theory discussed above may be useful in correlating microscopic and macroscopic properties, only 020406080100120140160 limited data exist in situ observation of fibre properties in CMCs. In particular, only a few studies have Distance from fibre surface(nm) determined in situ fibre properties for CMCs after testing Fig 3. Auger depth profiles for Tyranno* LoxM Si-Ti-C-o fibres in different atmospheres at elevated temperature. For that (a)standard, and(b) surface-modified [15]
crack front means that the ®bres are eectively tested at a gauge length, c, independent of the composite specimen gauge length and deduced to have the form [6,12]: c 4hhi l m rSo 5 with l m 0:716 1:36 m0:6 for m51 where hhi is the mean ®bre pull-out length after composite failure. Normalising So values for in situ ®bres would be expected to give reasonably accurate values for S 0 o at L0 o 10ÿ3 m as c Lo L0 o. However, some caution should be used when Lo L0 o as is often the case for ex situ ®bre strength where Lo may be typically 2510ÿ3 m. In such cases, researchers [13] have discovered appreciable dierences (up to 50%) when comparing values of So obtained from experimental gauge lengths Le Le L0 o and those predicted from Eq. (4) for a gauge length of L0 o after testing at a gauge length Lo Le Le. Once values of So and m have been established, they may be used to estimate the ultimate tensile strength of a composite, SUTS, that failed through multiple matrix cracking using the following equation [5]: SUTS VfSo 2 m 2 1 m1 m 1 m 2 6 where Vf is the ®bre volume fraction in the direction of loading. Comparison between experimental and predicted values for SUTS has met with some success, though deviations of 20±30% have been seen in other systems [9]. From Eq. (6) it may be observed that composite strength is determined essentially by the ®bre architecture, Vf, ®bre properties, So, and m, and also the interfacial shear strength, , which can be derived from rearrangement of Eq. (5) such that [6]: rl mS0 4hhi 7 Thus, in situ observation of ®bres makes possible the derivation of So, m, and , that eectively control most composite properties. Although the theory discussed above may be useful in correlating microscopic and macroscopic properties, only limited data exist concerning in situ observation of ®bre properties in CMCs. In particular, only a few studies have determined in situ ®bre properties for CMCs after testing in dierent atmospheres at elevated temperature. For that reason, the present paper will investigate in situ ®bre properties for composites based on the SiC/SiC system after tensile testing up to 1380C in vacuum and air. 2. Experimental procedure The composite investigated in this report was based on the SiC/SiC system with Tyranno1 Si±Ti±C±O LoxM ®bres being utilised that had been surface-modi®ed in order to improve interface properties. Ex situ strength data for standard and surface-modi®ed LoxM ®bres [14] has been presented in Table 1 whilst Auger depth pro- ®les at the ®bre surface [15] are shown in Fig. 3. Increases in So and m following surface modi®cation were presumably due to a reduction in the average Table 1 Ex situ strength parameters for Tyranno1 Si±Ti±C±O ®bres calculated from data in Ref. [14] Tyranno Si±Ti±C±O LoxM ®bre So (GPa) m S 0 o L0 o 10ÿ3 m (GPa) Standard 3.32 (0.01) 10.84 (0.32) 4.47 (0.13) Surface-modi®ed 4.23 (0.01) 12.54 (0.39) 5.47 (0.17) Fig. 3. Auger depth pro®les for Tyranno1 LoxM Si±Ti±C±O ®bres: (a) standard, and (b) surface-modi®ed [15]. I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811 803
I.. Davies et al. Composites Science and Technology 59(1999)801-811 defect size at the fibre surface caused by the surface tested specimens were similar at room temperature and reatment. The chemical depth profile in Fig 3(b)shows 1200oC(400 MPa), followed by a gradual decrease of the surface-modified LoxM fibre to possess a 10 nm approximately 50% until 1380oC. Tensile strain to fail- SiOx-rich layer at the surface surrounding an inner 40 ure was approximately 1. 2%for specimens tested at nm carbon-rich layer. The justification for such a sur- room temperature and up to 1380.C in vacuum face chemistry is that the outer SiOx- rich layer will bond Although a suitable strain measurement technique for strongly to the matrix with the 40 nm carbon layer specimens tested in air at elevated temperature was not effectively acting as the fibre/matrix interface. In this available, values of typically <0.05% would be expec- case, the fibres would be expected to fail at the carbon ted when considering the reduced tensile strength of layer within the fibre surface [16] rather than at the specimens tested in air at elevated temperature(Fig 3) actual fibre/matrix interface that is usually the case in However, it should be emphasised that specimens inves- CMCS tigated in this report possessed no oxidation protection Prior to matrix densification the fibres were woven system. Tensile tests at elevated temperature in air for nto an orthogonal 3-D structure with fibre volume similar specimens, but surface sealed using a proprietary fractions in the x, y, and z directions being 0.19, 0.19, glass-based technique, indicate tensile strength to be simi- and 0.02, respectively. Weaving technology was utilised lar to that for unsealed specimens tested in vacuum [4] for this composite as it possesses great versatility as Following tensile failure, specimen fracture surfaces regards shape and dimension control [17], which should were investigated using a JEOL JSM-6300F scanning reduce machining costs in final applications electron microscope(SEM). The general nature of the Matrix densification consisted of a polymer similar to composite fracture surface was assessed whilst a detailed polytitanocarbosilane(PTCS) that was impregnated study of fibre pull-out behaviour is reported elsewhere into the fibre preform and pyrolysed to form a matrix [19]. Fibre fracture surfaces were characterised, with the similar in chemistry to that of the fibres. Eight cycles of fracture mode and flaw mirror radius being noted impregnation and pyrolysis were required to maximise Between 100 and 800 fibres were examined for each test the composite density [16] condition whilst in situ fibre properties were derived Tensile testing was undertaken with the specimen with the aid of Eqs.(2H(7) axis parallel to the loading direction at temperatures between room temperature and 1380 C in vacuum and from room temperature to 1200 C in air For specimens 3. Results and discussion tested at elevated temperature, heating rates between 300C and the specified test temperature were approxi- 3.I. Microstructural observation mately 0. 75C s-I whilst failed specimens were furnace cooled at an estimated rate above 1000c of 3. 3C 3. .1. Room temperature and 1200 C in vacuum The total time spent at the test temperature was believed The majority of fibres within specimens tested at to be approximately 600 s-further experimental details room temperature and 1200C in vacuum possessed being given elsewhere [15, 18 fine-grained structures with a well-defined mirror zone The tensile strengths [2] of specimens examined in this and crackle region that originated at the fibre surface report are presented in Fig 4. The strengths of vacuum-(as indicated in Fig. 1). Surface flaws thus controlled fibre strength under these conditions with relatively few fibres failing as a result of internal flaws for the room temperature case. The percentage of fibres failing at internal flay her for the 1200@C case compared to room temperature as indicated in Table 2. However, this appeared to have no effect on composite tensile ￡30 strength shown in Fig. 4, which would be consistent with fibres failing at the surface and in the bulk having similar strength distributions; this will be the topic of ■ vacuum further research. An example of a fibre that failed 100 through an internal flaw during testing at 1200oC in vacuum is presented in Fig. 5 with a mirror zone also being present around the flaw. From Table 2 it can be 10001100120013001400 seen that the failure mode could not be determined for Test temperature(C) 13% of fibres that failed at room temperature Although these fibres had no fracture mirrors visible, it Fig. 4. Tensile strength [2] of SiC/SiC-based specimens tested in was concluded that they probably did fail due to surface vacuum and air up to 1380.C. Note the non linear temperature scale flaws and may have represented the strongest population
defect size at the ®bre surface caused by the surface treatment. The chemical depth pro®le in Fig. 3(b) shows the surface-modi®ed LoxM ®bre to possess a 10 nm SiOx-rich layer at the surface surrounding an inner 40 nm carbon-rich layer. The justi®cation for such a surface chemistry is that the outer SiOx-rich layer will bond strongly to the matrix with the 40 nm carbon layer eectively acting as the ®bre/matrix interface. In this case, the ®bres would be expected to fail at the carbon layer within the ®bre surface [16] rather than at the actual ®bre/matrix interface that is usually the case in CMCs. Prior to matrix densi®cation the ®bres were woven into an orthogonal 3-D structure with ®bre volume fractions in the x, y, and z directions being 0.19, 0.19, and 0.02, respectively. Weaving technology was utilised for this composite as it possesses great versatility as regards shape and dimension control [17], which should reduce machining costs in ®nal applications. Matrix densi®cation consisted of a polymer similar to polytitanocarbosilane (PTCS) that was impregnated into the ®bre preform and pyrolysed to form a matrix similar in chemistry to that of the ®bres. Eight cycles of impregnation and pyrolysis were required to maximise the composite density [16]. Tensile testing was undertaken with the specimen y-axis parallel to the loading direction at temperatures between room temperature and 1380C in vacuum and from room temperature to 1200C in air. For specimens tested at elevated temperature, heating rates between 300C and the speci®ed test temperature were approximately 0.75C sÿ1 whilst failed specimens were furnacecooled at an estimated rate above 1000C of 3.3C sÿ1 . The total time spent at the test temperature was believed to be approximately 600 sÐfurther experimental details being given elsewhere [15,18]. The tensile strengths [2] of specimens examined in this report are presented in Fig. 4. The strengths of vacuumtested specimens were similar at room temperature and 1200C (400 MPa), followed by a gradual decrease of approximately 50% until 1380C. Tensile strain to failure was approximately 1.2% for specimens tested at room temperature and up to 1380C in vacuum. Although a suitable strain measurement technique for specimens tested in air at elevated temperature was not available, values of typically <0.05% would be expected when considering the reduced tensile strength of specimens tested in air at elevated temperature (Fig. 3). However, it should be emphasised that specimens investigated in this report possessed no oxidation protection system. Tensile tests at elevated temperature in air for similar specimens, but surface sealed using a proprietary glass-based technique, indicate tensile strength to be similar to that for unsealed specimens tested in vacuum [4]. Following tensile failure, specimen fracture surfaces were investigated using a JEOL JSM-6300F scanning electron microscope (SEM). The general nature of the composite fracture surface was assessed whilst a detailed study of ®bre pull-out behaviour is reported elsewhere [19]. Fibre fracture surfaces were characterised, with the fracture mode and ¯aw mirror radius being noted. Between 100 and 800 ®bres were examined for each test condition whilst in situ ®bre properties were derived with the aid of Eqs. (2)±(7). 3. Results and discussion 3.1. Microstructural observation 3.1.1. Room temperature and 1200C in vacuum The majority of ®bres within specimens tested at room temperature and 1200C in vacuum possessed ®ne-grained structures with a well-de®ned mirror zone and crackle region that originated at the ®bre surface (as indicated in Fig. 1). Surface ¯aws thus controlled ®bre strength under these conditions with relatively few ®bres failing as a result of internal ¯aws for the room temperature case. The percentage of ®bres failing at internal ¯aws was higher for the 1200C case compared to room temperature as indicated in Table 2. However, this appeared to have no eect on composite tensile strength shown in Fig. 4, which would be consistent with ®bres failing at the surface and in the bulk having similar strength distributions; this will be the topic of further research. An example of a ®bre that failed through an internal ¯aw during testing at 1200C in vacuum is presented in Fig. 5 with a mirror zone also being present around the ¯aw. From Table 2 it can be seen that the failure mode could not be determined for 13% of ®bres that failed at room temperature. Although these ®bres had no fracture mirrors visible, it was concluded that they probably did fail due to surface ¯aws and may have represented the strongest population Fig. 4. Tensile strength [2] of SiC/SiC-based specimens tested in vacuum and air up to 1380C. Note the non-linear temperature scale. 804 I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811
I.. Davies et al. Composites Science and Technolog y 59(1999)801-811 Table 2 In situ fibre fracture surface characteristics for Tyranno* Si-THC-O fibres Test condition Fracture mirror Fracture mirror Flat fracture Undetermined Total (surface)%(N (internal)%(M surface %( %(N 75(114) 13(19) 100(151) 200°C/ vacuum 10(11) l(1 00(112) 1300° C/vacuum 3(5) 1200°C/air 100(788) a 3 um 4 um (b) (b) lm 2 um Fig. 5. Scanning electron micrographs illustrating the fracture surface Tyranno"ShTHC-O fibre tested in situ at 1200C in vacuum tha Fig. 6. Scanning electron micrographs illustrating the fracture surface failed due to an internal flaw:(a) general view, and (b) detailed view of f a Tyranno"Si-Ti-C-O fibre tested in situ at 1300.C in vacuum that fracture mirror failed due to a surface flaw:(a) general view, and (h) detailed view of fracture mirror of fibres that"shattered"upon failure. Likewise, fibres evolution of Co, with associated decreased tensile characterised as"flat fracture surface"in Table 2 for strength [21]. That this phenomenon was not observed the room temperature test condition possessed smooth in current specimens until 1300C was believed, in part features that suggested them to be the weakest group of to be due to the relatively short time above 1000C during testing compared to previous resea observed in Fig. 6 is a fracture mirror that was typical 3..2.l300° c in vacu of those seen in 84% of fibres at 1300C in vacuum Whereas fibres tested at 1200'C in vacuum indicated (Table 2). The number of fibres that failed due to inter- no obvious grain growth compared to room tempera- nal flaws was about 12% and only slightly larger than ture fibres, those tested at 1300@C exhibited significant that at 1200C (10%)indicating that the voids observed grain growth and finely distributed voids(Fig. 6). Such in Fig. 6 were not large enough to significantly further a phenomenon is known to occur in SiC-based fibres challenge the surface flaw-induced failure mode follow- held in an inert atmosphere above 1000.C[20] and ing the initial increase between room temperature and attributed to chemical decomposition of the fibre and 1200oC
of ®bres that ``shattered'' upon failure. Likewise, ®bres characterised as ``¯at fracture surface'' in Table 2 for the room temperature test condition possessed smooth features that suggested them to be the weakest group of ®bres. 3.1.2. 1300C in vacuum Whereas ®bres tested at 1200C in vacuum indicated no obvious grain growth compared to room temperature ®bres, those tested at 1300C exhibited signi®cant grain growth and ®nely distributed voids (Fig. 6). Such a phenomenon is known to occur in SiC-based ®bres held in an inert atmosphere above 1000C [20] and attributed to chemical decomposition of the ®bre and evolution of CO, with associated decreased tensile strength [21]. That this phenomenon was not observed in current specimens until 1300C was believed, in part, to be due to the relatively short time above 1000C during testing compared to previous researchers. Also observed in Fig. 6 is a fracture mirror that was typical of those seen in 84% of ®bres at 1300C in vacuum (Table 2). The number of ®bres that failed due to internal ¯aws was about 12% and only slightly larger than that at 1200C (10%) indicating that the voids observed in Fig. 6 were not large enough to signi®cantly further challenge the surface ¯aw-induced failure mode following the initial increase between room temperature and 1200C. Table 2 In situ ®bre fracture surface characteristics for Tyranno1 Si±Ti±C±O ®bres Test condition Fracture mirror (surface) % (N) Fracture mirror (internal) % (N) Flat fracture surface % (N) Undetermined % (N) Total % (N) Room temperature 75 (114) 1 (2) 11 (16) 13 (19) 100 (151) 1200C/vacuum 89 (100) 10 (11) 0 (0) 1 (1) 100 (112) 1300C/vacuum 84 (132) 12 (19) 1 (1) 3 (5) 100 (157) 1100C/air 18 (128) 0 (0) 82 (570) 0 (0) 100 (698) 1200C/air 18 (64) 0 (0) 92 (724) 0 (0) 100 (788) Fig. 6. Scanning electron micrographs illustrating the fracture surface of a Tyranno1 Si±Ti±C±O ®bre tested in situ at 1300C in vacuum that failed due to a surface ¯aw: (a) general view, and (h) detailed view of fracture mirror. Fig. 5. Scanning electron micrographs illustrating the fracture surface of a Tyranno1 Si±Ti±C±O ®bre tested in situ at 1200C in vacuum that failed due to an internal ¯aw: (a) general view, and (b) detailed view of fracture mirror. I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811 805
I.. Davies et al. Composites Science and Technology 59(1999)801-811 (b) 3 um 3 um c)2(d) 4 um Fig. 7. Scanning electron micrographs illustrating fracture surfaces of Tyranno" Si-TiC-O fibres tested in situ at 1350 and 1380C: (a) general view of fibre exhibiting"debris"on the fracture surface(1350 C), (b) general view of a fibre that failed at 1380C, (c)detailed view of a fibre that failed at 380C illustrating the grain structure, and (d) general view of a fibre which shows evidence of a fracture mirror centred on a surface flaw on the right hand side of the fibre(1380C) 3.1.3. 1350 and 1380.C in vacuum accurately measured. For this reason, fracture mirror Typical fracture surfaces for fibres within composite data was only obtained for fibres within specimens tes specimens tested at 1350 and 1380C in vacuum have ted up to 1300.C in vacuum. been presented in Fig. 7. Although broadly similar to fibres tested at 1300 C in vacuum, those at 1350 C pos- 3.1.4. 1100 and 1200C in air ssed a coarser grain structure together with additional Fracture surfaces for fibres in composite specimens features on the fracture surface that at first sight tested at 1000 and 1 C in air were essentially simi- appeared to be debris [Fig. 7(a)]. However, further lar-the main feature being that <20% of fibres investigation showed similar"debris"to be present on showed evidence of fracture mirrors. Instead, the vast the fracture surface of nearly all fibres tested at 1350.c majority of fibres possessed smooth fracture surfaces in vacuum, suggesting these features to be an integral accompanied by negligible fibre pull-out(Fig 8). Simi- part of the fibre structure. One point of note is that suc lar characteristics have been observed in previous features were not generally observed in fibres tested at CMCs following testing in air at elevated temperature the test temperatures on either side of 1350.C, i.e. 1300 and attributed to oxidation and removal of the carbon and 1380C, implying that temperature was not a sig- layer present at the fibre /matrix interface [22] and its nificant factor in their appearance. Further work replacement with a silica layer resulting from oxidation required to determine the exact nature of these features. of the fibre surface[8, 23]. Such a phenomenon would be The grain structure of fibres tested at 1380.C in expected to dramatically increase t due to the replace- vacuum [Fig. 7(b d)] was similar to that observed at ment of carbon, whose friction coefficient, u, is esti- 1200C apart from a noticeable increase in average mated to be 0.01 [24], with SiO2 that has u A03-0.8 rain size and attributed to the increased test tempera ture. Although some evidence of fracture mirrors was Previous work has shown matrix cracks perpend observed in the 1350 and 1380 C fibres [Fig. 7(d)]. the cular to the fibre to only be deflected at the fibre matrix/ features were generally not sufficiently distinct to be interface when the following inequality is satisfied [26]
3.1.3. 1350 and 1380C in vacuum Typical fracture surfaces for ®bres within composite specimens tested at 1350 and 1380C in vacuum have been presented in Fig. 7. Although broadly similar to ®bres tested at 1300C in vacuum, those at 1350C possessed a coarser grain structure together with additional features on the fracture surface that at ®rst sight appeared to be debris [Fig. 7(a)]. However, further investigation showed similar ``debris'' to be present on the fracture surface of nearly all ®bres tested at 1350C in vacuum, suggesting these features to be an integral part of the ®bre structure. One point of note is that such features were not generally observed in ®bres tested at the test temperatures on either side of 1350C, i.e. 1300 and 1380C, implying that temperature was not a signi®cant factor in their appearance. Further work is required to determine the exact nature of these features. The grain structure of ®bres tested at 1380C in vacuum [Fig. 7(b)±(d)] was similar to that observed at 1200C apart from a noticeable increase in average grain size and attributed to the increased test temperature. Although some evidence of fracture mirrors was observed in the 1350 and 1380C ®bres [Fig. 7(d)], the features were generally not suciently distinct to be accurately measured. For this reason, fracture mirror data was only obtained for ®bres within specimens tested up to 1300C in vacuum. 3.1.4. 1100 and 1200C in air Fracture surfaces for ®bres in composite specimens tested at 1000 and 1100C in air were essentially similarÐthe main feature being that <20% of ®bres showed evidence of fracture mirrors. Instead, the vast majority of ®bres possessed smooth fracture surfaces accompanied by negligible ®bre pull-out (Fig. 8). Similar characteristics have been observed in previous CMCs following testing in air at elevated temperature and attributed to oxidation and removal of the carbon layer present at the ®bre/matrix interface [22] and its replacement with a silica layer resulting from oxidation of the ®bre surface [8,23]. Such a phenomenon would be expected to dramatically increase due to the replacement of carbon, whose friction coecient, , is estimated to be 0.01 [24], with SiO2 that has 0:3±0.8 [25]. Previous work has shown matrix cracks perpendicular to the ®bre to only be de¯ected at the ®bre matrix/ interface when the following inequality is satis®ed [26]: Fig. 7. Scanning electron micrographs illustrating fracture surfaces of Tyranno1 Si±Ti±C±O ®bres tested in situ at 1350 and 1380C: (a) general view of ®bre exhibiting ``debris'' on the fracture surface (1350C), (b) general view of a ®bre that failed at 1380C, (c) detailed view of a ®bre that failed at 1380C illustrating the grain structure, and (d) general view of a ®bre which shows evidence of a fracture mirror centred on a surface ¯aw on the right hand side of the ®bre (1380C). 806 I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811
I.. Davies et al. Composites Science and Technology 59(1999)801-811 (a) 1.0 Ex situ standard Ex situ surface-modified 40 um 0.0 345678 Fibre strength(GPa) (b) Fig 9. Ex situ and in situ fibre dtrength of Tyranno Si-TiC-0 fibres normalised to a 10-3m gauge length(assuming Am=2.5 Pa m2) after tensile testing at room temperature an estimate and may be subject to revision at a later date. The smaller value of m for the in situ fibres indi- cated a wide distribution of faw sizes to have been introduced during composite manufacture. In situ fibre strength data for SiC-based fibres has been presented in Fig. 10 1. five sets of in data with different specimen test conditions were inves- tigated: (i) room temperature, (ii)1200oC in vacuum, Fig8.Scanning electron micrographs illustrating TyrannoSH-Ti-C-o (iii)1300oC in vacuum,(iv) 1100C in air, and(v) fibres tested in situ at 1100 and 1200'C in air: (a) general view of the 1200 C in air. Fig. 10(a)illustrates in situ strength, s, outer edge of a fibre bundle, and (b)detailed view of a fibre fracture after correcting the fracture mirror parameters using Fig. 2 [5 whilst data in Fig. 10(b) was normalised to a gauge length of Lo=10-3m. For comparison, Weibull (8) strength parameters have been presented in Table 3 for (i) uncorrected(S, m),(ii) corrected fracture mirror where G; and Gr are the critical strain energy release (So). Although Lo =10- m was confirmed to be a sui- rates for interface and fibre, respectively. A low value table gauge length for fibres tested in vacuum(Table 3) for t would also imply G; to be small whereas oxidation as 0.35x10-3<(h)<0.81x10-3m [19], the significantly of the fibre/matrix interface tends to increase t(due to lower (h) for specimens tested in air (0.06 and increased u)and hence increase Gi to the point where 0.07x10-3m) indicates a more appropriate value of La the inequality in Eq(8)is no longer satisfied. At this to have been 10-4 m for these latter specimens. How point, crack deflection at the fibre/matrix interface will ever, in order to accurately compare in situ strength for no longer be energetically favourable, allowing the crack different test conditions there is little option but to nor to continue through the fibre with resultant featureless malise the gauge length to a single value although the fibre fracture surfaces(Fig. &)and flat composite fracture comments above should be kept in mind when later surfaces comparing data between specimens tested in vacuum and that tested in air 3. 2. In situ fibre strength parameters The first point to note from Table 3 is that, to S, the value of So was approximately 1. 5% lower for 3.2.1. Effect of test temper fibres tested at room temperature and vacuum (1200, Fig 9 illustrates fibre strength for standard and sur- 1300C)and 3.5% higher for fibres tested in air(1100, face-modified fibres tested ex situ at room temperature 1200oC)due to differences in m. Likewise, compared [14]. The value of Am in Eq (2) for Tyranno" Si-Ti-C-o to m,, values of m were approximately 2% higher and fibres was estimated to be 2.5 MPa m/2 as this provided 3% lower for fibres tested at room temperature(and a 30% decrease for in situ strength compared to ex situ vacuum) and in air, respectively. The second point to strength as noted in previous ceramic fibre systems. note is that values of s and So generally had associated However, it is accepted that Am=2.5 MPa m/ is only uncertainties in the range 0.3% for room temperature
Gi Gf 4 1 4 8 where Gi and Gf are the critical strain energy release rates for interface and ®bre, respectively. A low value for would also imply Gi to be small whereas oxidation of the ®bre/matrix interface tends to increase (due to increased ) and hence increase Gi to the point where the inequality in Eq. (8) is no longer satis®ed. At this point, crack de¯ection at the ®bre/matrix interface will no longer be energetically favourable, allowing the crack to continue through the ®bre with resultant featureless ®bre fracture surfaces (Fig. 8) and ¯at composite fracture surfaces. 3.2. In situ ®bre strength parameters 3.2.1. Eect of test temperature and atmosphere Fig. 9 illustrates ®bre strength for standard and surface-modi®ed ®bres tested ex situ at room temperature [14]. The value of Am in Eq. (2) for Tyranno1 Si±Ti±C±O ®bres was estimated to be 2.5 MPa m1/2 as this provided a 30% decrease for in situ strength compared to ex situ strength as noted in previous ceramic ®bre systems. However, it is accepted that Am 2:5 MPa m1/2 is only an estimate and may be subject to revision at a later date. The smaller value of m for the in situ ®bres indicated a wide distribution of ¯aw sizes to have been introduced during composite manufacture. In situ ®bre strength data for SiC-based ®bres has been presented in Fig. 10. In total, ®ve sets of in situ data with dierent specimen test conditions were investigated: (i) room temperature, (ii) 1200C in vacuum, (iii) 1300C in vacuum, (iv) 1100C in air, and (v) 1200C in air. Fig. 10(a) illustrates in situ strength, S, after correcting the fracture mirror parameters using Fig. 2 [5] whilst data in Fig. 10(b) was normalised to a gauge length of L0 o 10ÿ3 m. For comparison, Weibull strength parameters have been presented in Table 3 for: (i) uncorrected S; m, (ii) corrected fracture mirror parameters So; m, and (iii) normalised to L0 o 10ÿ3 m (S0 o). Although L0 o 10ÿ3 m was con®rmed to be a suitable gauge length for ®bres tested in vacuum (Table 3) as 0.3510ÿ3 4hhi40.8110ÿ3 m [19], the signi®cantly lower hhi for specimens tested in air (0.06 and 0.0710ÿ3 m) indicates a more appropriate value of L0 o to have been 10ÿ4 m for these latter specimens. However, in order to accurately compare in situ strength for dierent test conditions there is little option but to normalise the gauge length to a single value although the comments above should be kept in mind when later comparing data between specimens tested in vacuum and that tested in air. The ®rst point to note from Table 3 is that, compared to S, the value of So was approximately 1.5% lower for ®bres tested at room temperature and vacuum (1200, 1300C) and 3.5% higher for ®bres tested in air (1100, 1200C)Ðdue to dierences in m. Likewise, compared to m, values of m were approximately 2% higher and 3% lower for ®bres tested at room temperature (and vacuum) and in air, respectively. The second point to note is that values of S and So generally had associated uncertainties in the range 0.3% for room temperature Fig. 9. Ex situ and in situ ®bre dtrength of Tyranno1 Si±Ti±C±O ®bres normalised to a 10ÿ3 m gauge length (assuming Am=2.5 Pa m1/2) after tensile testing at room temperature. Fig. 8. Scanning electron micrographs illustrating Tyranno1 Si±Ti±C±O ®bres tested in situ at 1100 and 1200C in air: (a) general view of the outer edge of a ®bre bundle, and (b) detailed view of a ®bre fracture surface. I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811 807
I.. Davies et al. Composites Science and Technology 59(1999)801-811 1200.C). This would also tend to increase the uncer- 1.0 Room temperature 1200°/ actin tainty for values of S"' and m 0.8 △1300°/ acuum One important point to note is the large difference in 1100°c/air 1200 C/air appearance between Fig. 10(a)and(bthis emphasises that comparison between in situ S, and m data should only be made after normalising to a standard gauge ngth. For example, data taken at different gauge lengths [Fig. 10(a)] indicates fibres tested in air to have 0.2 similar in situ fibre strength characteristics at 1 100 and 1200oC, both of which are approximately 80% that of room temperature and vacuum(1200, 1300oC)data 1.0 Legend:as above However, following normalisation to L=10-3m [ Fig. 10(b)] it is observed that fibres tested in air actually possess different strength characteristics(1200.C data is 9% lower compared to the 1100oC data)and further more. that values are about 35% that of room tem perature and vacuum (1200, 1300C)data. These substantial differences can be attributed to variations in m between specimens (Table 3)when applied to Eq(4) Table 3 indicates that s decreases with increased (b) vacuum. from 3.85 GPa at perature to 3. 18 GPa at 1300C. Conversely, m increa 0.50.7 234568 ses with increased temperature in vacuum, from 4.19 at Fibre strength(GPa) room temperature to 6.56 at 1300.C. One suggest mechanism for these changes in S and m would be Fig. 10. In situ fibre strength of Tyranno"Si-Ti-C-O fibres after interaction between intrinsic (surface)flaws within the tensile testing in vacuum and air up to 1380C: (a) after correction of fibre and voids introduced through heating the fibre to fracture mirror parameters according to Fig. 2 5]. and (b) normalised to a 10-3m gauge length(assuming Am= 2.5 MPa m 2). elevated temperature in vacuum. It is known that for- mation of voids within SiC-based fibres at the tempera ture tested in this report would tend to produce and vacuum fibres and 1.3% for fibres tested in air additional flaws distributed homogeneously within the whereas the uncertainty for m and m was typically 1.5 fibre bulk, and initially with a small average flaw size, and 4.3% for fibres tested at room temperature(and fw, and narrow distribution with largest size fm. with vacuum) and air respectively. This indicates differences increasing temperature and or time above a1000oC the between Eqs. (1)and (3)(S=S, m=m,)to be sig- value of fa will start to increase(Section 3. 1.3)until at nificant for fibres tested at room temperature and in some point the flaw distribution will approach and vacuum(1200, 1300oC)but within experimental error overlap the intrinsic flaw distribution, which has aver for fibres tested in air(1100, 1200 C). However, these age size fa(with fav >> fav) and a minimum value of uncertainty in S" and m depends to a large extent on failed due to flaws of size fmn (i.e. the strongest fibres) the quality of the data, in particular the number of data will start to instead fail due to flaws of size fmx. The points. Furthermore, Eqs. (1)and (3)assume a single result of this will be that the average strength of the flaw distribution which is suspected not to be the case fibres at a given gauge length, So, will decrease (as the for fibres tested in vacuum(1200, 1300C)and air(1100, average flaw size within the fibres will increase) whilst Table 3 In situ strength parameters and fibre/matrix interface shear stress for Tyranno*SH-Ti-C-O fibres after testing in vacuum and air up to 1380.C Test condition S(GPa) So(GPa) )(10-3m)2S(L=10-3m)r(MPa) Room temperature3.11(±0.01)4.18(±0.05)3.09(±0.01)4.19(±005)0.81(±0.02)1.29(±0.02)3.85(±0.13)494(±0.16) 200C/ actin3.12(±001)5.59(±0.10)307(±0.01)5.72(±0.10)0.59(±0.03)1.19(±0.02)3.45(±0.20)6.27(±0.35) 1300°C/ vacuum3.10(±0.01)6.36(±0.11)3.04(±0.01)6.56(±0.11)0.39(±0.04)1.16(±0.02)3.18(±0.34)9.13(±0.96 1100°C/air 240(±0.03)299(±0.14)247(±0.03)291(±0.13)007(±0.01)1.43(±0. 1.37(±0.15)54.75(±5 1120°C/air 2.38(±0.03)277(±0.12)248(±0.03)268(±0.11)0.06(±0.01)1.47(±0061.26(±0.18)60.50(±8.43) Note: Values of S, So and S, were calculated assuming Am=2.5 MPa m 2, whilst values of (h) have been taken from Ref. 2
and vacuum ®bres and 1.3% for ®bres tested in air whereas the uncertainty for m and m was typically 1.5 and 4.3% for ®bres tested at room temperature (and vacuum) and air respectively. This indicates dierences between Eqs. (1) and (3) (S S, m m) to be signi®cant for ®bres tested at room temperature and in vacuum (1200, 1300C) but within experimental error for ®bres tested in air (1100, 1200C). However, these conclusions cannot be applied generally to CMCs as the uncertainty in S and m depends to a large extent on the quality of the data, in particular the number of data points. Furthermore, Eqs. (1) and (3) assume a single ¯aw distribution which is suspected not to be the case for ®bres tested in vacuum (1200, 1300C) and air (1100, 1200C). This would also tend to increase the uncertainty for values of S and m. One important point to note is the large dierence in appearance between Fig. 10(a) and (b)Ðthis emphasises that comparison between in situ So and m data should only be made after normalising to a standard gauge length. For example, data taken at dierent gauge lengths [Fig. 10(a)] indicates ®bres tested in air to have similar in situ ®bre strength characteristics at 1100 and 1200C, both of which are approximately 80% that of room temperature and vacuum (1200, 1300C) data. However, following normalisation to L0 o 10ÿ3 m [Fig. 10(b)] it is observed that ®bres tested in air actually possess dierent strength characteristics (1200C data is 9% lower compared to the 1100C data) and furthermore, that values are about 35% that of room temperature and vacuum (1200, 1300C) data. These substantial dierences can be attributed to variations in m between specimens (Table 3) when applied to Eq. (4). Table 3 indicates that S0 o decreases with increased temperature in vacuum, from 3.85 GPa at room temperature to 3.18 GPa at 1300C. Conversely, m increases with increased temperature in vacuum, from 4.19 at room temperature to 6.56 at 1300C. One suggested mechanism for these changes in S 0 o and m would be interaction between intrinsic (surface) ¯aws within the ®bre and voids introduced through heating the ®bre to elevated temperature in vacuum. It is known that formation of voids within SiC-based ®bres at the temperature tested in this report would tend to produce additional ¯aws distributed homogeneously within the ®bre bulk, and initially with a small average ¯aw size, f a , and narrow distribution with largest size f max . With increasing temperature and/or time above 1000C the value of f a will start to increase (Section 3.1.3) until at some point the ¯aw distribution will approach and overlap the intrinsic ¯aw distribution, which has average size f a i (with f a i f a ) and a minimum value of f min i , i.e. f max 5f min i . At this point, ®bres that originally failed due to ¯aws of size f min i (i.e. the strongest ®bres) will start to instead fail due to ¯aws of size f max . The result of this will be that the average strength of the ®bres at a given gauge length, S 0 o, will decrease (as the average ¯aw size within the ®bres will increase) whilst Table 3 In situ strength parameters and ®bre/matrix interface shear stress for Tyranno1 Si±Ti±C±O ®bres after testing in vacuum and air up to 1380C Test condition S (GPa) m So (GPa) m hhi (10ÿ3 m) l S 0 o L0 o 10ÿ3 m (GPa) (MPa) Room temperature 3.11 (0.01) 4.18 (0.05) 3.09 (0.01) 4.19 (0.05) 0.81 (0.02) 1.29(0.02) 3.85 (0.13) 4.94 (0.16) 1200C/vacuum 3.12 (0.01) 5.59 (0.10) 3.07 (0.01) 5.72 (0.10) 0.59 (0.03) 1.19 (0.02) 3.45 (0.20) 6.27 (0.35) 1300C/vacuum 3.10 (0.01) 6.36 (0.11) 3.04 (0.01) 6.56 (0.11) 0.39 (0.04) 1.16 (0.02) 3.18 (0.34) 9.13 (0.96) 1100C/air 2.40 (0.03) 2.99 (0.14) 2.47 (0.03) 2.91 (0.13) 0.07 (0.01) 1.43 (0.07) 1.37 (0.15) 54.75 (5.40) 1120C/air 2.38 (0.03) 2.77 (0.12) 2.48 (0.03) 2.68 (0.11) 0.06 (0.01) 1.47 (0.06) 1.26 (0.18) 60.50 (8.43) Note: Values of S, So and S0 o were calculated assuming Am 2:5 MPa m1/2, whilst values of hhi have been taken from Ref. 2. Fig. 10. In situ ®bre strength of Tyranno1 Si±Ti±C±O ®bres after tensile testing in vacuum and air up to 1380C: (a) after correction of fracture mirror parameters according to Fig. 2 [5], and (b) normalised to a 10ÿ3 m gauge length (assuming Am 2:5 MPa m1/2). 808 I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811
I.. Davies et al. Composites Science and Technology 59(1999)801-811 the value of m will increase as the range of flaw sizes (i.e 1.2 fia-fm) decreases ■ Centre of fibre bundle Also observed from table 3 is that s and m were 10° Edge of fibre bundle reduced dramatically after exposure to air at 1100 and 1200C, in agreement with previous researchers[8, 27] and in conjunction with microstructural observations 吉06 outlined in section 31. 4. This was believed due to decomposition and oxidation damage to the fibres. More E04 specifically, it was believed that oxygen reacted with the Sic-based fibre with the following mechanisms [28]: 0.0 2SiC+302→2SiO2+2CO 121· Centre of fibre bundle 1.0 0 Edge of fibre bundle SICOy +(/2-y/2+1)O2- SiO2+ XCO (10) 0.8 2C+O2→2CO 9 0.6 The result of these reactions is the evolution of co from 0.4 the fibre and formation by diffusion of a Sio, layer, of 0.2 thickness y, at the fibre surface. It has been shown pre- (b) viously that the sioz layer acts as a defect of size y 0.0 within the fibre, which may significantly reduce the fibre 3456789 strength. Previous work had found the value of y to be 2 um after 1.08×10° s of exposure to650830° C in an Fibre strength(GPa oxidising environment [28] suggesting an average Fig. Il In situ strength(assuming Am=2.5 MPa m'n)of Tyranno" growth rate, y,, of 2x10-12 m s-l. Even taking into Si-Ti-C-O fibres for different regions within a single fibre bundle after account the higher temperatures used for the current tensile testing:(a)1300C in vacuum, and(b)1100C in air materials and that y' would probably have its highes value upon initial exposure, the short time of exposure The difference in behaviour between test conditions at 1100 and 1200@C in air for the current fibres indicates was attributed to ingression of oxygen into specimens y to be in the order of nanometres [28]. Such a small tested at 1100 and 1200oC in air. A suggested mechan value of y would be unlikely to reduce Se by the extent ism to explain the spatial dependence of in situ strength seen in Table 3 suggesting another phenomena to be characteristics for fibres at 1100 and 1200oC in air relies responsible for the observed rapid decrease in Sn such on the fact that fibres in the central portion along the fibre bundle main axis have effectively two sides exposed to the fibre bundle perimeter whereas fibres at the end 3.2. 2. Effect of fibre position within the fibre bundle of the fibre bundle main axis have three sides exposed to Data presented in Section 3. 2. 1 referred to the com- the fibre bundle perimeter. If it were to be assumed that bination of fibres measured at the end and central por- oxygen entered the fibre bundle mainly from the fibre tion of the fibre bundle main axis. However, Fig. ll bundle perimeter then it would be expected that regions illustrates in situ strength data for fibres separated into with the highest fraction of fibres close to the bundle two groups: (i) those located at the end, and (ii) those in perimeter would show a larger degree of oxidation the central portion, of the fibre bundle main axis From damage, at least initially. Such an assumption would Fig. 11(a) it may be observed that fibres tested at explain the lack of a spatial dependence for fibres in 1300 C in vacuum showed similar in situ strength char- specimens tested in vacuum but a significant difference acteristics at the end of the fibre bundle main axis(and in specimens tested in air. Although previous authors also the case for room temperature and 1200oC in have suggested that oxygen ingression for I-D compo- vacuum)whereas Fig. 11(b) shows in situ fibre strength sites occurs mainly along the length of individual fibre characteristics to be different for the two regions within matrix interfaces [22, 23] it was believed in the present he specimen tested at 1100C in air. Thus, fibres in specimens(which have a complex 3-D structure) that specimens tested at room temperature up to 1300C in oxygen entered into fibre bundles mainly via the fibre vacuum show no large-scale correlation between in situ bundle perimeter. The main evidence for such a scenario fibre strength characteristics and fibre bundle position was that fibre pull-out length and probability of fracture whereas fibres in specimens tested at 1100@C in air show mirror occurrence increased with distance away from the a significant spatial dependence fibre bundle perimeter [19, 29]. Although it was stated
the value of m will increase as the range of ¯aw sizes (i.e. f max i ÿ f min ) decreases. Also observed from Table 3 is that S0 o and m were reduced dramatically after exposure to air at 1100 and 1200C, in agreement with previous researchers [8,27] and in conjunction with microstructural observations outlined in Section 3.1.4. This was believed due to decomposition and oxidation damage to the ®bres. More speci®cally, it was believed that oxygen reacted with the SiC-based ®bre with the following mechanisms [28]: 2SiC 3O2 ! 2SiO2 2CO 9 SiCxOy x=2 ÿ y=2 1O2 ! SiO2 xCO 10 2C O2 ! 2CO 11 The result of these reactions is the evolution of CO from the ®bre and formation by diusion of a SiO2 layer, of thickness , at the ®bre surface. It has been shown previously that the SiO2 layer acts as a defect of size within the ®bre, which may signi®cantly reduce the ®bre strength. Previous work had found the value of to be 2 mm after 1.08106 s of exposure to 650±830C in an oxidising environment [28] suggesting an average growth rate, 0 , of 210ÿ12 m sÿ1 . Even taking into account the higher temperatures used for the current materials and that 0 would probably have its highest value upon initial exposure, the short time of exposure at 1100 and 1200C in air for the current ®bres indicates to be in the order of nanometres [28]. Such a small value of would be unlikely to reduce S 0 o by the extent seen in Table 3 suggesting another phenomena to be responsible for the observed rapid decrease in S 0 o such as thermal decomposition. 3.2.2. Eect of ®bre position within the ®bre bundle Data presented in Section 3.2.1 referred to the combination of ®bres measured at the end and central portion of the ®bre bundle main axis. However, Fig. 11 illustrates in situ strength data for ®bres separated into two groups: (i) those located at the end, and (ii) those in the central portion, of the ®bre bundle main axis. From Fig. 11(a) it may be observed that ®bres tested at 1300C in vacuum showed similar in situ strength characteristics at the end of the ®bre bundle main axis (and also the case for room temperature and 1200C in vacuum) whereas Fig. 11(b) shows in situ ®bre strength characteristics to be dierent for the two regions within the specimen tested at 1100C in air. Thus, ®bres in specimens tested at room temperature up to 1300C in vacuum show no large-scale correlation between in situ ®bre strength characteristics and ®bre bundle position whereas ®bres in specimens tested at 1100C in air show a signi®cant spatial dependence. The dierence in behaviour between test conditions was attributed to ingression of oxygen into specimens tested at 1100 and 1200C in air. A suggested mechanism to explain the spatial dependence of in situ strength characteristics for ®bres at 1100 and 1200C in air relies on the fact that ®bres in the central portion along the ®bre bundle main axis have eectively two sides exposed to the ®bre bundle perimeter whereas ®bres at the ends of the ®bre bundle main axis have three sides exposed to the ®bre bundle perimeter. If it were to be assumed that oxygen entered the ®bre bundle mainly from the ®bre bundle perimeter then it would be expected that regions with the highest fraction of ®bres close to the bundle perimeter would show a larger degree of oxidation damage, at least initially. Such an assumption would explain the lack of a spatial dependence for ®bres in specimens tested in vacuum but a signi®cant dierence in specimens tested in air. Although previous authors have suggested that oxygen ingression for l-D composites occurs mainly along the length of individual ®bre/ matrix interfaces [22,23] it was believed in the present specimens (which have a complex 3-D structure) that oxygen entered into ®bre bundles mainly via the ®bre bundle perimeter. The main evidence for such a scenario was that ®bre pull-out length and probability of fracture mirror occurrence increased with distance away from the ®bre bundle perimeter [19,29]. Although it was stated Fig. 11. In situ strength (assuming Am 2:5 MPa m1/2) of Tyranno1 Si±Ti±C±O ®bres for dierent regions within a single ®bre bundle after tensile testing: (a) 1300C in vacuum, and (b) 1100C in air. I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811 809
L. Davies et al sites Science and Technolog y 59(1999)801-811 earlier that the chemistry of fibres and matrix was attributed to oxygen ingression into the fibre bun approximately similar, only slight differences in coeff dles with only a relatively small number of fibres cients of thermal expansion(CTE) for fibre and matrix having an interface weak enough to allow"weak would be required to produce microcracking at the est link"mechanisms to apply interface between large regions of fibres and matrix, 3. Fibre strength characteristics normalised to a 10-3 particularly when taking into account the low value of t m gauge length indicated fibres tested in air at ele expected for these composites. Further investigation vated temperature to have significantly lower concerning the spatial dependence of in situ fibre prop strengths and average Weibull parameter, m, erties within single fibre bundles for these CMCs is compared to the room temperature, 1200C/ vacuum, and 1300 C/vacuum cases. This was attributed to oxygen damage of the fibre together 3.3. Fibre/matrix interface shear stress with oxidation of the fibre/matrix interface 4. Fibre characteristics for fibres tested at room Values of t shown in Table 3 were calculated from mperature, 1200C/vacuum, and 1300C/vacuum Eq(7)using values of (h and r established elsewhere showed no significant difference between fibre [2, 19]. It can be seen that t was approximately 5 MPa located at the end and centre of the fibre bundle for specimens tested at room temperature with a slight main axis. This was not the case for 1100 and increase to 9 MPa at 1300.C in vacuum. These values 1200 C/air specimens where a significant difference fall within the ran f 2-22 MP was observed and attributed to the difference in average of 9.47 MPa(+3. 50)] suggested by researchers ibre numbers closer to the fibre bundle perimeter for Sic-based fibres with carbon interfaces [6, 8, 27, 30- From this it was suggested that the majority of 33] tested at room temperature, and attributed to the oxygen ingress was from the fibre bundle perimeter low shear strength and friction coefficient of the carbon- 5. The fibre/matrix interface shear strength was low for he room temperature specimens and increased The increase in t with test temperature in vacuum was slightly with temperature when tested in vacuum suggested possibly due to the Cte mismatch being possibly due to changes in the thermal mismatch reduced for specimens tested closer to their processing between fibres and matrix. values for specimens tes temperature. An alternative reason might be slight grain ted at 1100 and 1200'C in air were an order of mag- coarsening at the fibre surface. Even for specimens tes- nitude greater than for room temperature specimens ted at 1300c in vacuum the value of t was still <10 indicating a significant degree of oxidation damage MPa which is normally associated with superior fibre at the fibre/matrix interface to have occurred. pull-out and composite mechanical properties However, expos air at ll00and1200°C increased t by an order of magnitude(55-60 MPa) Acknowledgements compared to that at room temperature. These results are within the range suggested for t after exposure to This work was supported by funding from the Scienc oxidising atmosphere at elevated temperature [8]. It is and Technolog possibly the case for the present materials that the rela- authors(IJ D)was supported as a Science and Technol tively short exposure time allowed production of only a ogy Agency Fellow. The authors gratefully acknowledge partial Sio2 layer at the fibre/matrix interface [30] so Y. Nomura and N. Suzuki for help with mechanical that t might be expected to increase further with addi- testing and Professor Curtin for useful discussion tional exposure time References 4. Conclusions Ishikawa T M, Hirokawa T. Stress/strain ehavior of 3D woven posites. in preparation. 1. In situ fibre fracture characteristics were investi 2] M, Hirokawa T. Optical gated for Si-Ti-C-O fibres after tensile testing up scopy of 3-D woven SiC/SiC-based composites. Comp Sci Tech- to 1380%C in vacuum and in air. The general fibre 3 Davies [J, Ishikawa T, Shibuya M, Hirokawa T. Damage char- morphology for specimens tested in vacuum indi- cterisation of 3-D woven SiC/SiC-based composites. In: Pro- cated fibre decomposition and void formation to be ceedings of the 2lst Symposium on Composite Materials, 31 noticeable at 1300.C with specimens tested above October-I November 1996, Toyama, Japan. Tokyo, Japan: The 1300 C not possessing distinct fracture mirrors 2. Specimens tested in air at 1100 and 1200C gen- 4 Davies IJ, Ishikawa T, Shibuya M. Hirokawa T, Gotoh J. Fibre and interfacial properties measured in situ for a 3-D woven erally had flat fracture surfaces with less than 20% Sic/SiC-based composite with glass sealant Composites Part A of the fibres exhibiting fracture mirrors. This was 199930(4):587-91
earlier that the chemistry of ®bres and matrix was approximately similar, only slight dierences in coe- cients of thermal expansion (CTE) for ®bre and matrix would be required to produce microcracking at the interface between large regions of ®bres and matrix, particularly when taking into account the low value of expected for these composites. Further investigation concerning the spatial dependence of in situ ®bre properties within single ®bre bundles for these CMCs is planned. 3.3. Fibre/matrix interface shear stress Values of shown in Table 3 were calculated from Eq. (7) using values of hhi and r established elsewhere [2,19]. It can be seen that was approximately 5 MPa for specimens tested at room temperature with a slight increase to 9 MPa at 1300C in vacuum. These values fall within the range of 2±22 MPa [and close to the average of 9.47 MPa (3.50)] suggested by researchers for SiC-based ®bres with carbon interfaces [6,8,27,30± 33] tested at room temperature, and attributed to the low shear strength and friction coecient of the carbonrich interface. The increase in with test temperature in vacuum was suggested possibly due to the CTE mismatch being reduced for specimens tested closer to their processing temperature. An alternative reason might be slight grain coarsening at the ®bre surface. Even for specimens tested at 1300C in vacuum the value of was still <10 MPa which is normally associated with superior ®bre pull-out and composite mechanical properties. However, exposure to air at 1100 and 1200C increased by an order of magnitude (55±60 MPa) compared to that at room temperature. These results are within the range suggested for after exposure to oxidising atmosphere at elevated temperature [8]. It is possibly the case for the present materials that the relatively short exposure time allowed production of only a partial SiO2 layer at the ®bre/matrix interface [30] so that might be expected to increase further with additional exposure time. 4. Conclusions 1. In situ ®bre fracture characteristics were investigated for Si±Ti±C±O ®bres after tensile testing up to 1380C in vacuum and in air. The general ®bre morphology for specimens tested in vacuum indicated ®bre decomposition and void formation to be noticeable at 1300C with specimens tested above 1300C not possessing distinct fracture mirrors. 2. Specimens tested in air at 1100 and 1200C generally had ¯at fracture surfaces with less than 20% of the ®bres exhibiting fracture mirrors. This was attributed to oxygen ingression into the ®bre bundles with only a relatively small number of ®bres having an interface weak enough to allow ``weakest link'' mechanisms to apply. 3. Fibre strength characteristics normalised to a 10ÿ3 m gauge length indicated ®bres tested in air at elevated temperature to have signi®cantly lower strengths and average Weibull parameter, m, compared to the room temperature, 1200C/ vacuum, and 1300C/vacuum cases. This was attributed to oxygen damage of the ®bre together with oxidation of the ®bre/matrix interface. 4. Fibre characteristics for ®bres tested at room temperature, 1200C/vacuum, and 1300C/vacuum showed no signi®cant dierence between ®bres located at the end and centre of the ®bre bundle main axis. This was not the case for 1100 and 1200C/air specimens where a signi®cant dierence was observed and attributed to the dierence in ®bre numbers closer to the ®bre bundle perimeter. From this it was suggested that the majority of oxygen ingress was from the ®bre bundle perimeter. 5. The ®bre/matrix interface shear strength was low for the room temperature specimens and increased slightly with temperature when tested in vacuum, possibly due to changes in the thermal mismatch between ®bres and matrix. Values for specimens tested at 1100 and 1200C in air were an order of magnitude greater than for room temperature specimens indicating a signi®cant degree of oxidation damage at the ®bre/matrix interface to have occurred. Acknowledgements This work was supported by funding from the Science and Technology Agency of Japan whilst one of the authors (I.J.D) was supported as a Science and Technology Agency Fellow. The authors gratefully acknowledge Y. Nomura and N. Suzuki for help with mechanical testing and Professor Curtin for useful discussion. References [1] Davies IJ, Ishikawa T, Shibuya M, Hirokawa T. Stress/strain behavior of 3-D woven SiC/SiC-based composites, in preparation. [2] Davies IJ, Ishikawa T, Shibuya M, Hirokawa T. Optical microscopy of 3-D woven SiC/SiC-based composites. Comp Sci Technol 1999;59(3):429±37. [3] Davies IJ, Ishikawa T, Shibuya M, Hirokawa T. Damage characterisation of 3-D woven SiC/SiC-based composites. In: Proceedings of the 21st Symposium on Composite Materials, 31 October±1 November 1996, Toyama, Japan. Tokyo, Japan: The Japan Composite Society. pp. 103±4. [4] Davies IJ, Ishikawa T, Shibuya M, Hirokawa T, Gotoh J. Fibre and interfacial properties measured in situ for a 3-D woven SiC/SiC-based composite with glass sealant. Composites Part A 1999;30(4):587±91. 810 I.J. Davies et al. / Composites Science and Technology 59 (1999) 801±811