E噩≈S Journal of the European Ceramic Society 22(2002)2777-2787 www.elsevier.com/locate/jeurcerar Intermediate temperature strength degradation in Sic/Sic composites G.N. Morschera.*, J.D. Cawley oHio Aerospace Institute(OAl), NASA Glenn Research Center, MS 106-5, Cleveland, OH 44135, US.A Department of Materials Science and Engineering, Case Western Reserve University, Cleveland, OH 44106, US.A Received 20 November 2001; received in revised form 5 February 2002: accepted 23 February 2002 Abstract Woven silicon carbide fiber-reinforced, silicon carbide matrix composites are leading candidate materials for an advanced jet engine combustor liner application. Although the use temperature in the hot region for this application is expected to exceed 1200C, a potential life-limiting concern for this composite system exists at intermediate temperatures(800=+200C), where sig- nificant time-dependent strength degradation has been observed under stress-rupture loading. A number of factors control the degree of stress-rupture strength degradation. the major factor being the nature of the interphase separating the fiber and the matrix. BN interphases are superior to carbon interphases due to the slower oxidation kinetics of BN. A model for the intermediate temperature stress-rupture of SiC/BN/SiC composites is presented based on the observed mechanistic process that leads to strength degradation for the simple case of through-thickness matrix cracks. The approach taken has much in common with that used by urtin and coworkers, for two different composite systems. The predictions of the model are in good agreement with the rupture data for stress-rupture of both precracked and as-produced composites. Also, three approaches that dramatically improve the intermediate temperature stress-rupture properties are described: Si-doped BN, fiber spreading, and"outside debonding".C 2002 Elsevier Science Ltd. All rights reserved Keywords: BN interfaces: Mechanical properties; Oxidation; SiC/SiC composites 1. Introduction liner I because the"cold side ' of the combustor liner would be exposed to this temperature range, and this Non-oxide ceramic matrix composites(CMCs)such would be the portion of the combustor liner under the as SiC fiber-reinforced SiC matrix composites are envi This is sioned for use as high-temperature,> 1200C, compo- the combustor liner would have to be attached to a nents of gas turbine engines. -However, there exists an metal frame. Therefore, it is important to understand intermediate temperature (600 to 1000 C) regime and predict the time-dependent mechanical behavior at where significant, time-dependent, strength degradation intermediate temperatures for design of these compo- can occur Depending on the application, the inter- sites in components, as well as for finding insights mediate temperature properties may be critical for suc- toward improvement of the intermediate temperature cess of that component. For example, the strength properties. degradation at intermediate temperatures could be an The primary cause for intermediate temperature issue for some applications, e.g. a cooled combustor strength degradation is the oxidation of a non-oxide interphase, usually C or BN, that separates the fibers from the matrix. For C interphases, rapid interphase Corresponding author. Tel: +1-216-433-5512; fax: +1-216-433. removal 5-7 can be associated with fiber strength degra- dation in the form of oxide scale formation 8 or pre- E-mail address: gmorscher(a grc. nasa. gov (G.N. morscher) ferential oxidation of carbon enriched areas on the fiber a By intermediate temperature, we mean an elevated temperature below the supposed use temperature at which the material exhibits a urface. For BN interphases, BN oxidizes to form minima in mechanical properties, i.e. what is sometimes called a liquid B2O3(boria) that leads to the formation of a glass at the interphase region and in the 0955-2219/02/S- see front matter C 2002 Elsevier Science Ltd. All rights reserved. PII:S0955-2219(02)00
Intermediate temperature strength degradation in SiC/SiC composites G.N. Morschera,*, J.D. Cawleyb a Ohio Aerospace Institute (OAI), NASA Glenn Research Center, MS 106-5, Cleveland, OH 44135, USA bDepartment of Materials Science andEngineering, Case Western Reserve University, Cleveland, OH 44106, USA Received 20 November 2001; received in revised form 5 February 2002; accepted 23 February 2002 Abstract Woven silicon carbide fiber-reinforced, silicon carbide matrix composites are leading candidate materials for an advanced jet engine combustor liner application. Although the use temperature in the hot region for this application is expected to exceed 1200 C, a potential life-limiting concern for this composite system exists at intermediate temperatures (800200 C), where significant time-dependent strength degradation has been observed under stress-rupture loading. A number of factors control the degree of stress-rupture strength degradation, the major factor being the nature of the interphase separating the fiber and the matrix. BN interphases are superior to carbon interphases due to the slower oxidation kinetics of BN. A model for the intermediate temperature stress-rupture of SiC/BN/SiC composites is presented based on the observed mechanistic process that leads to strength degradation for the simple case of through-thickness matrix cracks. The approach taken has much in common with that used by Curtin and coworkers, for two different composite systems. The predictions of the model are in good agreement with the rupture data for stress-rupture of both precracked and as-produced composites. Also, three approaches that dramatically improve the intermediate temperature stress-rupture properties are described: Si-doped BN, fiber spreading, and ‘‘outside debonding’’. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: BN interfaces; Mechanical properties; Oxidation; SiC/SiC composites 1. Introduction Non-oxide ceramic matrix composites (CMCs) such as SiC fiber-reinforced SiC matrix composites are envisioned for use as high-temperature, 51200 C, components of gas turbine engines.13 However, there exists an intermediate temperature (600 to 1000 C)a regime where significant, time-dependent, strength degradation can occur.4 Depending on the application, the intermediate temperature properties may be critical for success of that component. For example, the strength degradation at intermediate temperatures could be an issue for some applications, e.g. a cooled combustor liner,1 because the ‘‘cold side’’ of the combustor liner would be exposed to this temperature range, and this would be the portion of the combustor liner under the highest tensile stress. This is especially the case where the combustor liner would have to be attached to a metal frame. Therefore, it is important to understand and predict the time-dependent mechanical behavior at intermediate temperatures for design of these composites in components, as well as for finding insights toward improvement of the intermediate temperature properties. The primary cause for intermediate temperature strength degradation is the oxidation of a non-oxide interphase, usually C or BN, that separates the fibers from the matrix.4 For C interphases, rapid interphase removal 57 can be associated with fiber strength degradation in the form of oxide scale formation 8 or preferential oxidation of carbon enriched areas on the fiber surface.9 For BN interphases, BN oxidizes to form liquid B2O3 (boria) that leads to the formation of a borosilicate glass at the interphase region and in the 0955-2219/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0955-2219(02)00144-9 Journal of the European Ceramic Society 22 (2002) 2777–2787 www.elsevier.com/locate/jeurceramsoc * Corresponding author. Tel.: +1-216-433-5512; fax: +1-216-433- 5544. E-mail address: gmorscher@grc.nasa.gov (G.N. Morscher). a By intermediate temperature, we mean an elevated temperature below the supposed use temperature at which the material exhibits a minima in mechanical properties, i.e. what is sometimes called a ‘‘pest’’ condition.4
2778 G N. Morscher, J. D. Cawley/Journal of the European Ceramic Society 22(2002)2777-2787 matrix crack when the boria reacts with the SiC fibers 2. The process and factors affecting intermediate tem- and matrix 0-lI coupled with B removal from the oxi- perature stress-rupture of woven BN interphase com- dation product as volatile B-containing hydrated species posites in air form in the presence of water vapor. However, the intermediate temperature stress-rupture properties of It was established in Ref. 14 from chemical analysis of Sic/SiC composites with bn interphases have been fiber fracture surfaces(degree of fiber fracture surface shown to be superior to SiC/SiC composites with C oxidation) that whole areas of fibers in a matrix crack interphases when tested in air. failed at the same time during the course of the stress Since BN interphase composites are more durable rupture experiment. It was also discerned from fracture than C interphase composites in oxidizing environments mirror analysis of failed fibers that the amount of at intermediate temperatures, bN has been selected as degradation to the fiber strength was commensurate the interphase material for the earlier mentioned com- with the expected amount from single fiber stress-rup bustor liner application. Therefore, the factors and ture data. 16 That is, no additional strength degradation mechanisms that control intermediate temperature occurred to the population of strongly bonded fibers. stress-rupture for BN interphase composites will be the Therefore, failure under stress-rupture conditions at focus of this study. A model that accounts for some of intermediate temperatures occurs by local overloading these factors will be presented. It will be shown to pre- due to the stress concentration associated with the dict the intermediate temperature rupture data. Finally, strong bonding of fibers to the matrix. Not because the recent enhancements to the microstructure of SiC/BN/ fibers are weakened. The depth of this embrittled region Sic composites resulting in improved intermediate tem- grows as the oxidation front moves deeper into the perature composite performance will be presented. All matrix crack(Fig. 1). Eventually, one of the strongly of the stress-rupture data presented in this work that bonded fibers breaks and causes all the other strongly has not been published earlier was performed in the bonded fibers to fail due to the inability to globally same manner as described in Refs. 14, 15. Woven com- share the increased stress applied to the nearest neigh- posites, 150 mm in length, were tested in tension where bor fibers and unbridged crack growth. This view is the ends of the specimens were"cold-gripped"and a strongly supported by the pattern of fracture mirrors slotted furnace with a 15 mm hot zone was inserted in the from the strongly bonded regions of near fiber contact center region between the grips. The furnace was brought in the micrograph of Fig. l, which are indicative of to temperature prior to the application of the load correlated fiber failure. If the stress transferred to the fiber x(t) BN nnnnnnnnn Embrittled rea (Loc 5)x×a.点k1: O2 and H,0 O and Ho bonded Pristine fiber Area(Global Load Sharing) matrix Fiber break Fig. 1. Idealized schematic representation of oxygen ingress in a matrix crack and an individual fiber failure that leads to failure of all strongly bonded fibers. An example of which is given in the upper left hand corner for a HN/ BN/MI SiC composite
matrix crack when the boria reacts with the SiC fibers and matrix 1011 coupled with B removal from the oxidation product as volatile B-containing hydrated species form in the presence of water vapor.11 However, the intermediate temperature stress-rupture properties of SiC/SiC composites with BN interphases have been shown to be superior to SiC/SiC composites with C interphases when tested in air.9,12,13 Since BN interphase composites are more durable than C interphase composites in oxidizing environments at intermediate temperatures, BN has been selected as the interphase material for the earlier mentioned combustor liner application.1 Therefore, the factors and mechanisms that control intermediate temperature stress-rupture for BN interphase composites will be the focus of this study. A model that accounts for some of these factors will be presented. It will be shown to predict the intermediate temperature rupture data. Finally, recent enhancements to the microstructure of SiC/BN/ SiC composites resulting in improved intermediate temperature composite performance will be presented. All of the stress-rupture data presented in this work that has not been published earlier was performed in the same manner as described in Refs. 14,15. Woven composites, 150 mm in length, were tested in tension where the ends of the specimens were ‘‘cold-gripped’’ and a slotted furnace with a 15 mm hot zone was inserted in the center region between the grips. The furnace was brought to temperature prior to the application of the load. 2. The process and factors affecting intermediate temperature stress-rupture of woven BN interphase composites in air It was established in Ref. 14 from chemical analysis of fiber fracture surfaces (degree of fiber fracture surface oxidation) that whole areas of fibers in a matrix crack failed at the same time during the course of the stressrupture experiment. It was also discerned from fracture mirror analysis of failed fibers that the amount of degradation to the fiber strength was commensurate with the expected amount from single fiber stress-rupture data.16 That is, no additional strength degradation occurred to the population of strongly bonded fibers. Therefore, failure under stress-rupture conditions at intermediate temperatures occurs by local overloading due to the stress concentration associated with the strong bonding of fibers to the matrix. Not because the fibers are weakened. The depth of this embrittled region grows as the oxidation front moves deeper into the matrix crack (Fig. 1). Eventually, one of the strongly bonded fibers breaks and causes all the other strongly bonded fibers to fail due to the inability to globally share the increased stress applied to the nearest neighbor fibers and unbridged crack growth. This view is strongly supported by the pattern of fracture mirrors from the strongly bonded regions of near fiber contact in the micrograph of Fig. 1, which are indicative of correlated fiber failure. If the stress transferred to the Fig. 1. Idealized schematic representation of oxygen ingress in a matrix crack and an individual fiber failure that leads to failure of all strongly bonded fibers. An example of which is given in the upper left hand corner for a HN/BN/MI SiC composite. 2778 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787
G N. Morscher, J. D. Cawley/Journal of the European Ceramic Society 22(2002)2777-2787 pristine weakly bonded fibers cannot be carried by the for a fiber failure originating in the region exposed by a remaining fibers, then the composite fails. Two criteria matrix crack. The number of fibers per tow, number of must be met in order for a composite to fail according tows in a composite cross-section, and the size of the to this process: specimen for a given volume fraction of fibers will affect the total number of fibers in a matrix crack. The effec- 1. A critical number of fibers in a given matrix tive gage length of fibers will be controlled by the ability crack must be strongly bonded to one another or of the fibers to transfer load to the matrix, i. e. interfacial to the matrix. When these fibers fail in a matrix shear strength, the amount of interfacial recession that crack, the stress increase to the remaining may occur, and the number of matrix cracks that are unbroken fibers is sufficient to cause them to fail. exposed in the " hot zone""of the furnace. An increase in 2. An event has to occur to fail one or more of the effective length of fully-loaded fibers will increase those strongly bonded fibers to cause unbridged the likelihood that a strongly bonded fiber will fail in or crack growth. Most likely, this event is caused by near a matrix crack beginning the process of unbridged the failure of one strongly bonded fiber due to crack growth intrinsic fiber strength degradation(flaw growth) of a fiber that is relatively weak in the distribu tion of fiber strengths. It is also possible that 3. A model for intermediate temperature stress rupture fiber-degradation could occur from fiber oxida- of SiC/BN/SiC composites tion depending on the fiber-type and oxidizing environment In order to model this process, an approach con ceptually similar Curtin and coworkers 18-2 approach to model composite strength and individual The kinetics for fiber fusion or the depth into a matrix fiber fracture was employed. Only the simple case of rack away from the exposed surface that fibers are through-thickness cracks was considered. The model strongly bonded depends on the ingress of oxidizing was applied to two SiC fiber Bn interphase MI SiC duction, and the shortest distance between two fibers, for these systems. 4.I g available property information species into the matrix crack, i.e. the rate of oxide pro- matrix systems by usi i.e. the gap that must be filled by oxide. Ingress of oxi- dizing species can only occur if matrix cracks are pre 3. The model sent which intersect load-bearing fibers; therefore, fiber fusion will be dependent on the presence of matrix The stress on the fibers in a bridged matrix cracl cracks, and whether or not those cracks are through the can be found from the applied far-field composite stress, thickness of the specimen. The durability of the inter- 0, and the volume fraction of fibers in the loading phase will affect the rate for fiber-to-fiber fusion. It was direction, f. found for the woven Hi-Nicalon(Nippon Carbon, Co Japan)fiber(HN)reinforced, BN interphase, melt-infil trated (MI) SiC matrix composite of Ref. 14 that the thin carbon layer that exists between the fiber and the bn due to fiber decomposition during matrix processing 7 enhances crack growth and interphase oxidation I5 Also the closer fibers are to one another or the thinner Transfer e interphase coating and the method of interphase used in u= crack coating will be critical. It was found in the earlier study 4 model that over 95% of all the fibers were nearly in contact openIng with one another, i.e. are separated by less than 100 nm R displacement even though the average thickness of the interphase was 0.5 This is nce of woven structures, where the act of weaving tightens tows and forces fibers into intimate contact with one another oeo=o/f Fiber failure depends on the strength-distribution of the fibers in a matrix crack. the number of fibers in a Fig. 2. Schematic representation of stress-profile at and around a matrix crack, and the effective gage length of loaded a composite. sf. m fibers and the matri fibers. A wider distribution of fiber strengths for the rule of The model assumes same average strength will mean a greater probability [Eq(7)
pristine weakly bonded fibers cannot be carried by the remaining fibers, then the composite fails. Two criteria must be met in order for a composite to fail according to this process: 1. A critical number of fibers in a given matrix crack must be strongly bonded to one another or to the matrix. When these fibers fail in a matrix crack, the stress increase to the remaining unbroken fibers is sufficient to cause them to fail. 2. An event has to occur to fail one or more of those strongly bonded fibers to cause unbridged crack growth. Most likely, this event is caused by the failure of one strongly bonded fiber due to intrinsic fiber strength degradation (flaw growth) of a fiber that is relatively weak in the distribution of fiber strengths. It is also possible that fiber-degradation could occur from fiber oxidation depending on the fiber-type and oxidizing environment. The kinetics for fiber fusion or the depth into a matrix crack away from the exposed surface that fibers are strongly bonded depends on the ingress of oxidizing species into the matrix crack, i.e. the rate of oxide production, and the shortest distance between two fibers, i.e. the gap that must be filled by oxide. Ingress of oxidizing species can only occur if matrix cracks are present which intersect load-bearing fibers; therefore, fiber fusion will be dependent on the presence of matrix cracks, and whether or not those cracks are through the thickness of the specimen. The durability of the interphase will affect the rate for fiber-to-fiber fusion. It was found for the woven Hi-Nicalon (Nippon Carbon, Co., Japan) fiber (HN) reinforced, BN interphase, melt-infiltrated (MI) SiC matrix composite of Ref. 14 that the thin carbon layer that exists between the fiber and the BN due to fiber decomposition during matrix processing 17 enhances crack growth and interphase oxidation.15 Also, the closer fibers are to one another or the thinner the interphase, the faster fibers will fuse to one another or to the matrix, respectively. Therefore, the uniformity of the interphase coating and the method of interphase coating will be critical. It was found in the earlier study 14 that over 95% of all the fibers were nearly in contact with one another, i.e. are separated by less than 100 nm, even though the average thickness of the interphase was 0.5 mm. This is especially a consequence of woven structures, where the act of weaving tightens tows and forces fibers into intimate contact with one another. Fiber failure depends on the strength-distribution of the fibers in a matrix crack, the number of fibers in a matrix crack, and the effective gage length of loaded fibers. A wider distribution of fiber strengths for the same average strength will mean a greater probability for a fiber failure originating in the region exposed by a matrix crack. The number of fibers per tow, number of tows in a composite cross-section, and the size of the specimen for a given volume fraction of fibers will affect the total number of fibers in a matrix crack. The effective gage length of fibers will be controlled by the ability of the fibers to transfer load to the matrix, i.e. interfacial shear strength, the amount of interfacial recession that may occur, and the number of matrix cracks that are exposed in the ‘‘hot zone’’ of the furnace. An increase in the effective length of fully-loaded fibers will increase the likelihood that a strongly bonded fiber will fail in or near a matrix crack beginning the process of unbridged crack growth. 3. A model for intermediate temperature stress rupture of SiC/BN/SiC composites In order to model this process, an approach conceptually similar to Curtin and coworkers 1820 approach to model composite strength and individual fiber fracture was employed. Only the simple case of through-thickness cracks was considered. The model was applied to two SiC fiber BN interphase MI SiC matrix systems by using available property information for these systems.14,15 3.1. The model The stress on the fibers in a bridged matrix crack, sf, can be found from the applied far-field composite stress, s, and the volume fraction of fibers in the loading direction, f. Fig. 2. Schematic representation of stress-profile at and around a matrix crack in a composite. sf,m would represent the stress on the fibers where the fibers and the matrix share the load according to the rule of mixtures. The model assumes d/2 extends to sf=0 for simplicity [Eq. (7)]. G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787 2779
2780 G N Morscher, J.D. Cawley / Journal of the European Ceramic Society 22(2002)2777-2787 With time, fibers exposed by a matrix crack will fail. Eq (5a) then simply bec Fiber failure is assumed to follow a weibull distribution that can be used to determine the probability for fiber (m+1) failure φ(6)= (8) m+1 P(σ,L)=1-e-中 The fraction of fibers that fail in a matrix crack can be o is the fraction of failed fibers according to determined by summing Eqs.( 8)and(5b)and simplify ing with Eq(4) where m is the Weibull modulus, oo is the reference y, m+/x stress and Lo is the reference length that corresponds to the average fiber strength determined in single fiber tensile tests. L is the effective gage length in the matrix oo(t, T crack. It proved useful to adopt the formulation of urtin's s characteristic stress, Oe, and characteristic where Ks (10) gage length, 8c, where (oc, Sc)=l and Sc is twice the m+1 The operative mechanism to be modeled to R composite rupture is the failure of the strongly fiber that triggers the growth of an unbridged crack fiber slip length, by definition, R is the fiber diameter, through the embrittled fibers in a matrix crack. The time and t is the interfacial shear stress [Eq. (4) dependence for the depth of embrittlement into the The fiber stress around the matrix crack varies composite determines the number of embrittled fibers because of load transfer due to friction(Fig. 2). The available in a matrix crack. The region of fiber embrit- fibers are subject to the maximum fiber stress in the tlement often appears as a"picture frame"4 of strongly crack opening. To determine the total fraction of fiber bonded fibers around the rim of the cross-section of a failures, o can be integrated over the stress transfer specimen fracture surface for a composite with through length, Eo and added to the fraction of fiber failures in thickness matrix cracks. The time dependence for fiber the crack opening width, o to-fiber fusion was determined empirically from the depth of this"picture frame"by examination of rup a(z)"dz (5a) tured-specimen fracture surfaces for two different MI composite systems, one reinforced with HN fibers and =()(∞ Corp, Midland, MI) fibers \4is/mmic (Dow Corning that a semi-empirical parabolic time-dependence was an adequate description: where z is the stress transfer length and u is the crack opening width which can be approximated by: 21 (6) where Cox is an empirical coefficient that best fits the 4Tf2E measured oxidation depth data for a given composite system Since the fiber strength degradation of pu Eq (Sa)was solved by othersfor the case where z is fibers in a through-thickness cracked composite is con- qual to twice the fiber slip length, 8, assuming the far sistent with the measured degradation in fiber strengths field stress on the fibers to be zero. This is an appro- of as-produced fibers, the descriptive expression for time priate assumption because there is a negligible con- dependent fiber strength degradation developed by Yun tribution to from the low far field fiber stress(1/5 and DiCarlo 6 for the latter was used. Their data for ofc).8 Can then be approximated assuming a constant t rupture strength of three Sic type fibers are plotted in from the relationship(see Fig. 2) Fig 4 as a Larson-Miller plot. The conditions for our
f ¼ f ð1Þ With time, fibers exposed by a matrix crack will fail. Fiber failure is assumed to follow a Weibull distribution that can be used to determine the probability for fiber failure: Pð; LÞ ¼ 1e ð2Þ is the fraction of failed fibers according to: ¼ L Lo f o m ð3Þ where m is the Weibull modulus, so is the reference stress and Lo is the reference length that corresponds to the average fiber strength determined in single fiber tensile tests. L is the effective gage length in the matrix crack. It proved useful to adopt the formulation of Curtin’s 18 characteristic stress, sc, and characteristic gage length, dc, where (sc,dc)=1 and dc is twice the c ¼ m o Lo R 1 mþ1 ; c ¼ Rc ð4Þ fiber slip length, by definition, R is the fiber diameter, and t is the interfacial shear stress [Eq. (4)]. The fiber stress around the matrix crack varies because of load transfer due to friction (Fig. 2). The fibers are subject to the maximum fiber stress in the crack opening. To determine the total fraction of fiber failures, can be integrated over the stress transfer length, and added to the fraction of fiber failures in the crack opening width, u: ¼ ðZ dz ¼ ðZ ðzÞ o m dz Lo ð5aÞ u ¼ u Lo f o m ð5bÞ where z is the stress transfer length and u is the crack opening width which can be approximated by:21 u ¼ 2R 4f 2Ef 1 þ Ef f Emð Þ 1f ð6Þ Eq. (5a) was solved by others 22 for the case where z is equal to twice the fiber slip length, d, assuming the far field stress on the fibers to be zero. This is an appropriate assumption because there is a negligible contribution to from the low far field fiber stress (1/5 sfc). Can then be approximated assuming a constant t from the relationship (see Fig. 2): ¼ Rf ð7Þ Eq. (5a) then simply becomes: ðÞ ¼ f c ð Þ mþ1 m þ 1 ð8Þ The fraction of fibers that fail in a matrix crack can be determined by summing Eqs. (8) and (5b) and simplifying with Eq. (4): t;T ¼ f c ð Þ mþ1 m þ 1 þ u Lo f o m ¼ 1 Lo f oðt;TÞ m m þ 1 þ u ¼ K m oðt;TÞ ð9Þ where K ¼ m f Lo m þ 1 þ u ð10Þ The operative mechanism to be modeled to predict composite rupture is the failure of the strongly bonded fiber that triggers the growth of an unbridged crack through the embrittled fibers in a matrix crack. The time dependence for the depth of embrittlement into the composite determines the number of embrittled fibers available in a matrix crack. The region of fiber embrittlement often appears as a ‘‘picture frame’’ 4 of strongly bonded fibers around the rim of the cross-section of a specimen fracture surface for a composite with throughthickness matrix cracks. The time dependence for fiberto-fiber fusion was determined empirically from the depth of this ‘‘picture frame’’ by examination of ruptured-specimen fracture surfaces for two different MI composite systems, one reinforced with HN fibers and the other reinforced with Sylramic (Dow Corning Corp., Midland, MI) fibers 14,15 (Fig. 3). It was found that a semi-empirical parabolic time-dependence was an adequate description: x ¼ Coxt 1=2 ð11Þ where Cox is an empirical coefficient that best fits the measured oxidation depth data for a given composite system. Since the fiber strength degradation of pulled out fibers in a through-thickness cracked composite is consistent with the measured degradation in fiber strengths of as-produced fibers, the descriptive expression for time dependent fiber strength degradation developed by Yun and DiCarlo 16 for the latter was used. Their data for rupture strength of three SiC type fibers are plotted in Fig. 4 as a Larson–Miller plot. The conditions for our 2780 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787
G N. Morscher, J. D. Cawley/Journal of the European Ceramic Society 22(2002)2777-2787 198 One important factor for a fiber strength determina 0.215t tion is the actual strength of the fibers in the composite after processing. The fiber strength of both the Hn and E0.7 SYL fibers at room temperature found by Yun and DiCarlo was 2800 MPa. 6 However, some strength degradation may occur due to composite processing Curtin et al. have established a composite ultimate strength failure criterion based on global load sharing assumptions 2(m+1)=/m+1 for 28> (13) (m+2) Fig 3. Depth of oxidation into the specimen from the surface(face)of the composite versus rupture time for -815C rupture of BN inter- Eq(13)is for matrix crack saturation where pe is the phase, MI SiC composites. The n widths were approximately 2 crack density and pe would be the crack spacing. The m.4, I5 The arrow for the Hi-Nicalon data point was for a specimen room temperature ultimate strengths of all of the com- that did not have through-thickness cracks, i.e. the data point indi- cates that oxygen ingress was at least that deep, three plies, into the posites modeled in this study are known. Therefore, the specimen. 4 The closed symbols indicate 815C experiments where the ultimate strength of the fibers in the composites could specimens failed in the hot zone. The open symbols indicate specimens be estimated by solving for o. from Eqs.(13)and ( that were tested at 960 oC which had failed outside of the hot zone region at a lower temperature estimated to be -870oC 14 m(m+2)R/m+2 study are indicated in Fig 4. This data was best fit, re 2(m+1)Lom+ plotted on a stresss-time plot and re-fitted to fit the common form Eq. 12 is based on the room temperature ultimate (12) fiber strength, Oo(Rn, of 2800 MPa. Assuming the flaw growth mechanism that causes time-dependent fiber where Cr is the coefficient that best fits the fiber rupture strength-degradation rate at intermediate temperatures data and n is the rupture exponent; both are dependent were the same as for single fiber tests and depends on on the fiber type. This then becomes the time-dependent starting flaw size, the time-dependent fiber strength of reference stress for Eq.(9) fibers in the composite can be estimated from Eq . (12) Sylramic 8 Nicalon Hi-Nicalon Yun and Di Carlo [16] 800c1000C1200c 100h100h100h 0.1 500010000150002000025000300003500040000 Larson Miller Parameter, q=T[log t+ 22],(K, hr ig. 4. Rupture strength in a Larson-Miller format for different SiC fibers from Yun and DiCarlo. 6
study are indicated in Fig. 4. This data was best fit, replotted on a stresss-time plot and re-fitted to fit the common form: 0ðt;TÞ ¼ Cf t 1=n ð12Þ where Cf is the coefficient that best fits the fiber rupture data and n is the rupture exponent; both are dependent on the fiber type. This then becomes the time-dependent reference stress for Eq. (9). One important factor for a fiber strength determination is the actual strength of the fibers in the composite after processing. The fiber strength of both the HN and SYL fibers at room temperature found by Yun and DiCarlo was 2800 MPa.16 However, some strength degradation may occur due to composite processing. Curtin et al.19 have established a composite ultimatestrength failure criterion based on global load sharing assumptions: ult ¼ c 2ð Þ m þ 1 m mð Þ þ 2 1 mþ1 m þ 1 m þ 2 ; for 2>1 c ð13Þ Eq. (13) is for matrix crack saturation where c is the crack density and c 1 would be the crack spacing. The room temperature ultimate strengths of all of the composites modeled in this study are known. Therefore, the ultimate strength of the fibers in the composites could be estimated by solving for so from Eqs. (13) and (4): oðcompositeÞ ¼ m mð Þ þ 2 2ð Þ m þ 1 R Lo m þ 2 m þ 1 ult mþ1 " #1 m ð14Þ Eq. 12 is based on the room temperature ultimate fiber strength, so(RT), of 2800 MPa. Assuming the flaw growth mechanism that causes time-dependent fiber strength-degradation rate at intermediate temperatures were the same as for single fiber tests and depends on starting flaw size, the time-dependent fiber strength of fibers in the composite can be estimated from Eq. (12): Fig. 4. Rupture strength in a Larson–Miller format for different SiC fibers from Yun and DiCarlo.16 Fig. 3. Depth of oxidation into the specimen from the surface (face) of the composite versus rupture time for 815 C rupture of BN interphase, MI SiC composites. The specimen widths were approximately 2 mm.14,15 The arrow for the Hi-Nicalon data point was for a specimen that did not have through-thickness cracks, i.e. the data point indicates that oxygen ingress was at least that deep, three plies, into the specimen.14 The closed symbols indicate 815 C experiments where the specimens failed in the hot zone. The open symbols indicate specimens that were tested at 960 C which had failed outside of the hot zone region at a lower temperature estimated to be 870 C.14 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787 2781
G N. Morscher, J.D. Cawley /Journal of the European Ceramic Society 22(2002) 2777-2787 Table I Composite and constituent properties (16b) HN SYL Composite physical properties At this point it is illustrative to show the probability Tow ends per cm for fiber failure from Eqs. (16a)and (16b) and the hickness. mm number of predicted fiber failures in and around a single matrix crack for various applied rupture stress condi- No fibers per tow tions for the HN-BN-MI 4 composite system as a function of time and depth of embrittlement(Fig 5a and b, respectively). The variables used are listed in Table 1. most fiber failure would occur at short times and diminish with increasing time(Fig 5a); however, it takes a period of time to embrittle most fibers. For thi Room temperature cult reason, when predicting whether or not a fiber failure will occur for an embrittled fiber, it is absolutely neces- sary to take into account the probability that fibers had already failed prior to being embrittled. Fig. 5 depicts Fiber strength distribution properties e situation where rupture time reaches 21.5 hours(the 2800 2800 time to fully embrittle the HN-BN-MI composite at mm 815C in air). At a depth of 0. 2 mm from the composite urface, fibers were not embrittled for 0.9h. However, a Time dependent propertie greater fraction of fibers would be expected to fail prior 0215 0.125 to fiber embrittlement at 0. 2 mm depth, r, compared to the fraction of fibers that would be expected to fail after 56.5 embrittlement, emb. If fibers fail prior to embrittlemen hen the load shed from that fiber is shared globally, unbridged crack growth will not occur, and those fibers (15) that fail prior to embrittlement are removed from the OO(RT population of weak fibers that could fail. Therefore, the fraction of embrittled fibers that fail at a given The fraction of fibers that fail in a matrix crack as a region in a matrix crack can be estimated according to function of time can then be determined from Eqs. (10) the construct of Fig. 5 using Eq.(16b) for embrittle and(15)Eq(16a)]. However, for the purpose of deter- ment depth mining the fraction of fibers that fail as a function of the depth of oxidation embrittlement, it is advantageous convertino depth,x,from Eq.(1)asu(b.ab=pn、_的mCc2m1 ](17) (1)mn where max is the fraction of fibers that fail at the ●000015 000005 中 0246810121416182022000 0200.40 Time, hours (b Fig. 5. Predicted fraction of fiber failures for HN in a matrix crack of an HN/BN/MI composite at 815C (Table 1)as a function of (a)time and(b) embrittlement depth for an apl mposite stress of 150 MPa
0ðt;TÞ ¼ oðcompositeÞ oðRTÞ Cf t 1=n ¼ Cfrupturet 1=n ð15Þ The fraction of fibers that fail in a matrix crack as a function of time can then be determined from Eqs. (10) and (15) [Eq. (16a)]. However, for the purpose of determining the fraction of fibers that fail as a function of the depth of oxidation embrittlement, it is advantageous to convert t into depth, x, from Eq. (11) as well [Eq. (16b)]. t;T ¼ K Cm frupture ð Þt m=n ð16aÞ t;T ¼ K Cm frupture x Cox 2m=n ð16bÞ At this point it is illustrative to show the probability for fiber failure from Eqs. (16a) and (16b) and the number of predicted fiber failures in and around a single matrix crack for various applied rupture stress conditions for the HN-BN-MI 14 composite system as a function of time and depth of embrittlement (Fig. 5a and b, respectively). The variables used are listed in Table 1. Most fiber failure would occur at short times and diminish with increasing time (Fig. 5a); however, it takes a period of time to embrittle most fibers. For this reason, when predicting whether or not a fiber failure will occur for an embrittled fiber, it is absolutely necessary to take into account the probability that fibers had already failed prior to being embrittled. Fig. 5 depicts the situation where rupture time reaches 21.5 hours (the time to fully embrittle the HN–BN–MI composite at 815 C in air). At a depth of 0.2 mm from the composite surface, fibers were not embrittled for 0.9 h. However, a greater fraction of fibers would be expected to fail prior to fiber embrittlement at 0.2 mm depth, t, compared to the fraction of fibers that would be expected to fail after embrittlement, emb. If fibers fail prior to embrittlement then the load shed from that fiber is shared globally, unbridged crack growth will not occur, and those fibers that fail prior to embrittlement are removed from the population of weak fibers that could fail. Therefore, the fraction of embrittled fibers that fail at a given region in a matrix crack can be estimated according to the construct of Fig. 5 using Eq. (16b) for embrittlement depth: emb ¼ tmax t ¼ K Cm frupture C2m=n ox x2m=n max x2m=n ð17Þ where tmax is the fraction of fibers that fail at the Table 1 Composite and constituent properties HN SYL Composite physical properties Tow ends per cm 6.7 7.1 No. Plies 8 8 Thickness, mm 2.1 2.1 Width, mm 12.5 10 No. fibers per tow 500 800 R, mm 6.5 5 F 0.17 0.17 Composite mechanical properties Ec, GPa 215 265 n 0.15 0.15 Room temperature sult 390 340 t, MPa 30 60 Ef, GPa 280 380 Em, GPa 202 242 Fiber strength distribution properties so, MPa 2800 2800 M 75 Lo, mm 25.4 25.4 Time dependent properties Cox 0.215 0.125 Cf 1761 2169 n 56.5 122 Fig. 5. Predicted fraction of fiber failures for HN in a matrix crack of an HN/BN/MI composite at 815 C (Table 1) as a function of (a) time and (b) embrittlement depth for an applied composite stress of 150 MPa. 2782 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787
G N. Morscher, J D Cawley/ Journal of the European Ceramic Society 22(2002) 2777-2787 2783 maximum time and is the fraction of fibers that failed tested in rupture have several matrix cracks exposed to prior to the time of embrittlement. The number of the hot zone depending on the crack-density associated embrittled fiber failures can then be determined from with the applied stress condition. An increase in crack the product of the number of fibers per unit thickness, density increases the effective gage length of strongly Nx, and the integration of emb over the maximum bonded fibers, and thereby the fraction of fiber failures depth of oxidation embrittlement, To account for this, all that is required is to multiply femb by the number of cracks, Nc, which results in NcNemb(single crack) (23) NxCe-XGa)1-2m (18) where Nemb( single crack) This equation is valid up to the time that all of the is the solution to either Eq(18)or(19), depending on fibers throughout the entire matrix cross-section have the time and Nc can be found from been embrittled, femb. If no embrittled fibers failed prior to the time it takes to strongly bond all of the fibers in Nc=peg (24) through-thickness crack. the number of embrittled fiber failures would be equal to where Lg is the gage length, i.e. the length of specimen in the hot zone N盐bm=Mm+Nr(中1-a) 3. 2. Applying the model It was decided to model two regimes of rupture (m/n-im/n for t>temb (19) behavior rather than the entire rupture curve:(a)"oxi dation kinetics"controlled rupture and(b)"single fiber failure""controlled rupture. The first case is where many where N is the total number of fibers in the composite fibers break during the rupture condition due to higher cross-section applied stresses and high crack densities. The latter When a strongly bonded fiber fails at time tail, it will condition is the case where the first embrittled fiber to be assumed that all of the embrittled fibers fail in the break in a composite causes ultimate composite failure. cross-section of the matrix crack. Then, if the load shed The entire rupture curve could be modeled similar to onto the remaining pristine fibers cannot be carried by Lara-Curzio in an iterative fashion where a computer the remaining fibers, the composite will rupture. The program essentially continues to increase time in dis- ultimate stress-criterion 19 for most conditions will be crete steps and solves the above equations in order to for the case of a composite that is not crack-saturated determine if the equation offal> Cult is fulfilled for a given stress/crack-density condition. However, for the Cult ace/m+l; for 28Cult, and one would have to"wait"until another fiber fails in that specific matrix crack for ultimate failure to occur. It is If affil>cult, then the composite fails possible that the time for the lower stress condition first embrittled fiber failure to be less than the second fiber failure of the higher stress One further consideration must be taken into condition and sufficient for ultimate composite failure. 23(It should be account. All of the analysis up to this point has been for noted that the determination of Emb in Ref 23 is inappropriate under the case of a single matrix crack. Most of the specimens this scenario, the analysis used here should be used instead
maximum time and t is the fraction of fibers that failed prior to the time of embrittlement. The number of embrittled fiber failures can then be determined from the product of the number of fibers per unit thickness, N x; and the integration of emb over the maximum depth of oxidation embrittlement, xmax: Nttemb femb ¼ Nttemb ð19Þ where Nf is the total number of fibers in the composite cross-section. When a strongly bonded fiber fails at time tffail, it will be assumed that all of the embrittled fibers fail in the cross-section of the matrix crack. Then, if the load shed onto the remaining pristine fibers cannot be carried by the remaining fibers, the composite will rupture. The ultimate stress-criterion 19 for most conditions will be for the case of a composite that is not crack-saturated: ult ¼ ce1=mþ1 ; for 2sult, then the composite fails. One further consideration must be taken into account. All of the analysis up to this point has been for the case of a single matrix crack. Most of the specimens tested in rupture have several matrix cracks exposed to the hot zone depending on the crack-density associated with the applied stress condition. An increase in crack density increases the effective gage length of strongly bonded fibers, and thereby the fraction of fiber failures. To account for this, all that is required is to multiply femb by the number of cracks, Nc, which results in: Nfemb ¼ NcNfembð Þ single crack ð23Þ where Nfembð Þ single crack is the solution to either Eq. (18) or (19), depending on the time, and Nc can be found from: Nc ¼ cLg ð24Þ where Lg is the gage length, i.e. the length of specimen in the hot zone. 3.2. Applying the model It was decided to model two regimes of rupture behavior rather than the entire rupture curve: (a) ‘‘oxidation kinetics’’ controlled rupture and (b) ‘‘single fiber failure’’ controlled rupture. The first case is where many fibers break during the rupture condition due to higher applied stresses and high crack densities. The latter condition is the case where the first embrittled fiber to break in a composite causes ultimate composite failure. The entire rupture curve could be modeled similar to Lara-Curzio 8 in an iterative fashion where a computer program essentially continues to increase time in discrete steps and solves the above equations in order to determine if the equation sf fail>sult is fulfilled for a given stress/crack-density condition.23 However, for the case of several matrix cracks in a hot zone, after first fiber failure, each crack has to be treated independently and there exists an intermediate stress range where higher applied stresses will yield longer rupture times than lower applied stress conditions.b This may actually be indicative of some of the scatter in rupture results; however, what is usually considered to be of greatest importance is the ‘‘run-out’’ stress condition, i.e. case b For example, for a higher stress condition, first embrittled fiber failure occurs at a time less than at a lower stress condition. However, the shorter-time-first-embrittled-fiber-failure may not be at a condition where ultimate composite failure would occur, i.e. x is too small to satisfy sffail>sult, and one would have to ‘‘wait’’ until another fiber fails in that specific matrix crack for ultimate failure to occur. It is possible that the time for the lower stress condition first embrittled fiber failure to be less than the second fiber failure of the higher stress condition and sufficient for ultimate composite failure.23 (It should be noted that the determination of Øemb in Ref.23 is inappropriate under this scenario, the analysis used here should be used instead.) G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787 2783
2784 G N. Morscher, J D Cawley/ Journal of the European Ceramic Society 22(2002) 2777-2787 SYLMIb o precracked data(200 MPB) 140 Time, hours Time, Hours Fig. 6. Stress-rupture at 815C in air of BN interphase composites with(a)HN fibers and(b)SYL fibers b)above, which the single fiber failure condition would ultimate strength of the fibers in the as-produced com- predict. Therefore, rupture curves will be predicted for posite, oofcomposite) would be underestimated from the these two extremes and the switch from the kinetic- composite ultimate strength, cult [Eq. (14)], which was dependence to single fiber failure controlled rupture based on global load sharing assumptions. This would occurs at the point where single fiber failure rupture effectively reduce the estimated time-dependent fiber predicts longer times for rupture. strength and predict shorter rupture times for a given Case(a)can simply be determined by finding the time stress that satisfies the condition where Eq (20)and Eq (21) The model predicts a fiber gage-length dependence for are equal and is controlled by the time-dependent stress-rupture. This provides an opportunity to inde- embrittlement depth. Case(b) was determined by sol- pendently test model predictions. This was verified by ving for the case where Eq (23)is equal to l, i.e. the precracking composites at room temperature so that first embrittled fiber failure in a matrix crack Case(b) they possess a larger crack density than that from stress requires an accurate measure of crack-density as a rupture at an applied stress less than the precrack con- function of the stress-state. This information was avail- dition. For the HN-BN-MI specimen, some precracked able for the HN-BN-MI SiC and SYL-BN-MI SiC experiments were performed. A precrack stress of 200 systems. 4. I5 Fig. 6 shows the predictions for the two MPa was performed at room temperature and then the composite systems and the actual rupture data from specimen was subjected to stress-rupture at a lower Refs. 14, 15. Table I lists the experimentally determined stress. This precrack condition resulted in a crack den- variables that went into the model for both systems and sity of 2 cracks/mm. The increased lengths of loaded Fig. 7 shows the stress-dependent crack density as fibers resulted in significantly shorter rupture times at determined from measured crack densities from some of lower stresses than the specimens that were not pre- rupture mens cracked as the model also predicted fairly well(Fig. 6a) The two extremes predict the rupture behavior rela- The run-out stress was slightly underestimated by the tively well. The kinetic-limit seems to overestimate rup- model. However, the model effectively predicted the ture time slightly. One possible reason for this decrease in stress-rupture time with increasing crack overestimate is the presence of a possible stress-con- density centration on the outer perimeter of bridging fibers in a matrix crack. 24,25 which was not taken into account ne model. The model also slightly underestimates the rupture times for the SYL composites for the single fiber limit. One issue with SYL fiber composites in general is a relatively high t and the possibility that minor to moderate local load sharing conditions exist even for room temperature failure. o If this is the case, the SYL Composite For the model developed by Evans et aL. 25 composite system, fiber degradation was due to oxide scale growth and unbridged crack growth was due to fiber failure at the perimeter of the 100120140160180200220240260 unbridged crack. A fundamental difference with the model proposed Fig. 7. Stress-dependence for matrix crack density for SYL and HN here is that an interior strongly bonded fiber can trigger the failure of composites. The curves were based on post-test measurements of failed the entire region of strongly bonded fibers specimens
(b) above, which the single fiber failure condition would predict. Therefore, rupture curves will be predicted for these two extremes and the switch from the kineticdependence to single fiber failure controlled rupture occurs at the point where single fiber failure rupture predicts longer times for rupture. Case (a) can simply be determined by finding the time that satisfies the condition where Eq. (20) and Eq. (21) are equal and is controlled by the time-dependent embrittlement depth. Case (b) was determined by solving for the case where Eq. (23) is equal to 1, i.e. the first embrittled fiber failure in a matrix crack. Case (b) requires an accurate measure of crack-density as a function of the stress-state. This information was available for the HN–BN–MI SiC and SYL–BN–MI SiC systems.14,15 Fig. 6 shows the predictions for the two composite systems and the actual rupture data from Refs. 14,15. Table 1 lists the experimentally determined variables that went into the model for both systems and Fig. 7 shows the stress-dependent crack density as determined from measured crack densities from some of the rupture specimens. The two extremes predict the rupture behavior relatively well. The kinetic-limit seems to overestimate rupture time slightly. One possible reason for this overestimate is the presence of a possible stress-concentration on the outer perimeter of bridging fibers in a matrix crack,24,25 which was not taken into account in the model.c The model also slightly underestimates the rupture times for the SYL composites for the single fiber limit. One issue with SYL fiber composites in general is a relatively high t and the possibility that minor to moderate local load sharing conditions exist even for room temperature failure.26 If this is the case, the ultimate strength of the fibers in the as-produced composite, so(composite), would be underestimated from the composite ultimate strength, sult [Eq. (14)], which was based on global load sharing assumptions. This would effectively reduce the estimated time-dependent fiber strength and predict shorter rupture times for a given stress. The model predicts a fiber gage-length dependence for stress-rupture. This provides an opportunity to independently test model predictions. This was verified by precracking composites at room temperature so that they possess a larger crack density than that from stressrupture at an applied stress less than the precrack condition. For the HN–BN–MI specimen, some precracked experiments were performed. A precrack stress of 200 MPa was performed at room temperature and then the specimen was subjected to stress-rupture at a lower stress. This precrack condition resulted in a crack density of 2 cracks/mm. The increased lengths of loaded fibers resulted in significantly shorter rupture times at lower stresses than the specimens that were not precracked as the model also predicted fairly well (Fig. 6a). The run-out stress was slightly underestimated by the model. However, the model effectively predicted the decrease in stress-rupture time with increasing crack density. Fig. 7. Stress-dependence for matrix crack density for SYL and HN composites. The curves were based on post-test measurements of failed specimens. Fig. 6. Stress-rupture at 815 C in air of BN interphase composites with (a) HN fibers and (b) SYL fibers. c For the model developed by Evans et al. 25 of a C interphase composite system, fiber degradation was due to oxide scale growth and unbridged crack growth was due to fiber failure at the perimeter of the matrix crack due to the increased stress-concentration of the unbridged crack. A fundamental difference with the model proposed here is that an interior strongly bonded fiber can trigger the failure of the entire region of strongly bonded fibers. 2784 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787
G N. Morscher, J D. Cawley/ Journal of the European Ceramic Society 22(2002)2777-2787 4. Improvements in intermediate temperature stress- All three of the earlier approaches have been demon rupture of SiC/BN SiC composites strated for SYL reinforced composites. Fig 8 shows the dramatic improvements in intermediate stress rupture Based on the understanding of the mechanistic pro- properties for the three approaches described above in cess leading to intermediate temperature stress-rupture, comparison to the data from the material modeled in in part derived from the development and verification of Fig. 7. 14,15 500-h rupture stresses in excess of 200 MPa the earlier model, a few approaches have been employed were common for all three of the approaches, the"out to improve intermediate temperature stress-rupture. The side debonding"and Si-doped bn interphase compo- most desirable would to use a more durable interphase sites performed the best. Precrack experiments were than BN. However, to date, no real interphase has pre- performed for fiber-spread composites and for"outside sented itself, but the bn can be improved. For example, debonding "composites in Ref. 31( Fig. 9). The fiber composites with higher bn processing temperature or spread composites did show a decrease in stress-rupture bn doped with Si have shown greater resistance to properties with increased crack density whereas the oxygen and water containing environments at inter- outside debonding composites did not, in fact"out mediate temperatures. Si-doped BN was especially side debonding "composites showed improvement. In asistant to these environments; however, it requires other words, stress-rupture of fiber-spread composites processing temperatures on the order of 1400C, too still possess agage-length'"dependence whereas for high for processing of preforms. Some progress has outside debonding "this does not appear to be the been made with coating large individual pieces of woven case. The kinetic limit as determined for the composites loth with the higher processing temperature Si-doped modeled in Fig. Sb no longer applies to either of these BN interphase and then stacking the woven cloth to fabricate MI composites. 28 Since fiber separation is one of the controlling factors de Debonding-350 in fiber-to-fiber fusion, one logical technique to slow e-S-doped BN this process down would be to spread fibers further apart. By increasing the time for fiber fusion, more 14001Fiber-Spreading fibers would fail in a matrix crack under global load Spread SYL or SYL旧B L1200 sharing conditions, thereby reducing the available fibers that could fail after fibers are strongly bonded to one 5 1000 another or to the matrix. Two techniques have enabled fiber spreading that results in greater degree of fiber-to- fiber separation:(1)a proprietary approach to spread fibers in woven fabric and(2) heat-treating woven fabric to produce a-100 nm thick in-situ BN layer on the SYL fibers. 30 The former mechanically separates Time to fail, hr fibers resulting in fewer near fiber-to-fiber contacts and Fig. 8. Stress-rupture properties at 815oC in air of cor greater coverage of all the fibers with BN; the latter woven BN interphase composites and composites with produces a high-temperature BN layer that separates improvements fibers by an additional 200 nm. The in-situ BN Sylra mic R fibers are referred to as syl-iBN Another improvement has been to alter the interface outside debonding BN' where interface debonding and sliding occur within the interphase region from the fiber/BN interface to the N/matrix interface, i.e. "outside debonding. 3I This 350 MPa Precrack would be similar in concept to multi-layer coatings 32 where interphase oxidation is engineered to occur as far away from the fiber surface as possible. The multi-layer C/SiC interphase coating of NIC/SiC composites in Ref 37 does show slight improvement over other C-inter phase NC/SiC composites at intermediate tempera- tures. For"outside debonding"BN interphases, oxygen and water vapor do not have direct access to the fibers 30 In order to fuse fibers to the matrix or to one another Time. Hours oxidation must occur through the thickness of the bn, which is relatively slow because the boria reacts with the Fig. 9. Stress-rupture of fiber-spread composite and outside-debond- ing BN composite. Both composites had a fiber volume fraction in the Sic in the matrix crack to effectively seal the matrix crack loading direction of 0.2
4. Improvements in intermediate temperature stressrupture of SiC/BN/SiC composites Based on the understanding of the mechanistic process leading to intermediate temperature stress-rupture, in part derived from the development and verification of the earlier model, a few approaches have been employed to improve intermediate temperature stress-rupture. The most desirable would to use a more durable interphase than BN. However, to date, no real interphase has presented itself, but the BN can be improved. For example, composites with higher BN processing temperature or BN doped with Si have shown greater resistance to oxygen and water containing environments at intermediate temperatures.27 Si-doped BN was especially resistant to these environments; however, it requires processing temperatures on the order of 1400 C, too high for processing of preforms. Some progress has been made with coating large individual pieces of woven cloth with the higher processing temperature Si-doped BN interphase and then stacking the woven cloth to fabricate MI composites.28 Since fiber separation is one of the controlling factors in fiber-to-fiber fusion, one logical technique to slow this process down would be to spread fibers further apart. By increasing the time for fiber fusion, more fibers would fail in a matrix crack under global load sharing conditions, thereby reducing the available fibers that could fail after fibers are strongly bonded to one another or to the matrix. Two techniques have enabled fiber spreading that results in greater degree of fiber-to- fiber separation: (1) a proprietary approach to spread fibers in woven fabric 29 and (2) heat-treating woven fabric to produce a100 nm thick in-situ BN layer on the SYL fibers.30 The former mechanically separates fibers resulting in fewer near fiber-to-fiber contacts and greater coverage of all the fibers with BN; the latter produces a high-temperature BN layer that separates fibers by an additional 200 nm. The in-situ BN Sylramic1 fibers are referred to as SYL-iBN. Another improvement has been to alter the interface where interface debonding and sliding occur within the interphase region from the fiber/BN interface to the BN/matrix interface, i.e. ‘‘outside debonding’’. 31 This would be similar in concept to multi-layer coatings 32 where interphase oxidation is engineered to occur as far away from the fiber surface as possible. The multi-layer C/SiC interphase coating of NIC/SiC composites in Ref. 37 does show slight improvement over other C-interphase NIC/SiC composites at intermediate temperatures. For ‘‘outside debonding’’ BN interphases, oxygen and water vapor do not have direct access to the fibers. In order to fuse fibers to the matrix or to one another, oxidation must occur through the thickness of the BN, which is relatively slow because the boria reacts with the SiC in the matrix crack to effectively seal the matrix crack. All three of the earlier approaches have been demonstrated for SYL reinforced composites. Fig. 8 shows the dramatic improvements in intermediate stress rupture properties for the three approaches described above in comparison to the data from the material modeled in Fig. 7.14,15 500-h rupture stresses in excess of 200 MPa were common for all three of the approaches, the ‘‘outside debonding’’ and Si-doped BN interphase composites performed the best. Precrack experiments were performed for fiber-spread composites and for ‘‘outside debonding’’ composites in Ref. 31 (Fig. 9). The fiberspread composites did show a decrease in stress-rupture properties with increased crack density whereas the ‘‘outside debonding’’ composites did not, in fact ‘‘outside debonding’’ composites showed improvement. In other words, stress-rupture of fiber-spread composites still possess a ‘‘gage-length’’ dependence whereas for ‘‘outside debonding’’ this does not appear to be the case. The kinetic limit as determined for the composites modeled in Fig. 5b no longer applies to either of these Fig. 8. Stress-rupture properties at 815 C in air of conventional woven BN interphase composites and composites with interphase improvements. Fig. 9. Stress-rupture of fiber-spread composite and outside-debonding BN composite. Both composites had a fiber volume fraction in the loading direction of 0.2. G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787 2785
G N Morscher, J.D. Cawley / Journal of the European Ceramic Society 22(2002)2777-2787 ses(Fig. 9)and a ne ew ulme-de pendence for fiber ment of Ceramic Components in a MS9001FA Gas Turbine, embrittlement would have to be determined in order to ASME.98-GT-186,1998 model these composites. It obviously takes a longer time 3. Kameda, T, Itoh, Y, Hishata. T. and Okamura, T. Develop- to fuse fibers together that have longer separation dis ment of Continuous Fiber Reinforced Reaction Sintered Silicon Carbide Matrix Composite for Gas Turbine Hot Parts Applica- tances. "Outside debonding"composites appear to be tion.(ASME,2000-GT67)2000. contro olled by the time it takes to oxidize through xidize through the 4. Heredia, F. E. McNulty. J. C, Zok, F. W. and E BN interphase layer, approximately 100 h at 815Cin Oxidation embrittlement probe for ceramic-matrix composites. J.Am.Cera.Soc.,1995,78.209 5. Filipuzzi. L, Camus, G. Naslain, R and Thebault, J, Oxidation mechanisms and kinetics of lD-SiC/C/SiC composite materials: I 5. Conclusions 6. Eckel, A J, Cawley, J. D and Parthasarathy, T, Oxidation of a ntermediate temperature strength degradation of ntinuous carbon phase in a nonreactive matrix. J. Am. Ceram. Sic/Sic composites is due to a"pest condition pri- Soc.1995,78,972-980. 7. Lara- Curzio. E. Ferber. M. K. and Tortorelli. P. F. Interface marily caused by the oxidation of the interphase separ- xidation and stress-rupture of Nicalon/SiC CFCC's at inter- ating the fibers and the matrix. Although, BN ediate temperatures. In Key Engineering Materials, Vols. 127 interphases are superior to carbon interphase compo- 131. Trans. Tech Publications, Switzerland, 1997, pp. 1069-1082. sites, they still exhibit significant degradation in stress- 8. Lara-Curzio, E, Stress-rupture of Nicalon/SiC continuous fiber rupture properties at intermediate temperatures. The cramic matrix composites in air at 950C.. Am 1997.80.32683272 main factor causing this strength degradation is the 9. Martinez-Fernandez, J and Morscher. G. N. Room and elevated fusion of fibers to one another in a matrix crack that is temperature tensile properties of single tow Hi-Nicalon, carbon exposed to the oxidizing environment. The amount of terphase CVI SiC matrix strength degradation is dependent on the kinetics for 20.2627-2636 Fusion of fibers to one another the number of matrix un.E. Y, Nutt. S. R. and Brennan, JJ. Oxi dation of BN-coated SiC fibers in ceramic-matrix composites. cracks, and the applied stress state. It was shown that Am Cera. Soc. 1996 the stress-rupture properties of SiC/BN/SiC composites 11. Jacobson, N.S., Morscher, GN,Bryant, D. R and Tressler, could be effectively modeled using an approach that R. E, High-temperature oxidation of boron nitride: Il, boron considers the probability of fiber failure in relation te ayers in composites. JAnm. Ceram. Soc.,1999.82.1473- the likelihood that the fiber had already been fused to its 12. Lin. H. T. and Becher. P. F. Effect of coating on lifetime of neighbor or the matrix. One important aspect of the calon fiber-silicon carbide composites in air. Materials Science model that was verified was the increased susceptibility d engineerin,1997,A231,143-150. to stress-rupture for composites with a greater number 13. Morscher, G. N, Tensile stress-rupture of SiC/SiCm mini- of matrix cracks composites with carbon and boron nitride interphases at elevated ecently, improvements have been made for BN- mperatures in air. J. Am. Ceram. Soc., 1997, 80, 2029-2042 14. Morscher. G. N. Hurst. J and Brewer. D. Intermediate-tem- nterphase composites. These include, Si-doped BN, perature stress rupture of a woven Hi-Nicalon, BN-interphase composites with more effective fiber spreading, and BN Sic-matrix composite in air. J. Am. Ceram Soc., 2000, 83, 1441- nterphases wh bonding Iding occur between the bn layer and the matrix rather than the 15. Morscher. G. N. and Hurst, J, Stress-rupture and stress-relaxa- and the bn layer. Consistent with the mechanistics tion of Sic/SiC composites at intermediate temperature Ceram Eng.Sci.Proc2001,22,539546. assumed in the model, for composites made with these 16. Yun, H. M. and DiCarlo, J. A, Time/temperature dependent modifications, the 500-h rupture stress increased from tensile strength of Sic and Al203-based fibers. In in Ceramic about 155 MPa for conventional composites to over 200 ansactions, ol. 74. Advances in Ce Matrix Composit MPa. For"outside debonding 500-h rupture stresses l. ed. N. P. Bansal and J. P. Singh. American Ceramic Society close to 250 MPa have been attained. this does not Westerville OH, 1996, pp. 17-26 necessarily eliminate the"pest regime"for these com- 17. Ogbuji, L. U.J. T, Identification of a Carbon Sublayer in a Hi- calon/ BN/SiC Composite. J. Mater. Sci. Letters, 199 posites; however, these approaches would significantly increase the stress range these composites could with 18. Curtin, w.A., Multiple matrix crack spacing in brittle matrix stand at intermediate temperatures in oxidizing envir- omposites. J. Am. Ceram Soc., 1991, 74, 2837. onments 19. Curtin, w. A. Ahn. B K. and Takeda. N, Modeling brittle and tough stress-strain behavior in unidirectional ceramic matrix opposites. Acta Mater., 1998, 46. 3409-3420 20. lyengar, N. and Curtin, w.A., Time-dependent failure in fiber References reinforced composites by fiber degradation. Acta Mater., 1997 21. Marshall. D. B. Cox.B. N. and Evans.A.G. The mechanics of 1. Brewer, D, HSR/EPM Combustor Materials Development Pro- matrix cracking in brittle-matrix fiber composites. Acta Metal. gram. Mater. Sci. Eng. 4,, 1999, A261, 284-291 1985.33.2013-2021 2. Grondahl, C. M. and Tsuchiya, T. Performance Benefit Assess- 22. Cao, H. and Thouless, M. D, Tensile tests of ceramic-matrix
cases (Fig. 9) and a new time-dependence for fiber embrittlement would have to be determined in order to model these composites. It obviously takes a longer time to fuse fibers together that have longer separation distances. ‘‘Outside debonding’’ composites appear to be controlled by the time it takes to oxidize through the BN interphase layer, approximately 100 h at 815 C in air. 5. Conclusions Intermediate temperature strength degradation of SiC/SiC composites is due to a ‘‘pest’’ condition primarily caused by the oxidation of the interphase separating the fibers and the matrix. Although, BN interphases are superior to carbon interphase composites, they still exhibit significant degradation in stressrupture properties at intermediate temperatures. The main factor causing this strength degradation is the fusion of fibers to one another in a matrix crack that is exposed to the oxidizing environment. The amount of strength degradation is dependent on the kinetics for fusion of fibers to one another, the number of matrix cracks, and the applied stress state. It was shown that the stress-rupture properties of SiC/BN/SiC composites could be effectively modeled using an approach that considers the probability of fiber failure in relation to the likelihood that the fiber had already been fused to its neighbor or the matrix. One important aspect of the model that was verified was the increased susceptibility to stress-rupture for composites with a greater number of matrix cracks. Recently, improvements have been made for BNinterphase composites. These include, Si-doped BN, composites with more effective fiber spreading, and BN interphases where the debonding and sliding occur between the BN layer and the matrix rather than the fiber and the BN layer. Consistent with the mechanistics assumed in the model, for composites made with these modifications, the 500-h rupture stress increased from about 155 MPa for conventional composites to over 200 MPa. For ‘‘outside debonding’’, 500-h rupture stresses close to 250 MPa have been attained. This does not necessarily eliminate the ‘‘pest regime’’ for these composites; however, these approaches would significantly increase the stress range these composites could withstand at intermediate temperatures in oxidizing environments. References 1. Brewer, D., HSR/EPM Combustor Materials Development Program. Mater. Sci. Eng. A,, 1999, A261, 284–291. 2. Grondahl, C. M. and Tsuchiya, T. Performance Benefit Assessment of Ceramic Components in a MS9001FA Gas Turbine, ASME. 98-GT-186, 1998. 3. Kameda, T., Itoh, Y., Hishata, T., and Okamura, T. Development of Continuous Fiber Reinforced Reaction Sintered Silicon Carbide Matrix Composite for Gas Turbine Hot Parts Application. (ASME, 2000-GT-67) 2000. 4. Heredia, F. E., McNulty, J. C., Zok, F. W. and Evans, A. G., Oxidation embrittlement probe for ceramic-matrix composites. J. Am. Ceram. Soc., 1995, 78, 2097. 5. Filipuzzi, L., Camus, G., Naslain, R. and Thebault, J., Oxidation mechanisms and kinetics of 1D-SiC/C/SiC composite materials: I, an experimental approach. J. Am. Ceram. Soc., 1994, 77, 459– 466. 6. Eckel, A. J., Cawley, J. D. and Parthasarathy, T., Oxidation of a continuous carbon phase in a nonreactive matrix. J. Am. Ceram. Soc., 1995, 78, 972–980. 7. Lara-Curzio, E., Ferber, M. K., and Tortorelli, P. F., Interface oxidation and stress-rupture of Nicalon/SiC CFCC’s at intermediate temperatures. In Key Engineering Materials, Vols. 127– 131. Trans. Tech. Publications, Switzerland, 1997, pp. 1069–1082. 8. Lara-Curzio, E., Stress-rupture of Nicalon/SiC continuous fiber ceramic matrix composites in air at 950 C. J. Am. Ceram. Soc., 1997, 80, 3268–3272. 9. Martinez-Fernandez, J. and Morscher, G.N., Room and elevated temperature tensile properties of single tow Hi-Nicalon, carbon interphase CVI SiC matrix minicomposites. J. Eur. Ceram. Soc., 20, 2627–2636. 10. Sheldon, B. W., Sun, E. Y., Nutt, S. R. and Brennan, J. J., Oxidation of BN-coated SiC fibers in ceramic-matrix composites. J. Am. Ceram. Soc., 1996, 79, 539–543. 11. Jacobson, N. S., Morscher, G. N., Bryant, D. R. and Tressler, R. E., High-temperature oxidation of boron nitride: II, boron nitride layers in composites. J. Am. Ceram. Soc., 1999, 82, 1473– 1482. 12. Lin, H. T. and Becher, P. F., Effect of coating on lifetime of Nicalon fiber-silicon carbide composites in air. Materials Science andEngineering, 1997, A231, 143–150. 13. Morscher, G. N., Tensile stress-rupture of SiCf/SiCm minicomposites with carbon and boron nitride interphases at elevated temperatures in air. J. Am. Ceram. Soc., 1997, 80, 2029–2042. 14. Morscher, G. N., Hurst, J. and Brewer, D., Intermediate-temperature stress rupture of a woven Hi-Nicalon, BN-interphase, SiC-matrix composite in air. J. Am. Ceram. Soc., 2000, 83, 1441– 1449. 15. Morscher, G. N. and Hurst, J., Stress-rupture and stress-relaxation of SiC/SiC composites at intermediate temperature. Ceram. Eng. Sci. Proc., 2001, 22, 539–546. 16. Yun, H. M. and DiCarlo, J. A., Time/temperature dependent tensile strength of SiC and Al2O3-based fibers. In in Ceramic Transactions, Vol. 74. Advances in Ceramic-Matrix Composites III, ed. N. P. Bansal and J. P. Singh. American Ceramic Society, Westerville OH, 1996, pp. 17–26. 17. Ogbuji, L. U. J. T., Identification of a Carbon Sublayer in a HiNicalon/BN/SiC Composite. J. Mater. Sci. Letters., 1999, 18, 1825–1827. 18. Curtin, W. A., Multiple matrix crack spacing in brittle matrix composites. J. Am. Ceram. Soc., 1991, 74, 2837. 19. Curtin, W. A., Ahn, B. K. and Takeda, N., Modeling brittle and tough stress–strain behavior in unidirectional ceramic matrix composites. Acta Mater., 1998, 46, 3409–3420. 20. Iyengar, N. and Curtin, W. A., Time-dependent failure in fiberreinforced composites by fiber degradation. Acta Mater., 1997, 45, 1489. 21. Marshall, D. B., Cox, B. N. and Evans, A. G., The mechanics of matrix cracking in brittle-matrix fiber composites. Acta Metal., 1985, 33, 2013–2021. 22. Cao, H. and Thouless, M. D., Tensile tests of ceramic-matrix 2786 G.N. Morscher, J.D. Cawley / Journal of the European Ceramic Society 22 (2002) 2777–2787