《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_OXIDE-OXIDE-31

Availableonlineatwww.sciencedirectcom ScienceDirect E噩≈RS ELSEVIER Joumal of the European Ceramic Society 27(2007)4603-4611 www.elsevier.comlocate/jeurceramsoc Creep damage resistance of ceramic-matrix composites B Wilshire MR Bache Materials Research Centre, School of Engineering, University of Wales Swansea, Singleton Park, Swansea SA2 8PP UK Received 22 November 2006: received in revised form 20 March 2007: accepted 25 March 2007 Available online 12 June 2007 The tensile creep and creep fracture properties in air at 1300C are documented for two ceramic fibre-reinforced ceran (CFCMCS). These recently developed materials were produced with woven bundles of Hi-Nicalon fibres reinforcing either SiBC matrices, allowing data comparisons to be made with similar CFCMCs having different fibre-matrix combinations. The r the longitudinal fibres govern the rates of strain accumulation and crack growth, but the fracture characteristics are determined by fibre failure aused by oxygen penetration as matrix cracks develop. The analysis then suggests that carbon fibre-reinforced doloma-matrix composites could offer a combination of creep-resistant fibres and creep damage-resistant matrices suitable for long-term load-bearing in high-temperature oxidizing environments. 2007 Elsevier ltd. all rights reserved. Keywords: Composites; Creep: Fracture; Damage resistance; Structural applications 1. ntroduction mined in air at 1300C are documented for two advanced sic fibre-reinforced products. These results are then discussed with Since the 1950s, the introduction of new metallic materi- reference to data sets reported for several related materials als and improved manufacturing technologies have supported In this way, a straightforward evaluation can be made of the major advances in the power, efficiency and reliability of creep performance improvement achieved using aeroengines, underpinning impressive increases in the range, performance and safety of civil and military aircraft. In seek-(a) different fibre types for composites with comparable matri- ing further gains in thrust-to-weight ratio and fuel economy, ces. fibre-matrix interfaces and fibre architectures and ceramic-matrix composites reinforced with continuous ceramic (b) different matrices for composites having similar fibre type fibres(CFCMCs)represent a recognized material development fibre configurations and fibre-matrix interfaces opportunity. Even so, particularly as the civil aviation sec- tor becomes increasingly cost conscious, emphasis must be By clarifying the principal creep life-limiting features of this directed towards product cost reduction as well as component product range, proposals are made for a relatively low-cost com life enhancement posite which may provide the creep damage resistance needed Although a wide variety of CFCMCs have been developed, for high-temperature load-bearing applications involving pro- special attention has been focussed on composites with either longed exposure in oxidizing environme Sic or Al2 O3 matrices, reinforced with interwoven bundles of silicon carbide fibres. However, the creep fracture char- 2. Experimental procedures acteristics of these types of composite at high temperatures are this reason, in the present snody, creep property values deter- produced with high Hi-Nicalon TM ibrer, d for two C fibres, adversely affected by oxidizing atmospheres,bi.ethe service To quantify the effects of changing the reinforcing fibres, conditions typically experienced by aeroengine components For creep property measurements now recorde re compared with results available for similar composites containing Nicalon M NLM202 fibres(Nippon Carbon Co., Japan). Hi-NicalonTM Corresponding author. Tel. +44 1792 295243: fax: +44 1792 295244 fibres are characterized by superior elastic moduli and creep E-mail address: b wilshire@swansea. ac uk(B. wilshire) istance, achieved through processing operations adopted to 0955-2219/S-see front matter o 2007 Elsevier Ltd. All rights reserved. doi: 10. 1016/j-jeurceramsoc. 2007.03.029

Journal of the European Ceramic Society 27 (2007) 4603–4611 Creep damage resistance of ceramic–matrix composites B. Wilshire ∗, M.R. Bache Materials Research Centre, School of Engineering, University of Wales Swansea, Singleton Park, Swansea SA2 8PP, UK Received 22 November 2006; received in revised form 20 March 2007; accepted 25 March 2007 Available online 12 June 2007 Abstract The tensile creep and creep fracture properties in air at 1300 ◦C are documented for two ceramic fibre-reinforced ceramic–matrix composites (CFCMCs). These recently developed materials were produced with woven bundles of Hi-NicalonTM fibres reinforcing either A12O3 or enhanced SiBC matrices, allowing data comparisons to be made with similar CFCMCs having different fibre–matrix combinations. The results confirm that the longitudinal fibres govern the rates of strain accumulation and crack growth, but the fracture characteristics are determined by fibre failure caused by oxygen penetration as matrix cracks develop. The analysis then suggests that carbon fibre-reinforced doloma–matrix composites could offer a combination of creep-resistant fibres and creep damage-resistant matrices suitable for long-term load-bearing service in high-temperature oxidizing environments. © 2007 Elsevier Ltd. All rights reserved. Keywords: Composites; Creep; Fracture; Damage resistance; Structural applications 1. Introduction Since the 1950s, the introduction of new metallic materi￾als and improved manufacturing technologies have supported major advances in the power, efficiency and reliability of aeroengines, underpinning impressive increases in the range, performance and safety of civil and military aircraft.1,2 In seek￾ing further gains in thrust-to-weight ratio and fuel economy, ceramic–matrix composites reinforced with continuous ceramic fibres (CFCMCs) represent a recognized material development opportunity. Even so, particularly as the civil aviation sec￾tor becomes increasingly cost conscious, emphasis must be directed towards product cost reduction as well as component life enhancement. Although a wide variety of CFCMCs have been developed, special attention has been focussed on composites with either SiC or A12O3 matrices, reinforced with interwoven bundles of silicon carbide fibres.2,3 However, the creep fracture char￾acteristics of these types of composite at high temperatures are adversely affected by oxidizing atmospheres,4–6 i.e. the service conditions typically experienced by aeroengine components. For this reason, in the present study, creep property values deter- ∗ Corresponding author. Tel.: +44 1792 295243; fax: +44 1792 295244. E-mail address: b.wilshire@swansea.ac.uk (B. Wilshire). mined in air at 1300 ◦C are documented for two advanced SiC fibre-reinforced products. These results are then discussed with reference to data sets reported for several related materials.7–11 In this way, a straightforward evaluation can be made of the creep performance improvement achieved using (a) different fibre types for composites with comparable matri￾ces, fibre–matrix interfaces and fibre architectures and (b) different matrices for composites having similar fibre types, fibre configurations and fibre–matrix interfaces. By clarifying the principal creep life-limiting features of this product range, proposals are made for a relatively low-cost com￾posite which may provide the creep damage resistance needed for high-temperature load-bearing applications involving pro￾longed exposure in oxidizing environments. 2. Experimental procedures To quantify the effects of changing the reinforcing fibres, creep property measurements now recorded for two CFCMCs produced with high Hi-NicalonTM fibres are compared with results available for similar composites containing NicalonTM NLM202 fibres (Nippon Carbon Co., Japan). Hi-NicalonTM fibres are characterized by superior elastic moduli and creep resistance, achieved through processing operations adopted to 0955-2219/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2007.03.029

4604 B. wilshire, M.R. Bache /Journal of the European Ceramic Society 27(2007)4603-4611 eliminate the amorphous silicon oxy-carbide phase (SiC:O,) operations, all matrices had porosities of 15% or more Specifi which reduces the creep strength of NicalonM NLM20 ly, small pores were present in the matrix regions within fibres 12, 3 fibre bundles, with large pores(termed macro-pores) between To consider the effects of changing the matrix type, one of the plies and at yarn intersections within the plies the present Hi- NicalonM-reinforced composites had an alu Having selected composites with similar'macro-structures mina matrix, giving a material referred to as HNSiCr-Al2O3. comparable test procedures were used to determine the creep and The other composite was prepared with an 'enhanced'Sic creep fracture properties. Thus, for the present HNSiCr-A1203 matrix, containing boron-based additives which form a sealant and HNSiCr-SiBC materials, tensile creep tests were carried out glass to limit oxygen penetration into the material during creep in air at 1300 C using a servo-hydraulic machine in load control exposure.+ This product is designated as HNSiCr-SiBC mode. The load train included hydraulic wedge grips, an align- For tensile creep tests carried out in air at 1300C, ment fixture, a twin-zone split furnace and a high-temperature the behaviour patterns displayed by the HNSiCr-Al2O3 and extensometer. The tests were undertaken using flat specimens HNSiCr-SiBC samples are analyzed in relation to data obtained of 2 mm thickness and mm width, with 40 mm gauge lengths under the same conditions for two groups of CFCMCs machined such that the tensile stress axes were parallel (0o) to one of the 0/90 fibre directions. These procedures are the (a)The consequences of changing only the fibre type Same as those adopted to test the SiCr-SiC, SiCr-SiBCIO and arent by the creep properties of HNSiCr-Sic products. The present experimental methods HNSiCe-AlO with those of an alumina-matrix com- also gave results indistinguishable from those reported when posite reinforced with NicalonTM NLM202 fibres. called constant-load creep machines were used with the SiCe-sice SiCf-Al2O3. With both materials, the fibres were given and SiCt-A12O3 composites a thin boron nitride coating before 5 um thick SiC coat ings were deposited by chemical vapour infiltration(CVI), 3. Results and discussion resulting in double BN/SiC interfaces. The A12O3 matrices were then formed by in situ directional oxidation of liq Over the stress ranges investigated at 1300 C for the uid aluminium, as previously described for the SicCr-Al2O3 HNSiCr-Al3O3 and HNSiCr-SiBC samples, the general man- ner in which the creep strain(E)increases with time (t)is (b) The HNSiCr-SiBC samples relate to three different com- similar to that reported previously for all other CFCMCs now posites with SiC fibres reinforcing SiC matrices. All of these considered. 6- Thus, as illustrated by the e/t trajectories for the products were fabricated with carbon interfaces(0.5 um HNSiCf-SiBC material in Fig. 1, following the initial loading thick), before CVI processing to introduce the polycrys- strain, the creep rate decays continuously, reaching a minimum alline SiC matrices. However, the fibre-matrix combination rate(Em) just prior to failure. Little or no period of accelerat- differed. with ing tertiary creep is then apparent before fracture occurs after a Nicalon NLM202 fibres reinforcing,a standard Sic time(tr)when the total creep strain reaches the limiting creep matrix(called Sicr-Sic), ductility(Et). Hence, to compare the creep and creep fracture Nicalon M NLM202 fibres reinforcing an'enhanced properties of CFCMCs produced with NicalonM NLM202 or SiC matrix(called SiCr-SiBC)and Hi-NicalonTM fibres reinforcing Al2 O3, SiC or SiBC matrices Hi-Nicalon"M fibres reinforcing"a standard SiC matrix the values of Em, tr and ef were determined from each creep (called HNSiCr-SiC). curve The principal distinguishing features of these CFCMCs are 3. 1. Relative strengths of fibres and matrices summarized in Table 1. Each composite contained approx imately 40 vol %o fibres with average diameters of 15 um, Fig. 2 shows the variations of the mi incorporated as bundles of about 500 fibres woven to obtain rate with stress at 1300 C for SiCr-Al2O3'and the 2D layers of fabric. The woven layers or plies were then aligned present HNSiCr-AlO3 material, together with results for and stacked to produce preforms having balanced 0/90 archi- an alumina-matrix composite reinforced with 25 vol. Sic tectures. Moreover, after the fibre coating and densification whiskers,now termed SiCw-Al2O3. Also included in Fig. 2 Distinguishing features of fibre-reinforced composites Material designation Fibre type Matrix material Interface type Reference Al O3 Al2O3 BN/SIC Al2O3 BN/SiC Hi-Nicalon Enhanced SiC Carbon Carbon SiCe-SIBC Nicalon NLM 20 Enhanced SiC Carbon HNSiCe-SiC Hi-NicalonTM Carbon

4604 B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 eliminate the amorphous silicon oxy-carbide phase (SiCxOy) which reduces the creep strength of NicalonTM NLM202 fibres.12,13 To consider the effects of changing the matrix type, one of the present Hi-NicalonTM-reinforced composites had an alu￾mina matrix, giving a material referred to as HNSiCf–A12O3. The other composite was prepared with an ‘enhanced’ SiC matrix, containing boron-based additives which form a sealant glass to limit oxygen penetration into the material during creep exposure.14 This product is designated as HNSiCf–SiBC. For tensile creep tests carried out in air at 1300 ◦C, the behaviour patterns displayed by the HNSiCf–A12O3 and HNSiCf–SiBC samples are analyzed in relation to data obtained under the same conditions for two groups of CFCMCs. (a) The consequences of changing only the fibre type becomes apparent by comparing the creep properties of HNSiCf–A12O3 with those of an alumina–matrix com￾posite reinforced with NicalonTM NLM202 fibres, called SiCf–A12O3. 7 With both materials, the fibres were given a thin boron nitride coating before ∼5m thick SiC coat￾ings were deposited by chemical vapour infiltration (CVI), resulting in double BN/SiC interfaces. The A12O3 matrices were then formed by in situ directional oxidation of liq￾uid aluminium, as previously described for the SiCf–Al2O3 specimens.7 (b) The HNSiCf–SiBC samples relate to three different com￾posites with SiC fibres reinforcing SiC matrices. All of these products were fabricated with carbon interfaces (∼0.5m thick), before CVI processing to introduce the polycrys￾talline SiC matrices. However, the fibre–matrix combination differed, with • NicalonTM NLM202 fibres reinforcing6,9 a standard SiC matrix (called SiCf–SiC), • NicalonTM NLM202 fibres reinforcing10 an ‘enhanced’ SiC matrix (called SiCf–SiBC) and • Hi-NicalonTM fibres reinforcing11 a standard SiC matrix (called HNSiCf–SiC). The principal distinguishing features of these CFCMCs are summarized in Table 1. Each composite contained approx￾imately 40 vol.% fibres with average diameters of ∼15m, incorporated as bundles of about 500 fibres woven to obtain 2D layers of fabric. The woven layers or plies were then aligned and stacked to produce preforms having balanced 0/90◦ archi￾tectures. Moreover, after the fibre coating and densification operations, all matrices had porosities of 15% or more. Specifi- cally, small pores were present in the matrix regions within the fibre bundles, with large pores (termed macro-pores) between the plies and at yarn intersections within the plies.7 Having selected composites with similar ‘macro-structures’, comparable test procedures were used to determine the creep and creep fracture properties. Thus, for the present HNSiCf–A12O3 and HNSiCf–SiBC materials, tensile creep tests were carried out in air at 1300 ◦C using a servo-hydraulic machine in load control mode. The load train included hydraulic wedge grips, an align￾ment fixture, a twin-zone split furnace and a high-temperature extensometer. The tests were undertaken using flat specimens of 2 mm thickness and 8 mm width, with 40 mm gauge lengths machined such that the tensile stress axes were parallel (0◦) to one of the 0/90◦ fibre directions. These procedures are the same as those adopted to test the SiCf–SiC,9 SiCf–SiBC10 and HNSiCf–SiC products.11 The present experimental methods also gave results indistinguishable from those reported when constant-load creep machines6 were used with the SiCf–SiC6 and SiCf–A12O3 composites.7 3. Results and discussion Over the stress ranges investigated at 1300 ◦C for the HNSiCf–A13O3 and HNSiCf–SiBC samples, the general man￾ner in which the creep strain (ε) increases with time (t) is similar to that reported previously for all other CFCMCs now considered.6–11 Thus, as illustrated by the ε/t trajectories for the HNSiCf–SiBC material in Fig. 1, following the initial loading strain, the creep rate decays continuously, reaching a minimum rate (ε˙m) just prior to failure. Little or no period of accelerat￾ing tertiary creep is then apparent before fracture occurs after a time (tf) when the total creep strain reaches the limiting creep ductility (εf). Hence, to compare the creep and creep fracture properties of CFCMCs produced with NicalonTM NLM202 or Hi-NicalonTM fibres reinforcing A12O3, SiC or SiBC matrices, the values of ε˙m, tf and εf were determined from each creep curve. 3.1. Relative strengths of fibres and matrices Fig. 2 shows the variations of the minimum creep rate with stress at 1300 ◦C for SiCf–A12O3 7 and the present HNSiCf–Al2O3 material, together with results for an alumina–matrix composite reinforced with 25 vol.% SiC whiskers,15 now termed SiCw–A12O3. Also included in Fig. 2 Table 1 Distinguishing features of fibre-reinforced composites Material designation Fibre type Matrix material Interface type Reference HNSiCf–Al2O3 Hi-NicalonTM Al2O3 BN/SiC SiCf–Al2O3 NicalonTM NLM 202 Al2O3 BN/SiC 7 HNSiCf–SiBC Hi-NicalonTM Enhanced SiC Carbon SiCf–SiC NicalonTM NLM 202 SiC Carbon 6,9 SiCf–SIBC NicalonTM NLM 202 Enhanced SiC Carbon 10 HNSiCf–SiC Hi-NicalonTM SiC Carbon 11

B Wilshire, M.R. Bache /Journal of the European Ceramic Sociery 27(2007)4603-4611 4605 130 MPa 120 MPa 1051△scA203 0.015 100 MPa 110 MPa ▲ HNSICf-A203 0.010 0.005 aooEEcs 0.000 Time(ks) Stress(MPa) are Em measurements reported for Nicalon TM NLM202 fibres HNSiC-Sicll and HNSiCr-SiBC samples tested in air at C sips for Fig. 1. The variations of the creep strain with time for HNSiCr-SiBC samples sted at various stresses in air at 1300oC iCe-Al, O3 an to the stress axes occupy approximately one fifth of the testpiece The presence of 25 vol %o Sic whiskers significantly cross-sectional areas. Thus, the load-bearing capabilities of the increases the creep resistance of alumina. The porous Al2O3 alumina-matrix composites are governed by the longitudinal matrices formed by in situ oxidation of liquid aluminium must therefore have creep strengths much lower than that of the fully s are ss lev- or While the creep resistance of the fibres clearly exceeds that dense SiCw-Al2O3 samples. Yet, at comparable stress lev- of the matrices with the SiCr-Al2O3 and HNSiCr-Al2O3 sam- els, the Em values for the SiCw-Al2O3 specimens are several ples, it has been claimed that the matrices are more creep orders of magnitude faster than those for the HNSiCf-Al2O3 resistant than the fibres with the SiC-SiC, SiCf-SiBCO and and SiCr-Al2O3 products Consequently, the alumina matrices HNSiCr-SiC materials However, as evident from Fig 3, the make little contribution to the overall creep resistance of the Em values recorded for the SiCr-A12O3 testpieces are equal to fibre-reinforced composites. those for the SiCf-SiBC and lower than those for the SiCf-Sic <It is also apparent from Fig. 2 that stresses about five times specimens, with all three materials being reinforced with com- gher must be applied to the Nicalon NLM202 and Hi- able volume fractions of 0/90 NicalonMNLM202 fibre NicalonM fibres to obtain creep rates comparable with those for Similarly, with equivalent Hi-Nicalon reinforcement, the Em val the SiCf-Al2O3 and HNSiCr-Al2O3 specimens, respectively. ues for HNSiCf-A12O3 are equal to those for HNSiCf-SiBC and This result would be expected because these composites contain lower than those for the HNSiCr-SiC sample(Fig 3). He O vol. of interwoven 0/900 fibre bundles, so the fibres parallel porous SiC and SiBC matrices must be characterized by a creep resistance at least as poor as that for the weak A12O3 matrices Irrespective of whether the present series of CFCMCs were produced with A12O3, SiC or SiBC matrices, it is therefore clear oSiCw-Al203 that the creep strengths of the matrices are markedly inferior to those of the fibres. On this basis, from the results presented Cao- Mao in Figs. 3 and 4, the improvement in creep and creep rupture resistance achieved by replacing Nicalon NLM202 with Hi- Nicalon fibres can be quantified easily 3. 2. Creep data comparisons CFCMCs obviously display a stochastic strength response because of their essentially brittle character, coupled with the Stress(MPa) near-random nature of the size and distribution of the macro- scopic and microscopic flaws which are present. Even so, Fig. 2. The stress dependences of the minimum creep rates recorded for the recognizing that the em and tf measurements in Figs. 3 and 4 HNSiCr-Al2O3 composite in air at 1300 C compared with data available for SiCr-Al20,7 and a SiC whisker-reinforced composite, SiCw-Al2O3.15 Also were obtained for a range of composites tested in two diffe included are results reported at 1300C for Nicalon TM NLM202 6 and Hi. ent laboratories 6-11 the recorded data sets reveal remarkably NicalonTM fibres 17 In addition, stress-creep rate values at 1200.C are shown consistent patterns of property variation as the fibre-matrix com- for synthetic Cao-50% Mgo samples(Table 2)

B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 4605 Fig. 1. The variations of the creep strain with time for HNSiCf–SiBC samples tested at various stresses in air at 1300 ◦C. are ε˙m measurements reported for NicalonTM NLM202 fibres16 and one ε˙m value found for Hi-NicalonTM fibres.17 The presence of 25 vol.% SiC whiskers significantly increases the creep resistance of alumina.15 The porous A12O3 matrices formed by in situ oxidation of liquid aluminium must therefore have creep strengths much lower than that of the fully dense SiCw–A12O3 samples. Yet, at comparable stress lev￾els, the ε˙m values for the SiCw–A12O3 specimens are several orders of magnitude faster than those for the HNSiCf–A12O3 and SiCf–A12O3 products. Consequently, the alumina matrices make little contribution to the overall creep resistance of the fibre-reinforced composites.8 It is also apparent from Fig. 2 that stresses about five times higher must be applied to the NicalonTM NLM202 and Hi￾NicalonTM fibres to obtain creep rates comparable with those for the SiCf–A12O3 and HNSiCf–A12O3 specimens, respectively. This result would be expected because these composites contain 40 vol.% of interwoven 0/90◦ fibre bundles, so the fibres parallel Fig. 2. The stress dependences of the minimum creep rates recorded for the HNSiCf–Al2O3 composite in air at 1300 ◦C compared with data available for SiCf–Al2O3 7 and a SiC whisker-reinforced composite, SiCw–Al2O3. 15 Also included are results reported at 1300 ◦C for NicalonTM NLM20216 and Hi￾NicalonTM fibres.17 In addition, stress–creep rate values at 1200 ◦C are shown for synthetic CaO–50% MgO samples (Table 2). Fig. 3. Comparisons of the stress/minimum creep rate relationships for SiCf–Al2O3 7 and HNSiCf–Al2O3, as well as for SiCf–SiC,6,9 SiCf–SiBC,10 HNSiCf–SiC11 and HNSiCf–SiBC samples tested in air at 1300 ◦C. to the stress axes occupy approximately one fifth of the testpiece cross-sectional areas. Thus, the load-bearing capabilities of the alumina–matrix composites are governed by the longitudinal (0◦) fibres.8 While the creep resistance of the fibres clearly exceeds that of the matrices with the SiCf–A12O3 and HNSiCf–A12O3 sam￾ples, it has been claimed that the matrices are more creep resistant than the fibres with the SiCf–SiC,9 SiCf–SiBC10 and HNSiCf–SiC11 materials. However, as evident from Fig. 3, the ε˙m values recorded for the SiCf–A12O3 testpieces are equal to those for the SiCf–SiBC and lower than those for the SiCf–SiC specimens, with all three materials being reinforced with com￾parable volume fractions of 0/90◦ NicalonTM NLM202 fibres. Similarly, with equivalent Hi-Nicalon reinforcement, the ε˙m val￾ues for HNSiCf–A12O3 are equal to those for HNSiCf–SiBC and lower than those for the HNSiCf–SiC sample (Fig. 3). Hence, the porous SiC and SiBC matrices must be characterized by a creep resistance at least as poor as that for the weak A12O3 matrices. Irrespective of whether the present series of CFCMCs were produced with A12O3, SiC or SiBC matrices, it is therefore clear that the creep strengths of the matrices are markedly inferior to those of the fibres.8 On this basis, from the results presented in Figs. 3 and 4, the improvement in creep and creep rupture resistance achieved by replacing NicalonTM NLM202 with Hi￾NicalonTM fibres can be quantified easily. 3.2. Creep data comparisons CFCMCs obviously display a stochastic strength response because of their essentially brittle character, coupled with the near-random nature of the size and distribution of the macro￾scopic and microscopic flaws which are present. Even so, recognizing that the ε˙m and tf measurements in Figs. 3 and 4 were obtained for a range of composites tested in two differ￾ent laboratories,6–11 the recorded data sets reveal remarkably consistent patterns of property variation as the fibre–matrix com￾binations are changed

4606 B. wilshire, M.R. Bache /Journal of the European Ceramic Society 27(2007)4603-4611 o与gL9g日 HNSiCf-Al2O3 ccc vHNSICr-SiBC NSIC:-Az0 5075100 104 Stress(MPa) Minimum Creep Rate x Time to Fracture Fig. 4. Comparisons of the stress/creep life relationships for SiCr-Al2O,, and Fig. 6. The relationship between the product, Emf, and the creep ductility (eD HNSiCf-Al2O3, as well as for SiCr-SiC,6.9 SiCr-SiBC, 10 HNSiCe-Sicll and for SiCr-Al2O3'and HNSiCr-Al2O3, as well as for SiCr-SiC, 6. 9 SiCr-SiBC,10 HNSiCr-SiBC samples tested in air at 1300oC. HNSiCr-Sicll and HNSiCe-SiBC samples tested in air at 1300oC. As shown in Fig. 2, with the AlzO3-matrix composites, the As the applied stress is decreased over the ranges cov creep rates at a given stress are reduced by a factor of about 5 ered at 1300C, Emff increases from 0.0002 to 0.002 by replacing NicalonM NLM202 with stronger Hi-Nicalon TM with the SiCr SiC samples, 6.9 from -0.002 to -0.04 with fibres. Similarly, this change in fibre type results in an equiv- the SiCr-Al2O3, SiCr-SiBCiO and HNSiCr-Sicll and from alent enhancement in creep resistance with composites having 0.008 to -0.04 with the present HNSiCr-AlO3 and either sic or sibc matrices. as evident from Fig. 3. Moreover. HNSiCr-SiBC products(Fig. 5). Moreover, these increases in comparisons of the data sets in Figs. 3 and 4 demonstrate that Emf are matched by increases in creep ductility(ef)as the stress fibre-matrix combinations which improve creep resistance also is reduced(Fig. 6), indicating that lead to substantially longer creep lives. This result would be expected because the creep life is often inversely proportional Emlf=XEf to the minimum creep rate(Fig. 5), such that with x increasing from0. 4 to 0. 7 with increasing test duration Emtf and Ef, as well as the resu em签 constant ( presented in Figs. 2-6, can then be explained& by reference to showing that the rates at which creep damage develops to cause the e/t trajectories included in Fig. 7 fracture are determined by the rates of creep strain accumula tion, i.e. creep failure is strain controlled. Yet, with the present 3.3. Variations in creep curve shape CFCMC'S, the magnitude of Emtr is both material and test con- dition sensitive Under uniaxial tension, the creep rupture life(tr) can be defined conveniently as the time taken for the accumulated creep strain(E)to become equal to the limiting creep ductility, speci 0.012 t=843 ks 0013 SiCr-Al2O3 0.004 10-7 HNSiCp-SiBc oocoooo SiCcf-SiC 0.002 HNSICE. SBC 10110210310410510610 0.000 100 200 Time to Fracture(s) Time (ks) SiCr-Al2O3 and HNSiCr-Al2O3, as well as for SiC-Sic.9 and SiCp-SiBC, o Fig. 7. Creep strain-time curves recorded for SiCp-SiC, SiCr-SiBCIOat HNSiCr-Sic and HNSiCt-siBC samples tested in air at 1300C. 90 MPa and a HNSiCr-SiBC specimen at 100 MPa for tests in air at 1300C

4606 B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 Fig. 4. Comparisons of the stress/creep life relationships for SiCf–Al2O3 7 and HNSiCf–Al2O3, as well as for SiCf–SiC,6,9 SiCf–SiBC,10 HNSiCf–SiC11 and HNSiCf–SiBC samples tested in air at 1300 ◦C. As shown in Fig. 2, with the Al2O3–matrix composites, the creep rates at a given stress are reduced by a factor of about 5 by replacing NicalonTM NLM202 with stronger Hi-NicalonTM fibres. Similarly, this change in fibre type results in an equiv￾alent enhancement in creep resistance with composites having either SiC or SiBC matrices, as evident from Fig. 3. Moreover, comparisons of the data sets in Figs. 3 and 4 demonstrate that fibre–matrix combinations which improve creep resistance also lead to substantially longer creep lives. This result would be expected because the creep life is often inversely proportional to the minimum creep rate (Fig. 5), such that ˙mtf ∼= constant (1) showing that the rates at which creep damage develops to cause fracture are determined by the rates of creep strain accumula￾tion, i.e. creep failure is strain controlled. Yet, with the present CFCMC’s, the magnitude of ε˙mtf is both material and test con￾dition sensitive. Fig. 5. The dependences of the rupture life on the minimum creep rate for SiCf–Al2O3 7 and HNSiCf–Al2O3, as well as for SiCf–SiC6,9 and SiCf–SiBC,10 HNSiCf–SiC11 and HNSiCf–SiBC samples tested in air at 1300 ◦C. Fig. 6. The relationship between the product, ε˙mtf, and the creep ductility (εf) for SiCf–Al2O3 7 and HNSiCf–Al2O3, as well as for SiCf–SiC,6,9 SiCf–SiBC,10 HNSiCf–SiC11 and HNSiCf–SiBC samples tested in air at 1300 ◦C. As the applied stress is decreased over the ranges cov￾ered at 1300 ◦C, ε˙mtf increases from ∼0.0002 to 0.002 with the SiCf–SiC samples,6,9 from ∼0.002 to ∼0.04 with the SiCf–Al2O3, 7 SiCf–SiBC10 and HNSiCf–SiC11 and from ∼0.008 to ∼0.04 with the present HNSiCf–Al2O3 and HNSiCf–SiBC products (Fig. 5). Moreover, these increases in ε˙mtf are matched by increases in creep ductility (εf) as the stress is reduced (Fig. 6), indicating that ε˙mtf = χεf (2) with χ increasing from ∼0.4 to 0.7 with increasing test duration. This relationship between ε˙mtf and εf, as well as the results presented in Figs. 2–6, can then be explained8 by reference to the ε/t trajectories included in Fig. 7. 3.3. Variations in creep curve shape Under uniaxial tension, the creep rupture life (tf) can be defined conveniently as the time taken for the accumulated creep strain (ε) to become equal to the limiting creep ductility, speci￾Fig. 7. Creep strain–time curves recorded for SiCf–SiC,6 SiCf–SiBC10 at 90 MPa and a HNSiCf–SiBC specimen at 100 MPa for tests in air at 1300 ◦C.

B. Wilshire, M.R. Bache/ Journal of the European Ceramic Sociery 27(2007)4603-4611 fied as the total creep strain to failure(Ef) Continuously decaying creep strain/time curves of the form shown in Figs. I and 7 therefore terminate when t=tf ande 2 Em when a=Ef. In this context, with the SiCr-SiC composite, the creep duc tilities are very low(Fig. 6)so the creep curves terminate early (Fig. 7), giving high Em and low tr values(Figs. 3 and 4) Although the initial rates of creep strain accumulation are similar because the fibre reinforcement is essentially the same, the creep ductilities of the SiCf-SiBC material are higher than those for the Cr-SiC samples(Fig. 6). Creep therefore continues until the larger Er values are attained, so lower creeprates and much longer creep lives are displayed under the same test conditions( Fig. 7) Then, by replacing the NicalonTM NLM 202 with stronger Hi- 1 mm Nicalon fibres, the rates of creep strain accumulation are also decreased(Fig. 7), so the large Ef values(Fig. 6) lead to the creep and creep rupture strength of the HNSiCr-SiBC compos- (b) ite being considerably superior to the properties displayed by the Sicr-SiBC specimens(Figs. 3 and 4). Yet, while substan- tial performance gains would be anticipated by incorporation of stronger fibres the matrices contribute little to the stress- bearing capabilities of these CFCMCs Even so, with nominally identical fibre reinforcement, replacing SiC with either SiBC or Al2 O3 matrices leads to significant strength enhancements (Figs. 3 and 4). This observation can then be interpreted by con- sidering the deformation and damage processes governing strain accumulation and failure 3.4. Creep deformation and damage processes 100μm On applying a tensile load to a'textile'CFCMC, the inter- Fig 8. Scanning electron micrographs showing crack development through the woven longitudinal fibre bundles extend and straighten in the transverse(90 )fibre bundles, by-passing the fibres as the cracks grow through stress direction However as with the individual fibres. b the (a) the alumina matrix of the HNSiCr-Al2O3 and(b) the SiBC matrix of the creep strengths of fibre bundles also vary. 8 Hence, the weak- HNSiCr-SiBC composites In both cases, the tensile stress axis is vertical est fibre regions deform most easily, so the high initial creep rates decrease with time as load is transferred progressively to fracture takes place by fibre pull-out(Fig. 9b). In this way, stronger bundles. This deformation is accompanied by crack for- the longitudinal 0 fibres control the rates of strain accumu- crack development has little effect on the overall strength of the oxidation-assisted fibre failure and creep ductility. 8 rates c mation in the brittle matrices but, because the matrices are weak, lation and crack growth, while the matrices affect the rates of composite. Consequently, the creep rate continues to decrease With the SiC-SiC samples,6 the dominant macro-crack with time(Figs. I and 7), with the rate of strain accumulation nucleates at surface macro-pores, with direct oxygen penetra decreasing as the stress and temperature decrease. tion along the crack leading to low ductility failure(Fig. 7). In As the longitudinal bundles extend and straighten, the result- contrast, with the SiCr-SiBC material, glass formation limits ing complex stress state leads to crack formation within and oxidation-assisted fibre failure, giving higher creep ductilities between the 00 and 90 fibre tows, although composite failure is (Fig. 7). Further performance gains are then achieved with governed by the growth of"'macro-cracks'along planes normal the HNSiCr-SiBC product, when the benefits of the enhanced to the tensile axis. 8 These'tunnellingcracks easily by-pass SiBC matrices are combined with the reduced rates of strain fibres in the 90 bundles(Fig. 8)but, on penetrating into the accumulation achieved by replacing NicalonM NLM202 with 0 tows, the cracks become bridged by unbroken longitudinal Hi-Nicalon M fibres(Fig. 7) fibres. Cracks can then open and extend only at rates determined With alumina matrices. residual stress-induced micro-cracks by the creep resistance of the bridging fibres, accounting for the are present in the as-produced samples, so many small cracks dependence of tr on Em(Fig. 5) develop throughout the gauge length of the AlO3-matrix com- Unfortunately, in oxidizing environments, matrix cracking posites as creep proceeds. Indirect oxygen ingress through promotes oxygen ingress, causing premature failure of the crack- the micro-cracked matrices, coupled with the more oxidation bridging fibres and accelerating crack growth. 8 In general, the resistant double BN/SiC interfaces, then results in a resistance dominant crack causing failure is surface nucleated(Fig. 9a), to oxidation-assisted fibre failure equivalent to that of sibC developing until the stress on the remaining unbroken cross- matrices. Hence, similar creep and creep rupture strengths section of the composite reaches the critical level at which are exhibited by the Sicr-AlzO3 and Sicr-SiBC materials, as

B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 4607 fied as the total creep strain to failure (εf). Continuously decaying creep strain/time curves of the form shown in Figs. 1 and 7 therefore terminate when t = tf and ε˙ ∼= ε˙m when ε = εf. In this context, with the SiCf–SiC composite, the creep duc￾tilities are very low (Fig. 6) so the creep curves terminate early (Fig. 7), giving high ε˙m and low tf values (Figs. 3 and 4). Although the initial rates of creep strain accumulation are similar because the fibre reinforcement is essentially the same, the creep ductilities of the SiCf–SiBC material are higher than those for the SiCf–SiC samples (Fig. 6). Creep therefore continues until the larger εf values are attained, so lower creep rates and much longer creep lives are displayed under the same test conditions (Fig. 7). Then, by replacing the NicalonTM NLM 202 with stronger Hi￾NicalonTM fibres, the rates of creep strain accumulation are also decreased (Fig. 7), so the large εf values (Fig. 6) lead to the creep and creep rupture strength of the HNSiCf–SiBC compos￾ite being considerably superior to the properties displayed by the SiCf–SiBC specimens (Figs. 3 and 4). Yet, while substan￾tial performance gains would be anticipated by incorporation of stronger fibres, the matrices contribute little to the stress￾bearing capabilities of these CFCMCs. Even so, with nominally identical fibre reinforcement, replacing SiC with either SiBC or Al2O3 matrices leads to significant strength enhancements (Figs. 3 and 4). This observation can then be interpreted by con￾sidering the deformation and damage processes governing strain accumulation and failure. 3.4. Creep deformation and damage processes On applying a tensile load to a ‘textile’ CFCMC, the inter￾woven longitudinal fibre bundles extend and straighten in the stress direction. However, as with the individual fibres,16 the creep strengths of fibre bundles also vary.18 Hence, the weak￾est fibre regions deform most easily, so the high initial creep rates decrease with time as load is transferred progressively to stronger bundles. This deformation is accompanied by crack for￾mation in the brittle matrices but, because the matrices are weak, crack development has little effect on the overall strength of the composite. Consequently, the creep rate continues to decrease with time (Figs. 1 and 7), with the rate of strain accumulation decreasing as the stress and temperature decrease. As the longitudinal bundles extend and straighten, the result￾ing complex stress state leads to crack formation within and between the 0◦ and 90◦ fibre tows, although composite failure is governed by the growth of ‘macro-cracks’ along planes normal to the tensile axis.8 These ‘tunnelling’ cracks easily by-pass fibres in the 90◦ bundles (Fig. 8) but, on penetrating into the 0◦ tows, the cracks become bridged by unbroken longitudinal fibres. Cracks can then open and extend only at rates determined by the creep resistance of the bridging fibres, accounting for the dependence of tf on ε˙m (Fig. 5). Unfortunately, in oxidizing environments, matrix cracking promotes oxygen ingress, causing premature failure of the crack￾bridging fibres and accelerating crack growth.8 In general, the dominant crack causing failure is surface nucleated (Fig. 9a), developing until the stress on the remaining unbroken cross￾section of the composite reaches the critical level at which Fig. 8. Scanning electron micrographs showing crack development through the transverse (90◦) fibre bundles, by-passing the fibres as the cracks grow through: (a) the alumina matrix of the HNSiCf–Al2O3 and (b) the SiBC matrix of the HNSiCf–SiBC composites. In both cases, the tensile stress axis is vertical. fracture takes place by fibre pull-out (Fig. 9b). In this way, the longitudinal 0◦ fibres control the rates of strain accumu￾lation and crack growth, while the matrices affect the rates of oxidation-assisted fibre failure and creep ductility.8 With the SiCf–SiC samples,6 the dominant macro-crack nucleates at surface macro-pores, with direct oxygen penetra￾tion along the crack leading to low ductility failure (Fig. 7). In contrast, with the SiCf–SiBC material, glass formation limits oxidation-assisted fibre failure, giving higher creep ductilities (Fig. 7). Further performance gains are then achieved with the HNSiCf–SiBC product, when the benefits of the enhanced SiBC matrices are combined with the reduced rates of strain accumulation achieved by replacing NicalonTM NLM202 with Hi-NicalonTM fibres (Fig. 7). With alumina matrices, residual stress-induced micro-cracks are present in the as-produced samples,19 so many small cracks develop throughout the gauge length of the Al2O3–matrix com￾posites as creep proceeds.7 Indirect oxygen ingress through the micro-cracked matrices, coupled with the more oxidation￾resistant double BN/SiC interfaces, then results in a resistance to oxidation-assisted fibre failure equivalent to that of SiBC matrices. Hence, similar creep and creep rupture strengths are exhibited by the SiCf–Al2O3 and SiCf–SiBC materials, as

4608 B. wilshire, M.R. Bache /Journal of the European Ceramic Society 27(2007)4603-4611 w HNSiCf-SiBC 口-SC1400C b E10 200pm 125150175200225 Fig. 10. The stress dep the minimum creep rates recorded for the HNSiCr-AlzO3 and HNSiCp-SiBC composites in air at 1300C, compared with results reported for 2. 5D Cr-Si les" tested under low pressure argon at 1200-1400°C. ognized as showing creep brittle behaviour, i.e. ErlEp= l where Ep is the primary creep strain and Ef is the total creep strain to failure 20 (3)Although large cracks can develop before creep failure finally occurs under low applied stresses, so that the present CFCMCs can be regarded as 'crack tolerant, these creep brittle materials have low creep damage tolerance values 500pm a), where A can be defined2las Fig 9. Scanning electron micrographs showing: (a)nucleation of cracks normal to the tensile axis in the SiBC surface layer and (b)regions of fibre pull-out on the fracture surface of the HNSiCr-SiBC composite tested in air at 1300C so A=l because Et =0(Figs. 1 and 7). The creep dam- age tolerance value is important in practical situations when well as by the HNSiCf-Al2O3 and HNSiCr-SiBC composites materials must withstand local strain concentrations, say, in regions where a change in component cross-section leads to stress concentrations. Values of A in the range 5-10 are 3.5. Creep in oxidizing environments then thought to ensure that the strain concentrations encoun- tered during service will not lead to premature failure. 22 As demonstrated in Figs. 3 and 4, product development ini Hence, with the SiCr-reinforced composites now consid- tiatives such as the introduction of Hi-NicalonTM fibres and ered, a values near unity may represent a severe design partially self-sealing SiBC matrices have resulted in substan constraint. For this reason. alternative fibre-matrix combi tial improvements in the creep and creep fracture strength of nations should be considered for safety-critical aeroengine SiC fibre-reinforced composites. Even so, for aeroengine and and related applications related applications, the resulting component life enhancement be as impressive as the increased product costs might Interestingly, even when compared with the results now imply for several reasons. documented for the high-performance HNSiCr-Al2O3 and HNSiCr-SiBC products, the impressive creep resistance of a (1)With the problem of oxidation-assisted fibre failure caused carbon fibre-reinforced Sic-matrix composite-44 is illustrated predominantly by cracking of the weak porous matrices, a in Fig. 10. The carbon fibres were again introduced as 2D low design stress limit must be imposed for components woven bundles, but with the carbon fabric layers interlinked erving in non-protective atmospheres to avoid delamination, giving 2. 5D Cr-SiC testpieces. However, (2)Because NicalonM NLM202, Hi-Nicalon"M and other the data sets for the 2. D Cr-Sic samples were determined under types of SiC fibres display continuously decaying creep low-pressure argon. Hence, a major reduction in creep life and curves, 6, I7 similar curve shapes are exhibited by SiC fibre- ductility will occur when crack development in the brittle Sic reinforced materials(Figs. I and7). In the absence of clearly matrix allows oxygen penetration during creep in non-protective defined tertiary stages, these woven composites must be rec- atmospheres

4608 B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 Fig. 9. Scanning electron micrographs showing: (a) nucleation of cracks normal to the tensile axis in the SiBC surface layer and (b) regions of fibre pull-out on the fracture surface of the HNSiCf–SiBC composite tested in air at 1300 ◦C. well as by the HNSiCf–Al2O3 and HNSiCf–SiBC composites (Figs. 3 and 4). 3.5. Creep in oxidizing environments As demonstrated in Figs. 3 and 4, product development ini￾tiatives such as the introduction of Hi-NicalonTM fibres and partially self-sealing SiBC matrices have resulted in substan￾tial improvements in the creep and creep fracture strength of SiC fibre-reinforced composites. Even so, for aeroengine and related applications, the resulting component life enhancement may not be as impressive as the increased product costs might imply for several reasons. (1) With the problem of oxidation-assisted fibre failure caused predominantly by cracking of the weak porous matrices, a low design stress limit must be imposed for components serving in non-protective atmospheres. (2) Because NicalonTM NLM202, Hi-NicalonTM and other types of SiC fibres display continuously decaying creep curves,16,17 similar curve shapes are exhibited by SiC fibre￾reinforced materials (Figs. 1 and 7). In the absence of clearly defined tertiary stages, these woven composites must be rec￾Fig. 10. The stress dependences of the minimum creep rates recorded for the HNSiCf–Al2O3 and HNSiCf–SiBC composites in air at 1300 ◦C, compared with results reported for 2.5D Cf–SiC samples24 tested under low pressure argon at 1200–1400 ◦C. ognized as showing creep brittle behaviour, i.e. εf/εp ∼= 1, where εp is the primary creep strain and εf is the total creep strain to failure.20 (3) Although large cracks can develop before creep failure finally occurs under low applied stresses, so that the present CFCMCs can be regarded as ‘crack tolerant’, these creep brittle materials have low creep damage tolerance values (λ), where λ can be defined21 as λ = 1 + εt ε˙mtf  (3) so λ = 1 because εt ∼= 0 (Figs. 1 and 7). The creep dam￾age tolerance value is important in practical situations when materials must withstand local strain concentrations, say, in regions where a change in component cross-section leads to stress concentrations.22 Values of λ in the range 5–10 are then thought to ensure that the strain concentrations encoun￾tered during service will not lead to premature failure.22 Hence, with the SiCf-reinforced composites now consid￾ered, λ values near unity may represent a severe design constraint. For this reason, alternative fibre–matrix combi￾nations should be considered for safety-critical aeroengine and related applications. Interestingly, even when compared with the results now documented for the high-performance HNSiCf–Al2O3 and HNSiCf–SiBC products, the impressive creep resistance of a carbon fibre-reinforced SiC–matrix composite23,24 is illustrated in Fig. 10. The carbon fibres were again introduced as 2D woven bundles, but with the carbon fabric layers interlinked to avoid delamination, giving 2.5D Cf–SiC testpieces. However, the data sets for the 2.5D Cf–SiC samples were determined under low-pressure argon. Hence, a major reduction in creep life and ductility will occur when crack development in the brittle SiC matrix allows oxygen penetration during creep in non-protective atmospheres.

B. Wilshire, M.R. Bache/ Journal of the European Ceramic Sociery 27(2007)4603-4611 Research emphasis has also been directed to oxide fibre reinforced composites, with the oxide fibres seemingly less 5um liable to oxidation-assisted failure Yet, as expected for the 2.5D Cr-Sic product, creep data sets reported for an Al2O3-fibre reinforced SiC-matrix composite have emphasized the marked reduction in creep life and ductility caused by testing in air rather than vacuum.Moreover, in seeking to identify suitable oxide-oxide composites, the creep strengths of currently avail able oxide fibres are inferior to the values for the established sic fibres Clearly, irrespective of the fibre type chosen, the princi- pal creep life-limiting phenomenon encountered with woven CFCMCs operating under load in non-protective atmospheres at high temperatures is premature fibre failure associated with oxygen ingress as cracks develop in the brittle matrices. For this reason,to protect the vulnerable fibres and fibre-matrix inter- faces, it is now proposed that major benefits could be realized by considering composites fabricated with creep damage-resistant matrices, i.e. high-melting point ceramic matrices not prone to creep crack formation d 3.6. Creep of Cao-MgO ceramics Just as cracks develop in the brittle SiC, SiBC and Al2O matrices of the SiC fibre-reinforced composites, during tensile creep, intergranular cracks form extensively on the transverse grain boundaries of most polycrystalline ceramics produced i monolithic form as found for both sintered silicon carbide26 and alumina. 2 Even during creep in compression, intergran- ular damage accumulates on boundaries experiencing tensile Fig. Il. (a) Intergranular cracking of porous magnesia after a strain of 0.05 at hoop and radial stresses, in line with the stress distributions pre- 75 MPa and 1327 C and (b)the microstructure of natural doloma after a strain dicted using finite element methods. Yet, while creep cracks of 0.08 at 62MPa and 1 127C.The compression axis is vertical are evident after compressive creep strains of only a few per- cent with polycrystalline MgO(Fig. 1 la), as well as with tions, with testpieces containing 0, 25, 75 and 100% MgO,the Cao, crack formation was not- during compressive decaying primary stages give way to accelerating tertiary defor- creep of two-phase CaO-Mgo ( doloma)specimens containing mation as the crack incidence increases with increasing creep M42-50 wt% MgO (Fig. llb) strain. However, in the absence of crack formation or any other The Cao-Mgo system is a simple eutectic, with an eutectic damage process which can cause a tertiary acceleration with the omposition of Cao-32 wt%MgO and an eutectic temperature microstructurally stable Ca0-50% MgO samples, continuously of -2300 C With increasing MgO content, the microstructure changes from alime crystal matrix enclosing equiaxed magnesia Table 2 grains to a periclase grain network surrounding equiaxed lime Fabrication procedures, analyses(wt%)and microstructures of natural and syn- crystals. For a series of CaO-MgO samples varying in composi- thetic doloma tion from0 to 100 wt% MgO, 3 cracks were not discernible in specimens containing around 50 wt% MgO, whereas inter- Synthetic doloma granular damage was readily apparent with samples produced Starting material Whitwell dolomite with 0, 25, 75 and 100 wt% MgO. Thus, cracks evolve prefer- CaCO3·MgCO2) Mg(oh) CaCO3 entially on Cao-Cao and Mgo-Mgo boundaries rather than Sintering temperature(C) 1600 on Cao-MgO interfaces. In this context, it should be noted that porosity of sintered bars(%64-6 ~4-6 doloma testpieces produced with around 40-50 wt%o MgO were Average crystal size(um) 3-5 2-16 not prone to creep crack formation, irrespective of whether the Composition samples were fabricated using synthetic CaO and MgO powders Cao or natural dolomite( CacO3 MgCO3), as listed in Table 2 The composition dependence of creep crack development the synthetic Cao-Mgo ceramics was confirmed31 by differ- A2 03 ences in the shapes of the E/t trajectories observed at 1327C Other oxides as illustrated in Fig. 12. Even under compressive creep condi

B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 4609 Research emphasis has also been directed to oxide fibre￾reinforced composites, with the oxide fibres seemingly less liable to oxidation-assisted failure.3 Yet, as expected for the 2.5D Cf–SiC product, creep data sets reported for an Al2O3-fibre￾reinforced SiC–matrix composite have emphasized the marked reduction in creep life and ductility caused by testing in air rather than vacuum.25 Moreover, in seeking to identify suitable oxide–oxide composites, the creep strengths of currently avail￾able oxide fibres are inferior to the values for the established SiC fibres.3 Clearly, irrespective of the fibre type chosen, the princi￾pal creep life-limiting phenomenon encountered with woven CFCMCs operating under load in non-protective atmospheres at high temperatures is premature fibre failure associated with oxygen ingress as cracks develop in the brittle matrices. For this reason, to protect the vulnerable fibres and fibre–matrix inter￾faces, it is now proposed that major benefits could be realized by considering composites fabricated with creep damage-resistant matrices, i.e. high-melting point ceramic matrices not prone to creep crack formation. 3.6. Creep of CaO·MgO ceramics Just as cracks develop in the brittle SiC, SiBC and Al2O3 matrices of the SiC fibre-reinforced composites, during tensile creep, intergranular cracks form extensively on the transverse grain boundaries of most polycrystalline ceramics produced in monolithic form, as found for both sintered silicon carbide26 and alumina.27 Even during creep in compression, intergran￾ular damage accumulates on boundaries experiencing tensile hoop and radial stresses, in line with the stress distributions pre￾dicted using finite element methods.28 Yet, while creep cracks are evident after compressive creep strains of only a few per￾cent with polycrystalline MgO (Fig. 11a), as well as with CaO, crack formation was not observed29–31 during compressive creep of two-phase CaO–MgO (doloma) specimens containing ∼42–50 wt% MgO (Fig. 11b). The CaO–MgO system is a simple eutectic, with an eutectic composition of CaO–32 wt% MgO and an eutectic temperature of ∼2300 ◦C. With increasing MgO content, the microstructure changes from a lime crystal matrix enclosing equiaxed magnesia grains to a periclase grain network surrounding equiaxed lime crystals. For a series of CaO–MgO samples varying in composi￾tion from 0 to 100 wt% MgO,31 creep cracks were not discernible in specimens containing around 50 wt% MgO, whereas inter￾granular damage was readily apparent with samples produced with 0, 25, 75 and 100 wt% MgO. Thus, cracks evolve prefer￾entially on CaO–CaO and MgO–MgO boundaries rather than on CaO–MgO interfaces. In this context, it should be noted that doloma testpieces produced with around 40–50 wt% MgO were not prone to creep crack formation, irrespective of whether the samples were fabricated using synthetic CaO and MgO powders or natural dolomite (CaCO3·MgCO3), as listed in Table 2. The composition dependence of creep crack development in the synthetic CaO–MgO ceramics was confirmed31 by differ￾ences in the shapes of the ε/t trajectories observed at 1327 ◦C, as illustrated in Fig. 12. Even under compressive creep condi￾Fig. 11. (a) Intergranular cracking of porous magnesia after a strain of 0.05 at 75 MPa and 1327 ◦C and (b) the microstructure of natural doloma after a strain of 0.08 at 62 MPa and 1127 ◦C.29 The compression axis is vertical. tions, with testpieces containing 0, 25, 75 and 100% MgO, the decaying primary stages give way to accelerating tertiary defor￾mation as the crack incidence increases with increasing creep strain. However, in the absence of crack formation or any other damage process which can cause a tertiary acceleration with the microstructurally stable CaO–50% MgO samples, continuously Table 2 Fabrication procedures, analyses (wt%) and microstructures of natural and syn￾thetic doloma30 Natural doloma Synthetic doloma Starting material Whitwell dolomite (CaCO3·MgCO2) Analar Mg(OH)2CaCO3 Calcination temperature (◦C) 1300 1300 Sintering temperature (◦C) 1600 1800 Porosity of sintered bars (%) ∼4–6 ∼4–6 Average crystal size (m) 3–5 12–16 Compositions CaO 58.02 56.53 MgO 40.53 42.32 SiO2 0.49 0.76 Al2O3 0.03 0.10 Fe2O3 0.77 0.13 Other oxides 0.15 0.16

B. wilshire, M.R. Bache /Journal of the European Ceramic Society 27(2007)4603-4611 compatibility of Cao-MgO ceramics with carbon in aggres- sive oxidizing environments at 1600c and above is evident from the dominant selection of carbon-bearing doloma lin- ings for the original Thomas or basic Bessemer process and 0.08 graphite-bearing magnesia linings for modern basic oxygen steel-making vessels The potential success of Cr-doloma composites would epend critically on fabrication methods being devised to obtain low-porosity matrices, free from the inter-connected po which would provide channels for oxygen ingress. If low porosity creep damage-resistant Ca0-45 wt% MgO matrices can be developed, the impressive creep properties of the Cr-Sic material in argon(Fig. 10)may be attainable with Cr-doloma composites in oxidizing atmospheres. Fig. 12. The effect of variations in MgO content on the creep strain/time curves 4. Conclusions recorded for synthetic CaO-MgO samples tested under compressive stresses of 62 MPa at1327°C3 The improvements in creep and creep rupture strength achieved through different fibre-matrix combinations are decaying creep curves were always recorded, i.e. after loading, assessed for composites produced with Sic or SiBC matrices the creep rate decreases gradually and creep cracking was not (with carbon interfaces )or Al2O3 matrices(with double BN/Sic discernible29-3 after true creep strains up to 0. 25(when the tests interfaces ) reinforced with either Nicalon TM NLM202 or Hi- were discontinued ). On this basis, doloma appears to represent Nicalon"M fibres. This comparison of tensile data sets recorded a creep damage-resistant option as a matrix for fibre-reinforced in air at 1300 C demonstrates the substantial performance gains derived when hi-Nicalon fibres are used with either sibc e Al2O3 matrices. 3.7. Potential doloma-matrir composites For the product range considered, the fibres aligned parallel to the tensile stress axes govern the rates of creep strain accumula Although creep damage resistance is a critical requirement, tion and crack growth. However, in non-protective atmospheres, several additional property features are relevant to consideration the principal creep life-limiting phenomenon is premature fibre of doloma as a candidate matrix material for high-performance failure associated with oxygen penetration as cracks develop in CFCMCS the brittle matrices. Yet, while crack formation occurs during creep of the Sic, SiBC and Al2O3 matrix materials, as well (a) With synthetic CaO-MgO ceramics, the creep resistance with most monolithic ceramics, such damage is not discernible increases with increasing MgO content,3I with the natural with two-phase Cao 45 wt% MgO(doloma). Provided that pro- product having a lower cree th than the heti cess routes can be devised to obtain fine-grain lot Cao-50% MgO material due to the marginally lower Mgo doloma matrices, free from interconnected pores allowing oxy- level, higher impurity levels and finer average crystal size gen ingress, Cr-doloma composites could represent a relatively Table 2). At temperatures around 1300C, the synthetic inexpensive creep damage-resistant option for high temperature Cao-50% MgO samples exhibit creep strengths better than applications involving long-term service under load in oxidizing those recorded for the SiCw-Al2O3 ceramic, as evident environmen from Fig. 1. However, the present analysis confirms that the weak SiC, SiBC and al2 O3 matrices make little contri- References bution to the overall creep resistance of the present set of 1. Ruffles. P. Aerospace Structural Materials: Present and Future. The Insti- SiC fibre-reinforced composites, 8 so the creep strength of tute of Materials. London. 1995 the Cao-45 wt %o MgO samples is non-critical 2. Miller, S, Advanced materials means advanced engines. Interdiscipl. Sci. (b) Calcined natural doloma products can deteriorate by ' per- Re,1996,21,2 ishing'during storage, which occurs as a result of moisture 3. Parlier, M and Ritt, M H. State of the art and perspectives for oxide-oxide attacking the lime phase. Yet, with fully sintered samples 4.Heredia, E E McNulty, JC, Zok, E.Wand Evans,A.G,Oxidation of both natural and synthetic doloma(Table 2), no evidence of surface attack was apparent even after several years of 1995,78,2097 atmospheric exposure at room temperature. 5. Jones, R. H, Henager Jr, C. H. and Windish Jr, C. F, High temperature (e) Unfortunately, doloma is likely to react with both silicon Mater Sci eng. 4. 1995.A198103 carbide and oxide fibres at high temperatures, indicating 6. wilshire. B. Carreno. E and Percival, M G. L, Tensile creep and creep that CaO-MgO matrices would be better employed with acture of a fibre-reinforced SiC-SiC composite. Sc Mater, 1998, 39, less-expensive carbon fibre reinforcement. Certainly, the

4610 B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 Fig. 12. The effect of variations in MgO content on the creep strain/time curves recorded for synthetic CaO–MgO samples tested under compressive stresses of 62 MPa at 1327 ◦C.31 decaying creep curves were always recorded, i.e. after loading, the creep rate decreases gradually and creep cracking was not discernible29–31 after true creep strains up to 0.25 (when the tests were discontinued). On this basis, doloma appears to represent a creep damage-resistant option as a matrix for fibre-reinforced composites. 3.7. Potential doloma–matrix composites Although creep damage resistance is a critical requirement, several additional property features are relevant to consideration of doloma as a candidate matrix material for high-performance CFCMCs. (a) With synthetic CaO–MgO ceramics, the creep resistance increases with increasing MgO content,31 with the natural product having a lower creep strength than the synthetic CaO–50% MgO material due to the marginally lower MgO level, higher impurity levels and finer average crystal size (Table 2). At temperatures around 1300 ◦C, the synthetic CaO–50% MgO samples exhibit creep strengths better than those recorded for the SiCw–Al2O3 ceramic, as evident from Fig. 1. However, the present analysis confirms that the weak SiC, SiBC and Al2O3 matrices make little contri￾bution to the overall creep resistance of the present set of SiC fibre-reinforced composites,8 so the creep strength of the CaO–45 wt% MgO samples is non-critical. (b) Calcined natural doloma products can deteriorate by ‘per￾ishing’ during storage, which occurs as a result of moisture attacking the lime phase. Yet, with fully sintered samples of both natural and synthetic doloma (Table 2), no evidence of surface attack was apparent even after several years of atmospheric exposure at room temperature. (c) Unfortunately, doloma is likely to react with both silicon carbide and oxide fibres at high temperatures, indicating that CaO–MgO matrices would be better employed with less-expensive carbon fibre reinforcement. Certainly, the compatibility of CaO–MgO ceramics with carbon in aggres￾sive oxidizing environments at 1600 ◦C and above is evident from the dominant selection of carbon-bearing doloma lin￾ings for the original Thomas or basic Bessemer process and graphite-bearing magnesia linings for modern basic oxygen steel-making vessels. The potential success of Cf–doloma composites would depend critically on fabrication methods being devised to obtain low-porosity matrices, free from the inter-connected pores which would provide channels for oxygen ingress. If low￾porosity creep damage-resistant CaO–45 wt% MgO matrices can be developed, the impressive creep properties of the Cf–SiC material in argon (Fig. 10) may be attainable with Cf–doloma composites in oxidizing atmospheres. 4. Conclusions The improvements in creep and creep rupture strength achieved through different fibre–matrix combinations are assessed for composites produced with SiC or SiBC matrices (with carbon interfaces) or Al2O3 matrices (with double BN/SiC interfaces), reinforced with either NicalonTM NLM202 or Hi￾NicalonTM fibres. This comparison of tensile data sets recorded in air at 1300 ◦C demonstrates the substantial performance gains derived when Hi-NicalonTM fibres are used with either SiBC or Al2O3 matrices. For the product range considered, the fibres aligned parallel to the tensile stress axes govern the rates of creep strain accumula￾tion and crack growth. However, in non-protective atmospheres, the principal creep life-limiting phenomenon is premature fibre failure associated with oxygen penetration as cracks develop in the brittle matrices. Yet, while crack formation occurs during creep of the SiC, SiBC and Al2O3 matrix materials, as well as with most monolithic ceramics, such damage is not discernible with two-phase CaO–45 wt% MgO (doloma). Provided that pro￾cess routes can be devised to obtain fine-grain low-porosity doloma matrices, free from interconnected pores allowing oxy￾gen ingress, Cf–doloma composites could represent a relatively inexpensive creep damage-resistant option for high temperature applications involving long-term service under load in oxidizing environments. References 1. Ruffles, P., Aerospace Structural Materials: Present and Future. The Insti￾tute of Materials, London, 1995. 2. Miller, S., Advanced materials means advanced engines. Interdiscipl. Sci. Rev., 1996, 21, 2. 3. Parlier, M. and Ritti, M. H., State of the art and perspectives for oxide–oxide composites. Aerospace Sci. Tech., 2003, 14, 1. 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. Jones, R. H., Henager Jr., C. H. and Windish Jr., C. F., High temperature corrosion and crack growth of SiC–SiC at variable oxygen partial pressures. Mater. Sci. Eng. A, 1995, A198, 103. 6. Wilshire, B., Carreno, F. and Percival, M. G. L., Tensile creep and creep ˜ fracture of a fibre-reinforced SiC–SiC composite. Scr. Mater., 1998, 39, 729

B. Wilshire, M.R. Bache/ Journal of the European Ceramic Sociery 27(2007)4603-4611 4611 7. Wilshire. B. and Carreno F. Deformation and sses during 18. Yun, H M. and DiCarlo, J. A, Thermo-mechanical behaviour of advanced eramic-matrix composites J. Eur SiC fibre multi-filament tows Ceram. Eng. Sci. Proc., 1996, 17A, 61 ceran.Soc.,2000,20,463 19. Heredia, F E, Evans, A G. and Andersson, C. A, Tensile and shear proper- 8. wilshire, B, Creep property comparisons for ceramic-fibre-reinforced es of continuous fibre-reinforced SiC/Al203 processed by melt oxidation. ceramic-matrix composites. J. Eur Ceram Soc., 2002, 22, 1329 JAm. Ceran.Soc.,1995,78,2790. 9. Zhu, S, Mizuno, M, Kagawa, Y, Cao, J, Nagano, Y and Kaya, H, Creep 20. Goodall, I. N, Cockcroft, R. D. H. and Chubb, E J, An approximate and fatigue behaviour of SiC fibre reinforced SiC composite at high tem- description of the creep rupture of structures. Int J. Mech. Sci., 1975, 17, peratures. Mater. Sci. Eng. A, 1997, 225, 69 351. 10. Zhu, S, Mizuno, M, Nagano, Y. Cao, J, Kagawa, Y and Kaya, H, Creep 21. Wilshire, B and Burt, H, Tertiary creep of metals and alloys. Z Metallkd and fatigue behaviour of an enhanced SiC-SiC composite at high tempera 2005,96,552 tures.J. Am. Cera. Soc., 1998, 81, 2269 22. Leckie, F. A and Hayhurst, D. R, Constitutive equations for creep rupture 11. Zhu, S, Mizuno, M, Kagawa, Y, Cao, J, Nagano, Y and Kaya, H Acta Meta.,1977,25,1059 Creep and fatigue behaviour in Hi-Nicalon-fibre-reinforced silicon car- 23. Botier, G Vicens, J and Chermant, J L, Understanding the creep behaviour high temperatures. J. Am. Ceram. Soc. 1999, 82, 117. f a 2.5D Cr-SiC composite I Morphology and microstructure of the as- 12. Bodet, R, Bourrat, X, Lamon, J and Naslain, R, Tensile creep behaviour received material. Mater Sci Eng.A, 2000, 279, 73 of a silicon-carbide-based fibre with a low oxygen content. J. Mater. Sci., 24. Botier, G Chermant, J. L and Vicens, J, Understanding the creep behaviour 1995,30,661. of a 2.5D Cr-SiC composite ll. Experimental specification and macroscopic 13. Challon, G, Pailler, R Naslain, R and Olry, P, Structure composition and mechanical creep responses. Mater. Sci. Eng. A, 2000, 289, 265. mechanical behaviour at high temperature of the oxygen-free Hi-Nicalon 25. Lamouroux, F, Steen, M. and Valles, J. L, Damage of a 2D Al2O3-SiC fibre. In High-temperature Ceramic-Matrix Composites. 11: Manufacturing composite during uniaxial creep. Comp. Sci. TechnoL, 1996, 56, 825 and Materials Development, ed. A G. Evans and R. Naslain Am Ceram. 26. Wilshire, B and Jiang, H, Deformation and failure processes during tensile Soc., Westerville, OH, 1995, p. 299 creep of sintered silicon carbide. Brit. Ceram. Trans., 1994, 93, 213. 14. Fox, D. S and Nguyan, Q. N, Oxidation kinetics of enhanced SiC-SiC. 27. Folweiler, R. C, Creep behaviour of pore-free polycrystalline aluminium Ceram. Eng. Sci. Proc 1995, 16, 877. oxide. J. Appl. Phys., 1961, 32. 773. 15. O"Meara, C, Suihkonen, T, Hansson, T and Warren, R, A microstruc- 28. Birch, J M, wilshire, B, Owen, D. R.J. and Shantaram, D, The influence tural investigation of the mechanisms of tensile creep deformation in an of stress distribution on the deformation and fracture behaviour of ceramic Al203-SiCw composite Mater. Sci. Eng. A, 1996. 209, 251 materials under compressive creep conditions.. Mater. Sci, 1976, 11, 1817. 16. Simon, G. and Bunsell, A. R, Creep behaviour and structural characteriza- 29. wilshire, B, Microstructure dependence of the creep and creep fracture tion at high temperature of Nicalon SiC fibres. J. Mater Sci., 1984, 19, behaviour of ceramic materals. Microsc 1981. 124. 249 30 Coath, J. A. and wilshire, B, Deformation processes during high- 17. Challon, G, Pailler, R, Naslain, R. and Olry, P, Correlation between mperature creep of lime, magnesia and doloma. Ceram. Int, 1977, 3, microstructure and mechanical behaviour at high temperatures of a Sic fibre with a low oxygen content(Hi-Nicalon). J. Mater. Sci, 1997, 32, 31. Coath, J.A. and wilshire, B, The influence of variations in composition on the creep behaviour of doloma. Ceram. Int. 1978, 4, 66

B. Wilshire, M.R. Bache / Journal of the European Ceramic Society 27 (2007) 4603–4611 4611 7. Wilshire, B. and Carreno, F., Deformation and damage processes during ˜ tensile creep of ceramic-fibre-reinforced ceramic–matrix composites. J. Eur. Ceram. Soc., 2000, 20, 463. 8. Wilshire, B., Creep property comparisons for ceramic-fibre-reinforced ceramic–matrix composites. J. Eur. Ceram. Soc., 2002, 22, 1329. 9. Zhu, S., Mizuno, M., Kagawa, Y., Cao, J., Nagano, Y. and Kaya, H., Creep and fatigue behaviour of SiC fibre reinforced SiC composite at high tem￾peratures. Mater. Sci. Eng. A, 1997, 225, 69. 10. Zhu, S., Mizuno, M., Nagano, Y., Cao, J., Kagawa, Y. and Kaya, H., Creep and fatigue behaviour of an enhanced SiC–SiC composite at high tempera￾tures. J. Am. Ceram. Soc., 1998, 81, 2269. 11. Zhu, S., Mizuno, M., Kagawa, Y., Cao, J., Nagano, Y. and Kaya, H., Creep and fatigue behaviour in Hi-NicalonTM-fibre-reinforced silicon car￾bide composites at high temperatures. J. Am. Ceram. Soc., 1999, 82, 117. 12. Bodet, R., Bourrat, X., Lamon, J. and Naslain, R., Tensile creep behaviour of a silicon-carbide-based fibre with a low oxygen content. J. Mater. Sci., 1995, 30, 661. 13. Challon, G., Pailler, R., Naslain, R. and Olry, P., Structure composition and mechanical behaviour at high temperature of the oxygen-free Hi-NicalonTM fibre. In High-temperature Ceramic–Matrix Composites. II: Manufacturing and Materials Development, ed. A. G. Evans and R. Naslain. Am. Ceram. Soc., Westerville, OH, 1995, p. 299. 14. Fox, D. S. and Nguyan, Q. N., Oxidation kinetics of enhanced SiC–SiC. Ceram. Eng. Sci. Proc., 1995, 16, 877. 15. O’Meara, C., Suihkonen, T., Hansson, T. and Warren, R., A microstruc￾tural investigation of the mechanisms of tensile creep deformation in an Al2O3–SiCw composite. Mater. Sci. Eng. A, 1996, 209, 251. 16. Simon, G. and Bunsell, A. R., Creep behaviour and structural characteriza￾tion at high temperature of NicalonTM SiC fibres. J. Mater. Sci., 1984, 19, 3658. 17. Challon, G., Pailler, R., Naslain, R. and Olry, P., Correlation between microstructure and mechanical behaviour at high temperatures of a SiC fibre with a low oxygen content (Hi-Nicalon). J. Mater. Sci., 1997, 32, 1133. 18. Yun, H. M. and DiCarlo, J. A., Thermo-mechanical behaviour of advanced SiC fibre multi-filament tows. Ceram. Eng. Sci. Proc., 1996, 17A, 61. 19. Heredia, F. E., Evans, A. G. and Andersson, C. A., Tensile and shear proper￾ties of continuous fibre-reinforced SiC/Al2O3 processed by melt oxidation. J. Am. Ceram. Soc., 1995, 78, 2790. 20. Goodall, I. N., Cockcroft, R. D. H. and Chubb, E. J., An approximate description of the creep rupture of structures. Int. J. Mech. Sci., 1975, 17, 351. 21. Wilshire, B. and Burt, H., Tertiary creep of metals and alloys. Z. Metallkd., 2005, 96, 552. 22. Leckie, F. A. and Hayhurst, D. R., Constitutive equations for creep rupture. Acta Metall., 1977, 25, 1059. 23. Botier, G., Vicens, J. and Chermant, J. L., Understanding the creep behaviour of a 2.5D Cf–SiC composite I. Morphology and microstructure of the as￾received material. Mater. Sci. Eng. A, 2000, 279, 73. 24. Botier, G., Chermant, J. L. and Vicens, J., Understanding the creep behaviour of a 2.5D Cf–SiC composite II. Experimental specification and macroscopic mechanical creep responses. Mater. Sci. Eng. A, 2000, 289, 265. 25. Lamouroux, F., Steen, M. and Valles, J. L., Damage of a 2D Al ´ 2O3–SiC composite during uniaxial creep. Comp. Sci. Technol., 1996, 56, 825. 26. Wilshire, B. and Jiang, H., Deformation and failure processes during tensile creep of sintered silicon carbide. Brit. Ceram. Trans., 1994, 93, 213. 27. Folweiler, R. C., Creep behaviour of pore-free polycrystalline aluminium oxide. J. Appl. Phys., 1961, 32, 773. 28. Birch, J. M., Wilshire, B., Owen, D. R. J. and Shantaram, D., The influence of stress distribution on the deformation and fracture behaviour of ceramic materials under compressive creep conditions. J. Mater. Sci., 1976, 11, 1817. 29. Wilshire, B., Microstructure dependence of the creep and creep fracture behaviour of ceramic materials. J. Microsc., 1981, 124, 249. 30. Coath, J. A. and Wilshire, B., Deformation processes during high￾temperature creep of lime, magnesia and doloma. Ceram. Int., 1977, 3, 103. 31. Coath, J. A. and Wilshire, B., The influence of variations in composition on the creep behaviour of doloma. Ceram. Int., 1978, 4, 66