Part A: applied scienc and manufacturing ELSEVIER Composites: Part A 30(1999)445-4 Ceramic composites: roles of fiber and interface J. P Singh", D. Singh, M. Sutaria nergy Technology Division, Argonne National Laboratory, Argonne, IL 60439, USA Abstract Results are presented that elucidate: (a) the effects of fiber coating on retained fiber strength and mechanical properties of Nicalon-fiber reinforced SiC matrix composites; and(b)the role of residual stresses in the interfacial bond strength of SiC-fiber-reinforced reaction-bonded Si3 N4 matrix composites. For Nicalon-fiber-reinforced SiC matrix composites that were fractured in a flexural mode, retained in-situ fiber strength, ultimate strength and work-of-fracture(wOF) of the composites increased with increasing thickness of the fiber coating and cached maximum values at a coating thickness of 0. 3 um. a direct correlation between the variation of in-situ fiber strength and the variation of ultimate strength and woF of the composites clearly indicates the critical role of the retained in-situ strength of reinforcing fibers in composites. Fiber pushout tests performed on SiC-fiber-reinforced reaction-bonded Si3N4 matrix composites indicate that both debonding and frictional shear stresses decreased with increasing fiber content. These variations are consistent with the variation of residual radial stress on fibers, as measured by neutron diffraction, i. e. residual stresses decreased with increasing fiber content. Because fracture behavior is strongly controlled by interfacial bond strength, which is proportional to the residual radial stress, appropriate control of residual stresses is critical in the design of composites with desired fracture properties. o 1999 Elsevier Science Ltd. All rights reserved Keywords: Composites; In-situ fiber strength, B. Residual/internal stress; B. Interface/interphase 1. Introduction The strength of the fiber/matrix interface bond should be optimized to facilitate desired fiber pullout during compo Continuous-fiber ceramic composites(CFCCs)are candi- site fracture, this in turn will lead to substantial energy date materials for structural applications in various indus- dissipation and improved fracture toughness. Bond strength tries, including automotive, aerospace and utilities, is controlled by both interfacial characteristics(such as fiber primarily because of their improved flaw tolerance, large coating [5])and residual stresses induced by thermal expan- ork of fracture(WOF)and noncatastrophic mode of failure sion mismatch between fibers and matrix and can be tailored [1, 2]. The mechanical behavior of these composites is by appropriately coating the fiber surface Fiber coating not greatly influenced by the strength of the reinforcing fibers, only controls bond strength but also protects the fiber from the characteristics of fiber/matrix interface and the residual being damaged by flaws generated during processing and stresses caused by thermal expansion mismatch between service. Therefore, an improved understanding of the role of fibers and matrix. Fiber strength is an important parameter fiber coating and residual stresses in fracture behavior will that controls the fracture behavior of CFCCs. High strength lead to the design and processing of CFCCs with reliable of reinforcing fibers is critical because once a matrix crack performance. In this paper, we present results that will eluc is initiated and extended, load is transferred from the matrix date: (1)the effects of fiber coating on flaw generation, to the fibers in the wake of the crack. Weak fibers fracture retained fiber strength and mechanical properties of Nica and lead to catastrophic failure of the composite, whereas lon-fiber-reinforced SiC matrix composites; and (2)the role strong fibers accommodate the stresses. Theoretical analysis of residual stresses in the interfacial bond strength of Sic and experimental observations have shown that the amount fiber-reinforced reaction-bonded Si, N4 matrix composites of fiber pullout(which contributes to the toughening of omposite)is strongly infuenced by the mean strength and the variability in strength of the reinforcing fibers [3]. Also, 2 Specimens and experimental procedure the ultimate load-bearing capacity of the composite is deter mined by fiber strength characteristics [4] To evaluate the effects of fiber coating, specimens of Nicalon-fiber-reinforced SiC matrix composites with Corresponding author. Tel: +1-630-252-5 123; fax: pproximately 42 vol. fiber reinforcements and carbon 3604 interlayer coating of varying thickness(0-1. 25 um)were 835X/99/- see front matter 1999 Elsevier Science Ltd. All rights reserved 1359-835X(98)00133-X
Ceramic composites: roles of fiber and interface J.P. Singh*, D. Singh, M. Sutaria Energy Technology Division, Argonne National Laboratory, Argonne, IL 60439, USA Abstract Results are presented that elucidate: (a) the effects of fiber coating on retained fiber strength and mechanical properties of Nicalon-fiberreinforced SiC matrix composites; and (b) the role of residual stresses in the interfacial bond strength of SiC-fiber-reinforced reaction-bonded Si3N4 matrix composites. For Nicalon-fiber-reinforced SiC matrix composites that were fractured in a flexural mode, retained in-situ fiber strength, ultimate strength and work-of-fracture (WOF) of the composites increased with increasing thickness of the fiber coating and reached maximum values at a coating thickness of <0.3 mm. A direct correlation between the variation of in-situ fiber strength and the variation of ultimate strength and WOF of the composites clearly indicates the critical role of the retained in-situ strength of reinforcing fibers in composites. Fiber pushout tests performed on SiC-fiber-reinforced reaction-bonded Si3N4 matrix composites indicate that both debonding and frictional shear stresses decreased with increasing fiber content. These variations are consistent with the variation of residual radial stress on fibers, as measured by neutron diffraction, i.e. residual stresses decreased with increasing fiber content. Because fracture behavior is strongly controlled by interfacial bond strength, which is proportional to the residual radial stress, appropriate control of residual stresses is critical in the design of composites with desired fracture properties. q 1999 Elsevier Science Ltd. All rights reserved. Keywords: Composites; In-situ fiber strength; B. Residual/internal stress; B. Interface/interphase 1. Introduction Continuous-fiber ceramic composites (CFCCs) are candidate materials for structural applications in various industries, including automotive, aerospace and utilities, primarily because of their improved flaw tolerance, large work of fracture (WOF) and noncatastrophic mode of failure [1,2]. The mechanical behavior of these composites is greatly influenced by the strength of the reinforcing fibers, the characteristics of fiber/matrix interface and the residual stresses caused by thermal expansion mismatch between fibers and matrix. Fiber strength is an important parameter that controls the fracture behavior of CFCCs. High strength of reinforcing fibers is critical because once a matrix crack is initiated and extended, load is transferred from the matrix to the fibers in the wake of the crack. Weak fibers fracture and lead to catastrophic failure of the composite, whereas strong fibers accommodate the stresses. Theoretical analysis and experimental observations have shown that the amount of fiber pullout (which contributes to the toughening of a composite) is strongly influenced by the mean strength and the variability in strength of the reinforcing fibers [3]. Also, the ultimate load-bearing capacity of the composite is determined by fiber strength characteristics [4]. The strength of the fiber/matrix interface bond should be optimized to facilitate desired fiber pullout during composite fracture, this in turn will lead to substantial energy dissipation and improved fracture toughness. Bond strength is controlled by both interfacial characteristics (such as fiber coating [5]) and residual stresses induced by thermal expansion mismatch between fibers and matrix and can be tailored by appropriately coating the fiber surface. Fiber coating not only controls bond strength but also protects the fiber from being damaged by flaws generated during processing and service. Therefore, an improved understanding of the role of fiber coating and residual stresses in fracture behavior will lead to the design and processing of CFCCs with reliable performance. In this paper, we present results that will elucidate: (1) the effects of fiber coating on flaw generation, retained fiber strength and mechanical properties of Nicalon-fiber-reinforced SiC matrix composites; and (2) the role of residual stresses in the interfacial bond strength of SiC- fiber-reinforced reaction-bonded Si3N4 matrix composites. 2. Specimens and experimental procedure To evaluate the effects of fiber coating, specimens of Nicalon-fiber-reinforced SiC matrix composites with approximately 42 vol.% fiber reinforcements and carbon interlayer coating of varying thickness (0–1.25 mm) were Composites: Part A 30 (1999) 445–450 1359-835X/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S1359-835X(98)00133-X * Corresponding author. Tel: 1 1-630-252-5123; fax: 1 1-630-252- 3604
obtained from Oak Ridge National Laboratory. The speci- and outer loading spans of 15 and 30 mm, respectively. The mens were fabricated by densifying multiple layers of two- flexure bar specimens were typically 3 X 4X 32 mm and dimensional plain-weave Nicalon fiber mats stacked in a they were loaded at a rate of 1. 27 mm min under ambient 0-30-60 layup sequence in a graphite die. Chemical conditions. The ultimate stress was determined from the vapor infiltration( CVI), under forced conditions of thermal peak load obtained from the load vs specimen displacement and pressure gradients, was used to densify the preforms plot. The WoF of the composites was estimated from the with Sic. The resulting composites were close to 90% area under the load vs specimen displacement plots, normal dense. Details of composite specimen fabrication are ized on the basis of unit cross-sectional area of the fractured described elsewhere [6,71 composites. The true specimen displacement was obtained The role of residual stresses in controlling the fiber/matrix by subtracting the system displacement from the total interfacial bond strength was evaluated by using unidirec- displacement. The system displacement was determined tional monofilament SiC-reinforced reaction-bonded silicon by measuring system compliance with the aid of a stiff nitride(rBsn) composite specimens that were fabricated at alumina piece the NaSa Lewis Research Center. Cleveland. Oh Fractured composite specimens were examined with a Commercially available uncoated (SCS-0)or carbon-coated scanning electron microscope(Model JXA-840A, JEOL (SCS-6) SiC monofilaments(Textron Specialty Materials Co, Ltd, Tokyo) to locate the failure origin and establish Division, Lowell, MA) were used as reinforcements. the associated characteristic fracture surface morphology of These monofilaments consist of a central carbon core. the fibers to estimate in-situ fiber strength 33 um in diameter, an intermediate layer of chemical vapor-deposited SiC and an optional outer 3 um thick carbon-rich double coating, resulting in an overall filament 3 Results and discussion diameter of s 142 um. Composites were fabricated by high-temperature consolidation and nitridation of alter- Fig. I shows the variation in measured fracture stress(i.e nately stacked SiC fiber mats and silicon cloth in a die. ultimate strength) with fiber coating thickness. The ultimate The details of the processing are described in Bhatt [81 strength increases with the coating thickness and reaches a Composite specimens with varying fiber content were peak value of 380 MPa at a coating thickness of =0.2- processed to change the interfacial residual stress. The 0.6 um. Further increase in coating thickness did not sig- effects of the changes in residual stress were estimated by nificantly affect the ultimate strength. A plot of WoF as a measuring interfacial bond strength using fiber pushout function of coating thickness, presented in Fig. 2, shows that tests WoF increases rapidly with fiber coating thickness and described by Bright et al. [9] Pushout specimens, / e reaches a peak value of 1.92 x 10*Nm m- at a coating The fiber pushout testing followed the procedur thickness of 0.13-0.3 um. The WoF did not change 4 mm thick. were cut normal to the fiber axis. The faces significantly with further increase in coating thickness normal to the fiber axis were ground flat and parallel to In-situ fiber strength of the composites was evaluated each other. Subsequently, one face was fine-polished from the characteristic markings on the fracture surface of 0. 25 um surface finish. These specimens were mounted on a the fibers in a Nicalon-fiber-reinforced SiC composite specimen holder with an =0.8 mm wide slot to allow the tested in the four-point-bend mode, as shown in Fig. 3 exit of fiber that is pushed out from the other end. The Characteristic features associated with brittle failure, such sample holder was fixed to an X-y micropositioning stage as mirror(smooth region around the fracture origin)and and the whole assembly was attached to a load cell mounted hackle( region of multiple fracture planes)are clearly obser- on the crosshead of a mechanical testing machine(Model vable on the surface of fractured fibers. SEM of the fibers 4505, Instron Corporation, Canton, MA). The pushout tests showed that most failed from defects or flaws on the fiber involved positioning of individual fibers(with the X-y surface. Using the measured values of mirror radii, we esti stage, as described in Bright et al. [9]) under an =60 um mated the tensile strength of the fibers from the following diameter flat tip tungsten carbide indenter fixed to the top empirical relationship, which is commonly used for cera- plate of the mechanical tester, followed by loading the fibers mics [10, Il] at a displacement rate of 0.05 mm min The load-displa- cement plots were recorded and the debonding and frictional shear stresses were calculated from the corresponding load where Im represents the mirror radius, or is the tensile and interface area. For each composite, 15-20 fibers were strength, and Am is the mirror constant, which is related to pushed out to account for the statistical variation in the local the fracture toughness of the material. In the present study. response of the materials Am is taken as 3.5 MPa m -, following the work of Thou Ultimate strength and WoF of Nicalon-fiber-reinforced less et al. [3] The strength of more than 30 Nicalon fibers for Sic matrix comp with fiber coating of varying thick- each set of composite specimens was determined by ness were obtained by fracturing composites in a four-point- measuring their fracture mirror radius and solving Eq. (1) bend mode on the mechanical testing machine with inner The strength distribution of fibers in the composites was
obtained from Oak Ridge National Laboratory. The specimens were fabricated by densifying multiple layers of twodimensional plain-weave Nicalon fiber mats stacked in a 08–308–608 layup sequence in a graphite die. Chemical vapor infiltration (CVI), under forced conditions of thermal and pressure gradients, was used to densify the preforms with SiC. The resulting composites were close to 90% dense. Details of composite specimen fabrication are described elsewhere [6,7]. The role of residual stresses in controlling the fiber/matrix interfacial bond strength was evaluated by using unidirectional monofilament SiC-reinforced reaction-bonded silicon nitride (RBSN) composite specimens that were fabricated at the NASA Lewis Research Center, Cleveland, OH. Commercially available uncoated (SCS-0) or carbon-coated (SCS-6) SiC monofilaments (Textron Specialty Materials Division, Lowell, MA) were used as reinforcements. These monofilaments consist of a central carbon core, 33 mm in diameter, an intermediate layer of chemicalvapor-deposited SiC and an optional outer 3 mm thick carbon-rich double coating, resulting in an overall filament diameter of < 142 mm. Composites were fabricated by high-temperature consolidation and nitridation of alternately stacked SiC fiber mats and silicon cloth in a die. The details of the processing are described in Bhatt [8]. Composite specimens with varying fiber content were processed to change the interfacial residual stress. The effects of the changes in residual stress were estimated by measuring interfacial bond strength using fiber pushout tests. The fiber pushout testing followed the procedure described by Bright et al. [9] Pushout specimens, < 1– 4 mm thick, were cut normal to the fiber axis. The faces normal to the fiber axis were ground flat and parallel to each other. Subsequently, one face was fine-polished to a 0.25 mm surface finish. These specimens were mounted on a specimen holder with an <0.8 mm wide slot to allow the exit of fiber that is pushed out from the other end. The sample holder was fixed to an X–Y micropositioning stage and the whole assembly was attached to a load cell mounted on the crosshead of a mechanical testing machine (Model 4505, Instron Corporation, Canton, MA). The pushout tests involved positioning of individual fibers (with the X–Y stage, as described in Bright et al. [9]) under an <60 mm diameter flat tip tungsten carbide indenter fixed to the top plate of the mechanical tester, followed by loading the fibers at a displacement rate of 0.05 mm min21 . The load–displacement plots were recorded and the debonding and frictional shear stresses were calculated from the corresponding load and interface area. For each composite, 15–20 fibers were pushed out to account for the statistical variation in the local response of the materials. Ultimate strength and WOF of Nicalon-fiber-reinforced SiC matrix composites with fiber coating of varying thickness were obtained by fracturing composites in a four-pointbend mode on the mechanical testing machine with inner and outer loading spans of 15 and 30 mm, respectively. The flexure bar specimens were typically 3 × 4 × 32 mm and they were loaded at a rate of 1.27 mm min21 under ambient conditions. The ultimate stress was determined from the peak load obtained from the load vs specimen displacement plot. The WOF of the composites was estimated from the area under the load vs specimen displacement plots, normalized on the basis of unit cross-sectional area of the fractured composites. The true specimen displacement was obtained by subtracting the system displacement from the total displacement. The system displacement was determined by measuring system compliance with the aid of a stiff alumina piece. Fractured composite specimens were examined with a scanning electron microscope (Model JXA-840A, JEOL Co., Ltd, Tokyo) to locate the failure origin and establish the associated characteristic fracture surface morphology of the fibers to estimate in-situ fiber strength. 3. Results and discussion Fig. 1 shows the variation in measured fracture stress (i.e. ultimate strength) with fiber coating thickness. The ultimate strength increases with the coating thickness and reaches a peak value of 380 MPa at a coating thickness of <0.2– 0.6 mm. Further increase in coating thickness did not significantly affect the ultimate strength. A plot of WOF as a function of coating thickness, presented in Fig. 2, shows that WOF increases rapidly with fiber coating thickness and reaches a peak value of 1.92 × 104 Nm m22 at a coating thickness of <0.13–0.3 mm. The WOF did not change significantly with further increase in coating thickness. In-situ fiber strength of the composites was evaluated from the characteristic markings on the fracture surface of the fibers in a Nicalon-fiber-reinforced SiC composite, tested in the four-point-bend mode, as shown in Fig. 3. Characteristic features associated with brittle failure, such as mirror (smooth region around the fracture origin) and hackle (region of multiple fracture planes) are clearly observable on the surface of fractured fibers. SEM of the fibers showed that most failed from defects or flaws on the fiber surface. Using the measured values of mirror radii, we estimated the tensile strength of the fibers from the following empirical relationship, which is commonly used for ceramics [10,11]: sfr 1=2 m Am 1 where rm represents the mirror radius, sf is the tensile strength, and Am is the mirror constant, which is related to the fracture toughness of the material. In the present study, Am is taken as 3.5 MPa m21/2, following the work of Thouless et al. [3] The strength of more than 30 Nicalon fibers for each set of composite specimens was determined by measuring their fracture mirror radius and solving Eq. (1). The strength distribution of fibers in the composites was 446 J.P. Singh et al. / Composites: Part A 30 (1999) 445–450
J.P. Singh et al. /Composites: ParT 4 30(1999)445-450 447 Fracture 300 200 100 15kU x7,500 m888 Coating Thickness (um) Fig 3. Photomicrograph of surface morphology of fractured fiber in a CVI Fig. 1. Dependence of fracture stress on fiber coating thickness. described by the Weibull function given by Eq (2) important implications for the determination of optimum fiber coating parameters and composite processing m=1-s( The in-situ fiber strength parameter(oo) were used to redict ultimate strength (ours) of composites using Curtins model [ 12], In Eq (2), F is the fracture probability at a given stress o is the Weibull modulus that characterizes the faw size distribution in the fibers, and o is the scale parameter that signifies a characteristic strength value of the fibers The variation of in-situ fiber strength(scale parameter In Eq.( 3), fi is the fiber volume fraction in the loading To)as a function of coating thickness is shown in Fig. 4, direction, oo and m were defined earlier in Eq(2). In deter- which clearly shows that fiber strength initially increases mining, only fibers in the loading direction(0 fiber axis) with coating thickness and reaches a peak value at a coating were considered. The fi was obtained to be =0.07. It is to thickness in the range of =0.13-0.3 um. We believe that be noted that this assumption does not account for the the initial increase in strength is caused by protection of the contribution of ofi-axis fibers(30% and 60%)and, thus, will fiber by the coating, which minimizes fiber surface damage provide a lower strength prediction. A comparison of the during processing and fabrication. Further increase in coat- predicted and measured values of ultimate strength of ing thickness does not increase the effectiveness of the coat composites is shown in Fig. 5. As observed in the figure, ing in protecting the fiber from damage. This result has very he predicted strength values are lower than the measured 6 ITTTTTTTTTTTTTT In situ Nicalon Fibers 2.5 右o3 5.oo0 rnrIrnlrn I 00.20.40.60.8 00.20.40.60.81 Fiber Coating Thickness(um) Fiber Coating Thickness (um) Fig. 2. Dependence of work-of-fracture on fiber coating thickness Fig. 4. Dependence of in-situ fiber strength on fiber coating thickness
described by the Weibull function given by Eq. (2), F s 1 2 exp 2 s so m 2 In Eq. (2), F is the fracture probability at a given stress s, m is the Weibull modulus that characterizes the flaw size distribution in the fibers, and s0 is the scale parameter that signifies a characteristic strength value of the fibers. The variation of in-situ fiber strength (scale parameter, s0) as a function of coating thickness is shown in Fig. 4, which clearly shows that fiber strength initially increases with coating thickness and reaches a peak value at a coating thickness in the range of <0.13–0.3 mm. We believe that the initial increase in strength is caused by protection of the fiber by the coating, which minimizes fiber surface damage during processing and fabrication. Further increase in coating thickness does not increase the effectiveness of the coating in protecting the fiber from damage. This result has very important implications for the determination of optimum fiber coating parameters and composite processing. The in-situ fiber strength parameter (s0) were used to predict ultimate strength (sUTS) of composites using Curtin’s model [12], sUTS f1 2 m 1 2 1= m11 m 1 1 m 1 2 3 In Eq. (3), f1 is the fiber volume fraction in the loading direction, s0 and m were defined earlier in Eq. (2). In determining f1, only fibers in the loading direction (08 fiber axis) were considered. The f1 was obtained to be <0.07. It is to be noted that this assumption does not account for the contribution of off-axis fibers (308 and 608) and, thus, will provide a lower strength prediction. A comparison of the predicted and measured values of ultimate strength of composites is shown in Fig. 5. As observed in the figure, the predicted strength values are lower than the measured J.P. Singh et al. / Composites: Part A 30 (1999) 445–450 447 Fig. 3. Photomicrograph of surface morphology of fractured fiber in a CVI SiC/SiC composite. Fig. 4. Dependence of in-situ fiber strength on fiber coating thickness. Fig. 1. Dependence of fracture stress on fiber coating thickness. Fig. 2. Dependence of work-of-fracture on fiber coating thickness
J.P. Singh et al./Composites: Part A 30(1999)445-450 1000 △ Measured H This work Fiber Coating Thickness (um) Coating Thickness (um) Fig. 7. Variation of interfacial shear strength with fiber coating thickness Fig. 5. Correlation between predicted and measured composite ultimate for high coating thickness. The increase in fiber pull out length is a result of decrease in interfacial shear strength values as a function of fiber coating thickness. This differ- with increase in coating thickness. Using in-situ strength ence is believed to result from two reasons. First, the parameter(oo) and fiber pull out length(h), the interfacial composites strength measurements were made in a four- shear strength was calculated by Eq(4)[121 point flexural mode but the model prediction is for tensile strength. The flexural strength is expected to be higher thanT the tensile strength. The second reason is related to the estimation of fiber volume fraction (1) in the loading direc In Eq (4), A is a constant which depends on the Weibull tion contributing to strength. As discussed before, fi was parameter(m)and r is the radius of the fiber. Based on estimated based on an assumption that fibers in off-axis Curtin [12], the value of A was taken to be 1. The values direction do not contribute to strength of composites Prob- of o and h were obtained from Figs. 4 and 6, respectively ably, there will be some contribution of these off-axis fibers and a value of 8 um was used for the fiber radius. The to the composite strength predicted values of interfacial shear strength as a function The in-situ strength parameter (oo) was also used to of fiber coating thickness is shown in Fig. 7. The measured predict the fiber/matrix interfacial shear strength (T). To values of shear strength reported in Lowden [5] have also this end, the surfaces of composites fractured in bending been included in the figure for the purpose of comparison mode were evaluated by microscopy to measure fiber pull The predicted data shows a similar trend in variation of out length(h). The variation of fiber pull out length with shear strength with coating thickness. Both sets of data coating thickness is shown in Fig. 6. As seen in the figure, indicate a general decrease of interfacial shear strength the fiber pull out length increases with increasing fiber coat- with an increase in coating thickness till an optimal coating ing thickness and reaches approximately a constant value thickness is reached. The difference in the magnitude of shear strength may be related to the difference in composite 500 processing in the two cases A comparison of Figs. I and 2 with Fig. 4 shows a direct corre lation between the in-situ fiber strength. ultimate strength and WoF of composites with varying coating thickness. These observations suggest a strong dependence of ultimate strength and woF on in-situ fiber strength char- acteristics. In addition, fiber coating may also partly result in improved fiber/matrix interfacial characteristics that lead to the observed increase in both ultimate strength and WOF A typical load vs displacement plot for the fiber pushout value Pa, at which fiber/matrix debonding initiates an( Fig. 8. Initially, load increases and reaches a maximi 0.40.60 indicated by instantaneous load drop. At this point the fiber/ Fiber Coating Thickness (um) matrix interface may be either partially or completely With fur Fig. 6. Variation of fiber pull out length with fiber coating thickness, because of additional debonding and frictional siding an
values as a function of fiber coating thickness. This difference is believed to result from two reasons. First, the composites strength measurements were made in a fourpoint flexural mode but the model prediction is for tensile strength. The flexural strength is expected to be higher than the tensile strength. The second reason is related to the estimation of fiber volume fraction ( f1) in the loading direction contributing to strength. As discussed before, f1 was estimated based on an assumption that fibers in off-axis direction do not contribute to strength of composites. Probably, there will be some contribution of these off-axis fibers to the composite strength. The in-situ strength parameter (s0) was also used to predict the fiber/matrix interfacial shear strength (t). To this end, the surfaces of composites fractured in bending mode were evaluated by microscopy to measure fiber pull out length (h). The variation of fiber pull out length with coating thickness is shown in Fig. 6. As seen in the figure, the fiber pull out length increases with increasing fiber coating thickness and reaches approximately a constant value for high coating thickness. The increase in fiber pull out length is a result of decrease in interfacial shear strength with increase in coating thickness. Using in-situ strength parameter (s0) and fiber pull out length (h), the interfacial shear strength was calculated by Eq. (4) [12], t l mrs0 4h 4 In Eq. (4), l is a constant which depends on the Weibull parameter (m) and r is the radius of the fiber. Based on Curtin [12], the value of l was taken to be 1. The values of s0 and h were obtained from Figs. 4 and 6, respectively, and a value of 8 mm was used for the fiber radius. The predicted values of interfacial shear strength as a function of fiber coating thickness is shown in Fig. 7. The measured values of shear strength reported in Lowden [5] have also been included in the figure for the purpose of comparison. The predicted data shows a similar trend in variation of shear strength with coating thickness. Both sets of data indicate a general decrease of interfacial shear strength with an increase in coating thickness till an optimal coating thickness is reached. The difference in the magnitude of shear strength may be related to the difference in composite processing in the two cases. A comparison of Figs. 1 and 2 with Fig. 4 shows a direct correlation between the in-situ fiber strength, ultimate strength and WOF of composites with varying coating thickness. These observations suggest a strong dependence of ultimate strength and WOF on in-situ fiber strength characteristics. In addition, fiber coating may also partly result in improved fiber/matrix interfacial characteristics that lead to the observed increase in both ultimate strength and WOF. A typical load vs displacement plot for the fiber pushout experiment, performed on RBSN specimens, is shown in Fig. 8. Initially, load increases and reaches a maximum value Pd, at which fiber/matrix debonding initiates and is indicated by instantaneous load drop. At this point the fiber/ matrix interface may be either partially or completely debonded. With further displacement, the load increases because of additional debonding and frictional siding and 448 J.P. Singh et al. / Composites: Part A 30 (1999) 445–450 Fig. 5. Correlation between predicted and measured composite ultimate strength values. Fig. 6. Variation of fiber pull out length with fiber coating thickness. Fig. 7. Variation of interfacial shear strength with fiber coating thickness
J.P. Singh et al. /Composites: ParT 4 30(1999)445-450 Embedded Fiber Length= 1.45 mm Displacement, d(mm) Fiber Content, V(Vol % Fig 8. Typical load-displacement plot for fiber pushout test, showing Pa and p reaches a value Pmax. At this point, the entire fiber is debonded and begins to slide out the other side of the matrix, indicated by a steady decrease in load Assuming a uniform distribution of stress along the embedded fiber length L, the debonding Td and interfacial frictional Tr stresses were calculated by the following equations 5 Ta= Pa/2al and T= Pmax/2TaL where a is the fiber radius. The variation of debonding and 0 interfacial shear stresses as a function of fiber content, shown in Fig. 9, indicates that both debonding and shear Fiber Content, v,(VoL. % stresses decrease with increasing fiber content, except for the composites with a fiber content of 8.4 vol % The Fig 9. Variation of: (a)debonding; and(b)frictional sliding stresses with lebonding and shear stresses are lower for composites fiber content in SIC/RBSN composites with 8.4 vol. fiber than for composites with 12.6 vol% fiber. This inconsistency in the trend of stress variation is the interface. Hence, any change in residual radial stress believed to be related to processing. a density measurement will cause a variation in interfacial bonding and shear by Archimedes method indicates a lower density (70% of stresses theoretical) for composites with 8.4 vol. fiber than the Fig. 10 shows a variation in axial and transverse(radial) density (76% of theoretical) of composites with residual strains in SiC fibers that were measured [14 by 12.6 vol% fiber neutron diffraction in the intense pulsed neutron source at The general trend of decrease in debonding and interfa- Argonne National Laboratory. These strains were cial shear stresses with fiber content is consistent with the measured function of fiber content in SiC/RBSN predicted variation of residual stresses on the fibers caused composites. It can be clearly seen in the figure that trans by the expansion mismatch between the fiber and the matrix. verse(radial) compressive strain decreases with increas- As reported in Bright et al. [9] and goettler and Faber [1 ng fiber content, a finding that agrees with the analytical the thermal expansion coefficient of a rBSn matrix(am)is prediction of Majumdar et al. [15]. The observed decrease 3.3x10/C whereas the expansion coefficients of a Sic in the transverse strain will result in a corresponding fiber in axial (afa)and transverse(aa) directions are 4.5 x decrease in transverse stress. Since interfacial shear stress 10/C and 2.63 10/C, respectively. Based on these is proportional to the transverse stress, a decrease in trans- expansion coefficients, it is expected that, during cooling of verse stress will lead to reduced interfacial shear stress omposites from processing temperature, the fibers will be These predictions, based on the measured residual strain subjected to tensile axial and compressive transverse in composites by neutron diffraction, confirm the interfa- adial) stresses. The compressive radial stresses on the cial bonding and frictional shear stresses obtained during fiber will directly contribute to frictional shear stress at fiber pushout testing
reaches a value Pmax. At this point, the entire fiber is debonded and begins to slide out the other side of the matrix, indicated by a steady decrease in load. Assuming a uniform distribution of stress along the embedded fiber length L, the debonding t d and interfacial frictional t f stresses were calculated by the following equations td Pd=2paL and tf Pmax=2paL 5 where a is the fiber radius. The variation of debonding and interfacial shear stresses as a function of fiber content, shown in Fig. 9, indicates that both debonding and shear stresses decrease with increasing fiber content, except for the composites with a fiber content of 8.4 vol.%. The debonding and shear stresses are lower for composites with 8.4 vol.% fiber than for composites with 12.6 vol.% fiber. This inconsistency in the trend of stress variation is believed to be related to processing. A density measurement by Archimedes method indicates a lower density (70% of theoretical) for composites with 8.4 vol.% fiber than the density (76% of theoretical) of composites with 12.6 vol.% fiber. The general trend of decrease in debonding and interfacial shear stresses with fiber content is consistent with the predicted variation of residual stresses on the fibers caused by the expansion mismatch between the fiber and the matrix. As reported in Bright et al. [9] and Goettler and Faber [13], the thermal expansion coefficient of a RBSN matrix (am) is 3.3 × 1026 /8C whereas the expansion coefficients of a SiC fiber in axial (afa) and transverse (aft) directions are 4.5 × 1026 /8C and 2.63 × 1026 /8C, respectively. Based on these expansion coefficients, it is expected that, during cooling of composites from processing temperature, the fibers will be subjected to tensile axial and compressive transverse (radial) stresses. The compressive radial stresses on the fiber will directly contribute to frictional shear stress at the interface. Hence, any change in residual radial stress will cause a variation in interfacial bonding and shear stresses. Fig. 10 shows a variation in axial and transverse (radial) residual strains in SiC fibers that were measured [14] by neutron diffraction in the intense pulsed neutron source at Argonne National Laboratory. These strains were measured as a function of fiber content in SiC/RBSN composites. It can be clearly seen in the figure that transverse (radial) compressive strain decreases with increasing fiber content, a finding that agrees with the analytical prediction of Majumdar et al. [15]. The observed decrease in the transverse strain will result in a corresponding decrease in transverse stress. Since interfacial shear stress is proportional to the transverse stress, a decrease in transverse stress will lead to reduced interfacial shear stress. These predictions, based on the measured residual strain in composites by neutron diffraction, confirm the interfacial bonding and frictional shear stresses obtained during fiber pushout testing. J.P. Singh et al. / Composites: Part A 30 (1999) 445–450 449 Fig. 9. Variation of: (a) debonding; and (b) frictional sliding stresses with fiber content in SiC/RBSN composites. Fig. 8. Typical load–displacement plot for fiber pushout test, showing Pd and Pmax
0.002 Energy, Advanced Research and Technology Development, Fossil Energy Material Program, and Office of Industrial g0.0015 Technology, Energy Efficiency and Renewable Energy, AXIAL W-31-109-Eng-38. The authors thank R. A. Lowden(Oak Ridge National Laboratory) for pro- ber-reinforced SiC matrix composit 0.0005 Ramakrishna T Bhatt (U.S. Army Propulsion Directorate NASA Lewis Research Center) for providing SIC/RBSN RANSVERSE composites, and grant Pollard for providing experimental -0.00 References Fiber Content, V,(vol % f Sic/rBSN composites as a function of fiber mic matrix composites. J Am Ceramic Soc 1985: 68(5): 225-231 tron diffraction in the intense pulsed neutron [2] Evans AG, Marshall DB. The mechanical behavior of ceramic matrix source at Argo onal Laboratory and analytically predicted by opposites, Overview No 85. Acta Metall 1989, 37(10): 2567-2583 Majumdar et al. [1 B Thouless MD, Sbaizero O, Sigl LS, Evans AG. Effect of interface mechanical properties on pullout in a Sic-fiber-reinforced lithium 4. Summary aluminum silicate glass ceramic. J Am Ceramic Soc 1989: 72(4) 25-532 1. Processing-induced damage of Nicalon fibers can be [4] Mah T, Mendiratta MG, Katz AP, Ruh R, Mazdiyasni ks Room minimized by the application of a carbon coating, mperature mechanical behavior of fiber-reinforced ceramic matri composites. J Am Ceramic Soc 1985: 68(1): C27-C30 which leads to a corresponding increase in the retained 5 Lowden RA Fiber coatings and the mechanical properties of a fiber in-situ strength of the fibers during composite processing inforced ceramic composite. Adv Composite Mater, Ceramic Trans 2. Ultimate strength and work of fracture(wOF)of comp 1991;19:619-630 sites were found to increase with increasing coating [6]Stinton DP, Caputo AJ, Lowden RA. Synthesis of fiber-reinforced SiC mposites by chemical vapor infiltration. Am Ceramic Soc Bull thickness up to = 0.3 um. Further increase in coating 98665(2):347-350 thickness did not significantly affect mechanical prop. [7 Stinton DP, Bessman TM, Lowden RA. Advanced ceramics erties, indicating an optimum coating thickness for deposition techniques. Am Ceramic Soc Bu processIng 867(2:350-355 3. A similarity between the dependence of in-situ fiber 8]Bhatt RT. Method of preparing fiber-reinforced ceramic materials. US atent No, 4689188 strength, ultimate strength and WoF on fiber coating [9] Bright JD, Shetty DK, Griffin CW, Limaye SY. Interfaci thickness suggests a direct correlation between retained and friction in silicon carbide (filament)-reinforced cera in-situ fiber strength and resulting mechanical properties lass-matrix composites. J Am Ceramic Soc 1989, 72(10) 4. The increase in ultimate strength and WoF with increas- [10] Kirchner HP, Gruver RM. Fracture mirror in alumina ceramics. Phil ing fiber coating thickness may also be partly caused by Mag1973:27:1433-1446 improved fiber/matrix interfacial characteristics [11] Mecholsky JJ, Freiman Sw, Rice RW. Fracture surface analysis of ceramics. J Mater Sci 1976: 11:- 1310-1319 5. For SiC-fiber-reinforced reaction-bonded SizNa matrix [12] Curtin WA. Theory of mechanical properties of ceramic-matrix composites, both debonding and frictional shear stresses composites. J Am Ceramic Soc 1991; 74(11): 2831-2845 generally decreased with increasing fiber content. These [13] Goettler RW, Faber KT Interfacial shear stresses in Sic and AL,O ariations are consistent with the variations in residual fiber-reinforced glasses. Ceramic Engng Sci Proc 1988; 9(7-8): 701 radial stress of fibers measured by neutron diffraction, the [14] Saigal A, Kupperman DS, Singh JP, Singh D, Ri J. Bhatt rt latter decreased with increasing fiber content. Thermal residual strains and stresses in silico forced silicon nitride composites. Composite 075-1086 Acknowledgements [15] Majumdar S, Singh D, Singh JP. Analysis of pushout tests on a SiC- fiber-reinforced reaction-bonded Si, N4 composite Composites Engng 1993;3(4)287-312 This work was supported by the Us Department of
4. Summary 1. Processing-induced damage of Nicalon fibers can be minimized by the application of a carbon coating, which leads to a corresponding increase in the retained in-situ strength of the fibers during composite processing. 2. Ultimate strength and work of fracture (WOF) of composites were found to increase with increasing coating thickness up to <0.3 mm. Further increase in coating thickness did not significantly affect mechanical properties, indicating an optimum coating thickness for processing. 3. A similarity between the dependence of in-situ fiber strength, ultimate strength and WOF on fiber coating thickness suggests a direct correlation between retained in-situ fiber strength and resulting mechanical properties. 4. The increase in ultimate strength and WOF with increasing fiber coating thickness may also be partly caused by improved fiber/matrix interfacial characteristics. 5. For SiC-fiber-reinforced reaction-bonded Si3N4 matrix composites, both debonding and frictional shear stresses generally decreased with increasing fiber content. These variations are consistent with the variations in residual radial stress of fibers measured by neutron diffraction, the latter decreased with increasing fiber content. Acknowledgements This work was supported by the US Department of Energy, Advanced Research and Technology Development, Fossil Energy Material Program, and Office of Industrial Technology, Energy Efficiency and Renewable Energy, under Contract W-31-109-Eng-38. The authors thank R. A. Lowden (Oak Ridge National Laboratory) for providing Nicalon-fiber-reinforced SiC matrix composites, Ramakrishna T. Bhatt (U.S. Army Propulsion Directorate, NASA Lewis Research Center) for providing SiC/RBSN composites, and Grant Pollard for providing experimental assistance. References [1] Marshall DB, Evans AG. Failure mechanisms in ceramic-fiber/ceramic matrix composites. J Am Ceramic Soc 1985;68(5):225–231. [2] Evans AG, Marshall DB. The mechanical behavior of ceramic matrix composites, Overview No 85. Acta Metall 1989;37(10):2567–2583. [3] Thouless MD, Sbaizero O, Sigl LS, Evans AG. Effect of interface mechanical properties on pullout in a SiC-fiber-reinforced lithium aluminum silicate glass ceramic. J Am Ceramic Soc 1989;72(4): 525–532. [4] Mah T, Mendiratta MG, Katz AP, Ruh R, Mazdiyasni KS. Room temperature mechanical behavior of fiber-reinforced ceramic matrix composites. J Am Ceramic Soc 1985;68(1):C27–C30. [5] Lowden RA. Fiber coatings and the mechanical properties of a fiberreinforced ceramic composite. Adv Composite Mater, Ceramic Trans 1991;19:619–630. [6] Stinton DP, Caputo AJ, Lowden RA. Synthesis of fiber-reinforced SiC composites by chemical vapor infiltration. Am Ceramic Soc Bull 1986;65(2):347–350. [7] Stinton DP, Bessman TM, Lowden RA. Advanced ceramics by chemical vapor deposition techniques. Am Ceramic Soc Bull 1988;67(2):350–355. [8] Bhatt RT. Method of preparing fiber-reinforced ceramic materials. US Patent No. 4689188. [9] Bright JD, Shetty DK, Griffin CW, Limaye SY. Interfacial bonding and friction in silicon carbide (filament)-reinforced ceramic- and glass-matrix composites. J Am Ceramic Soc 1989;72(10):1891–1898. [10] Kirchner HP, Gruver RM. Fracture mirror in alumina ceramics. Phil Mag 1973;27:1433–1446. [11] Mecholsky JJ, Freiman SW, Rice RW. Fracture surface analysis of ceramics. J Mater Sci 1976;11:1310–1319. [12] Curtin WA. Theory of mechanical properties of ceramic-matrix composites. J Am Ceramic Soc 1991;74(11):2831–2845. [13] Goettler RW, Faber KT. Interfacial shear stresses in SiC and Al2O3 fiber-reinforced glasses. Ceramic Engng Sci Proc 1988;9(7–8):701– 861. [14] Saigal A, Kupperman DS, Singh JP, Singh D, Richardson J, Bhatt RT. Thermal residual strains and stresses in silicon-carbide-fiber-reinforced silicon nitride composites. Composites Engng 1993;3(11): 1075–1086. [15] Majumdar S, Singh D, Singh JP. Analysis of pushout tests on a SiC- fiber-reinforced reaction-bonded Si3N4 composite. Composites Engng 1993;3(4):287–312. 450 J.P. Singh et al. / Composites: Part A 30 (1999) 445–450 Fig. 10. Residual strain of SiC/RBSN composites as a function of fiber content, measured by neutron diffraction in the intense pulsed neutron source at Argonne National Laboratory and analytically predicted by Majumdar et al. [15] (solid lines)