Availableonlineatwww.sciencedirect.com SCIENCE DIRECT O CARBON ELSEVIER Carbon4302005)1954-1960 www.elsevier.com/locate/carbon Influence of the matrix microstructure on the mechanical properties of Cvi-infiltrated carbon fiber felts M. Guellali *R. Oberacker M.J. Hoffman Institut fir Keramik im Maschinenbau, Universitat Karlsruhe, Haid-und-Neu-Str. 7, 76131 Karlsruhe, Germany Received 20 September 2004; accepted I March 2005 Available online 14 April 2005 Abstract The correlation between the matrix microstructure and the mechanical properties of CVI-infiltrated carbon fiber felts was studied by optical microscopy, scanning electron microscopy, and three-point bending tests. The results of these investigations show a cor- elation between(a)the content of highly textured pyrocarbon in the matrix and the quasi-ductile fracture behavior of the samples and(b)the thickness of the low textured pyrocarbon layers beneath the fibers and the measured flexural strengths. Fractographic Ivestigations using SEM showed that toughness increase results from multiple crack deflections at the interface between numerou sublayers' forming the highly textured pyrocarbon. The increase of flexural strength could be explained by a thicker, so called'vir- tual fiber c 2005 Elsevier Ltd. All rights reserved Keywords: Pyrolytic carbon; Optical microscopy, scanning electron microscopy: Mechanical properties, texture 1. Introduction more advanced service applications. In 1966 Bokros et al. [5] already reported that the mechanical properties Isothermal, isobaric chemical vapor infiltration(I of pyrolytic carbons deposited in fluidized beds are CVI) is widely used to densify porous carbon preforms strongly dependant on their different microstructures for high tech applications such as in aeronautical and resulting from various deposition and annealing condi space industries [1]. The microstructure of the deposited tions. Granoff et al. [6] showed a slight influence of the pyrocarbon matrix during the I-Cvi process can be var- deposition conditions(temperature and pressure)on ied from a nearly amorphous to a highly crystalline gra- the mechanical properties of C/C-carbon felts, though phatic state [24] by controlling the process parameters they focused by their explanation more or less exclu- (e.g. gas pressure, residence time, temperature). These sively on the influence of the heat treatment. Oh et al microstructural differences should influence the mechan- [7,8] studied the effect of the matrix structure of two ical properties of CVl-materials. Therefore, the knowl- dimensional C/C-composites prepared by thermal-gradi edge about the relationship between microstructure ent CVi on mechanical properties and fracture behavior. variations of the pyrolytic matrix and the mechanical They related the observed differences either to the pres- properties of these materials is essential in order to as- ence of matrix cracks or to different densities. Kimura sess their full potential operational performance for et al. [9] examined the fracture behavior of unidirec- tional C/C-composites prepared by a pressure gradient Corresponding author. Tel: +49 721 608 4248: fax: + method. They found that in the case of ISo + rC 8891 matrix (Iso= isotrop, RC= rough columnar)the Iso E-mail address: moez. guellali(@ikm. uni-karlsruhe de (M. guellali carbon adhered well to the fibers and that cracks 0008-6223/S. see front matter 2005 Elsevier Ltd. All rights reserved
Influence of the matrix microstructure on the mechanical properties of CVI-infiltrated carbon fiber felts M. Guellali *, R. Oberacker, M.J. Hoffmann Institut fu¨r Keramik im Maschinenbau, Universita¨t Karlsruhe, Haid-und-Neu-Str. 7, 76131 Karlsruhe, Germany Received 20 September 2004; accepted 1 March 2005 Available online 14 April 2005 Abstract The correlation between the matrix microstructure and the mechanical properties of CVI-infiltrated carbon fiber felts was studied by optical microscopy, scanning electron microscopy, and three-point bending tests. The results of these investigations show a correlation between (a) the content of highly textured pyrocarbon in the matrix and the quasi-ductile fracture behavior of the samples and (b) the thickness of the low textured pyrocarbon layers beneath the fibers and the measured flexural strengths. Fractographic investigations using SEM showed that toughness increase results from multiple crack deflections at the interface between numerous sublayers forming the highly textured pyrocarbon. The increase of flexural strength could be explained by a thicker, so called virtual fiber. 2005 Elsevier Ltd. All rights reserved. Keywords: Pyrolytic carbon; Optical microscopy, scanning electron microscopy; Mechanical properties, texture 1. Introduction Isothermal, isobaric chemical vapor infiltration (ICVI) is widely used to densify porous carbon preforms for high tech applications such as in aeronautical and space industries [1]. The microstructure of the deposited pyrocarbon matrix during the I-CVI process can be varied from a nearly amorphous to a highly crystalline graphitic state [2–4] by controlling the process parameters (e.g. gas pressure, residence time, temperature). These microstructural differences should influence the mechanical properties of CVI-materials. Therefore, the knowledge about the relationship between microstructure variations of the pyrolytic matrix and the mechanical properties of these materials is essential in order to assess their full potential operational performance for more advanced service applications. In 1966 Bokros et al. [5] already reported that the mechanical properties of pyrolytic carbons deposited in fluidized beds are strongly dependant on their different microstructures resulting from various deposition and annealing conditions. Granoff et al. [6] showed a slight influence of the deposition conditions (temperature and pressure) on the mechanical properties of C/C-carbon felts, though they focused by their explanation more or less exclusively on the influence of the heat treatment. Oh et al. [7,8] studied the effect of the matrix structure of twodimensional C/C-composites prepared by thermal-gradient CVI on mechanical properties and fracture behavior. They related the observed differences either to the presence of matrix cracks or to different densities. Kimura et al. [9] examined the fracture behavior of unidirectional C/C-composites prepared by a pressure gradient method. They found that in the case of ISO + RC matrix (ISO = isotrop, RC = rough columnar) the ISO carbon adhered well to the fibers and that cracks Carbon 43 (2005) 1954–1960 www.elsevier.com/locate/carbon 0008-6223/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2005.03.006 * Corresponding author. Tel.: +49 721 608 4248; fax: +49 721 608 8891. E-mail address: moez.guellali@ikm.uni-karlsruhe.de (M. Guellali).
M. Guellali et al Carbon 43(2005)1954-1960 1955 propagated in the RC matrix. In the case of a RC+ Sc Table I matrix(SC= smooth columnar) delamination parallel ies of the infiltrated felts to the fiber axes was observed kpA 0 kPa In a previous paper [10] we studied the infuence of Density, p[g/cm]1.72 multilayered pyrocarbon matrix on the mechanical porosity, Po 20.0 24.5 properties of C/C-felts. The results of these preliminary tests showed the significant influence of the so called virtual'fiber(fiber plus low or medium textured pyro- bulk densities and porosities of the four infiltrated felts are summarized in Table 1 pyrocarbon(rough laminar)on the flexural strength an The textures of the deposited pyrocarbons were deter fracture toughness, respectively. The aim of this work is a more general clarification of the correlation between (PLM) using an OLYMPUS AX70 microscope accord CVI-infiltrated carbon fiber felts and to validate our ing to their optical activity and the value of the extinc explanations proposed in our previous work [1O]. Felts on angle Ae as described by Bourrat et al.[12).The re suitable preforms for such studies, as their random corresponding pyrocarbon optical textures with a pro- fiber orientation and their excellent homogeneity makes gressive anisotropy degree are defined as isotropic microstructure-property-correlations easier than the (ISo, Ae <4), low textured or dark laminar (LT or common long fiber reinforced composite materials DL,4°≤A<129), medium textured or smooth lami nar( MT or SL,12°≤A<18°), and highly textured or rough laminar( ht Or RL,A≥18°)[12,13].The 2. Experimental thicknesses of the deposited pyrocarbon layers were measured on polished samples. The figures in Table 2 The investigated samples are carbon fiber felts are mean values of at least 20 measurements. The highly (CCKF 1001, Sintec, Germany) with an initial relative textured pyrocarbon contents in the matrix were calcu porosity of 88 vol. infiltrated by means of an I-CVI lated knowing the arrangement of the layers around the fiber and their thickne process at a temperature of 1100C. The infiltration was carried out in the group of Prof. Huttinger at the Three-point bending tests were carried out in order to Institute for Chemical Technology of the University of determine the mechanical properties of the composites For this gular bars of approximately Karlsruhe, Germany. The PAN-fibers forming the felts 8x35 x0.5 mm' were cut from the center of the sam- have a typical mean diameter of 12 um and are ran domly oriented. The first pair of felts (referred to ples using a diamond-wire saw. To avoid possible dam- aging of the samples(thickness 0.5 mm)their edges 20 kPa and 30 kPa below)were infiltrated using a meth- were not rounded as suggested in some ASTM proce- ane/hydrogen mixture of the ratio PCH, /PH:=7: I at to- dures for measuring strength of ceramics. The sample pair of felts(here named inlet and outlet)were infiltrated dimensions were measured using a micrometer screw in one reactor run using pure methane at a total pressure Tests were then carried out on a universal testing ma of 30 kPa. The first (inlet) was located at the bottom of chine (UTS, Germany). The specimens were placed on rolls, 3 mm in diameter. A span of 7 mm was used, so of the reactor(Fig. 1). Further details on the infiltration carried out with a constant cross-head speed of 10 mm min. At least twenty specimens were tested for each composite. The load and deflection values were recorded as a function of time. The nominal bending stress(a) gas outlet and the nominal outer fiber strain(a) were calculated acco rding to following equations [14 3·F·L gas flow C- felt 6·s·d nlet where Fis the load(N), L the span(mm), b the specimen width(mm), d the specimen height(mm), and s the di placement or deflection(mm) In order to compare the quasi-ductile fracture behav Fig. 1. Experimental setup ior of the samples, a ductility factor Fp is introduced. It
propagated in the RC matrix. In the case of a RC + SC matrix (SC = smooth columnar) delamination parallel to the fiber axes was observed. In a previous paper [10] we studied the influence of a multilayered pyrocarbon matrix on the mechanical properties of C/C-felts. The results of these preliminary tests showed the significant influence of the so called virtual fiber (fiber plus low or medium textured pyrocarbon layers directly around it) and the highly textured pyrocarbon (rough laminar) on the flexural strength and fracture toughness, respectively. The aim of this work is a more general clarification of the correlation between matrix microstructure and mechanical properties of CVI-infiltrated carbon fiber felts and to validate our explanations proposed in our previous work [10]. Felts are suitable preforms for such studies, as their random fiber orientation and their excellent homogeneity makes microstructure-property-correlations easier than the common long fiber reinforced composite materials. 2. Experimental The investigated samples are carbon fiber felts (CCKF 1001, Sintec, Germany) with an initial relative porosity of 88 vol.% infiltrated by means of an I-CVI process at a temperature of 1100 C. The infiltration was carried out in the group of Prof. Hu¨ttinger at the Institute for Chemical Technology of the University of Karlsruhe, Germany. The PAN-fibers forming the felts have a typical mean diameter of 12 lm and are randomly oriented. The first pair of felts (referred to 20 kPa and 30 kPa below) were infiltrated using a methane/hydrogen mixture of the ratio pCH4 =pH2 ¼ 7 : 1 at total pressures of 20 and 30 kPa, respectively. The second pair of felts (here named inlet and outlet) were infiltrated in one reactor run using pure methane at a total pressure of 30 kPa. The first (inlet) was located at the bottom of the reactor while the second (outlet) was at the top part of the reactor (Fig. 1). Further details on the infiltration procedure are given elsewhere [3,4,11]. The determined bulk densities and porosities of the four infiltrated felts are summarized in Table 1. The textures of the deposited pyrocarbons were determined on polished cross-sections under polarized light (PLM) using an OLYMPUS AX70 microscope according to their optical activity and the value of the extinction angle Ae as described by Bourrat et al. [12]. The corresponding pyrocarbon optical textures with a progressive anisotropy degree are defined as isotropic (ISO, Ae < 4), low textured or dark laminar (LT or DL, 4 6 Ae < 12), medium textured or smooth laminar (MT or SL, 12 6 Ae < 18), and highly textured or rough laminar (HT or RL, Ae P 18) [12,13]. The thicknesses of the deposited pyrocarbon layers were measured on polished samples. The figures in Table 2 are mean values of at least 20 measurements. The highly textured pyrocarbon contents in the matrix were calculated knowing the arrangement of the layers around the fiber and their thickness. Three-point bending tests were carried out in order to determine the mechanical properties of the composites. For this purpose rectangular bars of approximately 8 · 3.5 · 0.5 mm3 were cut from the center of the samples using a diamond-wire saw. To avoid possible damaging of the samples (thickness 6 0.5 mm) their edges were not rounded as suggested in some ASTM procedures for measuring strength of ceramics. The sample dimensions were measured using a micrometer screw. Tests were then carried out on a universal testing machine (UTS, Germany). The specimens were placed on rolls, 3 mm in diameter. A span of 7 mm was used, so giving a span-to-width ratio of about 14. The tests were carried out with a constant cross-head speed of 10 mm/ min. At least twenty specimens were tested for each composite. The load and deflection values were recorded as a function of time. The nominal bending stress (r) and the nominal outer fiber strain (e) were calculated according to following equations [14]: r ¼ 3 F L 2 b d2 ð1Þ e ¼ 6 s d L2 100 ð2Þ where F is the load (N), L the span (mm), b the specimen width (mm), d the specimen height (mm), and s the displacement or deflection (mm). In order to compare the quasi-ductile fracture behavior of the samples, a ductility factor FD is introduced. It cylindrical sample holder gas outlet gas inlet gas flow C- felt direction outlet inlet graphite foil Fig. 1. Experimental setup. Table 1 Bulk properties of the infiltrated felts 20 kPa 30 kPa Inlet Outlet Density, q [g/cm3 ] 1.72 1.62 1.57 1.52 Porosity, P0 [%] 15.0 20.0 21.5 24.5 M. Guellali et al. / Carbon 43 (2005) 1954–1960 1955
M. Guellali et al. Carbon 43(2005)1954-1960 Table 2 Optical texture (OT) and thickness(d) of the pyrocarbon layers of the investigated C/C-felts Layer d [um d [uml ISO DL(LT) SL(MT) SLOT 12345 RL(HT SLOT RL(HT) RL(HT) RL(HT) RL(HT) 32.0 190 18.8 OT: optical texture, d. layer thickness, ISo: isotrop, DL(LT): dark laminar or low textured, SL(MT): smooth laminar or medium textured, Rl(hT ough laminar or highly textured. slope of the linear part of the stress-strain curve) as it stress at failure is described in(Fig. 2)[10, 15, 16 The interpretation of the bending strength results was made with the help of the widely used Weibull distribu- tion [17, 18]. The Weibull parameters(m: shape parame- ductility factor Fo F=1-(E ter of the distribution and ao: scale parameter, i.e. the strength at a failure probability of 63.21%) were calci lated by the maximum likelihood method, according to[9] After flexural testing. fracture surfaces were exam ined using a scanning electron microscope (SEM) LEO 440C" without depositing conductive layer nominal strain Fig. 2. Definition of the ductility factor Fr is calculated from the ratio of the secant modulus(the Fig. 3 shows four optical micrographs of the exam- slope of the line from the origin to the stress at failure ined composites. They exhibit various types of matrix in the stress-strain curve) to the elastic modulus(the microstructure resulting from different infiltration con Fig 3. Polarized I optical micro ographs of the investigated C/C-felts
is calculated from the ratio of the secant modulus (the slope of the line from the origin to the stress at failure in the stress–strain curve) to the elastic modulus (the slope of the linear part of the stress–strain curve) as it is described in (Fig. 2) [10,15,16]. The interpretation of the bending strength results was made with the help of the widely used Weibull distribution [17,18]. The Weibull parameters (m: shape parameter of the distribution and r0: scale parameter, i.e. the strength at a failure probability of 63.21%) were calculated by the maximum likelihood method, according to [19]. After flexural testing, fracture surfaces were examined using a scanning electron microscope (SEM) LEO 440C without depositing conductive layers. 3. Results Fig. 3 shows four optical micrographs of the examined composites. They exhibit various types of matrix microstructure resulting from different infiltration conTable 2 Optical texture (OT) and thickness (d) of the pyrocarbon layers of the investigated C/C-felts 20 kPa 30 kPa Inlet Outlet Layer OT d [lm] OT d [lm] OT d [lm] OT d [lm] 1 ISO 3.1 DL(LT) 2.0 SL(MT) 7.0 SL(MT) 11.8 2 RL(HT) 30.0 SL(MT) 7.0 RL(HT) 12.0 RL(HT) 7.0 3 RL(HT) 9.7 4 SL(MT) 4.5 5 RL(HT) 8.8 R 33.1 32.0 19.0 18.8 OT: optical texture, d: layer thickness, ISO: isotrop, DL(LT): dark laminar or low textured, SL(MT): smooth laminar or medium textured, RL(HT): rough laminar or highly textured. nominal stress nominal strain stress at failure εt ductility factor FD: FD = 1-(Esecant/Eorigin) = 1-(εlin/εt) εlin Fig. 2. Definition of the ductility factor FD. Fig. 3. Polarized optical micrographs of the investigated C/C-felts. 1956 M. Guellali et al. / Carbon 43 (2005) 1954–1960
M. Guellali et al Carbon 43(2005)1954-1960 ditions. The matrix of the first felt (20 kPa)obtained (<10%)exhibited a purely linear stress-strain curve, cor- with the methane/hydrogen mixture under the lower responding to a brittle, catastrophic failure mode. The methane partial pressure(17. 5 kPa)consists of approxi remainder of the samples exhibited a quasi-ductile frac nately 95% of highly textured pyrocarbon as shown in ture behavior. In fact, their nominal stress-nominal he top-left micrograph of Fig. 3. The increase of the outer fiber strain curves showed a linear zone followed methane partial pressure to 26.25 kPa (second felt, by a clear deviation from the linear part. Fig. 4 shows 30 kPa) led to the formation of a multilayered matrix typical curves for the four investigated felts composed of alternating bands of carbon layers having The stress-strain curves of the 20 kPa and inlet felts The matrices of the third felt (inlet)and the fourth felt 30 kPa and outlet felts(Fig. 4). The average values of (outlet) are both formed of one medium textured pyro- the ratio of the ductility factor for the 20 kPa(16.4%) carbon layer followed by a highly textured carbon sheet. and inlet(11.9%)felts are distinctly higher than those The inlet felt, closer to the entrance of the reactant gas, for the 30 kPa(11.2%)and outlet( 8.3%)felts contains about 27% SL(MT) carbon. On contrast, the The Weibull diagrams for the four felts are given outlet felt, placed in the top part of the reactor, is Fig. 5. The dashed lines mark the 95% confidence ormed of about 53% SL(MT) carbon(Fig 3, bottom- intervals. The 20 kPa and inlet felts have significantly right). The optical texture of the individual carbon lay- lower flexural strengths than the 30 kPa and outlet ers and their thicknesses are summarized in Table 2. felts. The scale parameters, go, of the 30 kPa The couples 20 kPa/30 kPa and inlet/outlet felts exhibit (60 MPa) and outlet (66 MPa) felts are about 25% each nearly an equal total thickness of the deposited higher than those of the 20 kPa(48 MPa) and inlet pyrocarbon matrix(Table 2). The felts 20 kPa and inlet (53 MPa) felts. At first view these strength values seem exhibit higher bulk densities and lower porosity than the to be low. But taking the low fiber volume fraction in felts 30 kPa and outlet Table 1) the felt into account. these values become reasonable More than twenty samples of each felt were tested in Comparable values for infiltrated carbon felts were re- three-point bending mode. Only a few of the samples ported, e.g. by Benzinger [20] 10 1.6 nominal strain [ nominal strain [ Fig. 4. Typical nominal stress-nominal strain curves of the investigated composites. flexural strength [MPa flexural strength[MPa] Fig. 5. Flexural strength distributions of the investi
ditions. The matrix of the first felt (20 kPa) obtained with the methane/hydrogen mixture under the lower methane partial pressure (17.5 kPa) consists of approximately 95% of highly textured pyrocarbon as shown in the top-left micrograph of Fig. 3. The increase of the methane partial pressure to 26.25 kPa (second felt, 30 kPa) led to the formation of a multilayered matrix composed of alternating bands of carbon layers having various textures and thicknesses (Fig. 3, top-right). The matrices of the third felt (inlet) and the fourth felt (outlet) are both formed of one medium textured pyrocarbon layer followed by a highly textured carbon sheet. The inlet felt, closer to the entrance of the reactant gas, contains about 27% SL(MT) carbon. On contrast, the outlet felt, placed in the top part of the reactor, is formed of about 53% SL(MT) carbon (Fig. 3, bottomright). The optical texture of the individual carbon layers and their thicknesses are summarized in Table 2. The couples 20 kPa/30 kPa and inlet/outlet felts exhibit each nearly an equal total thickness of the deposited pyrocarbon matrix (Table 2). The felts 20 kPa and inlet exhibit higher bulk densities and lower porosity than the felts 30 kPa and outlet (Table 1). More than twenty samples of each felt were tested in three-point bending mode. Only a few of the samples (<10%) exhibited a purely linear stress–strain curve, corresponding to a brittle, catastrophic failure mode. The remainder of the samples exhibited a quasi-ductile fracture behavior. In fact, their nominal stress-nominal outer fiber strain curves showed a linear zone followed by a clear deviation from the linear part. Fig. 4 shows typical curves for the four investigated felts. The stress–strain curves of the 20 kPa and inlet felts show a larger quasi-ductile zone than those of the 30 kPa and outlet felts (Fig. 4). The average values of the ratio of the ductility factor for the 20 kPa (16.4%) and inlet (11.9%) felts are distinctly higher than those for the 30 kPa (11.2%) and outlet (8.3%) felts. The Weibull diagrams for the four felts are given in Fig. 5. The dashed lines mark the 95% confidence intervals. The 20 kPa and inlet felts have significantly lower flexural strengths than the 30 kPa and outlet felts. The scale parameters, r0, of the 30 kPa (60 MPa) and outlet (66 MPa) felts are about 25% higher than those of the 20 kPa (48 MPa) and inlet (53 MPa) felts. At first view these strength values seem to be low. But taking the low fiber volume fraction in the felt into account, these values become reasonable. Comparable values for infiltrated carbon felts were reported, e.g. by Benzinger [20]. 0 10 20 30 40 50 60 70 0 0.4 0.8 1.2 1.6 nominal strain [%] nominal strain [%] nominal stress [MPa] 20kPa 30kPa inlet outlet 0 10 20 30 40 50 60 70 0 0.5 1 1.5 2 2.5 3 nominal stress [MPa] 2 (a) (b) Fig. 4. Typical nominal stress-nominal strain curves of the investigated composites. 10 100 1 2 3 5 10 20 30 40 50 60 70 80 90 95 99 flexural strength [MPa] flexural strength [MPa] failure probability [%] failure probability [%] m = 5.7 σ0 m = 11.4 σ0 20kPa 30kPa = 48 = 60 10 100 99 9590 80 70 60 50 40 30 20 10 5 3 2 1 outlet inlet m = 4.9 σ0 m = 4.4 σ0= 66 = 53 (a) (b) Fig. 5. Flexural strength distributions of the investigated composites. M. Guellali et al. / Carbon 43 (2005) 1954–1960 1957
M. Guellali et al. Carbon 43(2005)1954-1960 outlet Fig. 6.(a) SEM micrographs of fracture surfaces after three-point bending tests with(b) their corresponding schematic drawings of fracture profiles. Fig 6 shows SEM micrographs of the fracture sur- delamination between sublayers within highly textured faces of the composites after bending tests and the cor- Rl(Ht) pyrocarbon can be seen(Fig. 7) responding schemes of their fracture profiles. For al felts fracture of the fibers and the low or medium tex- tured matrix layers in the fiber vicinity occur approxi- 4 mately in the same fracture plane. The rest of the matrix occupies distinctly another fracture plane level Structural investigations and mechanical testing An intensive fragmentation takes place within the highly yielded complementary information about the correla textured(Rl(HT)) layers(Fig. 6)resulting in a zig-zag tion between matrix microstructure and mechanical rough fracture surface. In contrast, fracture surfaces of properties of the investigated composites. Thickness, se- low or medium textured carbon layers are much quences and texture of pyrocarbon layers studied by smoother(Fig. 6) PLM, and variations of fracture surfaces visualized by iding between high and low or medium textured SEM clearly reflect the influence of different infiltration carbon layers was observed frequently for all felts result- conditions. The shape of the stress-strain curves corre- ing in pull-out(Fig. 6)of the fibers with the low or med- lates with the morphology of the fracture surfaces show ium textured pyrocarbon layers direct in its vicinity ing a quasi-ductile fracture behavior for almost al forming virtual fibers. Moreover, a very rigorous investigated samples Generally, the fracture behavior of C/C-composites is governed by the mechanical properties of the fibers, ma- trix and fiber/matrix interface. All investigated compos ites were obtained after infiltration of the same porous felt preform. The low fiber content and the short fibers present in such preforms result in a matrix dominated mechanical behavior, which can qualitatively be corre- lated with two controlling microstructural parameters Fig. 8). The fracture toughness correlates with the frac- tion of highly textured pyrocarbon, while the strength correlates with the thickness of the "virtual fiber". This m virtual fiber"was defined as the fiber plus its surround ing low or medium textured pyrocarbon layers, which Fig. 7. SEM micrographs showing the pronounced exfolation within are generally pulled out together during crack propaga- highly textured pyrocarbon compared to medium textured tion(Fig. 6) and which are well-bonded, according to other SEM investigations [21]. Fig 8, indeed, represents
Fig. 6 shows SEM micrographs of the fracture surfaces of the composites after bending tests and the corresponding schemes of their fracture profiles. For all felts, fracture of the fibers and the low or medium textured matrix layers in the fiber vicinity occur approximately in the same fracture plane. The rest of the matrix occupies distinctly another fracture plane level. An intensive fragmentation takes place within the highly textured (RL(HT)) layers (Fig. 6) resulting in a zig–zag rough fracture surface. In contrast, fracture surfaces of low or medium textured carbon layers are much smoother (Fig. 6). Sliding between high and low or medium textured carbon layers was observed frequently for all felts resulting in pull-out (Fig. 6) of the fibers with the low or medium textured pyrocarbon layers direct in its vicinity forming virtual fibers. Moreover, a very rigorous delamination between sublayers within highly textured RL(HT) pyrocarbon can be seen (Fig. 7). 4. Discussion Structural investigations and mechanical testing yielded complementary information about the correlation between matrix microstructure and mechanical properties of the investigated composites. Thickness, sequences and texture of pyrocarbon layers studied by PLM, and variations of fracture surfaces visualized by SEM clearly reflect the influence of different infiltration conditions. The shape of the stress–strain curves correlates with the morphology of the fracture surfaces showing a quasi-ductile fracture behavior for almost all investigated samples. Generally, the fracture behavior of C/C-composites is governed by the mechanical properties of the fibers, matrix and fiber/matrix interface. All investigated composites were obtained after infiltration of the same porous felt preform. The low fiber content and the short fibers present in such preforms result in a matrix dominated mechanical behavior, which can qualitatively be correlated with two controlling microstructural parameters (Fig. 8). The fracture toughness correlates with the fraction of highly textured pyrocarbon, while the strength correlates with the thickness of the ‘‘virtual fiber’’. This ‘‘virtual fiber’’ was defined as the fiber plus its surrounding low or medium textured pyrocarbon layers, which are generally pulled out together during crack propagation (Fig. 6) and which are well-bonded, according to other SEM investigations [21]. Fig. 8, indeed, represents Fig. 6. (a) SEM micrographs of fracture surfaces after three-point bending tests with (b) their corresponding schematic drawings of fracture profiles. Fig. 7. SEM micrographs showing the pronounced exfolation within highly textured pyrocarbon compared to medium textured pyrocarbon. 1958 M. Guellali et al. / Carbon 43 (2005) 1954–1960
M. Guellali et al. Carbon 43(2005)1954-1960 1959 thickness of the 'virtual fiber [um highly textured pyrocarbon content [% Fig. 8. Correlation between: (a) thickness of the 'virtual fiber and flexural strength of the investigated composites and (b) highly textured pyrocarbon content and quasi-ductile fracture behavior very simplified correlations, as other microstructural and the pores leading to crack branching and crack features, like porosity, are not taken into account. How- bridging and thus to a toughness increase [22] ever. the virtual fiber thickness seems to be the dominat Comparison of Fig &a/b shows that there was an in ing parameter with respect to strength. This can be verse relationship between strength and the ductility fac- derived from the strong increase of oo for the 20/ or, FD. 30 kPa couple, where the increase in virtual fiber thick ness is the only significant difference in microstructure the inlet/outlet couple fits well into this trend, although 5Summary one would expect a lower strength level due to its higher The results of these investigations prove clearly the According to our observations the differences in frac distinct influence of the matrix microstructure on the ture behavior are related to the differences in the matrix mechanical properties of CVI-infiltrated carbon fiber microstructure (architecture and texture of individual felts. On one hand, the fiexural strength of a compos carbon layers ite can be enhanced by increasing the thickness of The fractographic investigations show, that a ver low or medium textured pyrocarbon layers in the pronounced fragmentation takes place within the highly vicinity of the fiber. On the other hand, quasi-ductile fracture behavior of a composite can be improved by developed roughness(zig-zag shape)of the fracture sur- the presence of a matrix consisting of highly textured faces(Figs. 6, 7). In addition, an interfacial sliding be- fragmentized pyrocarbon layers. Therefore, through tween pyrocarbon layers with different textures the control of the type and the amount of the depos- commonly observed(Fig. 6). These two mechanisms ited pyrocarbon within the matrix one can produce a lead to an energy dissipation and thus contribute to composite with tailored mechanical properties oughness enhancement. The fracture resistance would Though the highest strength and highest toughness fraction) of highly textured pyrocarbon and with the sample. ch be expected to increase with increasing amount(volume cannot be achieved simultaneously, i.e. in the same number of layers with different textures. Additionally, the porosity respectively the total pyrocarbon volume Acknowledgment fraction should have an influer Nevertheless, the ductility factor can simply be corre- lated with the fraction of highly textured pyrocarbon, as This research was performed in the Collaborative Re- visible from Fig. 8. This fraction of RL(HT) is a normal search Center 551, Project No. D4. The financial sup- ized parameter, which accounts for both, the volume port of the DFG is gratefully acknowledged. The fraction of RL(HT)and the porosity. The Influence of authors thank K.J. Huttinger for the infiltration of the the porosity itself on the toughness of fiber composites samples, and B Reznik and D. Gerthsen for the fruitful with pores of different sizes and shapes (like it is the case discussions here) can hardly be estimated [22]. On the one hand,a porosity increase leads to a decrease of the energy needed to build the new interface and thus to a tough References ness decrease. On the other hand. the increase of the [ Papenburg U. Faserverstarkte Keramische Werkstoffe(CMC). In number of pores with different sizes and shapes leads Kriegesmann J, editor. Technische Keramische Werkstoffe, Chap. to an enhancement of the interaction between the crack ter 4410. Koln: Deutscher Wirtschaftsdienst: 1994
very simplified correlations, as other microstructural features, like porosity, are not taken into account. However, the virtual fiber thickness seems to be the dominating parameter with respect to strength. This can be derived from the strong increase of r0 for the 20/ 30 kPa couple, where the increase in virtual fiber thickness is the only significant difference in microstructure. the inlet/outlet couple fits well into this trend, although one would expect a lower strength level due to its higher porosity. According to our observations, the differences in fracture behavior are related to the differences in the matrix microstructure (architecture and texture of individual carbon layers). The fractographic investigations show, that a very pronounced fragmentation takes place within the highly textured pyrocarbon layer which leads to the highly developed roughness (zig–zag shape) of the fracture surfaces (Figs. 6, 7). In addition, an interfacial sliding between pyrocarbon layers with different textures is commonly observed (Fig. 6). These two mechanisms lead to an energy dissipation and thus contribute to toughness enhancement. The fracture resistance would be expected to increase with increasing amount (volume fraction) of highly textured pyrocarbon and with the number of layers with different textures. Additionally, the porosity respectively the total pyrocarbon volume fraction should have an influence. Nevertheless, the ductility factor can simply be correlated with the fraction of highly textured pyrocarbon, as visible from Fig. 8. This fraction of RL(HT) is a normalized parameter, which accounts for both, the volume fraction of RL(HT) and the porosity. The Influence of the porosity itself on the toughness of fiber composites with pores of different sizes and shapes (like it is the case here) can hardly be estimated [22]. On the one hand, a porosity increase leads to a decrease of the energy needed to build the new interface and thus to a toughness decrease. On the other hand, the increase of the number of pores with different sizes and shapes leads to an enhancement of the interaction between the crack and the pores leading to crack branching and crack bridging and thus to a toughness increase [22]. Comparison of Fig. 8a/b shows that there was an inverse relationship between strength and the ductility factor, FD. 5. Summary The results of these investigations prove clearly the distinct influence of the matrix microstructure on the mechanical properties of CVI-infiltrated carbon fiber felts. On one hand, the flexural strength of a composite can be enhanced by increasing the thickness of low or medium textured pyrocarbon layers in the vicinity of the fiber. On the other hand, quasi-ductile fracture behavior of a composite can be improved by the presence of a matrix consisting of highly textured fragmentized pyrocarbon layers. Therefore, through the control of the type and the amount of the deposited pyrocarbon within the matrix one can produce a composite with tailored mechanical properties. Though the highest strength and highest toughness cannot be achieved simultaneously, i.e. in the same sample. Acknowledgments This research was performed in the Collaborative Research Center 551, Project No. D4. The financial support of the DFG is gratefully acknowledged. The authors thank K.J. Hu¨ttinger for the infiltration of the samples, and B. Reznik and D. Gerthsen for the fruitful discussions. References [1] Papenburg U. Faserversta¨rkte Keramische Werkstoffe (CMC). In: Kriegesmann J, editor. Technische Keramische Werkstoffe, Chapter 4410. Ko¨ln: Deutscher Wirtschaftsdienst; 1994. 14 16 18 20 22 24 26 40 45 50 55 60 65 70 75 σ0 [MPa] thickness of the 'virtual' fiber [µm] 40 50 60 70 80 90 100 0 5 10 15 20 25 FD [%] highly textured pyrocarbon content [%] Filz III 20kPa inlet 30kPa outlet 20kPa inlet 30kPa outlet (a) (b) Fig. 8. Correlation between: (a) thickness of the virtual fiber and flexural strength of the investigated composites and (b) highly textured pyrocarbon content and quasi-ductile fracture behavior. M. Guellali et al. / Carbon 43 (2005) 1954–1960 1959
M. Guellali et al. Carbon 43(2005)1954-1960 2 Bokros JC. Deposition, structure and properties of pyrolytic [12] Bourrat x, Trouvat B G, Vignoles G. Doux F. bon. In: Thrower PA, editor. Chemistry and physics of carbon, electron diffraction and voL 5. New York: Dekker: 1969.p. 1-118 larized light. Journal of Research2000;l5(1):92-l01 3 Guellali M, Oberacker R, Hoffmann MJ. Influence of the [13]Reznik B, Huttinger KJ. On the therminology for pyrolytic filtration conditions on properties and microstructure of CVi rbon. Carbon2002;40:621-4. C/CComposites. Extended abstracts, Carbon 02. Beijing( China), [14] Carlsson LA, Pipes RB. In: Hochleistungsfaserverbundwerkst offe. Stuttgart: Teubner: 1989. p. 74-7 Abscheide. und Heilpressbedingunger [5 Guellali M, Reznik b, Oberacker R, Hoffmann MJ, Gerthsen D und eigenschaften von CvI-CFC-Wer Correlation between mechanical pro and microstructure of des Instituts fuir Keramik im Maschi- CVI-infiltrated carbon fiber felts. Extended abstracts. Carbon o1 niversitat Karlsruhe (TH), PhD thesis American Carbon Society, Lexington(Kentucky, USA), 200 2003 [16 Baron C. Mechanische Eigenschaften Kohlenstofffaserverstarkte []Bokros JC, Price RJ. Deformation and fracture of pyrolytic Kunststoffe (CFK)bei Variation der Matrixduktilitat und der carbons deposited in a fluidized bed. Carbon 1966: 3: 503-19 Bruchdehnung der Faser. Universitat Bremen. PhD thesis, 1990 [6 Granoff B, Pierson HO, Schuster DM. The effect of chemical [17 Beyerlein I, Phoenix L Statistics for the strength and size effects apor deposition conditions on the properties of carbon-carbon of microcomposites with four carbon fibers in epoxy resin. composites. Carbon 1973: 11: 177-87 Composites Science and Technology 1996: 56: 75-92. [ Oh SM, Lee JY. Effects of matrix structure on mechanical [18] Munz D, Fett T. Ceramics: mechanical proper properties of carbon/carbon composites Carbon 1988: 26: 769-76 haviour, materials selection. Berlin: Springer- Verl [8]Oh SM, Lee JY. Fracture behavior of two-dimensional carbon/ 9 DIN 51 110, Teil 3. Prufung von keramischen Ho carbon composites. Carbon 1989: 27: 423-30 9 Kimura S, Yasuda E. Takase N, Kasuya S Fracture behavior of Ermittlung der Weibull-Parameter, 1993 arbon-fiber/CVD carbon composites. High Temperatures-High 20 Benzinger w, Huttinger K. Chemistry and kinetics of chemical Pressures 1981: 13: 193-9 vapor infiltration of pyrocarbon-VI: Mechanical and structural [10] Reznik B, Guellali M, Gerthsen D, Oberacker R, Hoffmann MJ rties of infiltrated carbon fiber felt. Carbon Microstructure and mechanical properties of carbon-carbon 1999;37:1311-22. composites with multilayered pyrocarbon matrix. Materials Let 21 Reznik B, Gerthsen D, Huttinger K. Micro- and nanostructure ters2002:52:14-9 of the carbon matrix of infiltrated carbon fiber felts. Carbon [11] Benzinger W, Huttinger KJ. Chemistry and kinetics of chemical 2001:39:215-29 apor infiltration of pyrocarbon-V: Infiltration of carbon fiber [22] Rice RW. In: Porosity of ceramics. New York: Marcel Dekke felt. Carbon 1999- 37-941-6 1998.p.168-205
[2] Bokros JC. Deposition, structure and properties of pyrolytic carbon. In: Thrower PA, editor. Chemistry and physics of carbon, vol. 5. New York: Dekker; 1969. p. 1–118. [3] Guellali M, Oberacker R, Hoffmann MJ. Influence of the infiltration conditions on properties and microstructure of CVIC/C-Composites. Extended abstracts, Carbon02. Beijing (China), 2002. [4] Guellali M. Einfluss der Abscheide- und Heißpressbedingungen auf die Mikrostruktur und eigenschaften von CVI-CFC-Werkstoffen. Schriftenreihe des Instituts fu¨r Keramik im Maschinenbau, IKM 040. Universita¨t Karlsruhe (TH), PhD thesis, 2003. [5] Bokros JC, Price RJ. Deformation and fracture of pyrolytic carbons deposited in a fluidized bed. Carbon 1966;3:503–19. [6] Granoff B, Pierson HO, Schuster DM. The effect of chemical vapor deposition conditions on the properties of carbon–carbon composites. Carbon 1973;11:177–87. [7] Oh SM, Lee JY. Effects of matrix structure on mechanical properties of carbon/carbon composites. Carbon 1988;26:769–76. [8] Oh SM, Lee JY. Fracture behavior of two-dimensional carbon/ carbon composites. Carbon 1989;27:423–30. [9] Kimura S, Yasuda E, Takase N, Kasuya S. Fracture behavior of carbon-fiber/CVD carbon composites. High Temperatures-High Pressures 1981;13:193–9. [10] Reznik B, Guellali M, Gerthsen D, Oberacker R, Hoffmann MJ. Microstructure and mechanical properties of carbon–carbon composites with multilayered pyrocarbon matrix. Materials Letters 2002;52:14–9. [11] Benzinger W, Hu¨ttinger KJ. Chemistry and kinetics of chemical vapor infiltration of pyrocarbon—V: Infiltration of carbon fiber felt. Carbon 1999;37:941–6. [12] Bourrat X, Trouvat B, Limousin G, Vignoles G, Doux F. Pyrocarbon anisotropy as measured by electron diffraction and polarized light. Journal of Materials Research 2000;15(1):92–101. [13] Reznik B, Hu¨ttinger KJ. On the therminology for pyrolytic carbon. Carbon 2002;40:621–4. [14] Carlsson LA, Pipes RB. In: Hochleistungsfaserverbundwerkstoffe. Stuttgart: Teubner; 1989. p. 74–7. [15] Guellali M, Reznik B, Oberacker R, Hoffmann MJ, Gerthsen D. Correlation between mechanical properties and microstructure of CVI-infiltrated carbon fiber felts. Extended abstracts, Carbon01, American Carbon Society, Lexington (Kentucky, USA), 2001. [16] Baron C. Mechanische Eigenschaften Kohlenstofffaserversta¨rkter Kunststoffe (CFK) bei Variation der Matrixduktilita¨t und der Bruchdehnung der Faser. Universita¨t Bremen, PhD thesis, 1990. [17] Beyerlein IJ, Phoenix L. Statistics for the strength and size effects of microcomposites with four carbon fibers in epoxy resin. Composites Science and Technology 1996;56:75–92. [18] Munz D, Fett T. Ceramics: mechanical properties, failure behaviour, materials selection. Berlin: Springer-Verlag; 1999. [19] DIN 51 110, Teil 3. Pru¨fung von keramischen Hochleistungswerkst offen: 4-Punkt-Biegeversuch, Statistische Auswertung, Ermittlung der Weibull-Parameter, 1993. [20] Benzinger W, Hu¨ttinger KJ. Chemistry and kinetics of chemical vapor infiltration of pyrocarbon—VI: Mechanical and structural properties of infiltrated carbon fiber felt. Carbon 1999;37:1311–22. [21] Reznik B, Gerthsen D, Hu¨ttinger KJ. Micro- and nanostructure of the carbon matrix of infiltrated carbon fiber felts. Carbon 2001;39:215–29. [22] Rice RW. In: Porosity of ceramics. New York: Marcel Dekker; 1998. p. 168–205. 1960 M. Guellali et al. / Carbon 43 (2005) 1954–1960