ournal JAm.Cem.Soc,82l92465-73(1999 Hi-Nicalon/SiC Minicomposites with (Pyrocarbon/SiC)n Nanoscale Multilayered Interphases ebastien Bertrand, Philippe Forio, Rene Pailler, and Jacques Lamon Laboratoire des Composites Thermostructuraux, UMR 5801 CNRS-SEP/SNECMA-UB1, Allee de la Boetie, 33600 Pessac, france Sic/SiC minicomposites that comprise different pyro- Japan)areas iown to be thermodynamically unstable at high The SiC-O Nicalon fibers(Nippon Carbon Co., Tokyo, and a tow of SiC fibers(Hi-Nicalon) have been prepared and SiCnpulsed chemical vapor infiltration. Pyrocarbon very low,8 xygen content (<I wt%)exhibit improved thermal P ere deposited from propane and a CH3SiCl3/H2 stability mixture, respectively. The microstructure of the inter The present paper investigates SiC/SiC minicomposites that phases has been investigated using transmission electron have been processed via P-CVI and contain nanoscale(PyC/ microscopy. The mechanical tensile behavior of the mi SiC) multilayered interphases and Hi-Nicalon fibers, Data are mposites at room t mperature exhibits the classical fea- provided on the microstructure and properties of constituents, ures of tough composites, regardless of the characteristics the features of the tensile stress-strain behavior, and matrix of the(Py C/SiC)sequences. The interfacial shear stress has cracking been determined from the width of hysteresis loops upon unloading/reloading and from the crack-spacing distance at II. Experimental Procedure saturation. All the experimental data indicate that the strength of the fiber/interphase interfaces is rather weak SiCSiC Mimicomposites Processing (50 MPa). (A) P-CVI Apparatus: In the P-CVI process, the source gases, which react to produce the matrix and the interphases L Introduction are introduced during short pulses. Each pulse involves three steps:(i)the introduction of gases, (ii)a hold-on period( during HE mechanical behavior of ceramic-matrix composites which the gases diffuse inside the fiber preform and react), and (CMCs)with continuous fibers is dependent not only on (ii) the evacuation of the gaseous reaction products. A layer of the intrinsic properties of the fiber and the matrix but also on material is deposited during each pulse. The p-CVI technique 1) ng - To control th omits the deposition of thin films the F/M bonding, an additional phase is deposited on the fiber; The P-CVI apparatus is shown in Fig. 1. The reactor is a this phase is called the interphase. The most commonly used silica glass tube(internal diameter of 25 mm, height of 50 mm) interphase material is pyrocarbon( PyC). Pyrocarbon displays a heated uniformly (5 K) in an electrical furnace (2"in the layered microtexture and has been shown to lead to high- figure). The temperature is measure trength-high-toughness SiC/SiC composites. 4 However, PyC Mass and ball flowmeters control the flow rate of the source is not stable in an oxidizing environment. Consequently, coat- gases( propane, MTS(methyltrichlorosilane, CH3 SiCl3), and ing the Py C layer with a SiC layer has been suggested as a way hydrogen(the carrier gas when using MTS).(See"3" in the to protect the pyrolytic carbon against oxidation; this procedure igure. ) First, MTS is evaporated in a heated chamber(4 " in leads to the concept of multilayered(Py C/SiC) interphases. 5. the figure), and then it is mixed with hydrogen in a pre CMCs are commonly fabricated via isothermal-isobaric regulated heated tank (5"in the figure), where it is hemical vapor infiltration(I-CVI). 7 To reduce the fabrication before admission into the reactor. During a pulse, the time, various CVI processes have been developed, -ll such as forced CVI(F-CVI), which involves steep temperature and through inlet pneumatic valves. After the hold-on time(the pressure gradients, or pressure-pulsed CVI(P-CVI residence time, t), which allows deposition of a certain thick- use of ness of material, the gases are evacuated from the reactor to the of each sublayer being up to 100 nm)in two-dimensional (2D) vacuum pump("7"in the figure)through an outlet pneumatic oven SiC/SiC composites that have been prepared according valve and liquid nitrogen trap(s)(6" in the figure).A pro- the I-CVI route has been reported by Droillard and co- grammable logic controller regulates the open and workers. 4,3Then, Heurtevent developed SiC/SIC micro- he valves and dictates the number of pulses composites with nanoscale(PyC/SiC), multilayered inter (B) Processing Conditions and Materials: The Hi hases( the thicknesses of the sublayers being as small as 3 Nicalon tows consist of 500 single filaments. each with an that were deposited via pulsed chemical vapor deposition(P- average diameter of 13.5+ 1.5 um. Tows with a length of 50 CVD). Other than these pioneering works, to our knowledge, mm were mounted on SiC frames(six tows per frame)to no detailed studies on nanoscale(PyC/SiC), multilayered in- infiltrate the PyC and Sic sublayers and the SiC matrix. The terphases have been reported in the literature tows were slightly twisted at a constant angle(one turn per 5 cm), to decrease the porosity in the minicomposites and to increase their fiber volume fraction, The Pyc sublayers were deposited at a temperature(r)of 1223 K, under a pressure(P of 3 kPa, with t,=5 s. 4 The SiC sublayers were deposited at T= 1223 K and P=5 kPa, with a y4 and t, =l s(a= OH/OMTS, where OH, and OMTs are the respective gas flows of hydrogen and MT The infiltration conditions for the sic script No 19031. Received March 16, 1998: approved February 13, 1999. matrix were as follows: T= 1223 K, P=3 kPa, a= 6, and 2465
Hi-Nicalon/SiC Minicomposites with (Pyrocarbon/SiC)n Nanoscale Multilayered Interphases Se´bastien Bertrand, Philippe Forio, Rene´ Pailler,* and Jacques Lamon* Laboratoire des Composites Thermostructuraux, UMR 5801 CNRS–SEP/SNECMA-UB1, Allee de la Boetie, 33600 Pessac, France SiC/SiC minicomposites that comprise different pyrocarbon/silicon carbide ((PyC/SiC)n) multilayered interphases and a tow of SiC fibers (Hi-Nicalon) have been prepared via pressure-pulsed chemical vapor infiltration. Pyrocarbon and SiC were deposited from propane and a CH3SiCl3/H2 mixture, respectively. The microstructure of the interphases has been investigated using transmission electron microscopy. The mechanical tensile behavior of the minicomposites at room temperature exhibits the classical features of tough composites, regardless of the characteristics of the (PyC/SiC) sequences. The interfacial shear stress has been determined from the width of hysteresis loops upon unloading/reloading and from the crack-spacing distance at saturation. All the experimental data indicate that the strength of the fiber/interphase interfaces is rather weak (∼50 MPa). I. Introduction THE mechanical behavior of ceramic-matrix composites (CMCs) with continuous fibers is dependent not only on the intrinsic properties of the fiber and the matrix but also on the fiber/matrix (F/M) bonding.1–3 To control the strength of the F/M bonding, an additional phase is deposited on the fiber; this phase is called the interphase. The most commonly used interphase material is pyrocarbon (PyC). Pyrocarbon displays a layered microtexture and has been shown to lead to highstrength–high-toughness SiC/SiC composites.4 However, PyC is not stable in an oxidizing environment. Consequently, coating the PyC layer with a SiC layer has been suggested as a way to protect the pyrolytic carbon against oxidation; this procedure leads to the concept of multilayered (PyC/SiC)n interphases.5,6 CMCs are commonly fabricated via isothermal–isobaric chemical vapor infiltration (I-CVI).7 To reduce the fabrication time, various CVI processes have been developed,8–11 such as forced CVI (F-CVI),9 which involves steep temperature and pressure gradients, or pressure-pulsed CVI (P-CVI).11 The use of multilayered PyC/SiC interphases (the thickness of each sublayer being up to 100 nm) in two-dimensional (2D) woven SiC/SiC composites that have been prepared according to the I-CVI route has been reported by Droillard and coworkers.4,12,13 Then, Heurtevent14 developed SiC/SiC microcomposites with nanoscale (PyC/SiC)n multilayered interphases (the thicknesses of the sublayers being as small as 3 nm) that were deposited via pulsed chemical vapor deposition (PCVD). Other than these pioneering works, to our knowledge, no detailed studies on nanoscale (PyC/SiC)n multilayered interphases have been reported in the literature. The Si–C–O Nicalon fibers (Nippon Carbon Co., Tokyo, Japan) are known to be thermodynamically unstable at high temperatures.15–17 Conversely, SiC Hi-Nicalon fibers with a very low oxygen content (<1 wt%) exhibit improved thermal stability.18 The present paper investigates SiC/SiC minicomposites that have been processed via P-CVI and contain nanoscale (PyC/ SiC)n multilayered interphases and Hi-Nicalon fibers. Data are provided on the microstructure and properties of constituents, the features of the tensile stress–strain behavior, and matrix cracking. II. Experimental Procedure (1) SiC/SiC Minicomposites Processing (A) P-CVI Apparatus: In the P-CVI process, the source gases, which react to produce the matrix and the interphases, are introduced during short pulses. Each pulse involves three steps: (i) the introduction of gases, (ii) a hold-on period (during which the gases diffuse inside the fiber preform and react), and (iii) the evacuation of the gaseous reaction products. A layer of material is deposited during each pulse. The P-CVI technique permits the deposition of thin films. The P-CVI apparatus is shown in Fig. 1. The reactor is a silica glass tube (internal diameter of 25 mm, height of 50 mm) heated uniformly (±5 K) in an electrical furnace (“2” in the figure). The temperature is measured with a thermocouple. Mass and ball flowmeters control the flow rate of the source gases (propane, MTS (methyltrichlorosilane, CH3SiCl3), and hydrogen (the carrier gas when using MTS)). (See “3” in the figure.) First, MTS is evaporated in a heated chamber (“4” in the figure), and then it is mixed with hydrogen in a pressureregulated heated tank (“5” in the figure), where it is stored before admission into the reactor. During a pulse, the source gases are introduced almost instantaneously into the reactor through inlet pneumatic valves. After the hold-on time (the residence time, tr ), which allows deposition of a certain thickness of material, the gases are evacuated from the reactor to the vacuum pump (“7” in the figure) through an outlet pneumatic valve and liquid nitrogen trap(s) (“6” in the figure). A programmable logic controller regulates the open and closing of the valves and dictates the number of pulses. (B) Processing Conditions and Materials: The HiNicalon tows consist of 500 single filaments, each with an average diameter of 13.5 ± 1.5 mm. Tows with a length of 50 mm were mounted on SiC frames (six tows per frame) to infiltrate the PyC and SiC sublayers and the SiC matrix. The tows were slightly twisted at a constant angle (one turn per 5 cm), to decrease the porosity in the minicomposites and to increase their fiber volume fraction. The PyC sublayers were deposited at a temperature (T) of 1223 K, under a pressure (P) of 3 kPa, with tr 4 5 s.14 The SiC sublayers were deposited at T 4 1223 K and P 4 5 kPa, with a 4 1⁄4 and tr 4 1s(a 4 QH2 /QMTS, where QH2 and QMTS are the respective gas flows of hydrogen and MTS).14 The infiltration conditions for the SiC matrix were as follows: T 4 1223 K, P 4 3 kPa, a 4 6, and tr 4 2 s.14 T. A. Parthasarathy—contributing editor Manuscript No. 190319. Received March 16, 1998; approved February 13, 1999. Supported by SEP and CNRS, through a grant given to author SB. *Member, American Ceramic Society. J. Am. Ceram. Soc., 82 [9] 2465–73 (1999) Journal 2465
Journal of the American Ceramic Society-Bertrand et al. Vol. 82. No 9 dinal Young s modulus(EL). The Poisson's ratio(v) has been assumed to be equal to that estimated for the Nicalon fiber. 9 The shear modulus (G)has been estimated from E and v. The statistical parameters have been derived from the distribution of strength data, using the maximum-likelihood estimator. 20 coho The longitudinal coefficient of thermal expansion(a) has been determined from the comparison of the expansion of a single filament to that of a reference sintered SiC tube at tem- peratures of 293-1373 K. The filament and the tube were mounted in parallel in a furnace that was located in the cham- ber of a scanning electron microscopy(SEM) system(Mod 840, JEOL, Tokyo, Japan). 21 The fiber has an isotropic micro- structure, as already mentioned; thus, the transverse coefficient of thermal expansion(ar) is assumed to be equal to or. B cause the microstructures of the sic Hi-Nicalon fiber and the SiC interphase sublayer are similar 4, 18(they both consist of M SiC nanocrystallites and free carbon), the properties of the SiC interphase are assumed to be equal to those of Sic Hi-nicalon fiber. The statistical parameters that are pertinent to the p-Cv o SiC matrix have been derived from the statistical distributions of the matrix fragmentation stresses that were measured on minicomposites, using a tensile device in the chamber of the SEM equipment pq· Shut-off valve (2) Microstructural Characterization Mass flowmeter The chemical characterization of the sic hi-Nicalon fibers has been reported elsewhere. 8 The surface analysis of the Hi- Nicalon fibers has been performed using Auger electron spe Fig. 1. Schematic diagram showing the P-CV troscopy(AES)(Model 310F, VG Microscopes, West Sussex process the SiC brous preform; nace;“3," source gase , fi: vacuum at 1223 k in the reaction chamber. to reproduce the heated tank,“6,” conditions that were experienced by the fiber prior to inter ase deposition. The microstructure of PyC and SiC in the interphase has been assessed using(1) optical microscopy (Model MeF3, Reichert-Jung) in polarized light, on polished The(PyC/SiC)n multilayered interphases exhibited the fol- cross sections of minicomposites, to measure the extinction wing features: (i)the PyC and SiC sublayers had a uniform angle(Ae, which characterizes the PyC anisotropy (only per- hickness; (ii) the first sublayer that was deposited on the fiber formed on minicomposites with large interphase size of always consisted of PyC; (iii)the number of(Py C/SiC)bilayers C)),3(ii)X-ray diffractometry(XRD)(Model D5000, Sie- (n) was varied in the range of 10-30;(iv) the thickness of mens, Karlsruhe, Germany) to determine the apparent mean the PyC and SiC sublayers(respectively noted e(Pvc) and and grain size(noted as Lu) of the b-sic crystalline phase, and e(sic) was 3-20 nm and 10-50 nm, respectively; and (v) the (iii) SEM(Model S-4500, Hitachi, Tokyo, Japan) and trans- interphase thickness was 100-1800 nm. For comparison pur mission electronic microscopy(TEM)(Model CM30, Philips poses, a batch of reference minicomposites with a single Research Laboratories, Eindhoven, The Netherlands). For op nte (100 nm thick)also was prepared (Table I). The a microscopy studies, the minicomposite cross sections ber volume fraction in the minicomposites was%(+5% were polished using standard metallographic techniques. For The properties of the minicomposites constituents are listed the SEM observations, certain minicomposites were etched us- n Table Il. The mechanical properties of the Hi-Nicalon fibers Murakami's reagent, to get a selective attack of the SiC on single filaments(gauge lengths of 10, 25, and 65 mm). The palladium or carbon. The TEM analyses have been performed fiber has an isotropic microstructure; therefore, the transverse on cross sections of the minicomposites. The preparation of the Youngs modulus(Er) is assumed to be equal to the longitu- thin foils is detailed elsewhere Table L. Batches and Mechanical Properties of the SiC/(PyC/SiC)/SiC Minicomposites Mechanical properties nterfacial shear stress, T(MPa Fq(4) (10001100 41342710.091710.6911618 95 (3/10)1 810.17 1684 38 1041345820.131810.9 7535 10101040349730.121660.8157 343870.161530. 26710 37 560.111680 2875011036 0501039351880.121950.832085017535 (1050)30 PyC sut he stress 温邮你叫
The (PyC/SiC)n multilayered interphases exhibited the following features: (i) the PyC and SiC sublayers had a uniform thickness; (ii) the first sublayer that was deposited on the fiber always consisted of PyC; (iii) the number of (PyC/SiC) bilayers (n) was varied in the range of 10–30; (iv) the thickness of the PyC and SiC sublayers (respectively noted e (P yC) and e(SiC)) was 3–20 nm and 10–50 nm, respectively; and (v) the interphase thickness was 100–1800 nm. For comparison purposes, a batch of reference minicomposites with a single PyC interphase (100 nm thick) also was prepared (Table I). The fiber volume fraction in the minicomposites was ∼40% (±5%). The properties of the minicomposites constituents are listed in Table II. The mechanical properties of the Hi-Nicalon fibers have been measured using tensile tests, at ambient temperature, on single filaments (gauge lengths of 10, 25, and 65 mm). The fiber has an isotropic microstructure;18 therefore, the transverse Young’s modulus (ET) is assumed to be equal to the longitudinal Young’s modulus (EL). The Poisson’s ratio (n) has been assumed to be equal to that estimated for the Nicalon fiber.19 The shear modulus (G) has been estimated from E and n. The statistical parameters have been derived from the distribution of strength data, using the maximum-likelihood estimator.20 The longitudinal coefficient of thermal expansion (aL) has been determined from the comparison of the expansion of a single filament to that of a reference sintered SiC tube at temperatures of 293–1373 K. The filament and the tube were mounted in parallel in a furnace that was located in the chamber of a scanning electron microscopy (SEM) system (Model 840, JEOL, Tokyo, Japan).21 The fiber has an isotropic microstructure, as already mentioned; thus, the transverse coefficient of thermal expansion (aT) is assumed to be equal to aL. Because the microstructures of the SiC Hi-Nicalon fiber and the SiC interphase sublayer are similar14,18 (they both consist of SiC nanocrystallites and free carbon), the properties of the SiC interphase are assumed to be equal to those of SiC Hi-Nicalon fiber. The statistical parameters that are pertinent to the P-CVI SiC matrix have been derived from the statistical distributions of the matrix fragmentation stresses that were measured on minicomposites, using a tensile device in the chamber of the SEM equipment.22 (2) Microstructural Characterization The chemical characterization of the SiC Hi-Nicalon fibers has been reported elsewhere.18 The surface analysis of the HiNicalon fibers has been performed using Auger electron spectroscopy (AES) (Model 310F, VG Microscopes, West Sussex, U.K.). The as-received Hi-Nicalon fibers were heated under vacuum at 1223 K in the reaction chamber, to reproduce the conditions that were experienced by the fiber prior to interphase deposition. The microstructure of PyC and SiC in the interphase has been assessed using (i) optical microscopy (Model MeF3, Reichert–Jung) in polarized light, on polished cross sections of minicomposites, to measure the extinction angle (Ae, which characterizes the PyC anisotropy (only performed on minicomposites with large interphase size of PyC)),23 (ii) X-ray diffractometry (XRD) (Model D5000, Siemens, Karlsruhe, Germany) to determine the apparent mean grain size (noted as L111) of the b-SiC crystalline phase, and (iii) SEM (Model S-4500, Hitachi, Tokyo, Japan) and transmission electronic microscopy (TEM) (Model CM30, Philips Research Laboratories, Eindhoven, The Netherlands). For optical microscopy studies, the minicomposite cross sections were polished using standard metallographic techniques. For the SEM observations, certain minicomposites were etched using Murakami’s reagent, to get a selective attack of the SiC layers, and then they were coated with a thin layer of gold– palladium or carbon. The TEM analyses have been performed on cross sections of the minicomposites. The preparation of the thin foils is detailed elsewhere.24 Fig. 1. Schematic diagram showing the P-CVI apparatus used to process the SiC/SiC minicomposites. Legend is as follows: “1,” fibrous preform; “2,” furnace; “3,” source gases; “4,” drying oven; “5,” heated tank; “6,” liquid nitrogen trap; and “7,” vacuum pump. Table I. Batches and Mechanical Properties of the SiC/(PyC/SiC)n/SiC Minicomposites Batch Interphase parameters† Mechanical properties‡ Interfacial shear stress, t (MPa) e(PyC) (nm) e(SiC) (nm) n Vf (%) Ec (GPa) Fe (N) ee (%) FR (N) eR (%) DlRT (mm) sS (MPa) lS (mm) Eq. (1) Eq. (3) Eq. (4) (100/0)1 § 100 0 1 41 342 71 0.09 171 0.69 11 618 95 50 34 29 (3/10)10 3 10 10 52 329 81 0.17 164 0.69 16 840 90 40 38 32 (3/30)10 3 30 10 41 355 95 0.14 158 0.64 17 678 107 55 39 35 (3/50)10 3 50 10 41 345 82 0.13 181 0.92 24 615 75 35 49 44 (10/10)10 10 10 10 40 349 73 0.12 166 0.81 57 600 185 32 19 17 (10/50)10 10 50 10 45 343 87 0.16 153 0.72 26 710 98 37 34 29 (20/10)10 20 10 10 44 336 56 0.11 168 0.92 28 750 110 36 32 28 (20/30)10 20 30 10 40 350 69 0.12 173 0.77 22 95 (20/50)10 20 50 10 39 351 88 0.12 195 0.83 20 850 175 35 25 22 (10/50)30 10 50 30 41 346 76 0.14 165 0.77 22 817 125 70 44 40 † Interphase parameters include the thickness of each PyC sublayer (e(PyC)), the thickness of each SiC sublayer (e(SiC)), and the number of bilayers (n). ‡ Mechanical properties include the fiber volume fraction (Vf ), Young’s modulus (Ec), the force and strain at the proportional limit (Fe and ee, respectively) and at failure (FR and eR, respectively), the maximum permanent elongation at zero load (DlRT), the stress at saturation (sS), and the matrix spacing distance (lS). § Reference interphase. 2466 Journal of the American Ceramic Society—Bertrand et al. Vol. 82, No. 9
September 1999 Hi-Nicalon/SiC Minicomposites with(Pyrocarbon/SiC)n Nanoscale Multilayered Interphases Table Il. Properties of the Minicomposite Constituents oisson's ratio in Coefficient of thermal expansion, (GPa) plane axial conditions TE(X 10/C) Constitu o°(MPa) Ii-Nicalon sic fiber 280 464.71 4.6-4.7 PyC interphase 120.2 280 0.12 P-CVI SIC 400 400170170 0.20 5.7 scripts"T"and"L "denote transverse and longitudinal conditions, respectively. ' Data for the PyC interphase are taken from Bobet and Lamon, whereas data for the SiC 3) Tensile Tests on Minicomposites Uniaxial tension tests were performed at room temperature at a constant strain rate(50 um/min). The load was measured -SiC matrix using a 500 N load cell. The minicomposite elongation was measured using two parallel linear-variable differential trans- former(LVDT) extensometers that were mounted on the grips The minicomposites ends were glued within metallic tubes that were then gripped into the testing machine. The gauge length was 20 mm. The system compliance(Cs) was determined or Fiber- Interphase dry fiber tows with decreasing gauge lengths(Cs =0.3 um/N) Unloading-reloading cycles were conducted on a few speci mens of each batch to evaluate the residual strains and the F/m bonding. After ultimate failure, the test specimens were exam- d using SEM, to measure the matrix-crack-spacing distance II. Results ( Material Characterization Figure 2 shows a general view of the failure surface of a minicomposite,some fiber pullout and residual porosity are visible. Figure 3 shows a multilayered(PyC/SiC)n, interphase at different scales. The microstructure of the SiC-based sublayers quasi-amorphous and nanocrystallized (Lu =15 nm). In such a microstructure. the number and size of microstructural Matrix defects are reduced, in comparison to those observed in the SiC-based sublayers made via I-CVI. 3 The PyC exhibits a rough, laminar microstructure(Ae= 18.5). Because of the strong microstructural anisotrophy that is indicated by the value of the extinction angle, the PyC sublayers are expected to cause deviation of the matrix microcracks, as observed by Droillard and co-workers. 4, 12 Both the PyC and SiC sublayers are generally parallel to the fibers axis(Figs. 3 and 4)and Fig. 2. SEM micrographs of the fracture surface of SiC(Py C/SiC),/ continuous(Fig 3). Their thicknesses e(sic) and e(pyc) are con- SiC minicomposites, showing the constituents, fiber pullout, and stant(Fig 3) debonding between the fiber and the first PyC interfacial sublayer AES depth profiles performed on Hi-Nicalon fiber show the presence of an oxygen-enriched layer(15 nm thick)at the fiber surface. This Si-C-O phase consists of SiO2 and free carbon It is thought to be related to the heating of the tows, under The scatter of the stress-strain curves falls within the range vaccum, before the deposition of the interphase and the matrix hat is usually observed with CMCs. Features of the force- In the minicomposites, the first interfacial Py C sublayer that is elongation curves(Table D)do not indicate a significant differ deposited is bonded to this Si-C-O layer ence between the batches. Figure 5 shows that the presence of a multilayered interphase does not affect the tensile behavior, (2) Tensile Behavior of Minicomposites in comparison with the reference minicomposites with a single (A) Force-Deformation Curves: The force-deformation Py C layer as the interphase curves of the minicomposites(Fig. 5)are markedly nonlinear Figure 6 shows typical hysteresis loops that and indicate nonbrittle mechanical behavior. All the curves when unloading-reloading the minicomposites. The elastic display the following typical features modulus that is pertinent to the cracked minicomposite is giver (1) An initially linear region, reflecting the elastic defor- by the slope of the linear portion of the reloading curve(mini mation of the minicomposites. The proportional limit is ob- mum tangent modulus). The tangent to this linear portion gen- served to be -0.1% deformation erally intercepts the origin. The nt strain at zero lo (2) A nonlinear domain of deformations, resulting from includes contributions from misfit relief e* and sliding Es. The multiple matrix cracking wide hysteresis loops, and the large permanent elongations at (3) After saturation of matrix ng. a second linear zero load(10-20 um, as shown in Table I), region that is attributed to the elas gation of the fibers ence of rather weak F/M interactions. 12 suggest the I 4) A second, slightly nonlinea prior to maximum Figure 7 shows the typical elastic moduli that have beer load. which is attributed to individual fiber break measured during the tensile tests. For most minicomposites, the (5) Finally, the ultimate failure modulus decreases to a minimum value that coincides with the
(3) Tensile Tests on Minicomposites Uniaxial tension tests were performed at room temperature at a constant strain rate (50 mm/min). The load was measured using a 500 N load cell. The minicomposite elongation was measured using two parallel linear–variable differential transformer (LVDT) extensometers that were mounted on the grips. The minicomposites ends were glued within metallic tubes that were then gripped into the testing machine. The gauge length was 20 mm. The system compliance (CS) was determined on dry fiber tows with decreasing gauge lengths (CS 4 0.3 mm/N). Unloading–reloading cycles were conducted on a few specimens of each batch, to evaluate the residual strains and the F/M bonding. After ultimate failure, the test specimens were examined using SEM, to measure the matrix-crack-spacing distance. III. Results (1) Material Characterization Figure 2 shows a general view of the failure surface of a minicomposite; some fiber pullout and residual porosity are visible. Figure 3 shows a multilayered (PyC/SiC)n interphase at different scales. The microstructure of the SiC-based sublayers is quasi-amorphous and nanocrystallized (L111 4 15 nm).14 In such a microstructure, the number and size of microstructural defects are reduced, in comparison to those observed in the SiC-based sublayers made via I-CVI.13 The PyC exhibits a rough, laminar microstructure (Ae 4 18.5°). Because of the strong microstructural anisotrophy that is indicated by the value of the extinction angle, the PyC sublayers are expected to cause deviation of the matrix microcracks, as observed by Droillard and co-workers.4,12 Both the PyC and SiC sublayers are generally parallel to the fibers axis (Figs. 3 and 4) and continuous (Fig. 3). Their thicknesses e(SiC) and e(PyC) are constant (Fig. 3). AES depth profiles performed on Hi-Nicalon fiber show the presence of an oxygen-enriched layer (15 nm thick) at the fiber surface. This Si–C–O phase consists of SiO2 and free carbon. It is thought to be related to the heating of the tows, under vaccum, before the deposition of the interphase and the matrix. In the minicomposites, the first interfacial PyC sublayer that is deposited is bonded to this Si–C–O layer. (2) Tensile Behavior of Minicomposites (A) Force–Deformation Curves: The force–deformation curves of the minicomposites (Fig. 5) are markedly nonlinear and indicate nonbrittle mechanical behavior. All the curves display the following typical features: (1) An initially linear region, reflecting the elastic deformation of the minicomposites. The proportional limit is observed to be ∼0.1% deformation. (2) A nonlinear domain of deformations, resulting from multiple matrix cracking. (3) After saturation of matrix cracking, a second linear region that is attributed to the elastic elongation of the fibers. (4) A second, slightly nonlinear domain prior to maximum load, which is attributed to individual fiber breaks. (5) Finally, the ultimate failure. The scatter of the stress–strain curves falls within the range that is usually observed with CMCs. Features of the force– elongation curves (Table I) do not indicate a significant difference between the batches. Figure 5 shows that the presence of a multilayered interphase does not affect the tensile behavior, in comparison with the reference minicomposites with a single PyC layer as the interphase. Figure 6 shows typical hysteresis loops that are obtained when unloading–reloading the minicomposites. The elastic modulus that is pertinent to the cracked minicomposite is given by the slope of the linear portion of the reloading curve (minimum tangent modulus). The tangent to this linear portion generally intercepts the origin. The permanent strain at zero load includes contributions from misfit relief e* and sliding e0 S. The wide hysteresis loops, and the large permanent elongations at zero load (10–20 mm, as shown in Table I), suggest the presence of rather weak F/M interactions.12 Figure 7 shows the typical elastic moduli that have been measured during the tensile tests. For most minicomposites, the modulus decreases to a minimum value that coincides with the Table II. Properties of the Minicomposite Constituents† Constituent‡ Young’s modulus (GPa) Shear modulus (GPa) Poisson’s ratio in plane axial conditions Coefficient of thermal expansion, CTE (× 10−6/°C) Statistical parameters EL ET GL GT n12 n13 aT aT m s0 § (MPa) Hi-Nicalon SiC fiber 280 280 125 125 0.12 0.12 4.6–4.7¶ 4.6–4.7 4.2 6 PyC interphase 80 12 36 5 0.12 0.20 2 20–28 SiC interphase 280 280 125 125 0.12 0.12 3.9 3.9 P-CVI SiC 400 400 170 170 0.20 0.20 4.6 4.6 5.5 5.7 † Subscripts “T” and “L” denote transverse and longitudinal conditions, respectively. ‡ Data for the PyC interphase are taken from Bobet and Lamon,19 whereas data for the SiC interphase and P-CVI SiC are taken from Heurtevent14 (however, the statistical parameters for P-CVI SiC have been taken from Fredefon et al.22). § For a reference volume (V0) of 1 m3 . ¶ For a temperature range of 523–1073 K. From Cabot et al.21 Fig. 2. SEM micrographs of the fracture surface of SiC/(PyC/SiC)n / SiC minicomposites, showing the constituents, fiber pullout, and debonding between the fiber and the first PyC interfacial sublayer. September 1999 Hi-Nicalon/SiC Minicomposites with (Pyrocarbon/SiC)n Nanoscale Multilayered Interphases 2467
2468 Joumal of the American Ceramic Sociery-Bertrand et al. Vol. 82. No 9 100nm Fibe Interphase Matrix Fibe C3 Sicl 1 Fiber Fig. 3. TEM micrographs of a multilayered interphase(batch(3/30)o)(a)cross section of the minicomposite and( b) interfacial zone):a high-resolution TEM micrograph of the interfacial bond between the fiber and the interphase is shown in Fig. 3(c) quantity ErV(Er is the fiber Young's modulus, and Vr is the scribed by Droillard and Lamon. They are much larger than fiber volume fraction), which indicates that the applied load is that observed in the 2D woven Nincalon/SiC composites with borne only by the fibers. Therefore, the debonding of fibers multilayered interphases and strong fiber/interphase bonding complete at this stage. The minimum is reached at deforma-(-20 um) tions of-0.4%, which correspond to the deformations at matrix The longitudinal cracks were not detected before the tests cracking saturation that are indicated by the force-deformation They cannot be attributed to thermally induced stresses that are curves(Fig. 5). For certain minicomposites, the minimum generated while cooling from the processing temperature. Fur- seems to be smaller than Er Ve This discrepancy may be at thermore. the fiber and the matrix exhibit identical coefficients tributed to the presence of broken or bent fibers and/or uncer- of thermal expansion(CTEs)(Table II). As discussed in a later tainty in the data, including the modulus measurements, Ep section, their presence may be related to the initial twisting of and I the fiber tows (B) SEM Analysis: The matrix cracks that are detected at The pull-out lengths(100 um, Fig. 2)are similar to those the surface of the minicomposites after ultimate failure includ neasured in Nicalon/SiC minicomposites' and 2D woven transverse cracks as well as a few longitudinal cracks. The Nicalon/SiC composites with multilayered interphases and average spacing distance at saturation for the transverse cracks weak fiber/interphase bonding. Figure 8 shows the double (s)is 90-185 um(Table I). These values are similar to those deflection of a matrix crack, first in the interphase/matrix in- measured on the 2D woven Nicalon/SiC composites with mul- terface and then in the fiber/interphase interface. Figure 2 tilayered interphases and weak fiber/interphase bonding, de shows that the surface of the pulled-out fibers is smooth, which
quantity Ef ?Vf (Ef is the fiber Young’s modulus, and Vf is the fiber volume fraction), which indicates that the applied load is borne only by the fibers. Therefore, the debonding of fibers is complete at this stage. The minimum is reached at deformations of ∼0.4%, which correspond to the deformations at matrix cracking saturation that are indicated by the force–deformation curves (Fig. 5). For certain minicomposites, the minimum seems to be smaller than Ef ?Vf . This discrepancy may be attributed to the presence of broken or bent fibers and/or uncertainty in the data, including the modulus measurements, Ef , and Vf . (B) SEM Analysis: The matrix cracks that are detected at the surface of the minicomposites after ultimate failure include transverse cracks as well as a few longitudinal cracks. The average spacing distance at saturation for the transverse cracks (lS) is 90–185 mm (Table I). These values are similar to those measured on the 2D woven Nicalon/SiC composites with multilayered interphases and weak fiber/interphase bonding, described by Droillard and Lamon.4 They are much larger than that observed in the 2D woven Nincalon/SiC composites with multilayered interphases and strong fiber/interphase bonding4 (∼20 mm). The longitudinal cracks were not detected before the tests. They cannot be attributed to thermally induced stresses that are generated while cooling from the processing temperature. Furthermore, the fiber and the matrix exhibit identical coefficients of thermal expansion (CTEs) (Table II). As discussed in a later section, their presence may be related to the initial twisting of the fiber tows. The pull-out lengths (∼100 mm, Fig. 2) are similar to those measured in Nicalon/SiC minicomposites3 and 2D woven Nicalon/SiC composites with multilayered interphases and weak fiber/interphase bonding.4 Figure 8 shows the double deflection of a matrix crack, first in the interphase/matrix interface and then in the fiber/interphase interface. Figure 2 shows that the surface of the pulled-out fibers is smooth, which Fig. 3. TEM micrographs of a multilayered interphase (batch (3/30)10) ((a) cross section of the minicomposite and (b) interfacial zone); a high-resolution TEM micrograph of the interfacial bond between the fiber and the interphase is shown in Fig. 3(c). 2468 Journal of the American Ceramic Society—Bertrand et al. Vol. 82, No. 9
September 1999 Hi-Nicalon/SiC Minicomposites with(Pyrocarbon/SiC)n Nanoscale Multilayered Interphases 2469 41 gardless of the method and batch. The T values obtained from the crack-spacing distance at saturation are 17-49 MP whereas those determined from the hyst are 32-70 MPa. Therefore, because of the scatter usually observed Interphase oer with T measurements, the T values do not indicate any signifi- cant effect of multilayering the interphase. The T values are comparable to those measured on the SiC/SiC minicomposites with a single PyC layer(batch(100/0)1). They also are com- parable to those measured on SiC/SiC minicomposites tha have been reinforced with as-received Nicalon fiber 3 this agreement may be attributed to the presence of the SiC-O layer of at the surface of the Hi-Nicalon and Nicalon fibers which has been demonstrated to dictate the devia matrix cracks Fiber Table I indicates that the T values derived from the hysteresis oop widths are generally larger than those obtained from the matrix-crack ng distance. As discussed in a following sec- tion, this discrepancy may be attributed to an effect of the initial twisting of the tows which enhances F/M interactions pared via P-CVI(n= h of a nanoscale multilayered interphase pre- Fig. 4. SEM micrograp V. Discussion interphase. Recall the previously mentioned feature, that the n It has been emphasized that the tensile mechanical behavior confirms debonding in the interface between the fiber and the the minicomposites was not affected by the presence of region between the fiber and the interphase consists mainly of multilayered interphases that involved stiff sublayers of Sic Similar features have been identified, regardless of the inter- been identified in the composites that have been reinforced phases, including proportional limit, saturation stress, strain to ith as-received Nicalon fibers. In these composites, the F/M nteractions are weak. and deflection of the matrix cracks oc- Slightly different results have been obtained for Hi-N curs at the fiber/interphase interfaces. SiC microcomposites with(PyC/SiC)n nanoscale mult interphases. 4 That author found that the stress-strain 3) Interfacial Shear Stress of the microcomposites was dependent on the respective thick- The interfacial shear stress (T)was estimated from the fol- nesses of the sublayers. The best properties were observed for lowing data he following sublayer conditions: e(Byc)=3 nm, e(sic)= 30 d during nm, and n= 10. The corresponding(3/30)10 minicomposites nloading-reloading cycles. T is given by the following equa- exhibited an interfacial shear stress that seemed to be slightly ion, which has been established elsewhere for microcompos- larger than that obtained with the other minicomposites. How- because of the scatter generally observed with T data b2NI-aVeR e The estimates of the interfacial shear stress in the minicom- () posites must be regarded as comparable. They do not indicate an effect of multilayering the interphase. This is a satisfactory result, because the properties of the minicomposite s nave been degraded by the presence of an interphase that contains ayers of a stiff material such as Sic a1=EJEc Comparable shear stresses have been estimated and crack deviation at the fiber surface has been observed (1+v)Em[Er+(1-2v)E the batch. Furthermore. similar interfacial shear stresses and E[(1+v)Er+(1-v)E crack deviations have been obtained with sic/sic minicom- posites that have been reinforced with as-received Nicalon fi- where &A is the hysteresis loop width, o the corresponding ber(NL 202)3 These features characterize a weak interfacial applied stress during the unloading-reloading sequence, a, the bond, as opposed to the rather strong bonding observed in the initial stress level at unloading, Ec the Youngs modulus of the SiC/SiC composites that are reinforced with treated Nicalon minicomposite, Rr the fiber radius, and v the Poissons ratio fiber. A layer of SiO/anisotropic carbon has been identified at measured s v). T was derived from the 8A-o data that were the surface of the fibers in the minicomposites of the present during the last unloading-reloading sequence, before study and in those SiC/SiC minic ites and 2D woven the ultimate failure of the minicomposites. N, the number of composites that have been reinforced with as-received Nicalon atrix cracks, was determined from SEM inspection of the NL 202)fiber; these minicomposites and composites exhibit a minicomposites after failure (2) The spacing distance of the matrix cracks at saturation, the composites that have been reinforced with treated Nicalon fiber and exhibit strong interfacial bonds. In these latter com- UsRe posites, the crack deviation occurs within the interphase. It has T been demonstrated that the weakest link is located within the 2 I interface in the former composites. 4, n the fiber/interphase interphases in the latter composites and Therefore, the debonding may be considered to be dictated by the surface of the fiber in the minicomposites of the present (4) study. Because of the presence of a layer of Sio /anisotropi carbon, the fiber/interphase interface is the weakest link where os is the applied stress at matrix cracking saturation. interfacial regio Table I shows that comparable T values were obtained, re- Debonding at the matrix/interphase interface also has
confirms debonding in the interface between the fiber and the interphase. Recall the previously mentioned feature, that the region between the fiber and the interphase consists mainly of a SiO2/anisotropic carbon sublayer. Such sublayers also have been identified in the composites that have been reinforced with as-received Nicalon fibers. In these composites, the F/M interactions are weak, and deflection of the matrix cracks occurs at the fiber/interphase interfaces.1,4 (3) Interfacial Shear Stress The interfacial shear stress (t) was estimated from the following data: (1) First, the width of hysteresis loops measured during unloading–reloading cycles. t is given by the following equation, which has been established elsewhere for microcomposites:25 t = b2N~1 − al Vf! 2 Rf 2Vf 2 Em S sp 2 dDD F s spS1 − s sp DG (1) with a1 = Ef/Ec (2a) b2 = ~1 + n!Em@Ef + ~1 − 2n!Ec# Ef@~1 + n!Ef + ~1 − n!Ec# (2b) where dD is the hysteresis loop width, s the corresponding applied stress during the unloading–reloading sequence, sp the initial stress level at unloading, Ec the Young’s modulus of the minicomposite, Rf the fiber radius, and n the Poisson’s ratio (n 4 nm 4 nf ). t was derived from the dD–s data that were measured during the last unloading–reloading sequence, before the ultimate failure of the minicomposites. N, the number of matrix cracks, was determined from SEM inspection of the minicomposites after failure. (2) The spacing distance of the matrix cracks at saturation, using the following equations:26,27 t = sSRf 2 VflS S1 + Ef Vf EmVm D (3) t = sSRf Vm 2 VflS (4) where sS is the applied stress at matrix cracking saturation. Table I shows that comparable t values were obtained, regardless of the method and batch. The t values obtained from the crack-spacing distance at saturation are 17–49 MPa, whereas those determined from the hysteresis loop widths are 32–70 MPa. Therefore, because of the scatter usually observed with t measurements, the t values do not indicate any significant effect of multilayering the interphase. The t values are comparable to those measured on the SiC/SiC minicomposites with a single PyC layer (batch (100/0)1). They also are comparable to those measured on SiC/SiC minicomposites that have been reinforced with as-received Nicalon fiber.3 This agreement may be attributed to the presence of the Si–C–O layer of at the surface of the Hi-Nicalon and Nicalon fibers, which has been demonstrated to dictate the deviation of the matrix cracks.28 Table I indicates that the t values derived from the hysteresis loop widths are generally larger than those obtained from the matrix-crack-spacing distance. As discussed in a following section, this discrepancy may be attributed to an effect of the initial twisting of the tows, which enhances F/M interactions within the minicomposites. IV. Discussion It has been emphasized that the tensile mechanical behavior of the minicomposites was not affected by the presence of multilayered interphases that involved stiff sublayers of SiC. Similar features have been identified, regardless of the interphases, including proportional limit, saturation stress, strain to failure, etc. Slightly different results have been obtained for Hi-Nicalon/ SiC microcomposites with (PyC/SiC)n nanoscale multilayered interphases.14 That author found that the stress–strain behavior of the microcomposites was dependent on the respective thicknesses of the sublayers. The best properties were observed for the following sublayer conditions: e(PyC) 4 3 nm, e(SiC) 4 30 nm, and n 4 10. The corresponding (3/30)10 minicomposites exhibited an interfacial shear stress that seemed to be slightly larger than that obtained with the other minicomposites. However, the difference cannot be considered to be significant, because of the scatter generally observed with t data. The estimates of the interfacial shear stress in the minicomposites must be regarded as comparable. They do not indicate an effect of multilayering the interphase. This is a satisfactory result, because the properties of the minicomposites have not been degraded by the presence of an interphase that contains layers of a stiff material such as SiC. Comparable shear stresses have been estimated and crack deviation at the fiber surface has been observed, regardless of the batch. Furthermore, similar interfacial shear stresses and crack deviations have been obtained with SiC/SiC minicomposites that have been reinforced with as-received Nicalon fiber (NL 202).3 These features characterize a weak interfacial bond, as opposed to the rather strong bonding observed in the SiC/SiC composites that are reinforced with treated Nicalon fiber.4 A layer of SiO2/anisotropic carbon has been identified at the surface of the fibers in the minicomposites of the present study and in those SiC/SiC minicomposites and 2D woven composites that have been reinforced with as-received Nicalon (NL 202) fiber; these minicomposites and composites exhibit a weak interfacial bond. By contrast, this layer is not present in the composites that have been reinforced with treated Nicalon fiber and exhibit strong interfacial bonds. In these latter composites, the crack deviation occurs within the interphase. It has been demonstrated that the weakest link is located within the interphases in the latter composites and in the fiber/interphase interface in the former composites.1,4,28 Therefore, the debonding may be considered to be dictated by the surface of the fiber in the minicomposites of the present study. Because of the presence of a layer of SiO2/anisotropic carbon, the fiber/interphase interface is the weakest link in the interfacial region. Debonding at the matrix/interphase interface also has been Fig. 4. SEM micrograph of a nanoscale multilayered interphase prepared via P-CVI (n 4 30). September 1999 Hi-Nicalon/SiC Minicomposites with (Pyrocarbon/SiC)n Nanoscale Multilayered Interphases 2469