《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_j.1151-2916.1999.tb01894.x[1]

ournal J An. Ceran. Soc., 82 [5] 1187-95(1999) Influence of Isothermal Chemical Vapor Deposition and Chemical Vapor Infiltration Conditions on the deposition Kinetics and Structure of boron nitride Marc Leparouxt and Lionel Vandenbulcke Laboratoire de Combustion et Systemes Reactifs, Centre National de la Recherche Scientifique(CNRS) 45071 Orleans france Christian clinard Centre de Nanoscopie Electronique Analytique (CNEA), Faculte des Sciences, Universite d'Orleans, 45067 Orleans, france An experimental study has been performed to gain some fibers and the matrix during composite insight into the correlations between the deposition condi- carbon interphase that is deposited on the to matrIx tions and the structure of boron nitride(BN) coatings that manufacturing leads to this nonbrittle me behavior 5-k are used in ceramic-matrix composites BN has been depos Unfortunately, the efficiency of these int limited by ited at 700oC from BCly-NH3-H, mixtures on various sub- he oxidation phenomena.9, 10 strates, by using chemical vapor deposition(CVD)and iso. Efforts have been made to improve the oxidation resistance thermal-isobaric chemical vapor infiltration(ICvI) of CMCs with a carbon interphase. Initially, external sealant processes, simultaneously in the same reactor. a kinetic coatings that contained silicon and/or boron were pro study has shown that the Cvd process is governed either by posed 11-13 Another approach considered a multilayered C/SiC a combination of mass transfer with chemical kinetics at matrix or interphase that may permit more crack deflec low flow rates or by the heterogeneous kinetics only at high issues are examined together with SiC/SiBC/ flow velocities. In contrast, the limiting contribution of SiC sequenced matrices, 7 to delay the access of oxygen to the mass transfer always is observed for the ICvI process. The carbon interphase or the fiber/interphase interface. In other influence of diffusion cages that are positioned around the studies, the oxidation resistance of the pyrocarbon interphase fibrous preforms is reported. The structure of BN deposits itself is improved by some change of its composition, as in has been studied as a function of the various deposition boron-doped pyrocarbon or in a compositional gradient layer conditions via transmission electron microscopy. The cho- the composition of which varies continuously from the fiber sen CVD conditions lead to a poor organization of the Bn interface(carbon) to the matrix interface( Sic). deposits. fairly well-organized bn coatings are depos On the other hand, boron nitride(Bn) has been proposed as on all fibers of a fibrous preform via ICVI. The results are an alternative, because of its graphitelike structure and its bet- discussed in terms of supersaturation and deposition ter resistance to oxidation. 20 It has been successfully used in yields. The use of diffusion cages and the adjustment of the various composite systems. 21-29 Particular attention has been inlet composition and mass flow rate seem to be very im- given to the interfacial zones, in terms of chemistry and mi- portant to obtain the best BN organization and thickness crostructure 26-29 These studies revealed that many interfacial noforn carbon-and/or SiO2-rich sublayers can be present between the SiC fibers(NicalonTM(Nippon Carbon, Tokyo, Japan)or L. Introduction TyrannoTM(UBE Industries, Yamaguchi, Japan) and the BN film. To improve the mechanical properties, especially at high HE potential of fiber-reinforced ceramic-matrix composites temperature, it is fundamental either to protect the carbon-rich (CMCs)for thermomechanical applications is well estab- sublayers from the external atmosphere or, even better, to pre lished, because of their fracture toughness, damage tolerance. vent their development. This condition implies the formation of and low density. The mechanical properties of CMCs are de- ndent largely on the fiber-matrix bonding, which must be ganization to deflect the cracks inside the interphase. Such an weak enough to allow crack deflection along the interface, yet optimized structure can be examined by controlling the classi- strong enough to retain load transfer from the matrix to the al experimental deposition parameters. Together with the pro- cess type(either chemical vapor deposition(CVD)in a hot wall reactor or isothermal-isobaric chemical vapor infiltration CVi), these variables produce different deposition condi- tions at the gas-phase/substrate interface which also are a func- R. Naslain-contributing edito tion, in any case, of the total amount of fibers to be coated Moreover, adequate deposition conditions can prevent the de velopment of amorphous glassy sublayers as SiO,. In previous studies, 26-29the interfaces have been principally observed and Manuscript No. 191278. Received January 14, 1997, approved September 21, generally, the evolution of the structure in the bn deposit has Supported by the Societe Europeenne de Propulsion(SEP) and Region Centre, not been described, especially as a function of the deposition conditions In this edat700°C pResent address: Fraunhofer-Iws,WinterbergstraBe 28, 0127 Dresden, Ger- from the BCl-NHa-H2 gas mixture, which was shown to be pResent address: CNRS-CRMD, IB Rue de la Ferollerie, 45071 Orleans, france less aggressive than the recursor under the infiltra 1187

Influence of Isothermal Chemical Vapor Deposition and Chemical Vapor Infiltration Conditions on the Deposition Kinetics and Structure of Boron Nitride Marc Leparoux† and Lionel Vandenbulcke* Laboratoire de Combustion et Syste`mes Re´actifs, Centre National de la Recherche Scientifique (CNRS), 45071 Orle´ans, France Christian Clinard‡ Centre de Nanoscopie Electronique Analytique (CNEA), Faculte´ des Sciences, Universite´ d’Orle´ans, 45067 Orle´ans, France An experimental study has been performed to gain some insight into the correlations between the deposition condi￾tions and the structure of boron nitride (BN) coatings that are used in ceramic-matrix composites. BN has been depos￾ited at 700°C from BCl3–NH3–H2 mixtures on various sub￾strates, by using chemical vapor deposition (CVD) and iso￾thermal–isobaric chemical vapor infiltration (ICVI) processes, simultaneously in the same reactor. A kinetic study has shown that the CVD process is governed either by a combination of mass transfer with chemical kinetics at low flow rates or by the heterogeneous kinetics only at high flow velocities. In contrast, the limiting contribution of mass transfer always is observed for the ICVI process. The influence of diffusion cages that are positioned around the fibrous preforms is reported. The structure of BN deposits has been studied as a function of the various deposition conditions via transmission electron microscopy. The cho￾sen CVD conditions lead to a poor organization of the BN deposits. Fairly well-organized BN coatings are deposited on all fibers of a fibrous preform via ICVI. The results are discussed in terms of supersaturation and deposition yields. The use of diffusion cages and the adjustment of the inlet composition and mass flow rate seem to be very im￾portant to obtain the best BN organization and thickness uniformity. I. Introduction THE potential of fiber-reinforced ceramic-matrix composites (CMCs) for thermomechanical applications is well estab￾lished, because of their fracture toughness, damage tolerance, and low density. The mechanical properties of CMCs are de￾pendent largely on the fiber–matrix bonding, which must be weak enough to allow crack deflection along the interface, yet strong enough to retain load transfer from the matrix to the fibers.1–4 A carbon-rich layer that is formed in situ between the fibers and the matrix during composite processing or a pyro￾carbon interphase that is deposited on the fibers prior to matrix manufacturing leads to this nonbrittle mechanical behavior.5–8 Unfortunately, the efficiency of these interphases is limited by the oxidation phenomena.9,10 Efforts have been made to improve the oxidation resistance of CMCs with a carbon interphase. Initially, external sealant coatings that contained silicon and/or boron were pro￾posed.11–13 Another approach considered a multilayered C/SiC matrix or interphase that may permit more crack deflec￾tions.14–16 These issues are examined, together with SiC/SiBC/ SiC sequenced matrices,17 to delay the access of oxygen to the carbon interphase or the fiber/interphase interface. In other studies, the oxidation resistance of the pyrocarbon interphase itself is improved by some change of its composition, as in boron-doped pyrocarbon18 or in a compositional gradient layer, the composition of which varies continuously from the fiber interface (carbon) to the matrix interface (SiC).19 On the other hand, boron nitride (BN) has been proposed as an alternative, because of its graphitelike structure and its bet￾ter resistance to oxidation.20 It has been successfully used in various composite systems.21–29 Particular attention has been given to the interfacial zones, in terms of chemistry and mi￾crostructure.26–29 These studies revealed that many interfacial carbon- and/or SiO2-rich sublayers can be present between the SiC fibers (Nicalon™ (Nippon Carbon, Tokyo, Japan) or Tyranno™ (UBE Industries, Yamaguchi, Japan)) and the BN film. To improve the mechanical properties, especially at high temperature, it is fundamental either to protect the carbon-rich sublayers from the external atmosphere or, even better, to pre￾vent their development. This condition implies the formation of strong interfacial bonding and the optimization of the BN or￾ganization to deflect the cracks inside the interphase. Such an optimized structure can be examined by controlling the classi￾cal experimental deposition parameters. Together with the pro￾cess type (either chemical vapor deposition (CVD) in a hot￾wall reactor or isothermal–isobaric chemical vapor infiltration (ICVI)), these variables produce different deposition condi￾tions at the gas-phase/substrate interface, which also are a func￾tion, in any case, of the total amount of fibers to be coated. Moreover, adequate deposition conditions can prevent the de￾velopment of amorphous glassy sublayers as SiO2. In previous studies,26–29 the interfaces have been principally observed and, generally, the evolution of the structure in the BN deposit has not been described, especially as a function of the deposition conditions. In this paper, the BN deposition was performed at 700°C from the BCl3–NH3–H2 gas mixture, which was shown to be less aggressive than the BF3–NH3 precursor under the infiltra￾R. Naslain—contributing editor Manuscript No. 191278. Received January 14, 1997; approved September 21, 1998. Supported by the Socie´te´ Europe´enne de Propulsion (SEP) and Re´gion Centre, via a grant given to author ML. *Member, American Ceramic Society. † Present address: Fraunhofer–IWS, Winterbergstraße 28, 01277 Dresden, Ger￾many. ‡ Present address: CNRS–CRMD, 1B Rue de la Fe´rollerie, 45071 Orle´ans, France. J. Am. Ceram. Soc., 82 [5] 1187–95 (1999) Journal 1187

1188 Journal of the American Ceramic Society-Leparoiex et al. Vol. 82. No 5 tion conditions. 30 To gain some insight into the correlations The thickness of these fibrous substrates was always very low between the deposition conditions and the BN structure, fir 0. 4 mm); therefore, the deposition conditions that have beer deposition conditions where the CVd process is purely con oIled by the surface kinetics have been determined. 30,t These or similar to(for tows) those from the CVD process. Different onditions allow investigation of the influence of the different fibers have been used to observe the influence of the nature of let gaseous species on the surface kinetics. Then, how the the substrate: these fiber types include T300(ex-PAN, average deposition conditions are influenced by mass transfer and de diameter of 8 um, Torayca, produced by Soficar, Abidos, art a fur France) and P55(pitch-derived, average diameter of 10 um ion of either the process type (CVD or ICVI), processing Thornel, produced by Amoco Polymers, Chicago, IL) carbon parameters, inlet gas composi iber, as well as NicalonTM NLM 202 fiber(ex-PCS, average given in this study. The influence of diffusion cages that are diameter of 15 um) positioned around the fibrous preform also is reported. Finally, The infiltration process(ICvI) was investigated by us the bn structure is studied using transmission electron micros- three-dimensional (3D)woven architecture(NovoltexTM, Soci- copy(TEM), under these various CVD and ICVI condition ete Europeenne de Propulsion(SEP), Saint-Medard-en-Jalles and the variations of the uniformity of the deposit are reported. France)that was composed of carbon fibers that were 9 um in diameter(ex-PAN, treated by SEP), this NovoltexTM material Il. Experimental Procedure had an average pore diameter of 45 um and a specific surface area of -0.25 m/g. The fibrous preform (150 mm x 18 mm x BN was deposited by using different isothermal CVD pro- Suring 15 mm x 18 mm x 3.3 mm. Each element was [el (I Materials and Deposition Procedure 10 mm) was cut into thirty intermediate samples, each me cesses at a temperature(T)of 700oC, under a reduced pressure enced by using the nomenclature"m n, "where m=1-10 for (P)of 1.3 kPa, from BCI-NH -H, gas mixtures of different the longitudinal position and n -3 for the transverse pos inlet composition. The deposition apparatus and procedure tion(see Fig. 1 ). Thus, different parameters have been intro- have been described in detail previously 30-32 Considering the duced to characterize the uniformity of the deposit. For the poor stability of Bn that was obtai the deposits were always thermally treated at 1000 C at the end thickness uniformity, which is calculated as the ratio between of the deposition stage, directly in the reactor, for 2 h unde he lowest and the highest relative mass increases along the vacuum(residual pressure of I Pa preform(m = 1-10). The variable IU represents the infiltration bn deposition was performed simultaneously on bulk sub- uniformity, which is defined as the average of ten ratios be- strates and a fibrous preform, to investigate both CVD and tween the internal elemental relative mass gain(n= 2)and the ICVI, respectively(Fig. 1). One bulk substrate was composed of ten graphite rings that were mounted on a holder. The n= 3). The total uniformity (TU) is defined as TU =LU X deposition process also was applied to one-dimensional (ID) IU. For the bulk substrates, the variable U represents the lon- fiber substrates, which consisted of either small tows or mon gitudinal uniformity, which is calculated as the ratio between filaments that were mounted on a carbon fixture. and to two- he relative weight increase at the bottom and top of the ten dimensional(2D) single-plane-weave fabrics In any case, the graphite rings after smoothing the curve that reports the relative ain surface of the substrate was parallel to the gaseous flow. weight gain of the ten substrates In some cases, diffusion screens were placed in fron fiber substrates(they were not used to protect the ngs). These screens created a pure diffusive volume the substrates, which were separated from the convective floy In case of the Cvd process on monofilaments or small tor the screens were made of carbon sheets with many holes bored into them The screens were mounted on both sides of the fiber carbon-based cement. In of Icvi the diffusion screens were composed of graphite plates that com- etely surrounded the preform(Fig. 2). The gaseous phase could only diffuse through oblong holes that had been bored into the plates that were positioned in front of the main surface substrate of the substrate. The axis of these holes was perpendicular to the external convective-flow direction and their diameter was four times smaller than the thickness of the plate. In the fol- lowing text, the terms"protected-CVD"and"protected-ICVI are used to respectively denote CVD and ICVI processes that were performed with protection screens ( Structural Characterization The BN-coated fibers were characterized via TEM. The analyses were performed at CNEA(Universite d'Orleans)on thin foils that had been prepared in a conventional manner, by inclusion in a methylmethacrylate resin or in Araldit (Cib Specialty Chemicals Holding, Basel, Switzerland) and cutting normal to the fiber axis via ultramicrotomy. Transverse varia- tions of the BN structure were obtained by scanning different zones of the coating, from the interface at the fiber to the external portion of the Bn deposit. Generally, three sections were observed and one representative BN coating was choser to be studied on each n. The dimensions of the greatest coherent domain and the interlayer distance, which is normally 0.333 nm for hexagonal BN. were measured on the TEm im Fig. 1. Schematic of the bulk substrates and the fibrous preform with ages. In the reported results, La represents the dimension of the their holders in the deposition reactor. in-plane extension and Le represents the extension of the co-

tion conditions.30 To gain some insight into the correlations between the deposition conditions and the BN structure, first, deposition conditions where the CVD process is purely con￾trolled by the surface kinetics have been determined.30,31 These conditions allow investigation of the influence of the different inlet gaseous species on the surface kinetics. Then, how the deposition conditions are influenced by mass transfer and de￾part from this limiting case of pure kinetic control, as a func￾tion of either the process type (CVD or ICVI), processing parameters, inlet gas composition, and/or total gas flow rate, is given in this study. The influence of diffusion cages that are positioned around the fibrous preform also is reported. Finally, the BN structure is studied, using transmission electron micros￾copy (TEM), under these various CVD and ICVI conditions, and the variations of the uniformity of the deposit are reported. II. Experimental Procedure (1) Materials and Deposition Procedure BN was deposited by using different isothermal CVD pro￾cesses at a temperature (T) of 700°C, under a reduced pressure (P) of 1.3 kPa, from BCl3–NH3–H2 gas mixtures of different inlet composition. The deposition apparatus and procedure have been described in detail previously.30–32 Considering the poor stability of BN that was obtained at the given temperature, the deposits were always thermally treated at 1000°C at the end of the deposition stage, directly in the reactor, for 2 h under vacuum (residual pressure of 1 Pa). BN deposition was performed simultaneously on bulk sub￾strates and a fibrous preform, to investigate both CVD and ICVI, respectively (Fig. 1). One bulk substrate was composed of ten graphite rings that were mounted on a holder. The deposition process also was applied to one-dimensional (1D) fiber substrates, which consisted of either small tows or mono￾filaments that were mounted on a carbon fixture, and to two￾dimensional (2D) single-plane-weave fabrics. In any case, the main surface of the substrate was parallel to the gaseous flow. The thickness of these fibrous substrates was always very low (<0.4 mm); therefore, the deposition conditions that have been considered in this paper were the same as (for monofilaments) or similar to (for tows) those from the CVD process. Different fibers have been used to observe the influence of the nature of the substrate: these fiber types include T300 (ex-PAN, average diameter of 8 mm; Torayca, produced by Soficar, Abidos, France) and P55 (pitch-derived, average diameter of 10 mm; Thornel, produced by Amoco Polymers, Chicago, IL) carbon fiber, as well as Nicalon™ NLM 202 fiber (ex-PCS, average diameter of 15 mm). The infiltration process (ICVI) was investigated by using a three-dimensional (3D) woven architecture (Novoltex™, Soci￾e´te´ Europe´enne de Propulsion (SEP), Saint-Medard-en-Jalles, France) that was composed of carbon fibers that were 9 mm in diameter (ex-PAN, treated by SEP); this Novoltex™ material had an average pore diameter of 45 mm and a specific surface area of ∼0.25 m2 /g. The fibrous preform (150 mm × 18 mm × 10 mm) was cut into thirty intermediate samples, each mea￾suring 15 mm × 18 mm × 3.3 mm. Each element was refer￾enced by using the nomenclature “m.n,” where m 4 1–10 for the longitudinal position and n 4 1–3 for the transverse posi￾tion (see Fig. 1). Thus, different parameters have been intro￾duced to characterize the uniformity of the deposit. For the fibrous preforms, the variable LU represents the longitudinal thickness uniformity, which is calculated as the ratio between the lowest and the highest relative mass increases along the preform (m 4 1–10). The variable IU represents the infiltration uniformity, which is defined as the average of ten ratios be￾tween the internal elemental relative mass gain (n 4 2) and the average external mass (calculated from values for n 4 1 and n 4 3). The total uniformity (TU) is defined as TU 4 LU × IU. For the bulk substrates, the variable U represents the lon￾gitudinal uniformity, which is calculated as the ratio between the relative weight increase at the bottom and top of the ten graphite rings after smoothing the curve that reports the relative weight gain of the ten substrates. In some cases, diffusion screens were placed in front of the fiber substrates (they were not used to protect the graphite rings). These screens created a pure diffusive volume around the substrates, which were separated from the convective flow. In case of the CVD process on monofilaments or small tows, the screens were made of carbon sheets with many holes bored into them. The screens were mounted on both sides of the fiber fixture, using a carbon-based cement. In case of ICVI, the diffusion screens were composed of graphite plates that com￾pletely surrounded the preform (Fig. 2). The gaseous phase could only diffuse through oblong holes that had been bored into the plates that were positioned in front of the main surface of the substrate. The axis of these holes was perpendicular to the external convective-flow direction, and their diameter was four times smaller than the thickness of the plate. In the fol￾lowing text, the terms “protected-CVD” and “protected-ICVI” are used to respectively denote CVD and ICVI processes that were performed with protection screens. (2) Structural Characterization The BN-coated fibers were characterized via TEM. The analyses were performed at CNEA (Universite´ d’Orle´ans) on thin foils that had been prepared in a conventional manner, by inclusion in a methylmethacrylate resin or in Aralditt (Ciba Specialty Chemicals Holding, Basel, Switzerland) and cutting normal to the fiber axis via ultramicrotomy. Transverse varia￾tions of the BN structure were obtained by scanning different zones of the coating, from the interface at the fiber to the external portion of the BN deposit. Generally, three sections were observed and one representative BN coating was chosen to be studied on each specimen. The dimensions of the greatest coherent domain and the interlayer distance, which is normally 0.333 nm for hexagonal BN, were measured on the TEM im￾ages. In the reported results, La represents the dimension of the in-plane extension and Lc represents the extension of the co￾Fig. 1. Schematic of the bulk substrates and the fibrous preform with their holders in the deposition reactor. 1188 Journal of the American Ceramic Society—Leparoux et al. Vol. 82, No. 5

Influence of lsothermal CVd and CV Conditions on the Deposition Kinetics and Structure of Bn 0.30 0.15兰 1.0 051015202530354045555 Total now rate (I0"*kg/s) Fig 3. Influence of the total mass flow rate on the relative weight o deposit(Am/m)on the bulk substrate via CVd and in the internal and ternal portions of the fibrous preform via pre 1.33 kPa, NH / BCI,= 1, and H/BCI,= 1) ematic of the diffusion es with external convective noted as“l”) and pure diffusion( denoted as“2”) to the fibrous rent domain along the c-axis of hexagonal bn. The greatest herent domain is the domain with the area is defined as La x le). For the interlayer distance, a mini- mum of five measurements was made on each image. the between the lowest and the in the figures. as well as rsion. Well crystallized graphite was used as a to measure the interlayer distance. The maximal error, from the procedures that have been used for our data acquisition, image treatments and final measurements. was -0.03 nm 0510520253344555560 Total now rate (10 x kg/s) IIL. Results Variation of the uniformities with the total mass flow rate on Deposition Kinetics bulk substrate U and in the preforms(■)LU,●)IU,and(★) protected-ICVI The influence of the deposition temperature on the do tion rate in the internal and external portions of the preform (CVI)and on the graphite rings(CVD)already has been re especially slightly greater uniformities are obtained by using ported for equal partial pressures of BCl3, NH3, and H, and a he protected-ICVI process, in comparison to classical ICvI total mass flow rate of 7 x 106 kg/s 30, 31 The deposition rates (see Figs. 4 and 5) in the preform are dependent exponentially on the inverse of High flow rates lead to deposition conditions where the CVd process is only controlled by the surface kinetics; thus, the they stabilize and even decrease in the internal portion, thus CVD deposition mechanism can be studied and the deposition indicating a mass-transfer limitation, which leads to poor in conditions in the ICVI process can be compared. The nil order filtration uniformity (IU 65% and LU = 45% at 800C. with respect to argon, allows both the total flow rate and the instead of 80% and 70%, respectively, at 700oC). Such a mass transfer limitation was not observed in CVD. However, the sure of a species was changed, to determine the apparent re- following study was performed at 700oC to preserve fairly action order with respect to this species. 0 Under these condi- good deposit uniformities via ICVI tions, the residence time of the gaseous phase within the reactor The influence of the mass-flow -rate variations on the cvd was maintained constant. That point could be important if some deposition rate on the graphite rings shows a transition in the homogeneous intermediate reaction occurs. On the other hand process from a mass-transfer and chemical-kinetics control to a a constant total pressure allows restriction of the variations pure chemical kinetic limitation(Fig. 3). At flow rates greater the diffusion conditions to the compositional effect to the kinetics at the surface of the bulk substrates is shown either by product HCl, was observed to be zero. 30, 3I Moreover, HCI has a weak influence on the bn deposition rate in the preform; this high value(98%)of the longitudinal uniformity(Fig. 4). The result is certainly due to similar mass-transfer conditions of the protected-ICVI process seems to be al ways governed by a cor reactant species(the diffusion coefficient of argon and HCI are bination of mass transfer and chemical kinetics, because of the not very different). The experimental results that concern the high specific surface area of the preform, which adsorbs a large reactants NH3 and BCl3 permit the determination of two dis- amount of reactants(Fig. 3). a slightly lower growth rate and tinct domains in either case, as shown, for example, in Fig. 6

herent domain along the c-axis of hexagonal BN. The greatest coherent domain is the domain with the greatest surface (the area is defined as La × Lc). For the interlayer distance, a mini￾mum of five measurements was made on each image. The average between the lowest and the greatest interlayer value is depicted in the figures, as well as the dispersion. Well￾crystallized graphite was used as a standard, to measure the interlayer distance. The maximal error, from the procedures that have been used for our data acquisition, image treatments, and final measurements, was ∼0.03 nm. III. Results (1) Deposition Kinetics The influence of the deposition temperature on the deposi￾tion rate in the internal and external portions of the preform (ICVI) and on the graphite rings (CVD) already has been re￾ported for equal partial pressures of BCl3, NH3, and H2 and a total mass flow rate of 7 × 10−6 kg/s.30,31 The deposition rates in the preform are dependent exponentially on the inverse of temperature in the 500°–700°C range. At higher temperatures, they stabilize and even decrease in the internal portion, thus indicating a mass-transfer limitation, which leads to poor in￾filtration uniformity (IU 4 65% and LU 4 45% at 800°C, instead of 80% and 70%, respectively, at 700°C). Such a mass￾transfer limitation was not observed in CVD. However, the following study was performed at 700°C to preserve fairly good deposit uniformities via ICVI. The influence of the mass-flow-rate variations on the CVD deposition rate on the graphite rings shows a transition in the process from a mass-transfer and chemical-kinetics control to a pure chemical kinetic limitation (Fig. 3). At flow rates greater than ∼25 × 10−6 kg/s, the main control by the heterogeneous kinetics at the surface of the bulk substrates is shown either by the negligible influence of the total flow rate or by the very high value (98%) of the longitudinal uniformity (Fig. 4). The protected-ICVI process seems to be always governed by a com￾bination of mass transfer and chemical kinetics, because of the high specific surface area of the preform, which adsorbs a large amount of reactants (Fig. 3). A slightly lower growth rate and especially slightly greater uniformities are obtained by using the protected-ICVI process, in comparison to classical ICVI (see Figs. 4 and 5). High flow rates lead to deposition conditions where the CVD process is only controlled by the surface kinetics; thus, the CVD deposition mechanism can be studied and the deposition conditions in the ICVI process can be compared. The nil order, with respect to argon, allows both the total flow rate and the total pressure to be maintained constant while the partial pres￾sure of a species was changed, to determine the apparent re￾action order with respect to this species.30 Under these condi￾tions, the residence time of the gaseous phase within the reactor was maintained constant. That point could be important if some homogeneous intermediate reaction occurs. On the other hand, a constant total pressure allows restriction of the variations of the diffusion conditions to the compositional effect. The apparent reaction order, with respect to the gaseous product HCl, was observed to be zero.30,31 Moreover, HCl has a weak influence on the BN deposition rate in the preform; this result is certainly due to similar mass-transfer conditions of the reactant species (the diffusion coefficient of argon and HCl are not very different). The experimental results that concern the reactants NH3 and BCl3 permit the determination of two dis￾tinct domains in either case, as shown, for example, in Fig. 6, Fig. 2. Schematic of the diffusion cages with external convective flow (denoted as “1”) and pure diffusion (denoted as “2”) to the fibrous preform. Fig. 3. Influence of the total mass flow rate on the relative weight of deposit (Dm/m) on the bulk substrate via CVD and in the internal and external portions of the fibrous preform via protected-ICVI (T 4 700°C, P 4 1.33 kPa, NH3/BCl3 4 1, and H2/BCl3 4 1). Fig. 4. Variation of the uniformities with the total mass flow rate on (m) the bulk substrate U and in the preforms (j) LU, (d) IU, and (.) TU for protected-ICVI. May 1999 Influence of Isothermal CVD and CVI Conditions on the Deposition Kinetics and Structure of BN 1189

Journal of the American Ceramic Sociery-Leparouex et al. Vol. 82. No 5 100 rate-limited. and a combination with mass transfer should be considered. Optimal infiltration rates and uniformities are attained for PNH,>PBCI, ( Fig. 7) but also when PBClg n- 8 creases. 30 However. these conditions could become detrimental to the molecular diffusion inside the preform if the total pres- sure increases too much. Moreover, some homogeneous nucle- ation also could occur Finally, these different behaviors of the CVD and ICVI cesses may have an impor Results le in the growth mechanism and the deposit structure. Re on the variations of the struc- ture with the deposition conditions are reported in the next (2) Deposit Structure Regardless of the deposition conditions, BN presents a tur bostratic structure with layers that have a tendency to lie par- 510152025335 allel to the fiber, particularly near the fiber/BN interface. No (106×kg/s) noticeable influence of the fiber type(T300, P55,or NicalonTM) has been observed on the organization and orien- Fig. 5. Variation of the uniformities with the total mass flow rate on tation of the BN structure, especially near the fiber surface ▲) the bulk substrate U and in the preforms(■)LU,(●)IU,and(★) The influence of the local deposition conditions on the de- posit structure, with respect to the process type(CVD or ICVI), was studied under the following standard conditions: T 700C, P= 1.33 kPa, NH,/BCI3 = 1, H/BCIs 1, an NH/BCI total flow rate (f) of 7x 106 kg/. The samples were taken 0.52/3 1.432 from the middle of T300 tows and NovoltexTM preforms(these 0.01 samples are referenced as6.2") for the CVD and ICvi pro- cesses, respectively. Figure 8 presents the variations of the greatest coherent domain and the variations of the interlayer distance in the bn coating. both as a function of the distance from the fiber, for the different deposition processes. In both 1E-3 cases, the BN layers have a tendency to lie parallel to the fiber surface, especially near the fiber/BN interface; 3 however,no preferential growth appears that would have produced a planar (Le La) or columnar(Le> La) structure. The two CVD pro- cesses--protected or not--produced similar BN film microtex- tures. The coherent domains were very small, except in the 1E4 most-external portion(which is-10-20 nm thick). The inter- externa layer distances and the dispersion of the values clearly in- creased when the coherent domains were smaller in the central portion of the BN deposit. The ICVI processes allowed depo- of BN coatings that globally 1E5 internal tion with greater coherent domains and smaller interlayer dis- ances. These two parameters also were clearly correlated for 68 p NH3 Torr) Fig. 6. Influence of the inlet partial pressure of NH, on the weight of deposit per surface unit(Am/S)on the bulk substrate and in the internal and external portions of the fibrous preform for protected-ICVI (T 700oC, P= 1.06 kPa, PBCL= Pu,=0.266 kPa(2 torr), andf2 x 10-°kg/s) where the deposition rates per surface unit are reported as a function of the inlet NH, partial pressure (PNH, ).Simi- The apparent reaction orders are almost one for the minor reactant species and zero for the major reactant species that seems to be preferentially adsorbed. These results agree with previous study that showed a first-order dependence on the BCl3 partial pressure (PBCI) and a nil order, with respect to NH3, when BCl, was the minor reactant. In regard to the classical-ICVI or protected-ICVI processes the deposition kinetic behaviors are different from the CVD p NH3 (Torr) behavior, as a function of the reactant partial pressures Clearly, the external and internal ICVI rates continuously Fig. 7. Variation of the uniformities with the inlet partial pressure of ase as the NHg/BCl] ratio increases, as shown here for pro- NH, on(A)the bulk substrate U and in the preforms(LU, (O)IU tected-ICVI(Fig 6); however, these values remain much lowe and(★) TU for protected-CVI(T=700°C,P than the CVD rate. The ICVI process is not purely chemical a, PBcl: P1,=0.266kPa(2tor),andf≥28×106kgs)

where the deposition rates per surface unit are reported as a function of the inlet NH3 partial pressure (pNH3 ). Simi￾lar results have been obtained with respect to BCl3. 30 The apparent reaction orders are almost one for the minor reactant species and zero for the major reactant species that seems to be preferentially adsorbed. These results agree with a previous study that showed a first-order dependence on the BCl3 partial pressure (pBCl3 ) and a nil order, with respect to NH3, when BCl3 was the minor reactant.33 In regard to the classical-ICVI or protected-ICVI processes, the deposition kinetic behaviors are different from the CVD behavior, as a function of the reactant partial pressures. Clearly, the external and internal ICVI rates continuously in￾crease as the NH3/BCl3 ratio increases, as shown here for pro￾tected-ICVI (Fig. 6); however, these values remain much lower than the CVD rate. The ICVI process is not purely chemical￾rate-limited, and a combination with mass transfer should be considered. Optimal infiltration rates and uniformities are attained for pNH3 > pBCl3 (Fig. 7) but also when pBCl3 in￾creases.30 However, these conditions could become detrimental to the molecular diffusion inside the preform if the total pres￾sure increases too much. Moreover, some homogeneous nucle￾ation also could occur. Finally, these different behaviors of the CVD and ICVI pro￾cesses may have an important role in the growth mechanism and the deposit structure. Results on the variations of the struc￾ture with the deposition conditions are reported in the next section. (2) Deposit Structure Regardless of the deposition conditions, BN presents a tur￾bostratic structure with layers that have a tendency to lie par￾allel to the fiber, particularly near the fiber/BN interface. No noticeable influence of the fiber type (T300, P55, or Nicalon™) has been observed on the organization and orien￾tation of the BN structure, especially near the fiber surface. The influence of the local deposition conditions on the de￾posit structure, with respect to the process type (CVD or ICVI), was studied under the following standard conditions: T 4 700°C, P 4 1.33 kPa, NH3/BCl3 4 1, H2/BCl3 4 1, and a total flow rate ( f ) of 7 × 10−6 kg/s. The samples were taken from the middle of T300 tows and Novoltex™ preforms (these samples are referenced as “6.2”) for the CVD and ICVI pro￾cesses, respectively. Figure 8 presents the variations of the greatest coherent domain and the variations of the interlayer distance in the BN coating, both as a function of the distance from the fiber, for the different deposition processes. In both cases, the BN layers have a tendency to lie parallel to the fiber surface, especially near the fiber/BN interface;30 however, no preferential growth appears that would have produced a planar (Lc La) structure. The two CVD pro￾cesses—protected or not—produced similar BN film microtex￾tures. The coherent domains were very small, except in the most-external portion (which is ∼10–20 nm thick). The inter￾layer distances and the dispersion of the values clearly in￾creased when the coherent domains were smaller, in the central portion of the BN deposit. The ICVI processes allowed depo￾sition of BN coatings that globally exhibited better organiza￾tion with greater coherent domains and smaller interlayer dis￾tances. These two parameters also were clearly correlated for Fig. 5. Variation of the uniformities with the total mass flow rate on (m) the bulk substrate U and in the preforms (j) LU, (d) IU, and (.) TU for classical ICVI. Fig. 6. Influence of the inlet partial pressure of NH3 on the weight of deposit per surface unit (Dm/S) on the bulk substrate and in the internal and external portions of the fibrous preform for protected-ICVI (T 4 700°C, P 4 1.06 kPa, pBCl3 4 pH2 4 0.266 kPa (2 torr), and f $ 28 × 10−6 kg/s). Fig. 7. Variation of the uniformities with the inlet partial pressure of NH3 on (m) the bulk substrate U and in the preforms (j) LU, (d) IU, and (.) TU for protected-ICVI (T 4 700°C, P 4 1.06 kPa, pBCl3 4 pH2 4 0.266 kPa (2 torr), and f $ 28 × 10−6 kg/s). 1190 Journal of the American Ceramic Society—Leparoux et al. Vol. 82, No. 5

Influence of Isothermal CVD and CV Conditions on the Deposition Kinetics and Structure of BN 1191 12 64202 0.30 38 5o8 f=14x10°kos 04s 望 8642 050100150200250300350400 P。 sition from the fiber surfac·nm) 5040 Evolution of the greatest coherent domain and the interlayer Position from the fIber surface (nm) in the bn thickness as a function of the mass flow rate for the and La are shaded gray and white, respec- mples were taken in the middle of the fibrous preforms(the Fig. 8. Evolution of the greatest coherent domain and the interlayer cimen reference distance in the Bn thickness for different processes. Samples were taken in the middle of the tows or of the fibrous preforms(the 6.2 pecimen reference). Le and La data are shaded gray and white, re- spectively(T= 700C, P= 1.33 kPa, NH /BCl3= 1, H/BCl3= 1, La. In any case, the interlayer distances remain relativel andf=7×10°kg/s) small, with average values of -0.35 nm, all along the bn film (Fig. 12). An additional experiment that was conducted at a flow rate of 56 x 10 kg/s led to a bn structure that was similar to that obtained at a flow rate of 14 x 10-6 kg/s the Icvi processes. No"skin effect" was observed in the most At each position of the substrate to be coated, the deposition external portion, in contrast to the CVD coatings. The range of and growth conditions obviously are dependent on the partial the interlayer distances was narrow, and values as low as 0.33 ressures of all the species that are present. Therefore, the nm(which corresponds to the theoretical distance for a hex- structure of the bn coating that has been deposited using the agonal BN crystal)were observed. The protected-ICVI process standard conditions has been investigated for three different roduced the best organization, with lower standard deviation positions in the fibrous preform. For example, Fig. 13 shows of the interlay results that have been obtained with the protected -ICVI pro- The evolution of the structure of the deposits in the middle cess. For this fairly low mass flow rate, the results that have of the NovoltexTM preform has been observed for three total been obtained with classical ICVI are not very different. In any flow rates--7x 10 6, 14 x 10, and 28 x 106 kg/s--for both case, the 6.1(6.3) specimen presents the best organization, as ICVI processes. Figure 9 shows the variation of the greatest indicated by its greater coherent domains. However, all the coherent domain and the interlayer distance from the carbon- variations that can be noted are moderate in comparison to the fiber/BN interface to the external portion of a coating that has variations of the structure that has been produced by using the been deposited via classical ICVI. The two lowest mass flow other experimental parameters. This observation is confirmed rates lead to similar structures; however, the BN organization by the examination of the variations of the interlayer distances seems to be slightly better at a flow rate of 14 x 10"b kg/s. In The interlayer distances remain relatively small and are only contrast, the bn organization is altered for the highest floy slightly dispersed, regardless of the position and the infiltration (i.e, at 28x" kg/s), especially at distances greater than -50 process nm from the fiber surface. Figures 10(a)and(b) respectivel A particular skin effect was observed for the CVD processes illustrate the structures that have been obtained for this flo (Fig 8). Observation of the microtexture with and without the rate at a distance of -50 nm from the fiber and near the fiber/ thermal post-treatment clearly shows that this effect results BN interface. Figure 10(a) is representative of one of the poor- from the post-treatment. For ICVI processes, such an effect est BN lattice organizations. However, this organization never appeared. In addition to the standard treatment, different presents characteristics that are even better than the structure treatment times were tested at 1000C (I and 4 h)and a treat- that is generally observed in the CVD coatings, except in their ment at 1200 C was performed for 2 h. Moreover, other treat- external portion. In contrast, with the protected-ICVI process, ments were conducted at 1.33 kPa in various gaseous atmo- this relatively high flow rate produces the best BN organiza- spheres, which included hydrogen, argon, nitrogen, and tion, as shown in the micrographs in Fig. 11. Figure 12 con- helium. Unfortunately, no improvement of the BN lattice firms the large dimensions of the greatest coherent domain ganization was observed. An alternate CVD process that in- The crystal growth seems to be preferentially columnar, i.e., Le volved several short deposition steps, us

the ICVI processes. No “skin effect” was observed in the most￾external portion, in contrast to the CVD coatings. The range of the interlayer distances was narrow, and values as low as 0.33 nm (which corresponds to the theoretical distance for a hex￾agonal BN crystal) were observed. The protected-ICVI process produced the best organization, with lower standard deviations of the interlayer spacing. The evolution of the structure of the deposits in the middle of the Novoltex™ preform has been observed for three total flow rates—7 × 10−6 , 14 × 10−6 , and 28 × 10−6 kg/s—for both ICVI processes. Figure 9 shows the variation of the greatest coherent domain and the interlayer distance from the carbon￾fiber/BN interface to the external portion of a coating that has been deposited via classical ICVI. The two lowest mass flow rates lead to similar structures; however, the BN organization seems to be slightly better at a flow rate of 14 × 10−6 kg/s. In contrast, the BN organization is altered for the highest flow (i.e., at 28 × 10−6 kg/s), especially at distances greater than ∼50 nm from the fiber surface. Figures 10(a) and (b) respectively illustrate the structures that have been obtained for this flow rate at a distance of ∼50 nm from the fiber and near the fiber/ BN interface. Figure 10(a) is representative of one of the poor￾est BN lattice organizations. However, this organization presents characteristics that are even better than the structure that is generally observed in the CVD coatings, except in their external portion. In contrast, with the protected-ICVI process, this relatively high flow rate produces the best BN organiza￾tion, as shown in the micrographs in Fig. 11. Figure 12 con￾firms the large dimensions of the greatest coherent domain. The crystal growth seems to be preferentially columnar, i.e., Lc > La. In any case, the interlayer distances remain relatively small, with average values of ∼0.35 nm, all along the BN film (Fig. 12). An additional experiment that was conducted at a flow rate of 56 × 10−6 kg/s led to a BN structure that was similar to that obtained at a flow rate of 14 × 10−6 kg/s. At each position of the substrate to be coated, the deposition and growth conditions obviously are dependent on the partial pressures of all the species that are present. Therefore, the structure of the BN coating that has been deposited using the standard conditions has been investigated for three different positions in the fibrous preform. For example, Fig. 13 shows results that have been obtained with the protected-ICVI pro￾cess. For this fairly low mass flow rate, the results that have been obtained with classical ICVI are not very different. In any case, the 6.1 (6.3) specimen presents the best organization, as indicated by its greater coherent domains. However, all the variations that can be noted are moderate in comparison to the variations of the structure that has been produced by using the other experimental parameters. This observation is confirmed by the examination of the variations of the interlayer distances. The interlayer distances remain relatively small and are only slightly dispersed, regardless of the position and the infiltration process. A particular skin effect was observed for the CVD processes (Fig. 8). Observation of the microtexture with and without the thermal post-treatment clearly shows that this effect results from the post-treatment. For ICVI processes, such an effect never appeared. In addition to the standard treatment, different treatment times were tested at 1000°C (1 and 4 h) and a treat￾ment at 1200°C was performed for 2 h. Moreover, other treat￾ments were conducted at 1.33 kPa in various gaseous atmo￾spheres, which included hydrogen, argon, nitrogen, and helium. Unfortunately, no improvement of the BN lattice or￾ganization was observed. An alternate CVD process that in￾volved several short deposition steps, using standard condi￾Fig. 8. Evolution of the greatest coherent domain and the interlayer distance in the BN thickness for different processes. Samples were taken in the middle of the tows or of the fibrous preforms (the 6.2 specimen reference). Lc and La data are shaded gray and white, re￾spectively (T 4 700°C, P 4 1.33 kPa, NH3/BCl3 4 1, H2/BCl3 4 1, and f 4 7 × 10−6 kg/s). Fig. 9. Evolution of the greatest coherent domain and the interlayer distance in the BN thickness, as a function of the mass flow rate for the classical ICVI process. Lc and La are shaded gray and white, respec￾tively. Samples were taken in the middle of the fibrous preforms (the 6.2 specimen reference). May 1999 Influence of Isothermal CVD and CVI Conditions on the Deposition Kinetics and Structure of BN 1191

Journal of the American Ceramic Society-Leparoux et al. Vol. 82. No. 5 (b) Fig. 10. TEM lattice fringes of the BN coating obtained by classical ICVI ( 28 x 10-6 kg/s)((a)inside the Bn deposit and (b)near the fiber/BN interface); the electron-diffraction result also is shown. (a) (b) (d) Fig. 11. TEM lattice fringes of the BN coating obtained via protected-ICVI ( 28 x 10 kg/s)((a) the external portion of the deposit,(b) tween the external portion and the middle, (c)the middle of the coating, and(d) near the fiber/BN interface), the electron-diffraction result also tions, followed by thermal treatment steps did not permit either the reactant(PBCl, or PNH,, whichever has the process to improve the Bn organization all cases, the deposition conditions at the ga depart very much from the approximate eq between IV. Discussion he gas phase and the solid; i.e., the supersaturation is high. If we suppose that no intermediate precursors are produced via At 700oC, the CVd process is limited either by a homogeneous reactions, a local equilibrium at the interface transfer with the chemical kinetics at low fl would be represented by reaction(1) rates ne chemical kinetics alone at high flow rates where rate is only governed by the partial pressure of BCl3+ NH3 F+3HCI

tions, followed by thermal treatment steps did not permit either process to improve the BN organization. IV. Discussion At 700°C, the CVD process is limited either by a combina￾tion of mass transfer with the chemical kinetics at low flow rates or by the chemical kinetics alone at high flow rates where the deposition rate is only governed by the partial pressure of the reactant (pBCl3 or pNH3 , whichever has the lowest value). In all cases, the deposition conditions at the gas/solid interface depart very much from the approximate equilibrium between the gas phase and the solid; i.e., the supersaturation is high. If we suppose that no intermediate precursors are produced via homogeneous reactions, a local equilibrium at the interface would be represented by reaction (1): BCl3 + NH3 →← ,BN. + 3HCl (1) Fig. 10. TEM lattice fringes of the BN coating obtained by classical ICVI (f 4 28 × 10−6 kg/s) ((a) inside the BN deposit and (b) near the fiber/BN interface); the electron-diffraction result also is shown. Fig. 11. TEM lattice fringes of the BN coating obtained via protected-ICVI (f 4 28 × 10−6 kg/s) ((a) the external portion of the deposit, (b) between the external portion and the middle, (c) the middle of the coating, and (d) near the fiber/BN interface); the electron-diffraction result also is shown. 1192 Journal of the American Ceramic Society—Leparoux et al. Vol. 82, No. 5

Influence of Isothermal CVD and CV Conditions on the Deposition Kinetics and Structure of BN 1193 5E8 040 0.35 r=14x106ky 045 F do 22 6 0 f-7x10- kgs 180 nm 0.40 150200250:00350400 Position from the fiber surface Fig. 12. of the greatest coherent domain and the interlayer Fig. 13. Evolution of the greatest coherent domain and the interlayer nickness. as a function of the mass flow rate for the 6.2 specimen referer taken in the middle of the fibrous preforms( the preform for the protected-ICVI process. Le and La are shadd n in the distance in the bn thickness. as a function of the locatio ess. Le and La are shaded gray and white, respec- f=7×10kg/s) NH3 2N+EH2 K Total supersaturation of the gas phase at the gas/solid inter- face could be defined, as a function of the actual local partial ressures in the gaseous reactants and product, by using the The supersaturation in boron (2B)and nitrogen(2N)can be obtained when the actual partial pressures are compared to the boron and nitrogen partial pressures that are in equilibrium with solid BN at a given temperature(pB and pN) (pHcu'PBd (3) Thermodynamic calculations showed that the yield of reac- tion(1)was always very high at 700oC (99%, with respect to the reactant BCla or NHa, whichever had the lowest partial ressure).30 Therefore, one also could say that the del 阳唱唱 from equilibrium or supersaturation is high (or low) and that of completion of the reactants conversion is high(or low ). In the same manner. supersaturation with respect to the (10) elements of the bn solid can be defined if we assume a local uilibrium state in the supersaturated gas phase at the gas/ ace. which allows determination of the actual boron BCL,+3H,2B+3HCI bEN=K (12) K'PBCI (PH )3/2 where K= KBKN. These relations are obviously also related to he degree of completion of the reactants conversion. With regard to the CVD process and the structure of the

with Kp = ~pHCl eq ! 3 pBCl3 eq pNH3 eq (2) Total supersaturation of the gas phase at the gas/solid inter￾face could be defined, as a function of the actual local partial pressures in the gaseous reactants and product, by using the relation34 St = pBCl3 pNH3 ~pHCl eq ! 3 ~pHCl! 3 pBCl3 eq pNH3 eq (3) Thermodynamic calculations showed that the yield of reac￾tion (1) was always very high at 700°C (>99%, with respect to the reactant BCl3 or NH3, whichever had the lowest partial pressure).30 Therefore, one also could say that the departure from equilibrium or supersaturation is high (or low) and that the degree of completion of the reactants conversion is high (or low). In the same manner, supersaturation with respect to the elements of the BN solid can be defined if we assume a local equilibrium state in the supersaturated gas phase at the gas/ solid interface, which allows determination of the actual boron and nitrogen partial pressures: BCl3 + 3 2 H2 →← B ` 3HCl (4) with pB = K8 p pBCl3 ~pH2 ! 3/2 ~pHCl! 3 (5) and NH3 →← N + 3 2 H2 (6) with pN = K9 p pNH3 ~pH2 ! 3/2 (7) The supersaturation in boron (SB) and nitrogen (SN) can be obtained when the actual partial pressures are compared to the boron and nitrogen partial pressures that are in equilibrium with solid BN at a given temperature (p° B and p° N): SB = pB p° B = KBpBCl3 ~pH2 ! 3/2 ~pHCl! 3 (8) KB = K8 p p° B (9) SN = pN p° N = KNpNH3 ~pH2 ! 3/2 (10) KN = K9 p p° N (11) and SBSN = K pBCl3 pNH3 ~pHCl! 3 (12) where K 4 KBKN. These relations are obviously also related to the degree of completion of the reactants conversion. With regard to the CVD process and the structure of the deposits, high supersaturations and low temperatures are gen￾Fig. 12. Evolution of the greatest coherent domain and the interlayer distance in the BN thickness, as a function of the mass flow rate for the protected-ICVI process. Lc and La are shaded gray and white, respec￾tively. Samples were taken in the middle of the fibrous preforms (the 6.2 specimen reference). Fig. 13. Evolution of the greatest coherent domain and the interlayer distance in the BN thickness, as a function of the location in the preform for the protected-ICVI process. Lc and La are shaded gray and white, respectively (T 4 700°C, P 4 1.33 kPa, NH3/BCl3 4 1, H2/BCl3 4 1, and f 4 7 × 10−6 kg/s). May 1999 Influence of Isothermal CVD and CVI Conditions on the Deposition Kinetics and Structure of BN 1193

1194 Journal of the American Ceramic Society-Leparoux et al. Vol. 82. No. 5 erally believed to lead to poor crystal organization. 34,35 Thus, it CVD deposits. In all cases, any possible influence on the initial is not surprising that all the CVd deposits present a poor or- growth conditions would be hidden by the unfavorable CVD ganization at 700C(Fig. 8). Forced flow toward single fibers, conditions that have been used in this study. However, it seems thin tows, or 2D single cloth(either stationary or moving )car that the structure can be improved by using a high-temperature be used to obtain sufficiently uniform coatings that can be post-treatment. In fact, only poorly organized films exhibit a better organized at higher temperature. 6 The drawbacks of skin effect in the outermost 10-20 nm of the BN coating such a process, in terms of the amount of fibers that could be Although the graphitization of BN from a turbostratic structure coated and processing complexity, are clear only occurs at-2000C, it has been proved that the external With regard to the ICVI process, the supersaturation is al- reorganization was the result of the treatment at 1000oC under ways lower than that with CVD at 700C; however, the gas- vacuum. Lacrambe3 also has shown that moderate tempera- phase chemistry also can be changed by the influence of dif- tures(13000-1700C)lead to better crystallization of BN that ferent parameters. The residence time especially is greater, has been prepared via CVD. However, he attributed this im- because of diffusion inside the preform; the conversion yield provement to the evolution of his samples when previously also is much higher, because of the specific surface of the exposed to a moist environment In our case, the substrates porous preform. To first address the possible influence of the were kept in the reactor under vacuum between the deposition variation of the supersaturation, it is clear that some conditions stage and the thermal post-treatment. However, they do contain lead to high supersaturations and then to organization of oxygen as an impurity, and they exhibit resistance to the ICVI lental conditions that have moisture when they are poorly organized 30 It could then be been studied here. these conditions are classical ICvi with high suggested that the thermal treatment allows some decrease in flow rates and the external and upper positions in the preform he content of some impurities, such as oxygen and hydrogen n contrast, all the deposits that have been elaborated via pro atoms that are incorporated in the depos This phenomenon ts. When they are tected-ICVI exhibit greater coherent domains and smaller in- removed, the BN lattice can rearrange terlayer distances and dispersions, in comparison to similar would be favored by an initially very poor organization, and it tings that have been obtained from the classical ICVi pro- would be limited to the surface of the coatings. The structure of For both ICVI and protected-ICVI processes, an interme- each of the BN sequences that are deposited by the alternate liate flow rate leads to the best organization of the film: in the CVD process seems to be sufficiently organized to prevent case of protected-ICVI, the optimal organization is obtained for such a reorganization. However, additional investigations are a higher mass flow rate than that for classical ICVI(see Figs needed to valid these hypotheses and 12). This observation has been confirmed by the experi- ment that was conducted at a flow rate of 56 x 10 kg/s, which did not drastically debase the organization when protected VI was used, in contrast to the experiment that was con From the results that have been presented in Section Ill and ducted without screens at a flow rate of 28 x 106 kg/s discussed in Section IV, the following conclusions can be All these results show that the BN crystalline organization drawn can be optimized under conditions where the ICVI process D The chemical vapor deposition(CVD) process is gov departs both from a pure mass-transfer control (i.e, from al erned either by a combination of mass transfer with chemica most-total conversion at the gas/solid interface )and obviousl kinetics at low flow rates or by the heterogeneous kinetics from a pure kinetic control (which is obtained with CVd at alone at higher flow rates. In this latter case, the deposition rate h flow rates ). These particular infiltration is dependent only on the lowest partial pressure of the reactants ifficult to discuss, because this result might be attributed to BCIs or NH3), and a very good longitudinal uniformity of the different changes in the deposition mechanism. If the surface deposit thickness is obtained. At a deposition temperature of nechanism of the ICVi process remains unchanged from the the supersaturation is high in any case. For the isother imiting CVD process, the BN organization would be improved mal-isobaric chemical vapor infiltration (ICVI)process, the organization of the ICVI deposits that have been obtained at cause of diffusion of the gaseous phase in the fibrous substrate high flow rates without protection is classical, deterioration of and the great deposition yield(because of its high specific the structure at low flow rates is unusual. The poorest organi- surface area). Therefore, the influence of the concentration of zation of the bn deposits has been obtained at the lowest flow the inlet reactant species on the deposition rate is more com- rates, especially in downstream and inside locations(e.g, with plex. The deposition rate is lower than that observed in the high HCI concentration, which could hamper the BN organi- lower rocess, and, accordingly, the supersaturation is much a high deposition yield). This result might be attributed to a CVD (2) The structure of BN coatings that have been deposited for the heterogeneous reaction on the bulk substrates. On the fr other hand, the residence time of the gaseous species is osition conditions, including the processing type(CVD or hanged through the infiltration itself, the variation of the total ICVI, protected or not by diffusion screens ). The present CVD flow rate, and the use of diffusion screens. The homogeneous conditions lead to poor organization of the BN deposits with a formation of intermediate precursors, which leads to a different large interlayer spacing, in accordance to the high supersatu- surface mechanism that favors or hinders the organization of ration. The ICVI process allows deposition of a fairly well- the BN deposit, cannot be excluded either. Evolution of the gas organized BN coating on all fibers of a fibrous preform under hase with time already has been observed in the deposition of conditions where the supersaturation is intermediate. Poorer pyrocarbon via pulsed CvD, 7 which is a process that allows organization that is obtained at lower supersaturation could be the residence time to be monitored by using the pulse duration due to the high yield of deposition, which leads to important The lattice organization and orientation is improved by inter- HCI formation and very different deposition conditions at the mediate"maturation times"of the gaseous phase, because par- ticular intermediates could form. Therefore other thoroug diate species that are produced by some homogeneous reac- studies would be necessary to understand how the bn or tions cannot be excluded under conditions where the residence nization is improved. The influence of the amount of HCl time also is ve or the formation of intermediate species could especially be necessary to vary different. However, additional studies validate these hypothese examined (3) Good thickness uniformities can be obtained via ICVI The above-mentioned discussion about the influence of the despite the influence of mass transfer. In regard to that point, deposition conditions on the BN structure shows that the nature the gaseous product HCl has no influence on the deposition of the fiber obviously cannot influence the structure of the rate. The use of diffusion screens and the adjustment of the

erally believed to lead to poor crystal organization.34,35 Thus, it is not surprising that all the CVD deposits present a poor or￾ganization at 700°C (Fig. 8). Forced flow toward single fibers, thin tows, or 2D single cloth (either stationary or moving) can be used to obtain sufficiently uniform coatings that can be better organized at higher temperature.36 The drawbacks of such a process, in terms of the amount of fibers that could be coated and processing complexity, are clear. With regard to the ICVI process, the supersaturation is al￾ways lower than that with CVD at 700°C; however, the gas￾phase chemistry also can be changed by the influence of dif￾ferent parameters. The residence time especially is greater, because of diffusion inside the preform; the conversion yield also is much higher, because of the specific surface of the porous preform. To first address the possible influence of the variation of the supersaturation, it is clear that some conditions lead to high supersaturations and then to poor organization of the ICVI deposits (for the experimental conditions that have been studied here, these conditions are classical ICVI with high flow rates and the external and upper positions in the preform). In contrast, all the deposits that have been elaborated via pro￾tected-ICVI exhibit greater coherent domains and smaller in￾terlayer distances and dispersions, in comparison to similar coatings that have been obtained from the classical ICVI pro￾cess. For both ICVI and protected-ICVI processes, an interme￾diate flow rate leads to the best organization of the film: in the case of protected-ICVI, the optimal organization is obtained for a higher mass flow rate than that for classical ICVI (see Figs. 9 and 12). This observation has been confirmed by the experi￾ment that was conducted at a flow rate of 56 × 10−6 kg/s, which did not drastically debase the organization when protected￾ICVI was used, in contrast to the experiment that was con￾ducted without screens at a flow rate of 28 × 10−6 kg/s. All these results show that the BN crystalline organization can be optimized under conditions where the ICVI process departs both from a pure mass-transfer control (i.e., from al￾most-total conversion at the gas/solid interface) and obviously from a pure kinetic control (which is obtained with CVD at high flow rates). These particular infiltration conditions are difficult to discuss, because this result might be attributed to different changes in the deposition mechanism. If the surface mechanism of the ICVI process remains unchanged from the limiting CVD process, the BN organization would be improved normally first at lower supersaturations. Although the poor organization of the ICVI deposits that have been obtained at high flow rates without protection is classical, deterioration of the structure at low flow rates is unusual. The poorest organi￾zation of the BN deposits has been obtained at the lowest flow rates, especially in downstream and inside locations (e.g., with a high deposition yield). This result might be attributed to a high HCl concentration, which could hamper the BN organi￾zation during growth, even if HCl does not act as an inhibitor for the heterogeneous reaction on the bulk substrates. On the other hand, the residence time of the gaseous species is changed through the infiltration itself, the variation of the total flow rate, and the use of diffusion screens. The homogeneous formation of intermediate precursors, which leads to a different surface mechanism that favors or hinders the organization of the BN deposit, cannot be excluded either. Evolution of the gas phase with time already has been observed in the deposition of pyrocarbon via pulsed CVD,37 which is a process that allows the residence time to be monitored by using the pulse duration. The lattice organization and orientation is improved by inter￾mediate “maturation times” of the gaseous phase, because par￾ticular intermediates could form. Therefore, other thorough studies would be necessary to understand how the BN orga￾nization is improved. The influence of the amount of HCl or the formation of intermediate species could especially be examined. The above-mentioned discussion about the influence of the deposition conditions on the BN structure shows that the nature of the fiber obviously cannot influence the structure of the CVD deposits. In all cases, any possible influence on the initial growth conditions would be hidden by the unfavorable CVD conditions that have been used in this study. However, it seems that the structure can be improved by using a high-temperature post-treatment. In fact, only poorly organized films exhibit a skin effect in the outermost 10–20 nm of the BN coating. Although the graphitization of BN from a turbostratic structure only occurs at ∼2000°C, it has been proved that the external reorganization was the result of the treatment at 1000°C under vacuum. Lacrambe38 also has shown that moderate tempera￾tures (1300°–1700°C) lead to better crystallization of BN that has been prepared via CVD. However, he attributed this im￾provement to the evolution of his samples when previously exposed to a moist environment. In our case, the substrates were kept in the reactor under vacuum between the deposition stage and the thermal post-treatment. However, they do contain oxygen as an impurity,39 and they exhibit poor resistance to moisture when they are poorly organized.30 It could then be suggested that the thermal treatment allows some decrease in the content of some impurities, such as oxygen and hydrogen atoms that are incorporated in the deposits. When they are removed, the BN lattice can rearrange. This phenomenon would be favored by an initially very poor organization, and it would be limited to the surface of the coatings. The structure of each of the BN sequences that are deposited by the alternate CVD process seems to be sufficiently organized to prevent such a reorganization. However, additional investigations are needed to valid these hypotheses. V. Conclusions From the results that have been presented in Section III and discussed in Section IV, the following conclusions can be drawn: (1) The chemical vapor deposition (CVD) process is gov￾erned either by a combination of mass transfer with chemical kinetics at low flow rates or by the heterogeneous kinetics alone at higher flow rates. In this latter case, the deposition rate is dependent only on the lowest partial pressure of the reactants (BCl3 or NH3), and a very good longitudinal uniformity of the deposit thickness is obtained. At a deposition temperature of 700°C, the supersaturation is high in any case. For the isother￾mal–isobaric chemical vapor infiltration (ICVI) process, the limiting contribution of mass transfer is always observed, be￾cause of diffusion of the gaseous phase in the fibrous substrate and the great deposition yield (because of its high specific surface area). Therefore, the influence of the concentration of the inlet reactant species on the deposition rate is more com￾plex. The deposition rate is lower than that observed in the CVD process, and, accordingly, the supersaturation is much lower. (2) The structure of BN coatings that have been deposited from BCl3–NH3–H2 mixtures is highly dependent on the de￾position conditions, including the processing type (CVD or ICVI, protected or not by diffusion screens). The present CVD conditions lead to poor organization of the BN deposits with a large interlayer spacing, in accordance to the high supersatu￾ration. The ICVI process allows deposition of a fairly well￾organized BN coating on all fibers of a fibrous preform under conditions where the supersaturation is intermediate. Poorer organization that is obtained at lower supersaturation could be due to the high yield of deposition, which leads to important HCl formation and very different deposition conditions at the gas/solid interface. The influence of the formation of interme￾diate species that are produced by some homogeneous reac￾tions cannot be excluded under conditions where the residence time also is very different. However, additional studies are necessary to validate these hypotheses. (3) Good thickness uniformities can be obtained via ICVI, despite the influence of mass transfer. In regard to that point, the gaseous product HCl has no influence on the deposition rate. The use of diffusion screens and the adjustment of the 1194 Journal of the American Ceramic Society—Leparoux et al. Vol. 82, No. 5

May 1999 Influence of lsothermal CVD and CV Conditions on the Deposition Kinetics and Structure of BA 1195 let composition and the mass flow rate seem to be very im- ress in Adanced Materials and Mechanics. Edited by T. Wang and T.W t to obtain both the best uniformity and the best organ The use of diffusion screens especially could be gen- IR. A Lowden. K harz, and N. L d to other systems, to improve the control of the depo re C (4) A thermal post-treatment at nder vacuum in Chemical Vapor Co-deposition of C and Si, Ci-n"J Phys. 11, 3, 549-5 proves the organization of bn that he posited via CVd 20T. Matsuda, "Stability to Moisture for Chemically Vapour-Deposited Bo- 3-58(1989 only. However, the 10-20 nm of the layer, and it seems to ally very Strength of a Silicon Carbide/Silicon Nitride Composite, Mater. Res. Soc poor organization and the release of impurities 型PC:250.239492 is. and S.L. Suib rength of CVI SiC/BN/SiC Comp Acknowledgments: The authors acknowledge R Herbin, for her help ran.Eng. Sci. P 2R. Naslain, O. Dugne, A Guette, J. Sevely, C. Robin Brosse, J.-P. Roche nical contribution. The authors also are indebted to S. Goujard and C and J. Cotteret, ""Boron Nitride Robin-Brosse(SEP) for the supply of materials and fruitful discussions. " N. Ricca, A. Guette, G. Camus, and J M. Jouin, ""SiC (ex-PCS) MAS Composites with a BN Interphase: Microstructure, Mechanical Properties and see Ref. 9, pp. 455-6 SS. Prouhet, G. Camus, C. Labrugere, A, Guette, and E. Martin, "Mechanical England. 1971 interphase ?ation of Si-C(o) Fiber/Sic (CVD) Matrix Composites with a BN- in Brittle Composites w Manth: "Models of Fiber Debonding and Pull-out Chemistry of SiC/BN Dual-Coated Nicalon-Fiber-Reinforced Glass-Ceramic Mech.Maer,9,139-63(199 3A. G. Evans and D B The Mechanical Behavior of Ceramic Khin, J 2567-83(1989) Matrix composites,". Am. Ceram. Soc. 7 (5)1329-39(199A\ e, Ak Sevely, J-P. Rocher, and J. Cotteret, "Interface Characterization by TE ES R. W. Davidge."Fibre- Ceramics, "Composites(Guildford UK and SIMS in Tough SiC (ex-PCS) Fibre-SiC (CVI) Matrix Composites with a J.J. Brennan, "Interfacial Chemistry and Bonding in Fiber-Reinforced Glass -Vapor-Infiltrated Ceramic and Glass-Ceramic Matrix Composites, "Mater. Sci. Res, 21, 387-99(1986) Mater.sc.Eng.A,A126,203-23(1990 of Fibre- Matri 29B. A. Bender and T L Jessen, "A Comparison of the Interphase Develop Composites. An ent and Mechanical Properties of Nicalon and Tyranno SiC Fiber-Reinforced Electron Microscopy Study, J. Mater. Sci., 22, 3148-60(1987) The Netherlands). Edited by aux obtenus, Ph D. Thesis. Universite d orleans, Orleans, France. 995 rfaces in Ceramic Matrix the 13th International Conference on Chemical C. Vix-Guterl, J. Lahaye, and P. Ehrburger, " Reactivity in Oxygen of a chemical Society Proceedings, Vol. 96-5. Edited by T M. Best The rmostrustuera ciep t r, pa a2is-32 an ingh ne rature neram wood. Allendorf, McD. Robinson, and R. K. Ulrich. Electrochemical shing Limited, Abington, Cambridge, England, 1993 mical va Nitride Interphase in Ceramic Fiber Preforms: Discussion of Some Aspects of Metal Matrix Composites, "J Mater. Sci., 23, 311-28(1988 the Fundamentals of the Isothermal Chemical Vapor Infiltration Process, " J. M. P Bacos, Carbon-Carbon Composites: Oxidation Behavior and Coat- J. Phys.,3,1895903(1993) Patibandla and K L. Luthra, "Chemical Vapor Deposition of Boron 2s. Goujard, L. Vandenbulcke, and H. Tawil, "Oxidation Behavior of 2D Polylayer Coatings, "Thin Solid Films, 252, 120-30(1994) Vandenbulcke and G. Vuillard, "Structure of Deposits-Process Rela tionships in the Chemical Vapor Deposition of Boron, "J. Electrochem. Soc. 3A. W. More, M. B, Dowell, E R. Stover, and L D. Bentsen, " Oxidation J. M. Blocher Jr,""Structure/Property/Process Relations in Chemical Vapor Deposition CVD, J. Vac. Sci. Technol. 11, 680-86(1974) Stinton, T M. Bes- Ceramic Matrix Composites, Mater. Res. Soc. Symp. Proc., 120, 157-62 nann, and O J. Schwarz, "Nextel/SiC Composites Fabricated Using Forced L. Heraud, R. Naslain, and J. M. Quenisset, "Procede de Fabrication d't P. Dupel, R Pailler, and X. Bourrat, "Pulse-CVD and Infiltration of Pyro- Matrice Ceramique a Tenacite Amelioree, "Fr. Pat. No arbon in Model Pores with Rectangular Cross-Sections; Part Il, Study of the Infiltration, J. Mater. Sci., 29, 1056-66(1994)- (a)C. Droillard, "Elaboration et Caracterisation de Composites a Matrice 3G. Lacrambe, "Fabrication et "Graphitation'du Nitrure de Bore Obtenu par SiC et a Sequence C/SiC", Ph D. Thesis No 913. Bordeaux Uni se Temperature", Ph D. Thesis N eparoux, L. Vandenbulcke, V. Serin, and J. Sevely, " The Interphase terphases, "Mater. Res. Soc. Symp. Proc., 365, 371-76(1995 and Interface Microstructure and Chemistry of Isothermal/isobar Vandenbulcke and S. Goujard, "Multilayer Systems Based on B, B.C, Vapour Infiltration SiC/BN/SiC Composites: TEM and Electron Energy Loss SiC and SiBC for Environmental Composite Protection" pp. 1198-204 in Prog 595-602(1

inlet composition and the mass flow rate seem to be very im￾portant to obtain both the best uniformity and the best organi￾zation. The use of diffusion screens especially could be gen￾eralized to other systems, to improve the control of the deposit structure. (4) A thermal post-treatment at 1000°C under vacuum im￾proves the organization of BN that has been deposited via CVD only. However, the reorganization is limited to the outermost 10–20 nm of the layer, and it seems to require an initially very poor organization and the release of impurities. Acknowledgments: The authors acknowledge R. Herbin, for her help in the preparation of the samples and the manuscript, and Y. Boussant, for his technical contribution. The authors also are indebted to S. Goujard and C. Robin-Brosse (SEP) for the supply of materials and fruitful discussions. References 1 J. Aveston, G. Cooper, and A. Kelly, “Single and Multiple Fracture”; pp. 15–26 in Properties of Fibre Composites, Proceedings of the Conference of the National Physics Laboratory. IPC Science Technology Press, Ltd., Guildford, England, 1971. 2 J. W. Hutchinson and H. Jensen, “Models of Fiber Debonding and Pull-out in Brittle Composites with Friction,” Mech. Mater., 9, 139–63 (1990). 3 A. G. Evans and D. B. Marshall, “The Mechanical Behavior of Ceramic Matrix Composites,” Acta Metall., 37, 2567–83 (1989). 4 R. W. Davidge, “Fibre-Reinforced Ceramics,” Composites (Guildford UK), 18, 92–98 (1987). 5 J. J. Brennan, “Interfacial Chemistry and Bonding in Fiber-Reinforced Glass and Glass-Ceramic Matrix Composites,” Mater. Sci. Res., 21, 387–99 (1986). 6 R. F. Cooper and K. Chyung, “Structure and Chemistry of Fibre-Matrix Interfaces in Silicon Carbide Fibre-Reinforced Glass-Ceramic Composites: An Electron Microscopy Study,” J. Mater. Sci., 22, 3148–60 (1987). 7 J. M. Jouin, J. Cotteret, and F. Christin, “SiC–SiC Interphases: Case His￾tory”; pp. 191–204 in Proceedings of the 2nd European Colloquium (Designing Ceramic Interfaces) (Nov. 11–13, 1991, Petten, The Netherlands). Edited by S. D. Peteves. CEC, Petten, The Netherlands, 1993. 8 R. Naslain, “Fibre–Matrix Interphases and Interfaces in Ceramic Matrix Composites Processed by CVI,” Compos. Interfaces, 1, 253–86 (1993). 9 C. Vix-Guterl, J. Lahaye, and P. Ehrburger, “Reactivity in Oxygen of a Thermostructural Composite”; pp. 725–32 in High Temperature Ceramic Ma￾trix Composites. Edited by R. Naslain, J. Lamon, and D. Doumeingts. Wood￾head Publishing Limited, Abington, Cambridge, England, 1993. 10D. H. Grande, “Fibre–Matrix Bond Strength Studies of Glass, Ceramic, and Metal Matrix Composites,” J. Mater. Sci., 23, 311–28 (1988). 11M. P. Bacos, “Carbon–Carbon Composites: Oxidation Behavior and Coat￾ings Protection,” J. Phys. IV, 3, 1895–903 (1993). 12S. Goujard, L. Vandenbulcke, and H. Tawil, “Oxidation Behavior of 2D and 3D Carbon/Carbon Thermostructural Materials Protected by CVD Polylayer Coatings,” Thin Solid Films, 252, 120–30 (1994). 13A. W. More, M. B. Dowell, E. R. Stover, and L. D. Bentsen, “Oxidation Protection of Carbon–Carbon Composites by (B+Si)N Coatings,” Ceram. Eng. Sci. Proc., 16, 263–70 (1995). 14R. J. Diefendorf and R. R. Boisvert, “Processing of Polymeric Precursor Ceramic Matrix Composites,” Mater. Res. Soc. Symp. Proc., 120, 157–62 (1988). 15L. He´raud, R. Naslain, and J. M. Quenisset, “Proce´de´ de Fabrication d’un Mate´riau Composite a` Matrice Ce´ramique a` Te´nacite´ Ame´liore´e,” Fr. Pat. No. 89.02718, March 2, 1989. 16(a) C. Droillard, “Elaboration et Caracte´risation de Composites a` Matrice SiC et a` Interphase Se´quence´e C/SiC”; Ph.D. Thesis No. 913. Bordeaux Uni￾versity, Talence, France, 1993. (b) C. Droillard, J. Lamon, and X. Bourrat, “Strong Interface Bonding in CMCs, Conditions for Efficient Multilayered In￾terphases,” Mater. Res. Soc. Symp. Proc., 365, 371–76 (1995). 17L. Vandenbulcke and S. Goujard, “Multilayer Systems Based on B, B4C, SiC and SiBC for Environmental Composite Protection”; pp. 1198–204 in Prog￾ress in Advanced Materials and Mechanics. Edited by T. Wang and T.-W. Chou. Peking University Press, Beijing, China, 1996. 18R. A. Lowden, K. L. More, O. J. Schwarz, and N. L. Vaughn, “Improved Fiber–Matrix Interlayers for Nicalon/SiC Composites”; see Ref. 9, pp. 345–52. 19J. M. Agullo, F. Maury, and J. M. Jouin, “Mechanical Properties of SiC/SiC Composites with a Treatment of the Fiber/Matrix Interfaces by Metal-organic Chemical Vapor Co-deposition of C and SixC1−x,” J. Phys. IV, 3, 549–56 (1993). 20T. Matsuda, “Stability to Moisture for Chemically Vapour-Deposited Bo￾ron Nitride,” J. Mater. Sci., 24, 2353–58 (1989). 21K. P. Norton and H. H. Streckert, “Effect of BN Interfacial Coating on the Strength of a Silicon Carbide/Silicon Nitride Composite,” Mater. Res. Soc. Symp. Proc., 250, 239–44 (1992). 22M. A. Kmetz, J. M. Laliberte, W. S. Willis, and S. L. Suib, “Synthesis, Characterization, and Tensile Strength of CVI SiC/BN/SiC Composites,” Ce￾ram. Eng. Sci. Proc., 12, 2161–74 (1991). 23R. Naslain, O. Dugne, A. Guette, J. Sevely, C. Robin Brosse, J.-P. Rocher, and J. Cotteret, “Boron Nitride Interphase in Ceramic-Matrix Composites,” J. Am. Ceram. Soc., 74 [10] 2482–88 (1991). 24N. Ricca, A. Guette, G. Camus, and J. M. Jouin, “SiC (ex-PCS) MAS Composites with a BN Interphase: Microstructure, Mechanical Properties and Oxidation Resistance”; see Ref. 9, pp. 455–62. 25S. Prouhet, G. Camus, C. Labruge`re, A. Guette, and E. Martin, “Mechanical Characterization of Si-C(O) Fiber/SiC (CVI) Matrix Composites with a BN￾Interphase,” J. Am. Ceram. Soc., 77 [3] 649–56 (1994). 26E. Y. Sun, S. R. Nutt, and J. J. Brennan, “Interfacial Microstructure and Chemistry of SiC/BN Dual-Coated Nicalon-Fiber-Reinforced Glass-Ceramic Matrix Composites,” J. Am. Ceram. Soc., 77 [5] 1329–39 (1994). 27O. Dugne, S. Prouhet, A. Guette, R. Naslain, R. Fourmeaux, Y. Khin, J. Sevely, J.-P. Rocher, and J. Cotteret, “Interface Characterization by TEM, AES, and SIMS in Tough SiC (ex-PCS) Fibre–SiC (CVI) Matrix Composites with a BN Interphase,” J. Mater. Sci., 28, 3409–22 (1993). 28J. J. Brennan, “Interfacial Studies of Chemical-Vapor-Infiltrated Ceramic Matrix Composites,” Mater. Sci. Eng. A, A126, 203–23 (1990). 29B. A. Bender and T. L. Jessen, “A Comparison of the Interphase Develop￾ment and Mechanical Properties of Nicalon and Tyranno SiC Fiber-Reinforced ZrTiO4 Matrix Composites,” J. Mater. Res., 9, 2670–76 (1994). 30M. Leparoux, “Elaboration a` Partir de la Phase Gazeuse des Interphases BN et BNSi pour Composites SiC/BN(Si)/SiC, Structure et Proprie´te´s des Mate´ri￾aux Obtenus”; Ph.D. Thesis. Universite´ d’Orle´ans, Orleans, France, 1995. 31M. Leparoux and L. Vandenbulcke, “Influence of the CVD and ICVI Pro￾cesses on the Boron Nitride Deposition Kinetics”; pp. 594–98 in Proceedings of the 13th International Conference on Chemical Vapor Deposition, Electro￾chemical Society Proceedings, Vol. 96-5. Edited by T. M. Besmann, M. D. Allendorf, McD. Robinson, and R. K. Ulrich. Electrochemical Society, Pen￾nington, NJ, 1996. 32V. Cholet and L. Vandenbulcke, “Chemical Vapor Infiltration of Boron Nitride Interphase in Ceramic Fiber Preforms: Discussion of Some Aspects of the Fundamentals of the Isothermal Chemical Vapor Infiltration Process,” J. Am. Ceram. Soc., 76 [11] 2846–58 (1993). 33N. Patibandla and K. L. Luthra, “Chemical Vapor Deposition of Boron Nitride,” J. Electrochem. Soc., 139, 3558–65 (1992). 34L. Vandenbulcke and G. Vuillard, “Structure of Deposits–Process Rela￾tionships in the Chemical Vapor Deposition of Boron,” J. Electrochem. Soc., 124, 1937–42 (1977). 35J. M. Blocher Jr., “Structure/Property/Process Relations in Chemical Vapor Deposition CVD,” J. Vac. Sci. Technol. 11, 680–86 (1974). 36B. L. Weaver, R. A. Lowden, J. C. McLaughlin, D. P. Stinton, T. M. Bes￾mann, and O. J. Schwarz, “Nextel/SiC Composites Fabricated Using Forced Chemical Vapor Infiltration”; see Ref. 9, pp. 353–60. 37P. Dupel, R. Pailler, and X. Bourrat, “Pulse-CVD and Infiltration of Pyro￾carbon in Model Pores with Rectangular Cross-Sections: Part II, Study of the Infiltration,” J. Mater. Sci., 29, 1056–66 (1994). 38G. Lacrambe, “Fabrication et ‘Graphitation’ du Nitrure de Bore Obtenu par De´poˆt Chimique en Phase Vapeur a` Basse Tempe´rature”; Ph.D. Thesis No. 2232. Bordeaux University, Talence, France, 1988. 39M. Leparoux, L. Vandenbulcke, V. Serin, and J. Sevely, “The Interphase and Interface Microstructure and Chemistry of Isothermal/Isobaric Chemical Vapour Infiltration SiC/BN/SiC Composites: TEM and Electron Energy Loss Studies,” J. Mater. Sci., 32, 4595–602 (1997). h May 1999 Influence of Isothermal CVD and CVI Conditions on the Deposition Kinetics and Structure of BN 1195