6 INDUSTRIAL CARBON CHEMICAL VAPOR INFILTRATION (CVI) PROCESSES I.Golecki 1 Introduction Carbon-Carbon(C-C)fiber-matrix composite materials-C-Cs(Buckley and Edie,1993; Savage,1993;Thomas,1993)possess several extraordinary sets of properties.Foremost,C-Cs have densities in the range of 1.6-2.2 gcm3,much lower than those of most metals and ceramics.Lower densities can be translated into lower component weights,an important con- sideration,especially for applications in flying platforms.C-Cs also have excellent mechan- ical and thermal properties.The mechanical strengths of C-Cs increase with temperature,in contrast to the strengths of the majority of other fiber-matrix composites and those of metals and ceramics,which decrease with increasing temperature.C-Cs evidence high toughness and graceful failure under load,as do some other fiber-matrix composites,in con- trast to the more abrupt brittle behavior of most monolithic materials.No melting occurs in C-Cs at increasing temperatures,although carbon evaporates above=2,500C.Appropriately processed,pitch-fiber based C-C's possess higher thermal conductivities than copper and sil- ver and they exhibit,by far,the highest thermal conductivity per unit density among thermal management materials,e.g.400(Wm-k/(g cm-3)(Golecki et al.,1998).In addition,the thermal expansion coefficient of C-Cs in the fiber direction at 20C to =1,000C is between -1 and 1 ppm/C,much lower than that of most other structural materials.Consequently, thermal stresses in C-C articles are,in principle,also lower.All of these properties of C-Cs can be tailored by design,using different fibers,matrices,processing methods and schedules. While C-C articles can be used from cryogenic temperatures to over 3,000C,and in severe and chemically aggressive environments,they require protection from oxidation for continu- ous use above =350C.The practical effect of oxidation is to diminish the structural and functional integrity of the article.Effective engineering solutions for protecting C-C articles from oxidation have been demonstrated for different types of articles and different tempera- ture-time envelopes (Buckley and Edie,1993;Savage,1993;Sheehan et al.,1994;Golecki et al.,2000).Current and potential future applications of C-Cs include disc brake pads for commercial and military airplanes and racing cars,uncooled engine parts,rocket leading- edge sections,furnace heating elements,heat sinks for high-power electronics,heat exchang- ers,and bipolar plates for proton-exchange-membrane fuel cells. Carbon-Carbon articles generally comprise(a)continuous or chopped carbon fibers pro- duced from rayon,polyacrilonitrile(PAN)or pitch precursors and(b)one or more types of ©2003 Taylor&Francis
6 INDUSTRIAL CARBON CHEMICAL VAPOR INFILTRATION (CVI) PROCESSES I. Golecki 1 Introduction Carbon–Carbon (C–C) fiber-matrix composite materials – C–Cs (Buckley and Edie, 1993; Savage, 1993; Thomas, 1993) possess several extraordinary sets of properties. Foremost, C–Cs have densities in the range of 1.6–2.2 g cm3 , much lower than those of most metals and ceramics. Lower densities can be translated into lower component weights, an important consideration, especially for applications in flying platforms. C–Cs also have excellent mechanical and thermal properties. The mechanical strengths of C–Cs increase with temperature, in contrast to the strengths of the majority of other fiber–matrix composites and those of metals and ceramics, which decrease with increasing temperature. C–Cs evidence high toughness and graceful failure under load, as do some other fiber-matrix composites, in contrast to the more abrupt brittle behavior of most monolithic materials. No melting occurs in C–Cs at increasing temperatures, although carbon evaporates above ≈2,500 C. Appropriately processed, pitch-fiber based C–C’s possess higher thermal conductivities than copper and silver and they exhibit, by far, the highest thermal conductivity per unit density among thermal management materials, e.g. 400 (Wm1 k1 )/(g cm3 ) (Golecki et al., 1998). In addition, the thermal expansion coefficient of C–Cs in the fiber direction at 20 C to ≈1,000 C is between 1 and 1 ppm/C, much lower than that of most other structural materials. Consequently, thermal stresses in C–C articles are, in principle, also lower. All of these properties of C–Cs can be tailored by design, using different fibers, matrices, processing methods and schedules. While C–C articles can be used from cryogenic temperatures to over 3,000 C, and in severe and chemically aggressive environments, they require protection from oxidation for continuous use above ≈350 C. The practical effect of oxidation is to diminish the structural and functional integrity of the article. Effective engineering solutions for protecting C–C articles from oxidation have been demonstrated for different types of articles and different temperature-time envelopes (Buckley and Edie, 1993; Savage, 1993; Sheehan et al., 1994; Golecki et al., 2000). Current and potential future applications of C–Cs include disc brake pads for commercial and military airplanes and racing cars, uncooled engine parts, rocket leadingedge sections, furnace heating elements, heat sinks for high-power electronics, heat exchangers, and bipolar plates for proton-exchange-membrane fuel cells. Carbon–Carbon articles generally comprise (a) continuous or chopped carbon fibers produced from rayon, polyacrilonitrile (PAN) or pitch precursors and (b) one or more types of © 2003 Taylor & Francis
carbon matrices(Savage,1993).One of the most common fabrication methods of C-C arti- cles is densification of a porous preform having the desired shape,so-called near-net pro- cessing.The preform consists only or principally of fibers.The initial geometrical density of such preforms varies in the range 10-80%of the theoretical value at full density. Preforms may be fabricated using e.g.weaving of continuous fibers,lay-up of fibrous mats or fabrics,needle-punching,or mixing of chopped fibers with resins and powders,followed, as needed,by thermal treatment in the 200-1,000C range to evaporate organic binders or residues(Savage,1993).Fabricating,cutting and machining a porous preform is easier and faster than machining a fully-dense C-C bar,which often requires diamond tooling.The densification of the preform can be achieved either from the vapor phase,conventionally by means of isothermal isobaric chemical vapor deposition and infiltration(CVD/CVD),or by liquid-phase resin infiltration and annealing,or by a combination of these two approaches. Both approaches involve heating the preforms to about 1,000C,either during CVI or after resin has been placed inside the preforms.Conventionally,in either densification approach, several steps are required to effect sufficient densification of the preforms.Minimum final density values are necessary for achieving the desired mechanical and thermal properties. All densification methods leave typically 1-10%of voids in the composite.The average density of a composite article,Pav,is equal to: EpiXi=PrX+pmiXml +pm2Xm2+ (1) where the X;denote the respective volume fractions (X;=1-X)and the subscripts f,mi, and v denote fiber,matrix no.i,and void,respectively.After densification is complete,the C-C article may just need to be lightly machined to final tolerances.Additional processing steps involved in the manufacture of C-C articles often include high-temperature (1,000-3,000C)annealing in an inert ambient,called graphitization,in order to achieve desired properties,and oxidation protection treatment. In this chapter,a brief review is provided of CVI routes to the densification of C-C's, from an industrial perspective,with emphasis placed on published approaches demonstrated to reduce the processing time of functional components or that show potential for the same; a more extensive review of this subject was published recently (Golecki,1997). 2 Overview of carbon CVI CVI and CVD (Pierson,1992;Hitchman and Jensen,1993)involve flowing one or several streams of precursor vapors containing the desired element or compound,e.g.methane (CH4),over and around the porous parts,while keeping these parts in a furnace at a suffi- ciently high temperature to decompose the precursor.Each flow rate is usually controlled by means of an electronic mass flow controller and the total pressure in the furnace can be con- trolled independently by means of a throttle valve,which varies the flow conductance to the vacuum pumps.The terms CVD and CVI are often used interchangeably,but strictly speak- ing,they denote different processes.The purpose of CVD is to deposit a functional,thin coating on a dense substrate;e.g.the coating may serve as the active part of an electronic device or as a protective layer.The coating produced by CVD generally adds less than 1% of weight to the substrate and the deposition time is of the order of a few minutes to a few hours.The primary purpose of CVL,on the other hand,is to increase the density of a porous body by 100-900%,in order to obtain a material with desirable properties.In addition to achieving the desired density,a specific carbon matrix microstructure may be required, ©2003 Taylor&Francis
carbon matrices (Savage, 1993). One of the most common fabrication methods of C–C articles is densification of a porous preform having the desired shape, so-called near-net processing. The preform consists only or principally of fibers. The initial geometrical density of such preforms varies in the range 10–80% of the theoretical value at full density. Preforms may be fabricated using e.g. weaving of continuous fibers, lay-up of fibrous mats or fabrics, needle-punching, or mixing of chopped fibers with resins and powders, followed, as needed, by thermal treatment in the 200–1,000 C range to evaporate organic binders or residues (Savage, 1993). Fabricating, cutting and machining a porous preform is easier and faster than machining a fully-dense C–C bar, which often requires diamond tooling. The densification of the preform can be achieved either from the vapor phase, conventionally by means of isothermal isobaric chemical vapor deposition and infiltration (CVD/CVI), or by liquid-phase resin infiltration and annealing, or by a combination of these two approaches. Both approaches involve heating the preforms to about 1,000 C, either during CVI or after resin has been placed inside the preforms. Conventionally, in either densification approach, several steps are required to effect sufficient densification of the preforms. Minimum final density values are necessary for achieving the desired mechanical and thermal properties. All densification methods leave typically 1–10% of voids in the composite. The average density of a composite article, av, is equal to: iXi fXf m1Xm1 m2Xm2 … (1) where the Xi denote the respective volume fractions (Xi 1 Xv) and the subscripts f, mi, and v denote fiber, matrix no. i, and void, respectively. After densification is complete, the C–C article may just need to be lightly machined to final tolerances. Additional processing steps involved in the manufacture of C–C articles often include high-temperature (1,000–3,000 C) annealing in an inert ambient, called graphitization, in order to achieve desired properties, and oxidation protection treatment. In this chapter, a brief review is provided of CVI routes to the densification of C–C’s, from an industrial perspective, with emphasis placed on published approaches demonstrated to reduce the processing time of functional components or that show potential for the same; a more extensive review of this subject was published recently (Golecki, 1997). 2 Overview of carbon CVI CVI and CVD (Pierson, 1992; Hitchman and Jensen, 1993) involve flowing one or several streams of precursor vapors containing the desired element or compound, e.g. methane (CH4), over and around the porous parts, while keeping these parts in a furnace at a sufficiently high temperature to decompose the precursor. Each flow rate is usually controlled by means of an electronic mass flow controller and the total pressure in the furnace can be controlled independently by means of a throttle valve, which varies the flow conductance to the vacuum pumps. The terms CVD and CVI are often used interchangeably, but strictly speaking, they denote different processes. The purpose of CVD is to deposit a functional, thin coating on a dense substrate; e.g. the coating may serve as the active part of an electronic device or as a protective layer. The coating produced by CVD generally adds less than 1% of weight to the substrate and the deposition time is of the order of a few minutes to a few hours. The primary purpose of CVI, on the other hand, is to increase the density of a porous body by 100–900%, in order to obtain a material with desirable properties. In addition to achieving the desired density, a specific carbon matrix microstructure may be required, © 2003 Taylor & Francis
which is a function of the CVI process conditions.The infiltration time in conventional CVI methods is therefore much longer than the deposition time in CVD. The most important quantity describing a CVD process is the deposition rate of the coat- ing,r,which may vary from =10-2-103umh.The most important parameter influencing the deposition rate for a given material system is the substrate temperature.The deposition rate can be expressed (Grove,1967)as: r=Ce[kshg/(ks hg)l/Ns, (2) where C is the chemical concentration in the gas phase,ks is the rate constant for hetero- geneous decomposition of the chemical into the film on the surface of the substrate,hg is the gas-phase mass-transfer coefficient of the chemical to the substrate,and Ns is a normal- izing constant.This chemical may be,but often is not the input precursor (e.g.CH4)intro- duced into the reactor.Often a complex series of chemical reactions lead from the input gas to the solid film.The gas-phase mass-transfer coefficient can be expressed as h=D/dp, where D is the gas-phase diffusivity and do is the thickness of the boundary layer.When kshg,r=kC INs and the process is surface-reaction controlled.Under these conditions, the reaction rate constant ks and hence the deposition rate r usually increase exponentially with temperature according to the Arrhenius law: ks=ksoexp(-E/kT),r=(ksoCg/N3)p"exp(-E/kT), (3) where Ea is the activation energy for the controlling surface reaction,T is the absolute tem- perature in K,and Boltzmann's constant k=1.3805X 10-16 ergK-1=8.614X 10-5eVK-1 (see Fig.6.1).For carbon,E2-4eV/molecule.Pressure signifies the pressure in the reactor chamber. At higher temperature,when khr=hC/N,and the process is gas-phase diffusion controlled.Under these conditions,transport of the precursor through the gas phase to the substrate becomes the limiting factor,while the surface chemical kinetics are relatively more rapid.In this regime,the deposition rate increases with temperature much more slowly,as r=ATb,where b=0.5-1,due to the temperature dependence of the gas-phase diffusivity. At still higher temperature,the deposition rate decreases with increasing temperature,due to competing reactions,such as homogeneous gas-phase nucleation or powder formation and sometimes etching of the surface of the film.Generally,lower temperatures,lower pres- sures,increased dilution,and higher flow rates(i.e.milder CVD conditions)minimize unde- sirable processes at the expense of growth rate.The choice of precursor may also influence the deposition rate and the properties of the deposited coating(Tesner,1984;Duan and Don, 1995;Huttinger,2001). Gas-phaseGas-phase!Surface nucleation mass chemical transport kinetics 1T(K-) Figure 6.I Three temperature regimes in chemical vapor deposition. ©2003 Taylor&Francis
which is a function of the CVI process conditions. The infiltration time in conventional CVI methods is therefore much longer than the deposition time in CVD. The most important quantity describing a CVD process is the deposition rate of the coating, r, which may vary from ≈102 –103mh1 . The most important parameter influencing the deposition rate for a given material system is the substrate temperature. The deposition rate can be expressed (Grove, 1967) as: r Cg[kshg/(ks hg)]/Ns, (2) where Cg is the chemical concentration in the gas phase, ks is the rate constant for heterogeneous decomposition of the chemical into the film on the surface of the substrate, hg is the gas-phase mass-transfer coefficient of the chemical to the substrate, and Ns is a normalizing constant. This chemical may be, but often is not the input precursor (e.g. CH4) introduced into the reactor. Often a complex series of chemical reactions lead from the input gas to the solid film. The gas-phase mass-transfer coefficient can be expressed as hg D/db, where D is the gas-phase diffusivity and db is the thickness of the boundary layer. When ks > hg, r ≈ hgCg/Ns and the process is gas-phase diffusion controlled. Under these conditions, transport of the precursor through the gas phase to the substrate becomes the limiting factor, while the surface chemical kinetics are relatively more rapid. In this regime, the deposition rate increases with temperature much more slowly, as r ATb , where b ≈ 0.5 1, due to the temperature dependence of the gas-phase diffusivity. At still higher temperature, the deposition rate decreases with increasing temperature, due to competing reactions, such as homogeneous gas-phase nucleation or powder formation and sometimes etching of the surface of the film. Generally, lower temperatures, lower pressures, increased dilution, and higher flow rates (i.e. milder CVD conditions) minimize undesirable processes at the expense of growth rate. The choice of precursor may also influence the deposition rate and the properties of the deposited coating (Tesner, 1984; Duan and Don, 1995; Hüttinger, 2001). Figure 6.1 Three temperature regimes in chemical vapor deposition. 1/T (K–1) Gas-phase nucleation Gas-phase mass transport Surface chemical kinetics Log deposition rate, r (µm h–1) © 2003 Taylor & Francis
To sum,the basic steps in CVD are:(1)transport of the gaseous precursor from the center of the gas stream to the boundary layer,(2)diffusion of the precursor across the boundary layer to the surface of the substrate,and(3)decomposition of the precursor on the surface of the substrate to form the solid coating.The last step includes adsorption of precursor-derived moieties on the surface,desorption of other moieties from the surface, surface diffusion and chemical reactions.The conditions during CVD are usually far from thermodynamic equilibrium.Only a few systems,such as Si and GaAs,have been studied in relative depth.An accurate description of the CVD mechanism requires numerically solv- ing the combined chemical and flow equations for the particular reactor configuration.In spite of the progress achieved,quantitative modeling of CVD processes in general cases is limited.Porting of a process from one reactor to another reactor of different geometry and/or size may not be trivial.Therefore,reliable data must be acquired experimentally. In CVL,an additional gas-phase diffusion step needs to occur after the second step in CVD, namely diffusion from the surface of the preform into the interior pore.Such diffusion may be driven by a concentration gradient (isobaric CVD)or by a pressure gradient (forced-flow CVD). Porous fiber preforms generally have a complex pore size distribution,which may consist of several median size ranges.Continuous fibers,with diameters of 5-15 um,are arranged in bundles or tows,with 500-3,000 or more fibers per bundle (Lovell,1995).Pore sizes between individual fibers are of the order of 1-15 pm,those between fiber bundles and between cloth layers are typically 50-500 um.Because flow conductances are proportional to the opening diameter to the third or fourth power,the partial pressure of a precursor inside a small pore in a preform may be different than its value in the reactor.Also,the characteristic dimensions(s), a,for the reactor(1-500cm)and for the interior of the preform (1-500 um)differ by many orders of magnitude.For example,for a mean free path A =17um (methane molecules at 1,000C and 10 Torr=1.3 x 103 Pa),and assuming the same pressure in the pore as in the reactor,the flow would be laminar outside the preform(A<a)and laminar,mixed or molec- ular inside a pore in the preform.Therefore,both the gas-phase diffusion and the CVD depo- sition mechanisms may differ in the reactor and inside a pore:while on the surface of the preform,the CVD process may be in the surface-reaction controlled regime,within an interior pore there may be significant depletion of the precursor due to slow gas-phase diffusion in the molecular regime and the process may be gas-phase diffusion controlled.This state of affairs may lead to non-uniform densification observed in isothermal,isobaric CVI of thick preforms, where the outer surface of the preform has a higher density than the interior regions. A helpful dimensionless number in CVI is the Thiele modulus (Thiele,1939): 0=(kL21Da)0.5 (4) where a and L are the pore diameter and length,respectively.The Thiele modulus gives the relative importance of chemical reaction rate versus gas-phase diffusion.Solution of the per- tinent differential equations for mass transfer,diffusion and change of pore geometry under simplifying assumptions results in the following expression for the concentration profile, C(z),of the precursor species along the pore (0L): C(z)/C(0)=cosh[(1-2z/L)0]/cosh 0. (5) As shown in Fig.6.2,the gas-phase concentration within the pore and therefore the deposi- tion rate of the solid will be more uniform the smaller the Thiele modulus,i.e.the larger the gas-phase diffusivity compared to the surface reaction rate,and the smaller the aspect ratio L/a.It is generally preferable in CVD and CVI to operate in the surface-reaction controlled ©2003 Taylor&Francis
To sum, the basic steps in CVD are: (1) transport of the gaseous precursor from the center of the gas stream to the boundary layer, (2) diffusion of the precursor across the boundary layer to the surface of the substrate, and (3) decomposition of the precursor on the surface of the substrate to form the solid coating. The last step includes adsorption of precursor-derived moieties on the surface, desorption of other moieties from the surface, surface diffusion and chemical reactions. The conditions during CVD are usually far from thermodynamic equilibrium. Only a few systems, such as Si and GaAs, have been studied in relative depth. An accurate description of the CVD mechanism requires numerically solving the combined chemical and flow equations for the particular reactor configuration. In spite of the progress achieved, quantitative modeling of CVD processes in general cases is limited. Porting of a process from one reactor to another reactor of different geometry and/or size may not be trivial. Therefore, reliable data must be acquired experimentally. In CVI, an additional gas-phase diffusion step needs to occur after the second step in CVD, namely diffusion from the surface of the preform into the interior pore. Such diffusion may be driven by a concentration gradient (isobaric CVI) or by a pressure gradient (forced-flow CVI). Porous fiber preforms generally have a complex pore size distribution, which may consist of several median size ranges. Continuous fibers, with diameters of 5–15m, are arranged in bundles or tows, with 500–3,000 or more fibers per bundle (Lovell, 1995). Pore sizes between individual fibers are of the order of 1–15m, those between fiber bundles and between cloth layers are typically 50–500m. Because flow conductances are proportional to the opening diameter to the third or fourth power, the partial pressure of a precursor inside a small pore in a preform may be different than its value in the reactor. Also, the characteristic dimensions(s), a, for the reactor (1–500cm) and for the interior of the preform (1–500m) differ by many orders of magnitude. For example, for a mean free path 17m (methane molecules at 1,000 C and 10 Torr1.3 103Pa), and assuming the same pressure in the pore as in the reactor, the flow would be laminar outside the preform ( <<a) and laminar, mixed or molecular inside a pore in the preform. Therefore, both the gas-phase diffusion and the CVD deposition mechanisms may differ in the reactor and inside a pore: while on the surface of the preform, the CVD process may be in the surface-reaction controlled regime, within an interior pore there may be significant depletion of the precursor due to slow gas-phase diffusion in the molecular regime and the process may be gas-phase diffusion controlled. This state of affairs may lead to non-uniform densification observed in isothermal, isobaric CVI of thick preforms, where the outer surface of the preform has a higher density than the interior regions. A helpful dimensionless number in CVI is the Thiele modulus (Thiele, 1939): (ksL2 /Da) 0.5 (4) where a and L are the pore diameter and length, respectively. The Thiele modulus gives the relative importance of chemical reaction rate versus gas-phase diffusion. Solution of the pertinent differential equations for mass transfer, diffusion and change of pore geometry under simplifying assumptions results in the following expression for the concentration profile, C(z), of the precursor species along the pore (0 z L): C(z)/C(0) cosh[(1 2z/L) ]/cosh . (5) As shown in Fig. 6.2, the gas-phase concentration within the pore and therefore the deposition rate of the solid will be more uniform the smaller the Thiele modulus, i.e. the larger the gas-phase diffusivity compared to the surface reaction rate, and the smaller the aspect ratio L/a. It is generally preferable in CVD and CVI to operate in the surface-reaction controlled © 2003 Taylor & Francis
1.0 0=0.1 0.3 0.8 0.5 巨0.6 60.4 0.2 10 0.0420 0.00.10.20.30.40.5 z/L Figure 6.2 Calculated relative deposition rate along a pore axis,z,for different values of the Thiele modulus,0;L=pore length. regime,where D/ks is large.For Fickian diffusion,the diffusivity D is inversely proportional to pressure and thus operating at lower pressures will decrease 6 and result in a more uni- form infiltration profile (Kotlensky,1973;Naslain et al.,1989;Pierson,1992;Savage, 1993)and a more uniform microstructure (Pierson and Lieberman,1975)in the pore.The value of the Thiele modulus will change during infiltration,because both the gas-related quantities and the aspect ratio of the pore will change.The above simplifying assumptions include a first-order chemical reaction with no change in volume and no homogeneous gas-phase reactions;solutions for other cases were published (Thiele,1939).In many CVI processes,including carbon CVI,there is a net increase in the volume of the gaseous materials due to the decomposition of the precursor.Thus,fresh precursor species have to diffuse into the pores against an opposite,higher flow of by-product species.For carbon CVI,temperatures can vary,e.g,in the 600-2,000C range,depending on the par- ticular chemistry and system;total pressures in the reactor are generally in the range 0.13-1.3×105Pa(10-3-103Tor). CVI has several advantages compared to other densification methods.CVI allows penetra- tion of the desired atoms or molecules into the smallest pores of the preform and does not require post-densification treatment to remove organics.CVI produces uniform and conformal coatings around each accessible fiber and surface in the preform.The matrix produced by CVI is purer than that obtained by hot pressing.The final shape of an article densified by CVI is closest to the desired shape.CVI minimizes the mechanical damage to the fibers as a result of the much lower pressures and temperatures employed.The published methods described here of infiltrating composites using CVI are listed in Table 6.1 and depicted schematically in Fig. 6.3(Golecki,1997).These CVI methods can be divided into several categories,depending on: (i)spatially uniform temperature (isothermal)or not(thermal gradient); (ii)heating method-radiative or inductive; (iii)type of reactor-cold wall or hot wall; (iv)method of extraction of heat from the preform (e.g.radiative,convective,conductive); (v)pressure regime-atmospheric or low pressure; (vi)uniform pressure (isobaric)or pressure gradient (forced flow,pulsed pressure); (vii)whether a plasma is used; (viii)whether immersion in a liquid is required. ©2003 Taylor&Francis
regime, where D/ks is large. For Fickian diffusion, the diffusivity D is inversely proportional to pressure and thus operating at lower pressures will decrease and result in a more uniform infiltration profile (Kotlensky, 1973; Naslain et al., 1989; Pierson, 1992; Savage, 1993) and a more uniform microstructure (Pierson and Lieberman, 1975) in the pore. The value of the Thiele modulus will change during infiltration, because both the gas-related quantities and the aspect ratio of the pore will change. The above simplifying assumptions include a first-order chemical reaction with no change in volume and no homogeneous gas-phase reactions; solutions for other cases were published (Thiele, 1939). In many CVI processes, including carbon CVI, there is a net increase in the volume of the gaseous materials due to the decomposition of the precursor. Thus, fresh precursor species have to diffuse into the pores against an opposite, higher flow of by-product species. For carbon CVI, temperatures can vary, e.g, in the 600–2,000 C range, depending on the particular chemistry and system; total pressures in the reactor are generally in the range 0.13 1.3 105 Pa (103 103 Torr). CVI has several advantages compared to other densification methods. CVI allows penetration of the desired atoms or molecules into the smallest pores of the preform and does not require post-densification treatment to remove organics. CVI produces uniform and conformal coatings around each accessible fiber and surface in the preform. The matrix produced by CVI is purer than that obtained by hot pressing. The final shape of an article densified by CVI is closest to the desired shape. CVI minimizes the mechanical damage to the fibers as a result of the much lower pressures and temperatures employed. The published methods described here of infiltrating composites using CVI are listed in Table 6.1 and depicted schematically in Fig. 6.3 (Golecki, 1997). These CVI methods can be divided into several categories, depending on: (i) spatially uniform temperature (isothermal) or not (thermal gradient); (ii) heating method – radiative or inductive; (iii) type of reactor – cold wall or hot wall; (iv) method of extraction of heat from the preform (e.g. radiative, convective, conductive); (v) pressure regime – atmospheric or low pressure; (vi) uniform pressure (isobaric) or pressure gradient (forced flow, pulsed pressure); (vii) whether a plasma is used; (viii) whether immersion in a liquid is required. Figure 6.2 Calculated relative deposition rate along a pore axis, z, for different values of the Thiele modulus, ; L pore length. 1.0 0.8 0.6 0.4 C(z)/ C(0) 0.2 20 5 2 1 0.5 0.3 = 0.1 10 0.0 0.0 0.1 0.2 z /L 0.3 0.4 0.5 © 2003 Taylor & Francis
Table 6.Published characteristics of different chemical vapor infiltration methods used to densify C-C composite articles and described in this chapter CVI process Preform Process Reactor (Section no.) Multiple Thick/large Densification Pressure Precursor Cost of Special size time (w/r to adjustable efficiency capital fixturing (>2.5cm perform in all thickness) directions) Isothermal isobaric (2.2) Yes Yes Very long Yes Low High No Early thermal-gradient No Yes Long Yes Yes inductively-heated isobaric (2.3.1) Recent thermal-gradient Yes Yes Short Yes High Low No inductively-heated isobaric(2.3.2)】 Liquid-immersion No No Short No thermal-gradient inductively-heated isobaric atm. pressure (2.4) Forced-flow thermal- No No Short Very High Low Yes gradient (2.5) limited 3 用由 Hot w w 。。。象。。年 w H1EitiiGiittiiiititiiiii Cold Isothermal Thermal-gradient radiantly-heated radiantly-heated isobaric CVI(M) isobaric CVI(M) Liquid Cold Cold precursor 0 0 0 。 0 Cold 0 Cold Cold O Cold 0 0 Cold 0 0 Cold ● Liquid-immersion Thermal-gradient thermal-gradient inductively-heated inductively-heated isobaric CVI(M) isobaric CVI(S) Hot w 田田田 bbbpcccoapacmmceep进Y 444， Cold 是↑多↑3 Isothermal Thermal-gradient radiantly-heated radiantly-heated forced-flow CVI (S) forced-flow CVI(S】 Figure 6.3 Principles of the main chemical vapor infiltration methods (Golecki,1997); M=multiple preforms,S=single preform per run. ©2003 Taylor&Francis
Table 6.1 Published characteristics of different chemical vapor infiltration methods used to densify CøC composite articles and described in this chapter CVI process Preform Process Reactor (Section no.) Multiple Thick/large Densification Pressure Precursor Cost of Special size time (w/r to adjustable efficiency capital fixturing ( 2.5 cm perform in all thickness) directions) Isothermal isobaric (2.2) Yes Yes Very long Yes Low High No Early thermal-gradient No Yes Long Yes Yes inductively-heated isobaric (2.3.1) Recent thermal-gradient Yes Yes Short Yes High Low No inductively-heated isobaric (2.3.2) Liquid-immersion No No Short No thermal-gradient inductively-heated isobaric atm. pressure (2.4) Forced-flow thermal- No No Short Very High Low Yes gradient (2.5) limited Figure 6.3 Principles of the main chemical vapor infiltration methods (Golecki, 1997); M multiple preforms, S single preform per run. Isothermal radiantly-heated isobaric CVI (M) Thermal-gradient inductively-heated isobaric CVI (M) Liquid-immersion thermal-gradient inductively-heated isobaric CVI (S) Isothermal radiantly-heated forced-flow CVI (S) Thermal-gradient radiantly-heated forced-flow CVI (S) Thermal-gradient radiantly-heated isobaric CVI (M) Hot Hot Hot Cold Cold Cold Cold Liquid precursor Cold Hot Cold Cold Cold Cold Cold © 2003 Taylor & Francis
Different CVI methods are at different stages of technological and industrial maturity.Much data on infiltration of composites are unpublished or unavailable.Thus,the comparisons provided in this chapter are based on printed,public-domain studies,and patents.In the next sections,more detailed consideration is given to specific CVI methods used in fabrication of C-C articles. 3 Chemical vapor infiltration processes 3.1 Isothermal isobaric carbon CVI Isothermal isobaric CVI is in wide use since the 1960s for densification of C-Cs and other refractory composites(Kotlensky,1971;Savage,1993;Golecki,1997).In a common appli- cation,a large number of porous carbon disk brake-pad preforms,typically 15-55 cm in outer diameter (o.d.)by 2-3 cm in thickness,are placed in a hot-wall reactor,i.e.a furnace uniformly heated by radiation,at a temperature in the range 1,000-1,100C and exposed to a flow of reactant gas,e.g.CHa at 5-100 Torr=6.6 X 102-1.3 X 104 Pa (Thomas,1993); see Fig.6.4.In isothermal isobaric CVI,the densification kinetics of a thin (e.g.1-2mm thick)preform with initial density po follow an exponential approach with time to a"final" density value,pe(Loll et al.,1977;Marinkovic and Dimitrijevic,1987;Naslain et al.,1989): p(t)=po+(Pr-po)[1-exp(-t/T)] (6) where r is the time constant of the process;r decreases with increasing temperature accord- ing to an exponential (Arrhenius)relationship in the surface-reaction limited regime.For such a thin preform,Pr may equal the desired final density,but in a thicker preform the surface pores become clogged well before that density value is reached,requiring several interruptions of the infiltration to allow grinding the external surfaces in order to open the pores and enable further infiltration;see Fig.6.5.The total densification time is thus a strong supra-linear function of the thickness of the preform.For >2cm thick preforms, 600-2,000h may be required to achieve the desired density. Since from an economic viewpoint,it is desirable to minimize the densification time, higher temperature and precursor pressure would seem to be the direction to follow.However, Pump Preforms 区XX囚☒ X☒区XX☒ Heater 4一CH4 Figure 6.4 Simplified schematic diagram of a hot-wall,isothermal-isobaric chemical vapor infiltration reactor (not to scale). ©2003 Taylor&Francis
Different CVI methods are at different stages of technological and industrial maturity. Much data on infiltration of composites are unpublished or unavailable. Thus, the comparisons provided in this chapter are based on printed, public-domain studies, and patents. In the next sections, more detailed consideration is given to specific CVI methods used in fabrication of C–C articles. 3 Chemical vapor infiltration processes 3.1 Isothermal isobaric carbon CVI Isothermal isobaric CVI is in wide use since the 1960s for densification of C–Cs and other refractory composites (Kotlensky, 1971; Savage, 1993; Golecki, 1997). In a common application, a large number of porous carbon disk brake-pad preforms, typically 15–55 cm in outer diameter (o.d.) by 2–3 cm in thickness, are placed in a hot-wall reactor, i.e. a furnace uniformly heated by radiation, at a temperature in the range 1,000–1,100 C and exposed to a flow of reactant gas, e.g. CH4 at 5–100 Torr 6.6 102 1.3 104 Pa (Thomas, 1993); see Fig. 6.4. In isothermal isobaric CVI, the densification kinetics of a thin (e.g. 1–2mm thick) preform with initial density o follow an exponential approach with time to a “final” density value, f (Loll et al., 1977; Marinkovic and Dimitrijevic, 1987; Naslain et al., 1989): (t) o (f o)[1 exp( t/)] (6) where is the time constant of the process; decreases with increasing temperature according to an exponential (Arrhenius) relationship in the surface-reaction limited regime. For such a thin preform, f may equal the desired final density, but in a thicker preform the surface pores become clogged well before that density value is reached, requiring several interruptions of the infiltration to allow grinding the external surfaces in order to open the pores and enable further infiltration; see Fig. 6.5. The total densification time is thus a strong supra-linear function of the thickness of the preform. For 2 cm thick preforms, 600–2,000 h may be required to achieve the desired density. Since from an economic viewpoint, it is desirable to minimize the densification time, higher temperature and precursor pressure would seem to be the direction to follow. However, Figure 6.4 Simplified schematic diagram of a hot-wall, isothermal–isobaric chemical vapor infiltration reactor (not to scale). Heater CH4 Pump Preforms © 2003 Taylor & Francis
Time Figure 6.5 Schematic illustration of the densification kinetics of a thick preform in isothermal isobaric chemical vapor infiltration. 2.4 20 2.2 0.02 Laminar 2.0 15 aromatic 10 巴 1.8 Isotropic 10 sooty 1.6 40 1.4 1.2 Surface 150 Torr nucleatad 1.0 0 1,200 1,600 2,000 1,200 1.600 2000 Deposition temperature(C) Figure 6.6 Density and microstructure of infiltrated carbon versus temperature and pressure (Kotlensky,1971;reprinted by permission of the Society for the Advancement of Materials and Process Engineering). doing so may also lead to more premature closure of surface pores.Conversely,lower tem- peratures and pressures may reduce undesirable gas-phase nucleation and formation of tar and soot by-products.Lower pressure corresponds to higher gas-phase diffusivity,leading to more uniform distributions of density and microstructure within the composite.For exam- ple,the density minimum observed in carbon CVI(Kotlensky,1971)-see Fig.6.6-was interpreted as resulting from the incorporation of soot particles formed in the gas phase into the carbon deposited in the pores of the preform.If the pressure was reduced sufficiently, this density minimum could be eliminated.C-C composites having larger crystallite size, La,and improved thermal and electrical conductivities were produced at lower pressures (Stoller and Frye,1972).Rough-laminar(anisotropic)carbon,which is generally desired for many applications,was obtained at 35 Torr (4.7 X 103Pa),but not at 100-760 Torr (1.3 X 104-1.0 X 105 Pa).Rough-laminar carbon can be rendered more graphitic,i.e.more ordered microstructurally,by high-temperature treatment (graphitization)at =2,000C (Loll et al.,1977)and this may result in improved composite properties.Lower pressures also allow lower inlet precursor flow rates. The effect of different precursors on the carbon densification rate of 7.6 X 7.6 X 1.6cm carbon preforms was studied(Duan and Don,1995).Randomly oriented pitch-fiber tow preforms were impregnated with phenolic resin and densified at 1,000-1,150C and 15 Torr ©2003 Taylor&Francis
doing so may also lead to more premature closure of surface pores. Conversely, lower temperatures and pressures may reduce undesirable gas-phase nucleation and formation of tar and soot by-products. Lower pressure corresponds to higher gas-phase diffusivity, leading to more uniform distributions of density and microstructure within the composite. For example, the density minimum observed in carbon CVI (Kotlensky, 1971) – see Fig. 6.6 – was interpreted as resulting from the incorporation of soot particles formed in the gas phase into the carbon deposited in the pores of the preform. If the pressure was reduced sufficiently, this density minimum could be eliminated. C–C composites having larger crystallite size, La, and improved thermal and electrical conductivities were produced at lower pressures (Stoller and Frye, 1972). Rough-laminar (anisotropic) carbon, which is generally desired for many applications, was obtained at 35 Torr (4.7 103 Pa), but not at 100–760 Torr (1.3 104 1.0 105 Pa). Rough-laminar carbon can be rendered more graphitic, i.e. more ordered microstructurally, by high-temperature treatment (graphitization) at 2,000 C (Loll et al., 1977) and this may result in improved composite properties. Lower pressures also allow lower inlet precursor flow rates. The effect of different precursors on the carbon densification rate of 7.6 7.6 1.6 cm carbon preforms was studied (Duan and Don, 1995). Randomly oriented pitch-fiber tow preforms were impregnated with phenolic resin and densified at 1,000–1,150 C and 15 Torr Figure 6.5 Schematic illustration of the densification kinetics of a thick preform in isothermal isobaric chemical vapor infiltration. Time Density Figure 6.6 Density and microstructure of infiltrated carbon versus temperature and pressure (Kotlensky, 1971; reprinted by permission of the Society for the Advancement of Materials and Process Engineering). Deposition temperature (°C) 1,200 1,600 2,000 Density (g cm–3) 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 0.02 10 40 150 Torr 1,200 1,600 2,000 Pressure (Torr) 20 15 10 5 0 Laminar aromatic Isotropic sooty Surface nucleatad Continously nucleatad © 2003 Taylor & Francis
1,100C 830f1,150℃ 55 1,050°C 15 1.000C 10 0 sccm C3He CaHe=35sccm T=1,050C 0 0 40 80 1200 40 80 120 Time(h) Figure 6.7 Relative weight gain versus time in isothermal isobaric carbon CVI from CH4 and H2, showing the effects of temperature and C3H6 additions (Duan and Don,1995;reprinted with the authors'kind permission). (2.0X 103 Pa).The initial density and open porosity of the preforms were 1.25gcm-3 and 38 vol%,respectively.The preform weight was continuously recorded by means of an in-situ balance.CH4 at 400sccm was the primary precursor,with additions of 100sccm H2 and 0-75sccm propylene,C3H6.The hot zone of the furnace was 15.2cm in diameter by 30.5 cm long.The curves of fractional weight gain versus time had an exponential-type behavior and were fitted to the equation Am/m=a(pcvppo)In{1+az[exp(-a3t-a42)]-1), where the a,'s were constants.At 1,050C,the addition of increasing fractions of C3H6 resulted in increases of the initial deposition rate and of the final density,shown in Fig.6.7. The microstructure of the deposited carbon changed from isotropic in the absence of C3H6 to anisotropic when 35 sccm C3H6 was added.With 35 sccm C3H6,increasing the tempera- ture from 1,000 to 1,100C also increased the initial densification rate and the final density, whereas at 1,150C,premature surface pore closure and non-uniform deposition through the thickness were seen. Thus,the operating conditions of a hot-wall,isothermal isobaric CVI process involve a compromise which results in materials with the desired properties in the shortest possible time and with the minimum amount of labor,supplies,and energy. Advantages of isothermal isobaric CVI include: 。 The method is well-established and relatively well-understood. ● A large number of preforms can be densified simultaneously. The densification time per preform is relatively low for heavy loading of the reactor. Preforms of different and complex shapes and sizes can be readily densified in the same run,although the minimum dimension (usually the thickness)needs to be similar. The energy expenditure per part is relatively low. Disadvantages of isothermal isobaric CVI include: Premature surface crusting occurs before the desired bulk density is reached,necessi- tating several interruptions in the CVI process to grind surfaces. Very long hot processing time,typically 600-2,000h per batch(but the time does not depend on the number of parts being densified in a given reactor). The density within the article is generally highest at the surfaces and lowest in the inte- rior regions. ©2003 Taylor&Francis
(2.0 103 Pa). The initial density and open porosity of the preforms were 1.25 g cm3 and 38 vol%, respectively. The preform weight was continuously recorded by means of an in-situ balance. CH4 at 400 sccm was the primary precursor, with additions of 100 sccm H2 and 0–75 sccm propylene, C3H6. The hot zone of the furnace was 15.2 cm in diameter by 30.5 cm long. The curves of fractional weight gain versus time had an exponential-type behavior and were fitted to the equation m/ma1(CVD/o)ln{1a2[exp(a3ta4t 2 )]1}, where the aj ’s were constants. At 1,050 C, the addition of increasing fractions of C3H6 resulted in increases of the initial deposition rate and of the final density, shown in Fig. 6.7. The microstructure of the deposited carbon changed from isotropic in the absence of C3H6 to anisotropic when 35 sccm C3H6 was added. With 35 sccm C3H6, increasing the temperature from 1,000 to 1,100 C also increased the initial densification rate and the final density, whereas at 1,150 C, premature surface pore closure and non-uniform deposition through the thickness were seen. Thus, the operating conditions of a hot-wall, isothermal isobaric CVI process involve a compromise which results in materials with the desired properties in the shortest possible time and with the minimum amount of labor, supplies, and energy. Advantages of isothermal isobaric CVI include: ● The method is well-established and relatively well-understood. ● A large number of preforms can be densified simultaneously. ● The densification time per preform is relatively low for heavy loading of the reactor. ● Preforms of different and complex shapes and sizes can be readily densified in the same run, although the minimum dimension (usually the thickness) needs to be similar. ● The energy expenditure per part is relatively low. Disadvantages of isothermal isobaric CVI include: ● Premature surface crusting occurs before the desired bulk density is reached, necessitating several interruptions in the CVI process to grind surfaces. ● Very long hot processing time, typically 600–2,000 h per batch (but the time does not depend on the number of parts being densified in a given reactor). ● The density within the article is generally highest at the surfaces and lowest in the interior regions. Figure 6.7 Relative weight gain versus time in isothermal isobaric carbon CVI from CH4 and H2, showing the effects of temperature and C3H6 additions (Duan and Don, 1995; reprinted with the authors’ kind permission). Time (h) 0 40 80 120 0 40 80 120 Weight gain (%) 0 10 20 30 40 C3H6= 35 sccm 1,150 °C 1,100 °C 1,050 °C 1,000 °C 0 sccm C3H6 75 55 35 15 T = 1,050 °C © 2003 Taylor & Francis
Overall precursor conversion efficiency is low,of the order of 1-2%. Preforms having different minimum dimensions(usually the thickness)densify at very different rates. Capital cost is relatively high. Cost of inventory and potential re-work can be high. Process development or process changes may be slow to implement. From the previous sections,it follows that in order to significantly speed up the densifi- cation process of relatively thick parts,either the temperature or the pressure during CVI has to be increased,yet without clogging surface pores or degrading material properties.The next section describes approaches using a thermal gradient to achieve this goal. 3.2 Thermal-gradient inductively-heated isobaric CVI 3.2.I Early thermal-gradient isobaric CVI studies In an early thermal-gradient CVI process (Stoller and Frye,1972;Lieberman and Noles, 1973;Stoller et al.,1974;Lieberman et al.,1975),intended for rocket components (Thomas,1993),a porous carbon felt was mounted around a truncated-conical graphite sus- ceptor;see Fig.6.8.This electrically conductive susceptor was heated by a conical-shaped induction coil(frequency and power levels not provided).The felt was fabricated using 5 cm long,chopped oxidized PAN fibers,which were carded into batts(Lieberman et al.,1975). Alternate carded batts were cross-plied when they were felted by means of a needle loom, in an effort to achieve isotropic properties in the plane.The felts were nominally 118 X 118 X 3.2 cm in size and had a density of 0.14+0.02gcm-3.Each felt,which was initially flat, was made into a truncated hollow cone by sewing a single seam along the entire length of the cone with a rayon-based carbon yarn.After carbonization in flowing Ar to 1,300C to remove volatiles,the weight loss was 44%and the linear shrinkage 13.5-14.6%.The carbon deposition conditions of one preform were (Lieberman et al.,1975):susceptor temperature 1,325C,total pressure =8.4 X 10 Pa (630 Torr 1 atm in Albuquerque,NM),CH4 flow rate=61 Imin-,Ar flow rate=205 Imin-and free gas space between shroud and preform= 2.5cm.From the description given (Lieberman and Noles,1973;Stoller et al.,1974; Lieberman et al.,1975)it appears that the felt did not initially couple electrically to the coil, so there was no induced electrical current flowing around the circumference of the felt.The sewn construction of the felt may have resulted in a very high electrical resistivity in the cir- cumferential direction and thus a lack of initial electromagnetic coupling.The heating of the felt therefore occurred initially by thermal radiation.It was stated(Stoller et al.,1974)that carbon was deposited first at the surface of the felt adjacent to the susceptor and then dep- osition progressed radially through the preform as the densified preform became inductively heated.However,as reported (Stoller and Frye,1972;Stoller et al.,1974),a reduction by a factor of two only was achieved in the overall densification cycle time,compared to isother- mal CVI,and this technique was limited to densifying one preform at a time.Gas analysis data(Lieberman and Noles,1973)indicated that densification times were equal to or longer than 130h;infiltration times of six weeks (1,000h)were reported (Lackey,1989).This process also suffered from soot formation due to the combination of the very high temper- ature and high pressure employed.Spatial density distributions were obtained from samples cored through the thickness of the frustum.There was a density gradient from the inner ©2003 Taylor&Francis
● Overall precursor conversion efficiency is low, of the order of 1–2%. ● Preforms having different minimum dimensions (usually the thickness) densify at very different rates. ● Capital cost is relatively high. ● Cost of inventory and potential re-work can be high. ● Process development or process changes may be slow to implement. From the previous sections, it follows that in order to significantly speed up the densification process of relatively thick parts, either the temperature or the pressure during CVI has to be increased, yet without clogging surface pores or degrading material properties. The next section describes approaches using a thermal gradient to achieve this goal. 3.2 Thermal-gradient inductively-heated isobaric CVI 3.2.1 Early thermal-gradient isobaric CVI studies In an early thermal-gradient CVI process (Stoller and Frye, 1972; Lieberman and Noles, 1973; Stoller et al., 1974; Lieberman et al., 1975), intended for rocket components (Thomas, 1993), a porous carbon felt was mounted around a truncated-conical graphite susceptor; see Fig. 6.8. This electrically conductive susceptor was heated by a conical-shaped induction coil (frequency and power levels not provided). The felt was fabricated using 5 cm long, chopped oxidized PAN fibers, which were carded into batts (Lieberman et al., 1975). Alternate carded batts were cross-plied when they were felted by means of a needle loom, in an effort to achieve isotropic properties in the plane. The felts were nominally 118 118 3.2 cm in size and had a density of 0.14 0.02 g cm3 . Each felt, which was initially flat, was made into a truncated hollow cone by sewing a single seam along the entire length of the cone with a rayon-based carbon yarn. After carbonization in flowing Ar to 1,300 C to remove volatiles, the weight loss was 44% and the linear shrinkage 13.5–14.6%. The carbon deposition conditions of one preform were (Lieberman et al., 1975): susceptor temperature 1,325 C, total pressure 8.4 104 Pa ( 630 Torr 1 atm in Albuquerque, NM), CH4 flow rate61 lmin1 , Ar flow rate205 1min1 and free gas space between shroud and preform 2.5 cm. From the description given (Lieberman and Noles, 1973; Stoller et al., 1974; Lieberman et al., 1975) it appears that the felt did not initially couple electrically to the coil, so there was no induced electrical current flowing around the circumference of the felt. The sewn construction of the felt may have resulted in a very high electrical resistivity in the circumferential direction and thus a lack of initial electromagnetic coupling. The heating of the felt therefore occurred initially by thermal radiation. It was stated (Stoller et al., 1974) that carbon was deposited first at the surface of the felt adjacent to the susceptor and then deposition progressed radially through the preform as the densified preform became inductively heated. However, as reported (Stoller and Frye, 1972; Stoller et al., 1974), a reduction by a factor of two only was achieved in the overall densification cycle time, compared to isothermal CVI, and this technique was limited to densifying one preform at a time. Gas analysis data (Lieberman and Noles, 1973) indicated that densification times were equal to or longer than 130 h; infiltration times of six weeks (1,000 h) were reported (Lackey, 1989). This process also suffered from soot formation due to the combination of the very high temperature and high pressure employed. Spatial density distributions were obtained from samples cored through the thickness of the frustum. There was a density gradient from the inner © 2003 Taylor & Francis