1. Golecki/ Rapid vapor-phase densification of refractory composites 50 0.8 R=500 um,L=7cm 0.00.1020.3040.5 Z/L Fig 8.(a)Calculated relative deposition rate along a pore axis, z(L= pore length )for different values of the Thiele modulus, 8, from Eq (3).(b)Dependence of axial deposition late pofile of TiC in pore ull pressure and teinperature [[32]; reprinted with the authors'kind missIon].●,90mbar/l00℃C;,90mbar/l100°C;、,900mbar/1000°C. of the heterogeneous decomposition reaction on the surface and d is the diffusivity of the multi component gas mixture. Solutions of the pertinent differential equations for mass transfer, diffusion and change of pore geometry [31, 32] under simplifying assumptions result in Eq. (3)for the concen- tration profile, C(z), of the precursor species along the pore(OszsL) C(z)/C(0)=cosh[(1-2z/)0]/cosh of ll s shown in Fig 8, the gas-phase concentration within the pore and therefore the deposition rate he solid will be more uniform, the smaller the Thiele modulus; i.e. (1) the larger the gas-phase diffusivity compared to the surface reaction rate, and(2)the smaller the aspect ratio L/a. As noted previously, it is generally preferable in CVD and Cvi to operate in the surface-reaction controlled 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 0 and result in a more uniform infiltration profile [1, 29, 33, 34] and a more uniform microstructure [35] in the pore(see Fig. 8). The value of Thiele modulus will change during the infiltration, however, because both the gas-related quantities and the aspect ratio of the pore will change. The simplifying assumptions used in deriving Eq.(3 include a first-order chemical reaction with no change in volume and no homogeneous gas-phase reactions;solutions for other cases were given in Ref. [31] and more realistic models were published elsewhere [36,37]. A closely related dimensionless number is the second Damkohler number, Dall proportional to 0. An additional dimensionless number particularly useful in forced-flow CVi is the Peclet number, Pe=va/D, the physical significance of which is the ratio between convection and diffusion. Note that in most CVIreactors, conditions are far removed from thermodynamic equilibrium Advantages of CVI VS. other fabrication methods, such as hot pressing for densifying and fabricating composite materials are [29, 38] as follows CVi is CVi minimizes the mechanical damage to the fibers as a result of the much lower pressures and temperatures employed in CVI, compared to those in other fabrication routes The matrix produced by CVi is purer than that obtained by hot pressing, the latter leaving purities introduced as sintering additives However, all CVi methods, just like other processing methods, leave a certain amount of void fraction or unfilled porosity in the composite, typically of the order of 1-10%. Residual porosity may be open (i.e. accessible from the extermal surface) or closed, and interconnected or not Whether and how
I. GoIecki / Rapid vapor-phase densification of refractory composites 47 0.8 6 u’ . 0.6 3 . 5 0.4 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0 1 2 3 z/L z (cm) Fig. 8. (a) Calculated relative deposition rate along a pore axis, z (L= pore length) for different values of the Thiele modulus, 8, fromEq. (3). (b) Dependence of axial deposition rate profile of TiC in pore on pressure and temperature [ [ 321; reprinted with the authors’ kind permission]. l ,90 mbar/ 1000 ‘C; v, 90 mbar/llOO “C; V, 900 mbar/lOOO “C. of the heterogeneous decomposition reaction on the surface and D is the diffusivity of the multicomponent gas mixture. Solutions of the pertinent differential equations for mass transfer, diffusion and change of pore geometry [ 31,321 under simplifying assumptions result in Eq. (3) for the concentration profile, C(z) , of the precursor species along the pore (0 I z IL) : C(z)IC(O)=cosh[(l-2z/L)B]/cosh0 (3) As shown in Fig. 8, 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. (1) the larger the gas-phase diffusivity compared to the surface reaction rate, and (2) the smaller the aspect ratio L/a. As noted previously, it is generally preferable in CVD and CVI to operate in the surface-reaction controlled regime, where D/k, is large. For Fickian diffusion, the diffusivity D is inversely proportional to pressure and thus operating at lower pressures will decrease 8 and result in a more uniform infiltration profile [ 1,29,33,34] and a more uniform microstructure [35] in the pore (see Fig. 8). The value of the Thiele modulus will change during the infiltration, however, because both the gas-related quantities and the aspect ratio of the pore will change. The simplifying assumptions used in deriving Eq. (3) include a first-order chemical reaction with no change in volume and no homogeneous gas-phase reactions; solutions for other cases were given in Ref. [ 311 and more realistic models were published elsewhere [ 36,371. A closely related dimensionless number is the second Damkijhler number, DaII, proportional to 82. An additional dimensionless number particularly useful in forced-flow CVI is the Peclet number, Pe = va/D, the physical significance of which is the ratio between convection and diffusion. Note that in most CVI reactors, conditions are far removed from thermodynamic equilibrium. Advantages of CVI vs. other fabrication methods, such as hot pressing, for densifying and fabricating composite materials are [ 29,381 as follows. + CVI is a near-net-shape process. + CVI minimizes the mechanical damage to the fibers as a result of the much lower pressures and temperatures employed in CVI, compared to those in other fabrication routes. + The matrix produced by CVI is purer than that obtained by hot pressing, the latter leaving impurities introduced as sintering additives. However, all CVI methods, just like other processing methods, leave a certain amount of void fraction or unfilled porosity in the composite, typically of the order of l-10%. Residual porosity may be open (i.e. accessible from the external surface) or closed, and interconnected or not. Whether and how
. Golecki/ Rapid vapor-phase densification of refractory composites residual porosity attects the performance of the final product depends on the materials, processing and application. The overall average density of a composite article, Pav, is equal to the weighted arithmetic average of its constituents, Pay=2P; X: =p,Xr+ Pm XmI+Pm2X-m2+., where the Xi denote the respec- tive volume fractions(2X: =1-X )and the subscripts f, mi and v denote fiber, matrix no i and voi respectively( the voids have zero mass and density, of course) The principal published methods of infiltrating composites using CVI are listed in Table 3 and depicted schematically in Fig 9. These CVI methods can be divided into several categories, depending (a) the heating method -radiative or inductive (b)whether the temperature is spatially uniform(isothermal)or not( thermal gradient) (c)the type of reactor-cold wall or hot wall (d)the mcthod of extraction of heat from the preform(e, g radiative, convective, conductive) (e) the pressure regime-atmospheric or low pressure; (g)whether the pressure is uniform(isobaric) or there is an imposed pressure gradient across the preform(forced flow, pulsed pressure); (g) whether a plasma is used; (h) whether immersion in a liquid is required. Different infiltration 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 review are based on printed, public-domain studies and patents. In the next sections we describe each method of cvi with its advantages and disadvantages starting with the baseline or conventional method of isothermal, isobaric CVI 4. Isothermal isobaric CVI This is the most widcly uscd and the oldest approach(in use since the 1960s)for densification of refractory composites [1-4, 33, 37, 40-64, 32, 36, 65-76]. This method is used for the densification of C-C composites for aircraft brakes and SiC composites for aerospace components [29].The preforms are placed in a uniform-temperature zone of a hot-wall reactor, which is radiatively heated precursor gases flow around the preforms and delivery of the molecular or atomic species onto the surfaces and internal porosities of the preforms occurs through diffusion in the gas phase. Molecular or atomic by-products of the chemical reactions likewise have to diffuse out of the pores of the preform back into the reactor. In many CVI processes, there is a net increase in the volume of the gaseous materials due to the decomposition of the precursor. For example, the overall decomposition of methane to produce carbon can be written CH4(gas)+C(solid)+2H2(gas) (4) illustrating the fact that the final products consist of the solid coating plus two moles of gas vs. only one mole of gas in the precursor. Thus, fresh precursor species have to diffuse into the pores against an opposite, higher fow of by-product species. Depending on the particular process, the pressure in the reactor can be atmospheric or sub-atmospheric. The advantages of using low pressures in CVI are (i)a higher gas-phase diffusivity [29], leading to more uniform distributions of density and micros- tructure within the composite [1, 29, 33,35,321,(ii)a reduction in undesirable gas-phase nucleation and in the formation of deleterious by-products, such as tar and soot in the case of carbon CVI, and (iii)lower inlet precursor flow rates. The microstructure of the deposited matrix is generally also a function of the pressure as well as the temperature( see Fig 10)[40]. The density minimum observed
48 I. Golecki / Rapid vapor-phase densij5cation of refractory composites residual porosity affects the performance of the final product depends on the materials, processing and application. The overall average density of a composite article, pa”, is equal to the weighted arithmetic average of its constituents, pav = ZpiXi = pfXf+ pmlXml + fm2Xm2 + . . . , where the Xi denote the respective volume fractions ( ZXi = 1 -X,) and the subscripts f, mi and v denote fiber, matrix no, i and void, respectively (the voids have zero mass and density, of course). The principal published methods of infiltrating composites using CVI are listed in Table 3 and depicted schematically in Fig. 9. These CVI methods can be divided into several categories, depending on: (a) the heating method - radiative or inductive; (b) whether the temperature is spatially uniform (isothermal) or not (thermal gradient) ; (c) the type of reactor - cold wall or hot wall; (d) the method of extraction of heat from the preform (e.g. radiative, convective, conductive) ; (e) the pressure regime - atmospheric or low pressure; (g ) whether the pressure is uniform (isobaric) or there is an imposed pressure gradient across the preform (forced flow, pulsed pressure) ; (g) whether a plasma is used; (h) whether immersion in a liquid is required. Different infiltration 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 review are based on printed, public-domain studies and patents. In the next sections we describe each method of CVI with its advantages and disadvantages, starting with the baseline or conventional method of isothermal, isobaric CVI. 4. Isothermal, isobaric CVI This is the most widely used and the oldest approach (in use since the 1960s) for densification of refractory composites [ 1+33,37,40-64,32,36,65-761. This method is used for the densilication of C-C composites for aircraft brakes and Sic composites for aerospace components [29]. The preforms are placed in a uniform-temperature zone of a hot-wall reactor, which is radiatively heated. The precursor gases flow around the preforms and delivery of the molecular or atomic species onto the surfaces and internal porosities of the preforms occurs through diffusion in the gas phase. Molecular or atomic by-products of the chemical reactions likewise have to diffuse out of the pores of the preform back into the reactor. In many CVI processes, there is a net increase in the volume of the gaseous materials due to the decomposition of the precursor. For example, the overall decomposition of methane to produce carbon can be written as CH4 (gas) + C (solid)+2H, (gas) (4) illustrating the fact that the final products consist of the solid coating plus two moles of gas vs. only one mole of gas in the precursor. Thus, fresh precursor species have to diffuse into the pores against an opposite, higher flow of by-product species. Depending on the particular process, the pressure in the reactor can be atmospheric or sub-atmospheric. The advantages of using low pressures in CVI are: (i) a higher gas-phase diffusivity [ 291, leading to more uniform distributions of density and microstructure within the composite [ 1,29,33,35,32], (ii) a reduction in undesirable gas-phase nucleation and in the formation of deleterious by-products, such as tar and soot in the case of carbon CVI, and (iii) lower inlet precursor flow rates. The microstructure of the deposited matrix is generally also a function of the pressure as well as the temperature (see Fig. 10) [ 401. The density minimum observed
L. Golecki/ Rapid vapor-phase densification of refractory compositEs 、⊥ clz\\\3 \\\\\\ ⊥⊥ 、y ⊥、 员合
Table 3 Different published chemical vapor infiltration methods used to densify composite materials CVI process (and representative composite material) Temperature Preform heating method Pressure Plasma Reactor Unif. Grad. Radiative Inductive Unif. Grad. Atm Low N Y HW CW Isothermal, isobaric (C-C, Sic-Sic) v I/ I/ 1/ v v v Plasma-enhanced, low pressure (C-C, C-diamond) r/ r/ r/ v v v */r/ r/ Thermal gradient, radiantly heated, isobaric (C-C) y’ r/ J r/ r/ I/ 1/ Thermal gradient, inductively heated, isobaric (C-C) v I/ fl v fl H ti Liquid immersion, thermal gradient, inductively heated, isobaric, v v v r, I/ I/ atm. pressure (C-C, C-BN, SiCSiC, SiC-Si,N,) Isothermal, forced flow (C-C, C-SIC) v v v v I/ v v Isothermal, pulsed pressure (C-C, CSiC, C-TiN) v r/ J v v v v Thermal gradient, forced flow (C-C, Sic-Sic, C-SIC, CSi3N4, L/ v r/ v r/ I/ Sic-S&N,, Mullite-SiaN,) Microwave-heated, isobaric or forced-flow (SiCSiC, J r/ I/ I/ and/or r/ ti r/ J I/ r/ SiCSi,N,, A1203-A1203) capacitive Catalyst-enhanced, isothermal, isobaric (C-C) J v Lc r/ v v Particle-transport enhanced, isothermal, isobaric (not for v r/ J rc J w continuous-fiber composites) (TiOz-Ti02, Sic-Sic, Mo-Sic)
l. Golecki/ Rapid vapor-phase densification of refractory composites Cold Radiantly-Heated Isobaric CVI (M Cold Cold o 難腓:。 Cold cold Liquid Thermal-gradient adient Hot Cold ↑↑氵 Isothermal Thermal-Gradient Heated Fig 9. Principles of the main chemical vapor infiltration methods(augmented from Ref. [391) 15 10 14 Surface l.0 100012001400160018002000 l00012001400160018002000 Deposition Temperature) Fig. 10. Relationships of density and microstructure of infiltrated carbon with temperature and pressure [[40]; reprinted by permission of the Society for the Advancement of Materials and Process Engineering] in carbon CVI(Fig. 10)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[41]. If the pressure was reduced sufficiently, this density minimum could be eliminated [41 ]. Carbon-carbon composites having larger crystallite size, Le, and improved thermal and electrical conductivities were produced at lowerpressures [64], as shown in Table 4. Highly optically anisotropic (rough-laminar) carbon, which is generally desired for many applications, was obtained at 35 Torr, but not at 100-760 Torr Rough- laminar carbon can bc rendered more graphitic, i.c. more ordered microstructurally, by high-temperature treatment (HTT)at 22000C(graphitization)[42, 43]. HTT may result in improved composite properties The disadvantage of using low pressures is a slower deposition rate and therefore a longer densification time. The deposition rate can be increased by running at higher temperature and higher pressure,but
50 I. Golecki / Rapid vapor-phase densi’cation of refractory composites - co ‘Isot’ eimal” Radiant ‘t y-Heated Isobaric CVI (M) ld~~l~Co, ‘.‘!. :. .*t. . . . . a a ” Cold v Thermal-Gradient Inductive1 -Heated Isobaric ?VI (M) Q ; 0 0 Liquid-Immersion Thermal-Gradient Inductive1 -Heated Isobaric EVI (S) Isothermal Radiantly-Heated Thermal-Gradient Forced-Flow CVI (S) Radiantly-Heated Forced-Flow CVI (S) Thermal-Gradient Radiantly-Heated Isobaric CVI (M) Fig, 9. Principles of the main chemical vapor infiltration methods (augmented from Ref. [39]). 2.4 20 1000 1200 1400 1600 1800 2000 Deposition Temperature (C) loo0 1200 1400 1600 1800 2OOO Deposition Temperature (C) Fig. 10. Relationships of density and microstructure of infiltrated carbon with temperature and pressure [ [40]; reprinted by permission of the Society for the Advancement of Materials and Process Engineering]. in carbon CVI (Fig. 10) 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 [ 411. If the pressure was reduced sufficiently, this density minimum could be eliminated [41]. Carbon-carbon composites having larger crystallite size, L,, and improved thermal and electrical conductivities were produced at lower pressures [ 641, as shown in Table 4. Highly optically anisotropic (rough-laminar) carbon, which is generally desired for many applications, was obtained at 35 Torr, but not at 100-760 Torr. Rough-laminar carbon can be rendered more graphitic, i.e. more ordered microstructurally, by high-temperature treatment (HTT) at 2 2000 “C (graphitization) [ 42,431. HTT may result in improved composite properties. The disadvantage of using low pressures is a slower deposition rate and therefore a longer densification time. The deposition rate can be increased by running at higher temperature and higher pressure, but
1. Galecki/ Rapid vapor-phase densification of refractory composites 51 Table 4 pressure on the properties of carbon-carbon composites densified by isothermal, isobaric CVI at 1100C (Torr) dooz(A) Lc(A) Kh(W m-1k-) na micr at600℃C 34 HTT As-dep. HTT As-dcp HTT 3.36550 4016 03251.79 95 25 760SL 3.2 Initial fiber fraction, 9 vol %; hTT done at 3000"C for 2h; RL, rough-laminar; SL, smooth laminar [[64]: reproduced by permission of the Society for the Advancement of Materials and Process Engineering] doing so generally also leads to more premature surface pore closure and formation of undesirable by- products. The deposition rate is also influenced by the choice of the precursor gases used; for example, CHa is a stable molecule which may produce a relatively low deposition rate of carbon. Thus, the opcrating 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 anmount of labor, supplies and energy Some of the matrix materials which have been formed by isothermal, isobaric CVi include C SiC, B4C, TiC, Si3N4, BN, AL2O3 and Zro2 [33]. C-C composites have been studied relatively extensively. The densification kinetics of a thin preform with initial density po in isothermal CVI follow an exponential approach to a"density value, Pr 133, 42, 44, 49, 36, 76 p(t)=P,+(Pr-Po)[1-exp(t/r) (5) where r is the time constant of the process, which decreases with increasing temperature according to an Arrhenius relationship [44] in the surface-reaction limited regime. For a very thin preform, Pe may actually equal the desired final density value, hut in a thicker preform the surface pores hecome clogged well before that density value is reached, requiring several interruptions of the infiltration to allow grinding the external surfaces. Systematic carbon CVI experiments were carried out [44, 49,50] on 0. 2-0.45 cm thick preforms made of a variety of carbon materials. The preforms were made usin polyacrylonitrile(PAN) carbon fibers from those authors'lab or Torayca T-300 PAN fibers, cellulose carbon powder, natural graphite powder compacts, carbon felt from cellulose and polycrystalline graphite. The Pan preforms were infiltrated directly or after having been made rigid by means of impregnation with phenolic resin, then hot pressing and curing at 150C, followed by carbonization through slow heating to 900 C in Ar Carbon C VI was performed in flowing propylene( C3H6)diluted in Ar or He at atmospheric pressure. The propylene concentration(3-50 vol %), reactor temperature (700-880C)and gas residence time in the reactor, tr(16-240 s), were varied; tr=(react F)(273/T)(p/760), where Reactor is the reactor volume, Fis the volumetric flow rate, Tis in K and p in Torr. The infiltration rate for all these different preforms [44, 51, 52] was proportional to the actual open porosity P of the preform, -(dP/dt)=P/r, or In[ P(r)/P)]=-t/T. The time constant found in early experiments [51, 52] was 49.5 h for a cellulose carbon preform and 31.75 h for a compacted powder of natural graphite. The results were fitted to the equation mcyD(r)=(k,/k)pcvp P[1-exp ((t-t)/r) where mcvd was the mass of carbon infiltrated into the preform, k the overall kinetic constant, Po the initial porosity and PcvD=2.030 g cm the density of CVD carbon. Table 5 givcs the valucs of to and the kinetic constants k,(constant of pore filling)and k2(constant of pore closing)[50] calculated oin the measurements. The time constants obtained froIn mass gain measurements (Table 5)were shorter than those obtained earlier from porosity measurements. An activation energy of 2.05 eV
1. Golecki/ Rapid vapor-phase densifcation of refractory composites 51 Table 4 Effect of deposition pressure on the properties of carbon-carbon composites densified by isothermal, isobaric CVI at 1100 “C 0 Deposition pressure (Torr) &,2 (4 r, 6) q,, (W m-’ K-’ ) P and micro-structure at 600 “C (g cmA3) As-dep. HTT As-dep. HTT As-dep. HTT As-dep. HTT 35, FL 3.449 3.365 50 340 16 71 2.9 0.325 1.79 100, SL 3.445 3.417 37 95 12 25 3.0 2.5 1.80 760, SL 3.459 3.423 32 32 10 38 3.2 2.4 1.80 Initial fiber fraction, 9 vol.%; HTT done at 3ooO “C for 2 h; E&, rough-laminar; SL, smooth-laminar [ [ 641; reproduced by permission of the Society for the Advancement of Materials and Process Engineering]. _. doing so generally also leads to more premature surface pore closure and formation of undesirable byproducts. The deposition rate is also influenced by the choice of the precursor gases used; for example, CH4 is a stable molecule which may produce a relatively low deposition rate of carbon. 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. Some of the matrix materials which have been formed by isothermal, isobaric CVI include C, Sic, B&, Tic, Si3N4, BN, A1203 and ZrOz [33]. C-C composites have been studied relatively extensively. The densification kinetics of a thin preform with initial density p0 in isothermal CVI follow an exponential approach to a “final” density value, pf [ 33,42,44,49,36,76] P(t)=Po+(Pf-Po)[1-exp(-t/7)1 (5) where T is the time constant of the process, which decreases with increasing temperature according to an Arrhenius relationship [ 441 in the surface-reaction limited regime. For a very thin preform, pf may actually equal the desired final density value, 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. Systematic carbon CVI experiments were carried out [44,49,50] on 0.2-0.45 cm thick preforms made of a variety of carbon materials. The preforms were made using polyacrylonitrile (PAN) carbon fibers from those authors’ lab or Torayca T-300 PAN fibers, cellulose carbon powder, natural graphite powder compacts, carbon felt from cellulose and polycrystalline graphite. The PAN preforms were infiltrated directly or after having been made rigid by means of impregnation with phenolic resin, then hot pressing and curing at 150 “C, followed by carbonization through slow heating to 900 “C in Ar. Carbon CVI was performed in flowing propylene ( C3H6) diluted in Ar or He at atmospheric pressure. The propylene concentration (3-50 vol.%), reactor temperature (700-880 “C) and gas residence time in the reactor, t, (16-240 s), were varied; t,= ( Vreactor/ F) (273 /T) (p/760), where Kaactor is the reactor volume, F is the volumetric flow rate, T is in K and p in Torr. The infiltration rate for all these different preforms [ 44,5 1,521 was proportional to the actual open porosity P of the preform, - (dPldt) = Pl r, or ln[ P( t) /PO) ] = - t/T. The time constant Tfound in early experiments [ 5 1,521 was 49.5 h for a cellulose carbon preform and 3 1.75 h for a compacted powder of natural graphite. The results were fitted to the equation ~c”D(o=wlc)P CVD P, D- exp(- (t-tdM1 (6) where nzCVD was the mass of carbon infiltrated into the preform, k the overall kinetic constant, P, the initial porosity and pcvD = 2.030 g cmm3 the density of CVD carbon. Table 5 gives the values of to and the kinetic constants kl (constant of pore filling) and k2 (constant of pore closing) [ 501 calculated from the measurements. The time constants obtained from mass gain measurements (Table 5) were shorter than those obtained earlier from porosity measurements. An activation energy of 2.05 eV