1. Golecki/ Rapid wapor- phase densification of refractory composites Table 5 Kinetic constants of mass gain in isothermal, isobaric carbon CVI of 0. 2-0.45 cm thick carbon preforms [[44]: reproduced with kind permission from Editions Scientifiques, Elsevier] Substrate P。T k k2 content(gcm-3)(%)(h)(h)(10-3h)(10-3h)(h) Carbon fiber bound by II138 58.52345975 9 54.02381004.5 54 69 I-352 0967 4243381115.5 3.5 13.8 Unidirectional carbon fber bound IV-1 52 39.033 3.6 by CVD carbon Chopped carbon fiber bound by 0.270 9110 02 CVD carbon Carbon felt 0.120920689110 0.1 10 Carbon felt bound by CVD carbon V-2 12 0383 78.0 1650 224 22224 From porosity measurements [51,52 Molecule was obtained in the relatively narrow temperature range of 760-800C[44], lower than 2.47-2.65 eV molecule reported by others [53, 54]. The mass gain after 80 h of CVi at 800Cand t,=33 s, shown in Fig. Il, increased linearly with partial pressure of propylene in the range 30-90 Torr[44, 49]. As shown, at 44 Torr propylene, the mass gain decreased non-linearly with residence time. The conversion efficiency of propylene( the mass of infiltrated CVD carbon matrix divided by the amount of carbon introduced as propylene) increased with increasing temperature in the range 720-880C and generally with increasing residence time(33-100 s)or decreasing flow rate; the efficiency generally decreased with increasing propylene concentration [44, 491. Conversion efficien cies, generally in the 1-2% range, varied from a fraction of 1% to a maximum of 14% at 880 C, 6% propylene and t=100 S. These conversion efficiencies were averages over the entire infiltration runs However, the reported final densities of these C-C composites, 1. 1-1.5 g cm-3[44, 49, 50],were relatively low. The instantaneous conversion efficiency in isothermal, isobaric CVI is generally highest in the beginning of the run and decreases with time similarly to the rate of mass gain. Soot(an undesirable product of homogeneous gas-phase nucleation)was the main solid product found, reaching 73% at 720C. The soot yield increased with propylene concentration. The flexural strength of these C-C composites increased with their apparent density as a power law [44] o Residence Time(s) 1020304050 08078% E3≌z25 1.05% 1.5% 2.14% 0.5 30405060708090 ·C2 H Pressure(Tor) Fig. 11. Fractional mass gain in isothermal, isobaric carbon Cvi of thin carbon preform after 80 h at 800C in propylene: dependence on C3 Hs pressure(4" 33s)and residence time (p(C3 Hs)=44 Torr). The numbers are propylene conversion efficiencies in %[[49]; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 IGB, UK]
52 I. Golecki / Rapid vapor-phase densijication of refractory composites Table 5 Kinetic constants of mass gain in isothermal, isobaric carbon CVI of 0.2-0.45 cm thick carbon preforms [ [ 4.41; reproduced wifi kind permission from Editions Scientifiques, Elsevier] Substrate Fiber p,, r kz t content (gcmm3) k) ;h)’ (h) F;O-“h) ( 1O-3 h) (h) (vol.%) Carbon fiber bound by III-1 38 0.716 58.5 234 59 7.5 9.5 1.4 phenolic carbon III-2 42 0.793 54.0 238 100 4.5 5.4 6.9 III-3 52 0.967 42.4 338 111 5.5 3.5 13.8 Unidirectional carbon fiber bound IV-l 52 1.120 39.0 331 125 6.7 1.3 3.6 by CVD carbon Chopped carbon fiber bound by IV-2 9 0.270 85.0 91 1.0 0.2 17.9 CVD carbon Carbon felt V-l 8 0.120 92.0 68 91 1.0 0.1 10.4 Carbon felt bound by CVD carbon v-2 12 0.383 78.0 77 1.2 0.2 24.1 Polycrystalline graphite VI - 1.650 22.4 222 2.4 2.1 11.4 a From porosity measurements [51,52], molecule- ’ was obtained in the relatively narrow temperature range of 760-800 “C [ 441, lower than 2.47-2.65 eV molecule- ’ reported by others [ 53,541. The mass gain after 80 h of CVI at 800 “C and t,= 33 s, shown in Fig. 11, increased linearly with partial pressure of propylene in the range 30-90 Torr [ 44,491. As shown, at 44 Torr propylene, the mass gain decreased non-linearly with residence time. The conversion efficiency of propylene (the mass of infiltrated CVD carbon matrix divided by the amount of carbon introduced as propylene) increased with increasing temperature in the range 720-880 “C and generally with increasing residence time (33-100 s) or decreasing flow rate; the efficiency generally decreased with increasing propylene concentration [ 44,491. Conversion efficiencies, generally in the l-2% range, varied from a fraction of 1% to a maximum of 14% at 880 “C, 6% propylene and t, = 100 s. These conversion efficiencies were averages over the entire infiltration runs. However, the reported final densities of these C-C composites, 1.1-1.5 g cmB3 [44,49,50], were relatively low. The instantaneous conversion efficiency in isothermal, isobaric CVI is generally highest in the beginning of the run and decreases with time similarly to the rate of mass gain. Soot (an undesirable product of homogeneous gas-phase nucleation) was the main solid product found, reaching 73% at 720 “C. The soot yield increased with propylene concentration. The flexural strength of these C-C composites increased with their apparent density as a power law [44]. o Residence Time (s) 10 20 30 40 50 60 70 0.9 / I 4 30 40 50 60 70 80 9-O 100 l CsH6 Pressure (Torr) Fig. 11, Fractional mass gain in isothermal, isobaric carbon CVI of thin carbon preform after 80 h at 800 “C in propylene: dependence on C,H, pressure (t, = 33 s) and residence time (p(C,H,) =44 Torr). The numbers are propylene conversion efficiencies in % [ [49] ; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 lGB, UK]
L. Golecki/ Rapid vapor-phnse densification of refractory composite Time ig. 12. Schematic illustration of the densification of a thick preform vs time in isothermal, isobaric chemical vapor infiltration The bulk densification sequence of a thick(>2.5 cm) preform in this process can be depicted qualitatively as shown in Fig. 12. The approach to final density is composed of several approximately exponential steps. The exterior surfaces of the preforms must be ground between each step to open the surface pores and enable the gases to penetrate into the interior. Once the surface pores become sealed deposition continues but at a much reduced rate because the effective area is then equal to the geometrical surface area of the preforms. The total infiltration time required to reach the desired density will increase with preform thickness. The overall conversion efficiency for thick preforms will be lower than for thin ones, due to material lost in grindin The effect of different precursors on the carbon densification rate of carbon preforms was studied recently [55]. The randomly-oriented pitch-fiber tow preforms were impregnated with phenolic resin and were densified in a mixture of methane, propylene and hydrogen at 15 Torr. The preform size was 7.6X76X16cm,, its initial density was 1.25 g cm-3 and initial open porosity was 38 vol%.The preform weight was continuously recorded by means of an in-situ balance. Methane flowing at 400 sccm was used as the primary precursor, hydrogen was added at 100 sccm and c3H ato to 75 sccm in the temperature range 1000-1150C. The hot zone of the furnace was 15. 2 cm in diameter by 30.5 cm long. The curves of fractional weight gain vs. time had an exponential-type behavior and were fitted to the equation Am/m=a1(PcvD/po)In(1+a2[exp(-a3t-a4r)]-1, where the a, parameters were constants. At 1050C, the addition of increasing fractions of propylene resulted in increases of both the initial deposition rate and the final density, shown in Fig. 13. The microstructure of the deposited carbon changed from in the absence of C3hs to anisotropic [1] when 35 sccm C3H was added. With a C3H flow rate of 35 sccm, increasing the temperature from 1000 to 1100C CH=400 sccm 10°C =400 sccm Q 30FH2=100 sccm H=lC0 sccm °c 0 sccm C3H6 C3H6=358cm T=1050(C 020406080100120020406080100120 Fig. 13. Relative weight gain vs time in isothermal, isobaric carbon CVI from methane and H2, showing the effects of (a)temperature and (b) propylene additions [[55]; reprinted with the authors kind
I. Golecki / Rapid vapor-phase densijcation of refractory composites 53 Time Fig. 12. Schematic illustration of the densitication of a thick preform vs. time in isothermal, isobaric chemical vapor infiltration, The bulk densification sequence of a thick ( 2 2.5 cm) preform in this process can be depicted qualitatively as shown in Fig. 12. The approach to final density is composed of several approximately exponential steps. The exterior surfaces of the preforms must be ground between each step to open the surface pores and enable the gases to penetrate into the interior. Once the surface pores become sealed, deposition continues but at a much reduced rate because the effective area is then equal to the geometrical surface area of the preforms. The total infiltration time required to reach the desired density will increase with preform thickness. The overall conversion efficiency for thick preforms will be lower than for thin ones, due to material lost in grinding. The effect of different precursors on the carbon densification rate of carbon preforms was studied recently [ 551. The randomly-oriented pitch-fiber tow preforms were impregnated with phenolic resin and were densified in a mixture of methane, propylene and hydrogen at 15 Torr. The preform size was 7.6 X 7.6 X 1.6 cm3, its initial density was 1.25 g cmm3 and initial open porosity was 38 vol.%. The preform weight was continuously recorded by means of an in-situ balance. Methane flowing at 400 seem was used as the primary precursor, hydrogen was added at 100 seem and C3H6 at 0 to 75 seem in the temperature range 1000-l 150 “C. The hot zone of the furnace was 15.2 cm in diameter by 30.5 cm long. The curves of fractional weight gain vs. time had an exponential-type behavior and were fitted to the equation Am/m = a, (p cv,,/Po)ln{ 1 +a,[exp( -a3t-ua,?)] -l}, wheretheajparameters were constants. At 1050 “C, the addition of increasing fractions of propylene resulted in increases of both the initial deposition rate and the final density, shown in Fig. 13. The microstructure of the deposited carbon changed from isotropic in the absence of C3H6 to anisotropic [ 1 ] when 35 seem C3H6 was added. With a C3H6 flow rate of 35 seem, increasing the temperature from 1000 to 1100 “C Fig. (b) 40 10 0 0 20 40 60 80 loo 120 Oo 20 40 60 80 100 120 Time (h) Time (h) 13. Relative weight gain vs. time in isothermal, isobaric carbon CVI from methane and HZ, showing the effects of propylene additions [ [55] ; reprinted with the authors’ kind permission]. (4 temperature and
. Golecki/ Rapid vapor-phase densification of refractory composites increased the initial densification rate and the final density, whereas at 1150%C, premature surface closure and non-uniform deposition through the thickness were seen The deposition rate of CVD carbon(normalized to the number of C atoms in the precursor molecule and to the partial pressure )from different precursors on fat alumina substrates was found to increase with the precursor in the following order: CH4 methane, C2H4 ethylene, C2H2 acetylene C3H propylene, CsH benzene [56]. The temperature was 1100C for methane and 1000C for the other precursors. The residence time was varied between 0.1 and 1 s, the pressure was about 0.5 atm and the runs were 5-6 h long. The concentrations of the by-product gases were measured by chroma tography and the amount of carbon by a carbon mass balance. The deposition rate increased exponen- tially with tr for CHa(main by-products were H2, C2H] and condensibles), C2H4(by-products wcre H2, C2H2, condensibles and small amounts of CH4 )and C2H2(by-products were H2 and condensibles); for C31I6(by-products H2, condensibles, C2H4, C2H2 and CH)the deposition rale was constant at tr=0.5 S, increasing thereafter; for C6Hs(by-products H2 and condensibles), the deposition rate decreased with t, up to about o5s, and then increased Attr=l s, the deposition rate of all the precursors except C6H was similar and about 2.2 times lower than that of C6Hg. A comprehensive review of pyrolytic carbon deposition can be found in Ref. [18 and additional studies in Refs. [ 57- 64, 32, 36,65,66] Modeling results of isothermal, isobaric CVI are described in references already cited and in Refs. [67-74. Properties of composites densified by isothermal, isobaric CVI can be found in the references and in section 10 4.1. Advantages of isothermal, isobaric CVI (al) The method is well-established and relatively well-understo (a2)A large number of preforms can be densified simultaneously (a3)The densification time per preform is relatively low for heavy loading of the reactor. (a4)Preforms of different and complex shapes and sizes can be readily densified in the run, although the minimum dimension (usually the thickness )needs to be similar (a5) The energy expenditure per part is relatively low. 4.2. Disadvantages of isothermal, isobaric CVT ( d1) The total processing time at temperature is very long, typically 600-2000 h per batch(but the time does not depend on the number of parts being densified) (d2)Spatial density gradients exist within the composite, where the density is generally highest at the outer surfaces and lowest in the interior regions (d3) Premature surface crusting occurs before the desired bulk density is reached. (d4) The CVi process must be interrupted several times to enable grinding of the surfaces, hereby opening surface pores and allowing further infiltration (d5)Preforms having different minimum dimensions(usually the thickness) densify at dif- ferent rates; thinner preforms will densify faster than thicker ones (d6)The cost of capital is relatively high, since the method favors densification of a large number of parts per batch, thus requiring relatively large CVi reactors d7) The cost of inventory and potential re-work or scrap can be high, due to the large number of parts per batch. (d8)Process development or changes to a process may be slow to implement because of the long duration of the process (d9)The overall precursor conversion efficiency is relatively low, of the order of 1-2%
54 I. Golecki / Rapid vapor-phase densifrcation of refractory composites also increased the initial densification rate and the final density, whereas at 1150 “C, premature surface pore closure and non-uniform deposition through the thickness were seen. The deposition rate of CVD carbon (normalized to the number of C atoms in the precursor molecule and to the partial pressure) from different precursors on flat alumina substrates was found to increase with the precursor in the following order: CH4 methane, C2H4 ethylene, C2Hz acetylene, C3H6 propylene, C6H6 benzene [ 561. The temperature was 1100 “C for methane and 1000 “C for the other precursors. The residence time was varied between 0.1 and 1 s, the pressure was about 0.5 atm and the runs were 5-6 h long. The concentrations of the by-product gases were measured by chromatography and the amount of carbon by a carbon mass balance. The deposition rate increased exponentially with t, for CH4 (main by-products were HZ, C2H2 and condensibles) C2H4 (by-products were HZ, C2H2, condensibles and small amounts of CH4) and &Hz (by-products were H, and condensibles) ; for C3H5 (by-products Ha, condensibles, CzH4, CzH2 and CHJ the deposition rate was constant at t, = 0.5 s, increasing thereafter; for C6H6 (by-products Hz and condensibles), the deposition rate decreased with t, up to about 0.5 s, and then increased. At tr = 1 s, the deposition rate of all the precursors except C6H6 was similar and about 2.2 times lower than that of C6H6. A comprehensive review of pyrolytic carbon deposition can be found in Ref. [ 181 and additional studies in Refs. [57- 64,32,36,65,66]. Modeling results of isothermal, isobaric CVI aredescribed in references already cited and in Refs. [ 67-741. Properties of composites densified by isothermal, isobaric CVI can be found in the references and in Section 10. 4.1. Advantages of isothemal, isobaric CVl (al ) The method is well-established and relatively well-understood. (a2) A large number of preforms can be densified simultaneously. (a3) The densification time per preform is relatively low for heavy loading of the reactor. (a4) 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 - see disadvantage ( d5 ) be10 w . (a5) The energy expenditure per part is relatively low. 4.2. Disadvantages of isothermal isobaric CVI (dl) The total processing time at temperature is very long, typically 600-2000 h per batch (but the time does not depend on the number of parts being densified) . (d2) Spatial density gradients exist within the composite, where the density is generally highest at the outer surfaces and lowest in the interior regions. (d3) Premature surface crusting occurs before the desired bulk density is reached. (d4) The CVI process must be interrupted several times to enable grinding of the surfaces, thereby opening surface pores and allowing further infiltration. (d5) Preforms having different minimum dimensions (usually the thickness) densify at different rates; thinner preforms will densify faster than thicker ones. (d6) The cost of capital is relatively high, since the method favors densification of a large number of parts per batch, thus requiring relatively large CVI reactors. (d7) The cost of inventory and potential re-work or scrap can be high, due to the large number of parts per batch. (d8) Process development or changes to a process may be slow to implement because of the long duration of the process. (d9) The overall precursor conversion efficiency is relatively low, of the order of l-2%
I. Golecki/ Rapid vapor-phase densification of refractory composites 5. Plasma-enhanced, isothermal or thermal-gradient, low-pressure CVI This approach, which was used for the densification of C-C compos described in Refs [12, 14, 15]. The goal was to obtain a higher deposition rate and, in principle, a shorter densification time by producing excited neutral precursor species, radicals and ions in a plasma. In the earlier study [12],ad c. discharge in pure methane was used(Fig 14). The preforms were single-layer( thickness unspecified), 1.2X10 cm"satin-woven polyacrylonitrile(PAN)-based carbon substrates from Societe Europccnnc dc Propulsion(SEP). The substrates(one per run) were Joule heated by passing an electrical current through the warp yarns. That substrate surface which was mainly constituted of warp yarns faced the anode and was hotter Lhan the opposite surface, which was made up principally of weft yarns. Thus, there was an unspecified thermal gradient across the thickness of the substrate during the densification. The substrate temperature was measured on its hotter face pyrometrically and checked by a thermocouple. The preform floated electrically in the discharge. The conditions during carbon CVI were: methane pressure, 6 Torr; flow rate, 50 sccm; preform temperature, 850 C; d. c, electric field, 50 V cm d c input power, 35 W; electrode diameter, 8 cm; inter-electrode spacing, 6 cI these substrates were denoted"P". Comparative runs were performed without the plasma discharge at the same conditions(denoted"NP"')and also [751 at 1000C, 110 Torr, 500 sccm(denoted D"). Fig. 15 shows the rate of mass gain for the three types of processing conditions. After 100 h of densification, the infiltration rate in the plasma at 850C was 35 times higher compared to the value without plasma. This result is not too surprising, in view of the relatively low temperature. The infiltration rate with plasma at 850C, was, however, also higher by 25% compared to that without plasma at 1000C. The microstructure of the deposited carbon was measured using polarized-light microscopy and found to be smooth-laminar under all three proccssing conditions. The matrix density was determined by immersing 0. 1-0.2 mm diameter grains in a benzene-tetrabromoethane mixture the values were 1.76, 1.77 and 1.96 g cin for" P","NP"and"D"conditions, respectively. The density of the preforms(and thus the fibers), measured by the same technique was 1.73 g cm-3.The composite density was not given but the authors stated that it was lower in the""P'andNP''samples compared to the"D''samples. As seen in Fig. 16, the authors found by measuring the thickness of the carbon coating [14] through SEM that the density in the interior of the preform was apparently Anode Fig. 14. D.C. plasma CVI reactor for infiltrating C-C composites [[12]; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK]
1. Golecki /Rapid vapor-phase densi$cation of refractory composites 55 5. Plasma-enhanced, isothermal or thermal-gradient, low-pressure CVI This approach, which was used for the densification of C-C composites, is described in Refs. [ 12,14,15]. The goal was to obtain a higher deposition rate and, in principle, a shorter densification time by producing excited neutral precursor species, radicals and ions in a plasma. In the earlier study [ 121, a d.c. discharge in pure methane was used (Fig. 14). The preforms were single-layer (thickness unspecified), 1.2X 10 cm2 satin-woven polyacrylonitrile (PAN) -based carbon substrates from Societd Europeenne de Propulsion (SEP). The substrates (one per run) were Joule heated by passing an electrical current through the warp yarns. That substrate surface which was mainly constituted of warp yarns faced the anode and was hotter than the opposite surface, which was made up principally of weft yarns. Thus, there was an unspecified thermal gradient across the thickness of the substrate during the densification. The substrate temperature was measured on its hotter face pyrometrically and checked by a thermocouple. The preform floated electrically in the discharge. The conditions during carbon CVI were: methane pressure, 6 Torr; flow rate, 50 seem; preform temperature, 850 “C; d.c. electric field, 50 V cm-‘; d.c. input power, 35 W; electrode diameter, 8 cm; inter-electrode spacing, 6 cm; these substrates were denoted “P”. Comparative runs were performed without the plasma discharge at the same conditions (denoted “NP”) and also [75] at 1000 “C, 110 Torr, 500 seem (denoted “D”). Fig. 15 shows the rate of mass gain for the three types of processing conditions. After 100 h of densification, the infiltration rate in the plasma at 850 “C was 35 times higher compared to the value without plasma. This result is not too surprising, in view of the relatively low temperature. The infiltration rate with plasma at 850 “C, was, however, also higher by 25% compared to that without plasma at 1000 “C. The microstructure of the deposited carbon was measured using polarized-light microscopy and found to be smooth-laminar under all three processing conditions. The matrix density was determined by immersing 0.1-0.2 mm diameter grains in a benzene-tetrabromoethane mixture; the values were 1.76, 1.77 and 1.96 g cme3 for “P”, “NP” and “D” conditions, respectively. The density of the preforms (and thus the fibers), measured by the same technique was 1.73 g crnm3, The composite density was not given but the authors stated that it was lower in the “P” and “NP” samples compared to the “D” samples. As seen in Fig. 16, the authors found by measuring the thickness of the carbon coating [ 141 through SEM that the density in the interior of the preform was apparently . . . . . . . . . ..__..... I Flow meters / . . . . .._________.._.._( Needle valves ,....................... Pumping system . . . . . . . .._._...............____ Fig. 14. DC. plasma CVI reactor for infiltrating C-C composites [ [ 121; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 lGB, UK]
. Galecki/Rapi se densification of refractory composite D1000c NP850°C Deposition Time(h) Fig. 15. Rate of mass gain of single-layer PAN-based C-C composite infiltrated in a d. c. methane plasma( P)or without a plasma(NP, D) [[12]; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 IGB, UK] P near Cathode Anode trate Thickness Fig. 16. Axial thickness profile of carbon deposited in single- layer C-C composite infiltrated in a d. c. methane plasma(P)or without a plasma(NP)[[12]; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 IGB, UK] significantly lower than that at either exterior surface. This trend was similar to that observed in isothermal, isobaric CVI. The deposition rate on the surface facing the cathode was higher than on the opposing surface, presumably due to the thermal gradient across the preform. The carbon deposition rate in the interior of the preform was still higher by 4-10 times in the presence of the plasma, compared to the case without plasma. The authors stated that the ratio of carbon deposited on the external surfaces to that deposited in the interior could be reduced by an order of magnitude by varying the position of the preform in the discharge Results of isothermal a.c. plasma-enhanced carbon CVi are described in Refs. [14, 15]. The electrical discharge was generated in pure methane(8-50 sccm flow rate)or in a methane/hydrogen mixture flowing in a 4 cm diameter quartz tube, by means of a coil driven at 13. 56 MHz at a power level of 0-200W. The total pressure was in the range 0.. 6 Torr and the CH,/H2 ratio was measured with a gas chromatograph. The tube was part of a conventional, radiantly-heated furnace, constituting a hot-wall CVI reactor(see Fig 3). The temperature range studied was 720-1200C; the hot zone of the reactor, where the substrates were located, was 15 cm long and had a temperature uniformity within +5C. The plasma was generated upstream of the hot zone, and the porous carbon substrates were inside the discharge region. The excited neutral species in the plasma had the longest lifetime of the order of 1 s, which was similar to the gas residence time in the reactor (0.1-10 s); the free radicals had a much shorter lifetime of 0. 1-10 ms and the ionic species had the shortest lifetime, in the range
56 I. GoEecki/Rapid vapor-phase densification of refractory composites 0 100 200 Deposition Time (h) Fig. 15. Rate of mass gain of single-layer PAN-based C-C composite infiltrated in a d.c. methane plasma (P) or without a plasma (NP, D) [ [ 121; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 lGB, UK]. Substrate Thickness Fig. 16. Axial thickness profile of carbon deposited in single-layer C-C composite infiltrated in a d.c. methane plasma (P) or without a plasma (NP) [ 1121; reprinted with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 lGB, UK]. significantly lower than that at either exterior surface. This trend was similar to that observed in isothermal, isobaric CVI. The deposition rate on the surface facing the cathode was higher than on the opposing surface, presumably due to the thermal gradient across the preform. The carbon deposition rate in the interior of the preform was still higher by 4-10 times in the presence of the plasma, compared to the case without plasma. The authors stated that the ratio of carbon deposited on the external surfaces to that deposited in the interior could be reduced by an order of magnitude by varying the position of the preform in the discharge. Results of isothermal a-c. plasma-enhanced carbon CVI are described in Refs. [ 14,151. The electrical discharge was generated in pure methane (8-50 seem flow rate) or in a methane/hydrogen mixture flowing in a 4 cm diameter quartz tube, by means of a coil driven at 13.56 MHz at a power level of O-200 W. The total pressure was in the range 0.2-7.6 Torr and the CH4/H2 ratio was measured with a gas chromatograph. The tube was part of a conventional, radiantly-heated furnace, constituting a hot-wall CVI reactor (see Fig. 3). The temperature range studied was 720-1200 “C!; the hot zone of the reactor, where the substrates were located, was 15 cm long and had a temperature uniformity within f 5 “C. The plasma was generated upstream of the hot zone, and the porous carbon substrates were inside the discharge region. The excited neutral species in the plasma had the longest lifetime of the order of 1 s, which was similar to the gas residence time in the reactor (0.1-10 s) ; the free radicals had a much shorter lifetime of 0.1-10 ms and the ionic species had the shortest lifetime, in the range