Improvement of film boiling chemical vapor infiltration process
for fabrication of large size C/C composite
Ji-ping Wang⁎, Jun-min Qian, Guan-jun Qiao, Zhi-hao Jin
State Key Laboratory for Mechanical Behavior of Materials, Xi'an Jiaotong University, Xi'an 710049, People's Republic of China
Received 21 June 2005; accepted 5 November 2005
Available online 28 November 2005
An improved film boiling chemical vapor infiltration process was developed to fabricate a large size C/C composite with homogeneous density
and microstructure. The C/C composite was prepared by processing a disc-shaped carbon felt preform, whose upper and lower sides were fixed
and heated simultaneously by two flat surfaces of two heat sources, with kerosene as a precursor at 1050 °C for 3 h at an atmospheric pressure.
The in-situ temperature distribution along the radial direction of the preform upper surface was analyzed to get better information and control of
the process. Experimental results show that the average density of the composite of Φ 110×10 mm3size is about 1.72 g/cm3and its maximal
difference along radial direction is 0.05 g/cm3. Polarized light microscopy (PLM) and scanning electron microscopy (SEM) reveal that the carbon
fibers of the composite are surrounded by ring-shaped pyrocarbons with a thickness of ∼20 μm, and that pyrocarbons are delaminated to 4–6
layers. A schematic model is proposed to analyze the process by dividing the reactor into different regions associated with specific functions.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Carbon/carbon composite; Chemical vapor infiltration; Rapid densification; Microstructure
Carbon/carbon (C/C) composites are widely applied in many
fields for their low density, excellent thermal and mechanical
properties with smooth frictional behavior and good biocom-
patibility . Currently, the main method for fabricating C/C
composites in industry is the isothermal chemical vapor
infiltration (CVI) technique. However, it has a major intrinsic
drawback, namely, a long processing period is inevitable to
obtain desired density [2,3]. Fortunately, another method called
as film boiling chemical vapor infiltration (FBCVI) or chemical
liquid-vaporized infiltration (CLVI) [4–7] has been developed
to increase the deposition efficiency. It appears very attractive to
prepare C/C composite in a short processing time with a high
carbon yield which is about one order of magnitude larger than
by classic isothermal CVI.
It is known that the FBCVI method involves a strong thermal
gradient inside cold wall reactor. A mobile densification front is
created in a porous preform which is directly immersed into a
liquid hydrocarbon precursor. The principle, the experimental
device, and the influences of some basic parameters (temper-
ature, pressure, precursors, etc.) have already been well studied
[3,5,6]. Further investigations are carried out experimentally or
theoretically to reveal the complex chemical reactions leading to
the pyrocarbon matrix in a confined place and the role of the
heat and mass transfers inside porous preform [8–11].
Nevertheless, these studies were mainly carried out in
laboratory reactors. The prepared C/C composites are usually of
thin-walled tubular shape with small dimensions (the wall
thickness is below 35 mm). Moreover, spatial density gradients
exist in the composites, where the density at the interior regions
near the heat source is the highest and that at the outer surfaces
near the liquid precursor is the lowest [11,12]. Therefore, further
improvement of this process is necessary for preparing large
bulk C/C composite of more regular shape with homogeneous
density distribution and uniform microstructure.
For this purpose, a double heat source design is firstly
developed in the present work. A large size C/C composite disc
was fabricated by this improved FBCVI method. To get better
information and control of the whole process, the in-situ
temperature distribution in the preform was recorded and
Materials Letters 60 (2006) 1269–1272
⁎Corresponding author. Tel.: +86 29 82667942; fax: +86 29 82665443.
E-mail address: email@example.com (J. Wang).
0167-577X/$ - see front matter © 2005 Elsevier B.V. All rights reserved.
analyzed. The density, porosity and microstructures of the
prepared composite are characterized by Archimedes principle,
polarized light microscopy and scanning electron microscopy
techniques. Finally, a schematic model is proposed to study the
2. Experimental procedure
2.1. Material preparation
2.1.1. Preform and precursor
A PAN-based carbon felt (thickness: 10 mm, bulk density:
∼0.20 g/cm3, fiber diameter: Φ 9–13 μm) was used as a
preform in this study. A liquid hydrocarbon mixture, kerosene
(molecular formula: C10Hn–C16Hm) with a boiling temperature
range of 180–230 °C, was chosen as a precursor, which was
proven to be a feasible and efficient precursor .
2.1.2. Experimental set-up
The experimental device is schematically shown in Fig. 1. In
a cylindrical quartz glass reactor, two graphite cylinders (H1
and H2) with diameter of Φ110 mm were placed at the same axis
as the reactor and inductively heated by an inductor coil. The
preform cut in disc shape (Φ110 mm) was fixed between the
lower surface of H1 and the upper surface of H2 and heated by
both of them. The residual surfaces of the two graphite cylinders
were wrapped with a thermal insulator. Both the cylinders and
the preform were immersed into the liquid precursor. Three
thermocouples (T1, T2 and T3) were located at the interface of
H1 and the preform. Their distances to the axis are 0 mm, 25
mm and 50 mm, respectively. Thus the inner, middle and outer
temperatures of the upper surface of the preform can be
simultaneously measured by them during the process.
The deposition of the preform was performed at 1050 °C
(measured by T1 thermocouple) for 3 h at an atmospheric
pressure by the improved FBCVI. During the process, the
heating rate was controlled by adjusting the power of an
inductive generator. N2flowed through the reactor for safety
consideration. With the increasing deposition time, the
precursor was consumed and resupplied from the precursor
inlet to the reactor.
The dimensions of the as-prepared C/C composites are Φ
110×10 mm3. In order to study the homogeneity of density and
microstructure, three specimens labeled as S1, S2 and S3 were
cut off from the composite along the radial direction, which was
located at the distance of 0 mm, 25 mm and 50 mm to the axis
(corresponding to the T1, T2 and T3 location), respectively. The
densities and open porosities of the three specimens were
measured using Archimedes principle. Microscopy observa-
tions on the polished surfaces and the fracture surfaces of the
specimens were carried out by polarized light microscope
(PLM, Reichert, MeF3) and scanning electron microscope
(SEM, Hitachi, S-2700) operated at 20 kV and 20 mA,
3. Results and discussion
3.1. Temperature distribution in the preform
The key point of this process is the control of temperature
distribution in the preform, either in axial or in radial direction [5,11].
Therefore, we designed a double heat source. This creates two
extensive thermal gradients in the axial direction of the preform. Both
sides of the preform are heated simultaneously by H1 and H2. The
temperatures of the upper and lower surfaces of the preform are the
highest. The deposition firstly occurred on these two surfaces and then
the formed densification zones moved successively to the lower
temperature zone. Thus, a rapid single cycle densification of the
preform can be achieved due to the double high temperature gradients
along the axis.
In the radial direction of the preform, a uniform temperature of the
heat surface is required to obtain uniform thickness and microstructure
of the composite. Under the experimental conditions of the present
paper, the temperature distribution in the radial direction of the upper
surface of the preform is shown in Fig. 2, which was recorded
Fig. 1. Sketch of experimental device for preparing C/C composite disc.
Fig. 2. Temperatures vs. time of three thermocouples recorded during the
1270 J. Wang et al. / Materials Letters 60 (2006) 1269–1272
simultaneously by three thermocouples during the process. It can be
seen that a very high heating rate was obtained using inductive heating,
and the temperatures exceeded 900 °C in 10 min. During the deposition
process, when T1 temperature was kept at 1050 °C, T2 temperature
was about 30 °C higher and T3 temperature was about 40 °C lower than
T1 temperature. An expected even temperature distribution along the
upper surface of the preform was obtained.
The temperature distribution in radial direction of the preform was
the balance of the inductive heating and thermal exchanges between the
heaters and the surrounding environment in the reactor. Theoretical
calculations of the thermal exchange in the preform are very difficult
and complex . In this study, the temperatures mainly depend on the
height of H1, the thickness of the thermal insulator, and the inductive
heat frequency. Thus, the desired temperature distribution along the
upper surface of the preform can be obtained by adjusting such
3.2. Density and open porosity of the composite
The densities and open porosities of the three specimens are shown
in Table 1. The density increases up to 1.70 g/cm3from the initial
density ∼0.2 g/cm3after 3 h of densification. This result reveals that a
rapid densification of the preform is achieved by the improved FBCVI
Comparing the values of S1, S2 and S3, we can see that densities
along the radial direction of the composite are close and there is a
small difference of 0.05 g/cm3among them. This uniform density
distribution in the composite is the result of controlling the radial
temperature distribution of the preform. The S2 density is a little
higher than S1 and S3 because its corresponding deposition
temperature (T2) is higher, which leads to a higher densification
rate. The open porosity of the composite in Table 1 is higher than
10%. We can expect an increase of the density with increasing of
Densities and open porosities of S1, S2 and S3 specimens
Fig. 3. PLM micrographs of polished sections of S1 (a), S2 (b) and S3 (c)
Fig. 4. SEM micrographs of the fracture surface of S1 (a), S2 (b) and S3 (c)
1271 J. Wang et al. / Materials Letters 60 (2006) 1269–1272
3.3. Microstructure of the composite
It is known that the qualities of the C/C composites, such as density
and mechanical, electrical or thermal properties are determined by the
microstructure of pyrocarbon . Fig. 3 shows the PLM micrographs
of the polished surfaces of S1, S2 and S3 specimens. It can be seen that
the fibersare surrounded concentrically by a ringpyrocarbon whichhas
optically anisotropic domains (dark cross appears systematically in the
bright deposit). The thicknesses of pyrocarbons are all about 20 μm and
the textures are smooth laminar according to the classification method
proposed by Pierson and Lieberman [14–16]. Small cracks can also be
seeninpyrocarbonmatrix and atinterfaceoffiberandpyrocarbon. This
can be attributed to the thermal expanding coefficients mismatch
are shown in Fig. 4. It shows that the surface of the pyrocarbon
surrounding one fiber is rough and laid up with 5–6 layers.
Comparisons of the PLM or SEM micrographs of S1, S2 and S3
reveal that there are no distinct differences, which indicates that the
microstructures of pyrocarbon are similar.
3.4. Model of the improved FBCVI
In order to get a better understanding of the densification process, a
schematic model of the improved FBCVI is presented in Fig. 5. In this
model, the differences of temperature along the radial direction and that
in upper and lower surface of the preform are negligible.
We divided the reactor into different regions and endowed different
functions to each of the regions.
1. Densified preform: during the process, two high thermal gradients
exist in the axial direction of the preform due to the use of two heat
sources.Theupper andlowersurfaces ofthe preformcontacted with
the hot surface are firstly densified. The densification zones moved
successively to lower temperature zone along the thermal gradient.
2. Densification zone: the chemical reactions occurred in these zones
where steep thermal gradient exist. The decomposition of the
precursors is complex, including many unit reactions and a large
number of intermediate species [6,10,11].
3. Undensified zone: in the undensified porous preform, the heat and
mass exchange in a complex way. The precursor is fed from the
outside and the reaction products such as hydrogen have to be
4. Boiling precursor: the boiling precursor outside is considered as a
mass reservoir for carbon precursor surrounding the preform. The
carbon source for the infiltration and densification can be
transported to the preform and reaction zone rapidly.
Based on this schematic model, it can be concluded that the
improved FBCVI process has the following three advantages compared
with the classical one: (1) a more rapid densification rate because of the
existing double densification front; (2) the diameter of the disc is not
limited by the densification thickness; and (3) the as-prepared disc can
be easily removed from the graphite cylinder surface without
An improved FBCVI method was developed and used to
prepare a large size C/C composite disk of Φ 110×10 mm3at
1050 °C for 3 h at an atmospheric pressure. The average density
of the resulting composite is about 1.72 g/cm3, and the
maximum difference of density along the radial direction is 0.05
g/cm3. The carbon fibers of the composite are surrounded by
ring-shaped pyrocarbons with a thickness of ∼20 μm, and
pyrocarbon is delaminated to 4–6 layers.
A schematic model was proposed to analyze the process by
dividing the reactor into different regions associated with
specific functions. The double-heat source design allows a
double densification front which can accelerate the densifica-
tion theoretically. Moreover, the diameter of the C/C composite
disc is not limited by the densification thickness. Therefore, this
process will have a promising application for rapid preparation
of large size C/C composite.
This study was supported by the National Natural Science
Foundation of China (No. 50272051).
 E. Fitzer, L.M. Manocha, Carbon Reinforcements and Carbon/Carbon
Composites, Springer-Verlag, New York, 1998.
 I. Golecki, Mater. Sci. Eng., R Rep. 20 (1997) 37.
 P. Delhaes, Carbon 40 (2002) 641.
 M. Houdayer, J. Spitz, D. Tran-Van, U. S. Patent No. 4,472,454,
September 18, 1984.
 E. Bruneton, B. Narcy, A. Oberlin, Carbon 35 (1997) 1593.
 D.Rovillain,M.Trinquecoste, E.Bruneton,A. Derre,P. David,P.Delhaes,
Carbon 39 (2001) 1355.
 W.C. Sun, H.J. Li, H.M. Han, S.Y Zhang, K.Z. Li, Mater. Sci. Eng., A
A369 (2004) 245.
 E. Bruneton, B. Narcy, A. Oberlin, Carbon 35 (1997) 1599.
 P. Delhaes, M. Trinquecoste, A. Derre, D. Rovillain, P. David, Carbon Sci.
4 (2003) 163.
 H. Wang, H.F. Yang, X.Q. Ran, Q.Z. Shi, Z.Y. Wen, J. Mol. Struct.,
Theochem 571 (2001) 115.
 P. Delhaes, M. Trinquecoste, J.-F. Lines, A. Cosculluela, J.-M.
Goyheneche, M. Couzi, Carbon 43 (2005) 681.
 A.L. Ji, X.X. Huo, B.X. Ma, Guti Houjian Jishu 24 (2001) 68.
 A. Oberlin, Carbon 40 (2002) 7.
 M.L. Lieberman, H.O. Pierson, Carbon 12 (1974) 233.
 H.O. Pierson, M.L. Lieberman, Carbon 13 (1975) 159.
 B. Reznik, K.J. Huttinger, Carbon 40 (2002) 617.
Fig. 5. A schematic model of the improved FBCVI process.
1272 J. Wang et al. / Materials Letters 60 (2006) 1269–1272