Accepted Manuscript

CFD Coupled Kinetic Modeling and Simulation of Hot Wall Vertical Tubular
Reactor for Deposition of SiC Crystal from MTS

P.K. Mollick, R.Venugopalan, D. Srivastava

PII: S0022-0248(17)30421-9
DOI: http://dx.doi.org/10.1016/j.jcrysgro.2017.06.004
Reference: CRYS 24207

To appear in: Journal of Crystal Growth

Received Date: 16 March 2017

Revised Date: 15 May 2017
Accepted Date: 8 June 2017

Please cite this article as: P.K. Mollick, R.Venugopalan, D. Srivastava, CFD Coupled Kinetic Modeling and
Simulation of Hot Wall Vertical Tubular Reactor for Deposition of SiC Crystal from MTS, Journal of Crystal
Growth (2017), doi: http://dx.doi.org/10.1016/j.jcrysgro.2017.06.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 CFD Coupled Kinetic Modeling and Simulation of Hot
Wall Vertical Tubular Reactor for Deposition of SiC
Crystal from MTS

P. K. Mollick*, R. Venugopalan and D. Srivastava

Mechanical Metallurgy Division, Bhabha Atomic Research Centre, Trombay, Mumbai-400085

*Corresponding author. Tel: +91 22 25593937, Email: pmollick@barc.gov.in (P. K. Mollick)

ABSTRACT

Chemical Vapor Deposition (CVD) process is generally carried out in a hot wall reactor of vertical or
horizontal type keeping the substrate inside the chamber on which deposition is targeted. Present study
is focused to explain the role of hydrodynamics and temperature conditions on the overall coating rates
inside a hot wall vertical tubular reactor. Deposition of β-Silicon Carbide crystals from
Methytricholorosilane catalyzed by hydrogen is modeled here considering growth kinetics which can
be successfully described – using only two steps. Finite Element Method based simulation is performed
to obtain the flow and temperature profiles inside the hot wall reactor. Model equations for kinetics are
derived in differential form based on mass balance considering transport of species. Kinetic
parameters were approximated comparing the experimentally found coating rates as reported earlier.
Present model is seen to fit reasonably well for the wide variation of gas flow rates as well as
temperature. Apart from the flow rates of total fluid at inlet and initial wall temperature of reactor,
sample position and the inlet diameter of the reactor are found to be key important parameters for the
desired coating to take place. Model prediction thus can provide better knowledge in order to carefully
choose process parameters in designing the reactor for achieving optimized deposition rates by CVD
with desired properties.

Key words: silicon carbide, chemical vapor deposition, tubular reactor, computational fluid dynamics,
modeling and simulation, reaction kinetics, methyltrichlorosilane
1. Introduction:

Silicon carbide has low thermal expansion coefficient, low specific mass, high melting point, strength
and hardness due to its excellent covalent bond between silicon and carbon. In view of these favorable
properties, SiC has been widely accepted as an excellent structural ceramic material [1-5]. Engineering
materials are proven to fail under harsh environment and hence extensive interests on coating with SiC
was developed due to its anti-oxidant properties, provide safety from chemical corrosion and are also
used as pressure vessel for nuclear fuel containment [6-12]. It is also extensively accepted as
semiconductor material due to its wide band gap and high electron mobility [4].

Chemical vapor deposition is a common technique used for high purity SiC deposition and
requires high temperature environment [13-17]. For the various advance applications of thin films,
apart from carbide, many oxides, borides of various metals like copper, strontium and indium thin
films are seen to be grown by vapor deposition techniques sometimes using plasma at low temperature
and low pressure [18-24]. Hot wall vertical/ horizontal tubular reactor is generally used to provide the
required temperature where cracking of precursor takes place [9-10, 13-16, 25-27]. The deposition rate,
coating uniformity and its properties are highly controlled by various process parameters such as,
chemical kinetics with selectivity, temperature and pressure inside the chamber [6, 13, 25-27].
However, the effects of reactor geometry, hydrodynamics and temperature distribution inside the
coating chamber of various designs are reported recently in order to optimize coating process [6, 9-10,
14-16, 27-34].

Methyltrichlorosilane (MTS) is commercially used as conventional source for silicon carbide

and its decomposition is catalyzed by hydrogen. Hydrogen assisted pyrolysis followed by
heterogeneous deposition is associated with complex mechanism of MTS cracking resulting in
formation of many intermediates. Scientists have reported probable mechanisms for the deposition
reactions along with their corresponding activation energies considering those intermediates. However,
researchers paid more attention towards formation of C2H4, CH4 and SiCl2, SiCl4 as major
intermediates, and their reactions to form SiC with very high rate of formation among other
intermediates [9-10, 14-16, 29, 32, 37-42].

In the current work we primarily focus on estimating kinetic parameters which are
simultaneously affected by hydrodynamics as well as temperature profile inside the CVD chamber.
Kinetic parameters are estimated comparing earlier reported coating rates (experimental) at various
process conditions [1]. A very similar methodology which is reported earlier [25] to estimate kinetic
parameters by approximation and comparing with experimental data against all process parameters is
applied here.
This work also intended to demonstrate that the entire coating process can be assumed to
happen in only two steps comprising pyrolysis of MTS and deposition of SiC from intermediates
produced during pyrolysis. In literature, it is emphasized that the formation rate of C2H4 is higher than
CH4 which is majorly based on theoretical thermodynamic calculations. Formation of CH3* free
radicals which is more likely responsible for formation of CH4 with another H* radical or with H2 is
also highly possible [37-38]. Irrespective of many gaseous intermediates, here we considered only two
major intermediates namely CH4 and SiCl2 [14, 29-32, 37-43]. However, another reason for choosing
of CH4 intermediate for this case is to represent sets of equations stoichiometricaly correct in order to
constitute mass balance equations which will be seen extremely reasonable with the model predicted
results.
Apart from the variation in precursor concentration (or partial pressure) in the feed stream,
reactor wall temperature as well as flow rate of feed stream is seen to play key role for CVD type
reactions [30-34]. Estimated kinetic parameters are thus used in the model in order to predict coating
rates for various inlet gas flow conditions and wall temperatures inside the reactor considering
transportation of intermediate species up till the reacting surface of the substrate plate suspended inside
the chamber. Transportation of intermediates is seen to be affected adversely with the combined effect
of inlet gas flow as well as wall temperature inside the reactor resulting in huge variations in deposition
rates which were further compared with the experimental findings.
Novelty of this work is to find out temperature profile across the sample plate which is placed
perpendicular to the precursor inlet for various total gas flow conditions as well as wall temperatures
which justifies the reason for the selection of particular surface of the substrate plate for deposition. It
also demonstrates that entire coating process can be well described, explained and modeled in
simplified two steps only with the consideration of two major intermediates. Model simulations were
thus carried out in order to profile coating rates for various possible reactor geometries.
2. Model development and Simulation:
The proposed model considers two step coating process carried out over a horizontally placed C/C
composite plate suspended inside a hot wall vertical tubular reactor as shown in Schematic Fig. 1. The
first step of the coating process is the pyrolysis of MTS catalyzed with hydrogen and results in two
major intermediate species namely CH4 and SiCl2 which is well discussed and emphasized in the
earlier reported literature [9, 14, 29, 32]. In the second step, CH4 and SiCl2 simultaneously react on the
substrate surface to form SiC crystal on the surface of the plate. Both the steps can be described using
the following equations:
KP
CH3SiCl3 + H2 CH4 + SiCl2 + HCl (1)
KD
CH4 + SiCl2 + S* SiC + 2HCl + H2 (2)

Where KP and KD are rate constants for pyrolysis and deposition reaction respectively and S* indicates
active substrate site for the deposition.

The corresponding rate equations are:

dCH 4 dSiCl 2 a
rCH 4 = = rSiCl2 = = K P C MTS C bH 2
dt dt (3)

 EP 
− 
 RT 
K P = K P 0e (4)
dCSiC e f
rSiC ( solid ) = = K DC CH C SiCl Sg (5)
dz 4 2

 Ed 
− 
 RT 
K D = K D 0e (6)

The rate of SiC deposition (step 2) is proposed based on the assumption that the reaction
between CH4 and SiCl2 at the substrate sites is slow, whereas, the pyrolysis reaction (step 1) to form
CH4 and SiCl2 is fast and taking place isothermally inside the hot zone (below the substrate plate). The
gas stream (consists of argon as carrier and inert gas, hydrogen and vaporized MTS) enters from the
bottom inlet of the reactor and goes up, meanwhile MTS react with hydrogen and cracks to form CH4
and SiCl2 which finally arrives at the solid surface where the heterogeneous deposition takes place to
deposit as SiC crystals. However, it is expected that the substrate surface temperature varies with
temperature of the reactor (isothermal hot zone) as well as the mixed gas flow rates. Mixed gas has
relatively low temperature (500 K) and when it passes through the reactor with a high velocity it thus
further cools the substrate plate.

2.1 Hydrodynamics and deposition mechanism

The proposed two steps mechanism has been schematically shown in Fig. 1. It is shown in the
schematic that MTS and H2 along with carrier gas Ar are entering from the bottom of the reactor and
goes up vertically along the reactor length. The lower part of the reactor is identified as the pyrolysis
zone as it receives heat from the reactor wall (isothermal zone continues about 100 mm at the middle
of the furnace) to crack the precursor molecules and to result in intermediate species whose major
composition of which is considered to be CH4 and SiCl2. These intermediate products are carried by
the gas mixture to reach up to the sample plate. However, the transport of CH4 and SiCl2 species is
assumed to be highly affected by the flow vector. As it is obvious that the direction of flow of the gas
mixture would be diverted from its vertical path by the sample plate itself, hence it would form a
vortex just above the top surface of the plate. CH4 and SiCl2 which are already within the vortex would
cover the top surface of the plate and would initiate deposition reaction inside the zone identified for
deposition process. Once the deposition takes place, the gaseous products and unconverted precursors
along with carrier gas Ar leave the reactor core through exit.

2.2 Governing equations:

Model equations were derived based on the mass balance of the intermediate products viz., CH4 and
SiCl2 from precursor MTS, in a one dimensional frame. Variation of the two intermediate products
CH4 and SiCl2 with respect to the height of the reactor has been pre-determined using CFD simulation
to enable further calculations to obtain the conversion values for the coating reaction.

Differential mass balance equations within an assumed infinitesimally small ‘dz’ height of the reactor
(with proposed rate law) are;

    

(I) CH4(gas): = (7)

   

    

(II) SiCl2 (gas): = (8)

   

 
(III) SiC (solid): =      !
(9)



Continuity equation for total gas velocity;


(IV) =0 (10)



Where, a, b, e, f, g, are order of the corresponding reaction with respect to MTS, H2, CH4, SiCl2 and S
(substrate area) respectively. #$% , # '() #* represent superficial velocity of argon, hydrogen and
mixture of both gases respectively.

2.3 Assumptions:
Following assumptions are made to formulate the model.

i) Coating takes place at steady state within the specified isothermal hot zone at the surface of
the substrate plate.
ii) Contribution of gaseous by-products in change of volume of gas mixture is neglected as the
overall yield for SiC deposition is poor.
iii) Volume contribution of vaporized MTS is neglected in calculation of total volume of mixed
gas stream.
iv) Pyrolysis of MTS which takes place mainly below the substrate plate is only responsible for
deposition reaction at the hottest surface of substrate only.
v) Deposition reaction takes place over the top surface of the substrate plate only due to
following reasons (assumptions have been validated and reported in sub-section 4.1):
a. Vortex formation and back flow of intermediate species can happen only at the top of the
surface which is exactly at the opposite direction of the flow.
b. Dynamic pressure of gas mixture at the peripherals and the bottom is much higher than the top
surface.
c. Temperature of the top surface is reasonably higher than the bottom surface.

2.4 Boundary conditions and solutions of the governing equations:

This is an initial value problem. At steady state, initial concentrations (t=0) at the reactor inlet are;

(I) CSiC = 0;
(II) CCH4 = 0;
(III) CSiCl2 = 0;
(IV) CMTS =CMTS(i) and
(V) CH2 = CH2(i), suffix ‘i’ indicates initial.

At steady state condition, concentration profile of individual reactants, intermediates and

products do not vary with time but changes along the height of the reactor (one dimensional model)
being space dependent variables. In other words, at any particular height of the reactor, concentration
profile of individual reactants, intermediates and products will remain constant for infinite time at
steady state. Hence to achieve steady state solutions, model equations were solved making them
independent of time but dependent on space as given in Eqns. 7-10. One can also note that the rate
constants in the Eqns. 7-10 are essentially dependent of time when the time is actually expressed in
length scale (height of the reactor).

The model equations (eqn. 7-10) were solved using MATLAB software (v 7.1) with its inbuilt
ODE solver named ode15s. Simulation of the derived governing equations was performed using the
constant parameters listed in Table 1. These constant parameters were estimated so as to closely fit the
experimental and simulated coating rates (methodology reported earlier, [25]). The conversion and
corresponding coating rates were simultaneously calculated from the known position of the sample
plate, which is 20 cm above the gas inlet nozzle.

2.5 Computational Fluid Dynamics Simulation for Flow and Temperature Profiling

A Computational Fluid Dynamics simulation was carried out using Finite Element Method based
commercial COMSOL Multiphysics software (v. 4.4) choosing the in build Conjugate Heat Transfer
Module in order to priory generate temperature profile in and around the coating zone and to support
model assumptions. 2D-axisymmetric model was simulated and the same was further used to obtain
flow and temperature profiles inside the coating chamber (reactor). Dimension of the furnace was kept
similar with that of experimental one. A steady state simulation was performed with a constant
temperature of 1673 K at wall (as boundary condition) for a constant temperature zone of 100 mm at
the centre of the furnace. Subsequently top and bottom part of the furnace were also maintained at
constant temperature 973 K (was measured using IR pyrometer in experimental reactor). Inlet mixed
gas velocity was calculated and set as air inlet at the bottom of the furnace boundary condition to
initiate simulation. Air, instead of mixed gas was considered to find flow pattern and temperature
profile inside the reactor when actual mixed gas (reactant and carrier gases) properties were unknown.
Velocity of the air was kept similar as the total mixed gas used for the experiments. However, flow
contribution from MTS vapor was not considered for negligible volume fraction.

2.6. Simulation Methodology:

Two types of simulations were performed using COMSOL Multiphysics and MATLAB commercial
software in order to get flow and temperature patterns inside reactor (by FEM method) which was later
used to get yield profiles for SiC crystals (solving ordinary differential equations, Eqn. 7-10)
respectively.
Reactor and sample geometry was made in geometry window of COMSOL Multiphysics
software adhering with the dimension of experimental set up considering an axisymmetric case.
Physics controlled meshing was performed in the “build geometry” pan keeping grid size as fine as
0.12 mm with triangular pitch including all the edges of sample plate. In-build “Conjugate Heat
Transfer” module was used for FEM solution for various gas velocities at inlets and reactor wall
temperatures considered here. All the walls and sample plate were considered as graphite material from
materials library. Flow domain was selected for Air as material. In the drag down box of “Conjugate
Heat Transfer” heat transfer in solid was additionally incorporated to capture change in temperature
distribution in sample plate. As initial condition, a temperature of 500 K was set throughout the
domain. All reactor walls except top opening were set as “Wall” boundary condition including all the
edges of sample plate. Top opening was set as “Outlet” for gas to exit. Temperature value of individual
cases was set at wall (at the centre of the reactor, 100 mm in length). Top and bottom wall except
central zone (100 mm) was accordingly set as per experimental measurement using pyrometer. A
convective heat flux branch was added for the accountability of heat losses and gains by gas stream
inside and outside of reactor walls. Finally steady state solution was obtained carrying out the
simulation and thus various profiles were analyzed.
Average temperature across sample plate for various wall temperatures and flow conditions were noted
from the FEM simulation and set in ODEs (Eqn. 7-10) to get steady state solution for the yield of SiC
crystals.

3. Experimental:

Experimental coating rates for deposition of SiC crystal were reported [1], was used to compare and
validate the model. Dimension of CVD reaction chamber used in the current model is kept similar to
the experimental reactor used for the coating (schematic of the coating facility is shown in Fig.2).
Process parameters for the experiments have been used in the model to obtain coating rates for the
corresponding experimental conditions. Experiments were carried out in a hot wall graphite tubular
reactor of the following dimension: height- 40 cm, diameter-6 cm. Schematic of the experimental set
up is given in Fig.2. The reactor temperature was maintained using resistance heater and controlled by
feedback controller through a control panel. The coating reaction was carried out on a C/C composite
plate having the following average dimensions, length-3.4 cm, width- 3.4 cm and thickness-0.5 cm.
The plate was horizontally placed at the isothermal zone of the reactor (at centre, 20 cm above from the
gas inlet nozzle). A measured quantity of mixed gas stream along with MTS (vaporized) as precursor
was pre-heated up to 500 ±5K and was sent through the bottom inlet of the reactor. Here the mixed gas
is defined as the mixture of Ar and H2. The summation of individual flow rates is called the total flow
rate. Gas compositions were maintained using mass flow controller separately. The reactor was also
pre-heated up to 1673+/- 5 K before the gas enters into the reactor. Once the mixed gas enters the
reactor, pyrolysis of MTS followed by deposition of SiC takes place. After the deposition, sample plate
was cooled inside by stopping the power supply to resistance heater and collected at ambient
temperature and weighed. The difference in weight of coated and uncoated sample is regarded as the
weight of SiC coating and the coating rate was determined by dividing the weight change with coating
time. The SiC coatings obtained using different process parameters as given in Table 2 were
characterized by X-ray Diffraction (XRD) - Philips X pert pro XRD unit using Cu Kα radiation and
reported earlier in [1].
4. Results and discussion:

4.1 Computational Fluid Dynamics simulation by FEM

Fig. 3 (a) & (b) show typical contours of velocity and flow directional vectors inside the reactor for
total mixed gas flow rates (QT) 2 & 6 lpm respectively. Normalized flow directional vector across the
substrate plate clearly indicate that the flow direction of the pyrolyzed intermediates is towards the
centre of the plate favoring the deposition at the top surface of the substrate. Flow direction of the gas
stream below the plate is seen to be in the direction of wall followed by channeling through the
opening between substrate plate and the wall to come up finally reaching exit after taking part in
deposition reaction. Fig. 4 (a) & (b) show axial profiles of velocity magnitude. Velocity is seen to fall
till its zero value at the both surfaces of the substrate plate positioned at 200 mm height (plotted in x-
axis) of the reactor. It is worth noting that the difference in radial velocity magnitude above and below
the substrate surface is substantial as seen in Fig. 4 (c) & (d) leading to poor residence time for
deposition reaction to occur at the bottom surface of the plate. Fig. 4(e) shows the velocity contours of
gas stream near the substrate plate. It can be observed from Fig. 4(e) that the gas velocity is very high
below the bottom surface of the plate whereas it has reasonably low velocity beside peripheral surface
and almost become nil above the top surface. Flow directional vector (proportional with velocity
magnitude) colored with temperature of gas stream is shown in Fig. 4(f) which also proves that the
cold gas stream is in contact with bottom surface of the plate and moves towards radial direction where
as relatively hot gas stream moves upward with lower velocity above the top surface of the plate. A
swirl motion is also observed near the passage between sample plate and the reactor wall which is
responsible for back mixing and thus maintains a homogeneous concentration of gaseous species.
Typical contour and the axial profile of temperature inside the reactor are shown in Fig. 5(a)
& (b) respectively for the total gas flow rate of 2 lpm. It is clear from the Fig. 5 that the lower part of
the sample sees average temperature less than 1290 ± 5 K whereas, upper part of the sample maintains
an average temperature close to 1320± 5 K for the constant furnace wall temperature 1673 K at QT = 2
lpm. Top surface of the sample plate typically experience approximately higher average 30± 5 K
temperature than the bottom surface for total gas flow rate of 2 lpm. This further confirms that the
deposition reaction is certainly favorable at the upper surface of the substrate plate.
Average temperature of the top surface of the substrate is considered as the maximum
temperature for the deposition reaction and average temperature of the bottom surface is responsible
for pyrolysis reaction. Both the temperature values are updated in Arrhenius equations (Eqn. 4 & 6) in
order to estimate coverall deposition rates.
Thus a generalized mechanism can be described from the above results as follows. Relatively
low temperature gas stream enters from the bottom of the CVD chamber goes up and directly hits the
bottom surface of the substrate plate and hence heat transfer happens by convection from hot substrate
surface to cold gas stream. The gas stream further takes a circulative path towards bottom part of the
CVD chamber where further heat transfer happens from reactor wall to gas stream similarly by
convection. Fraction of gas stream which is relatively hot becomes lighter and hence goes up through
the passage between substrate plate and the reactor wall. A relatively high gas velocity is thus expected
while coming up from the passage (less area for flow) compared to average velocity of gas stream
inside reactor chamber as shown in Fig 4 (e) & (f). Later, a small fraction of the hot gas reaches top
surface of the sample plate for deposition reaction to take place. It is therefore intuitive that the heat
transfer from top surface of the plate is not very significant as compared to bottom surface and hence
top surface maintains relatively higher temperature.

4.2 Fitting of experimental and model predicted coating rates

The model equations (Eqn. 7-10) were derived based on the assumed rate law (assumed kinetic
parameters are listed in Table 1) and solved to closely fit our earlier reported experimental data [1].
Values of rate constants and order of reactions with respect to individual species are listed in Table 1
and were further approximated in order to best fit experimental data.

The first step involves pyrolysis of MTS assisted with hydrogen which is a fast reaction
compared to the second step where heterogeneous deposition reaction takes place. It is also found that
the overall order of the reaction for formation of two major intermediate species CH4 and SiCl2 is 1.1
while, the overall order for SiC deposition reaction is unity. Individually, the order of the pyrolysis
reaction (step 1) is unity with respect to MTS concentration and 0.1 with respect to hydrogen
concentration. This implies that hydrogen concentration has a very less role in SiC coating. On the
other hand, the order of the SiC deposition reaction (step 2) is 0.25 with respect to concentration of
both the intermediates, i.e., CH4 & SiCl2. The order of the deposition reaction is 0.5 with respect to
substrate site. This indicates that both the intermediates interact individually with the substrate site to
result in solid SiC deposition.
 Fig. 6 shows fitting of experimentally found coating rates at various experimental conditions
with that of simulated coating rates for corresponding parameters listed in Table 2 (serial no. 1-9). It is
seen from the Table 2 that the simulation results are fitting well with experimental data for the case of
MTS flow rate at 1 ml/min. at all flow rates of hydrogen (experimental range). Model estimated
deposition rate is seen to be slightly higher for MTS flow rate at 0.5 ml/min., whereas, marginally
lower estimation is seen for MTS flow rate at 2.0 ml/min. The trend of deviation with hydrogen flow
rates is exactly opposite for two different MTS flow rates indicating that the ratio of mole fraction of
MTS and hydrogen has a role in controlling the deposition rates [30] which promotes generation of
various intermediates and thus estimation of vapor pressure of such intermediates was not possible in
this model.

4.3. Effect of Wall Temperature:

Temperature profile inside the reactor is obtained and shown in Fig. 7(a) & (b) using CFD simulation
for wall temperature 1050 °C for a constant gas flow rate of 2 lpm. One can compare temperature
profile for the wall temperature 1400 °C which is shown in Fig. 5. It is very obvious that the substrate
temperature will fall with wall temperature, as opposite for cold wall reactor [30]. However, it is worth
noting that the temperature difference between wall and substrate plate is seen to rise with higher wall
temperature. This implies increase in wall temperature will not increase the substrate temperature
linearly. Hence reactor wall temperature may always be optimized considering deposition rates and the
energy consumption for achieving high wall temperature of the reactor.

Fig. 9 compares experimental deposition rates with model predicted rates for constant flow
rate of MTS is equal to 1.0 ml/min. and hydrogen flow rate at 1.2 lpm with argon flow of 0.8 lpm as
carrier gas. Deposition rate is seen to fall exponentially with the wall temperature as desired.
Agreement among experimental and simulated deposition rates certainly encourages us about the
correctness of the model equations and well approximation of the kinetic parameters with suitable
assumptions. The model is thus subjected for the simulation at various total gas flow rates in order to
compare deposition rates with that of experimental one expecting desired change of temperature profile
inside the reaction chamber with rate of transportation of intermediate species.
4.4 Effect of Mixed Gas Flow Rate:

The model equations were further simulated in order to predict the conversion and thereby the
deposition rates of SiC for various flow conditions. CFD simulations were performed to estimate the
temperature around the substrate surface at various mixed gas flow rates keeping constant wall
temperature (1400 °C) of the reactor. Temperature of the substrate surfaces is seen to reduce with flow
rates of mixed gas due to higher heat removal rate by cold gas stream with higher flow rates as seen in
Fig. 8 (a) & (b) comparing with the contour obtained in Fig 5. It was also seen that the difference in
temperature between top and bottom surfaces increases with mixed gas flow rates. Comparative
reduction of temperature among bottom and top surfaces is due to higher convective heat transfer rate
from plate to gas stream at higher gas flow which insignificantly affect top surface as very less fraction
of gas touches the top surface.

`Model equations describing the kinetics as well as transport of species (one dimensional)
were further solved in Matlab software updating the average temperatures for pyrolysis and deposition
reactions were estimated from CFD simulation. Updated transportation rates of gas (or various
intermediate species) were seen to influence the average temperature across substrate plate (Fig. 8).
The change in average temperature is thus used in the model equations to predict deposition rates. Fig.
10 compares model predicted and experimentally found deposition rates for MTS flow rate 0.5 ml/
min. with various hydrogen flow rates starting from 0 to 1.2 lpm (total flow rate of mixed gas (QT)
represents combination of hydrogen and argon gas in feed stream). Trend shows that the deposition
rate falls with the gas flow rates for the combined effect of the fall in average temperature inside the
chamber as well as substrate plate and also with the lower residence time of the gaseous intermediates.
Close prediction of deposition rates with that of experimental one is indeed acceptable for realizing
combined effect of gas velocity as well as change in average temperature for the gas velocity inside a
CVD chamber.

Various dimensions of the CVD chamber would expectedly change the hydrodynamics and
temperature profile of the chamber leading change in deposition rates.

4.5 Reactor design

Till now we have seen change of deposition rates based on the change in temperature profile due to
effect of various wall temperatures and flow of gas inside the chamber. It is thus understood that the
heat transfer rate and transportation of species with temperature and flow has significant role in
determining deposition rate of SiC from MTS. It is however other way interesting to profile
temperature for various dimensions of the CVD chamber or reactor. Dimension/ size of the reactor thus
can be decided in order to achieve efficient deposition of coated material with optimized properties.

Steady state solution of model equations were obtained for the same kinetic parameters
following the estimated temperature and flow profile using CFD solution for various diameter and inlet
size of tubular reactor in order to predict deposition rates which is shown in Fig. 11. It is seen that the
deposition rate steadily increases with the length of the tube (provided sample is kept at that length
with maintained temperature 1400 °C). Higher deposition rate can be achieved by narrowing down the
diameter of the tube. Smaller diameter of the tube with same temperature and gas flow rate favors to
achieve higher temperature across the sample plate (Fig. 12) thus results in higher deposition rates
enhancing overall kinetics. Higher plate temperature is achieved due to dominating heat transfer from
reactor wall to sample plate by radiation as the gap between wall and the plate is lesser for smaller
diameter tube. Gas inlet size of the reactor is also seen to affect the deposition rate. Overall kinetics is
seen to improve with increase in gas residence time in the pyrolysis zone (below the sample plate) thus
the deposition rate is also increases with higher gas inlet diameter (recalling, with constant gas flow
rate, velocity is higher for lower cross section of flow area and vice versa). Gas residence time below
the lower part of the sample increases with lower in the local gas velocity which is achieved with
higher gas inlet diameter. When gas inlet diameter is bigger for same gas flow rate, velocity of the gas
stream is lower following the principle of continuity for incompressible fluid. Higher gas residence
time in turn can extract more heat from reactor wall and increases its temperature. Higher gas
temperature should intuitively favor to maintain higher temperature of sample plate (shown in Fig. 12)
resulting higher deposition rate. [15, 28, 30]

However, in order to achieve maximum coating rate, positioning of the sample at maximum
reactor height is not advisable. This is due to practical difficulties to attain/ maintain maximum
temperature at the extreme top of the reactor. Therefore, one can position the sample plate at maximum
reactor length/ height simultaneously ensuring maximum temperature zone of the reactor to achieve
maximum deposition rate.
4.6. Coated surface morphology
The Fig. 13 (a), (b) and (c) show camera, microscopic and SEM images of the SiC deposited layer in
top, bottom and peripheral surfaces of the sample. The schematic of the sample surface orientation is
shown in Fig. 2. The cross sectional view of the C/C composite sample plate after coating is shown in
Fig 13(a). Microstructure showing dense deposition of SiC can be inferred. XRD pattern and EDAX
of deposited β-SiC can be seen in earlier reported work [1]. A thick deposition of SiC layer of about 40
µ m is seen from the optical microscopy image on the top surface of the sample. However, peripheral
sides as well as bottom surface do not show any significant deposition. This is probably due to the fact
that the bottom and peripheral surfaces of the plates are facing more turbulent flow in comparison to
the top surface as shown in this study. On the top surface, vortexes containing activated intermediates
are formed which results in greater residence time to react with top surface. Furthermore, top surface of
the plate is relatively hot compared to the bottom surface. The bottom surface is relatively cold due to
continuous interaction with relatively low temperature gas flow which is travelling along reactor length
from bottom inlet to top outlet.

5. Conclusions:

This work demonstrates a robust model to predict deposition rate of silicon carbide on a C/C composite
substrate plate using methytrichlorosilane and hydrogen by CVD process. A simplified two steps
mechanism for the heterogeneous deposition in a hot wall tubular reactor is proposed. The coating
process is modeled based on the mass balance of only two major intermediate species namely CH4 and
SiCl2. Model equations are simulated which fit well with experimental results assuming various kinetic
parameters. Later, kinetic parameters were used to compare deposition rates at various gas flow rates
and reactor wall temperature. CFD simulation was coupled to obtain profile of hydrodynamics thereby
transportation of intermediate species and also the temperature profile inside the reactor which are used
for kinetic solution. The rate law for proposed two steps reactions is good enough in order to
meaningfully explain the effect of reaction kinetics, wall temperature, gas flow rate, reactor geometry
and the hydrodynamics on the SiC deposition rates. Results infer that the efficient deposition with
desired properties can potentially be obtained by changing the reactor geometry, flow dynamics apart
from concentration of precursor and temperature.
Acknowledgement
All financial supports were received from Bbabha Atomic Research Centre, Mumbai, India to carry out
this work. Mr. Amit Kaushal, Mr. P. T. Rao and Dr. Jyoti Prakash are acknowledged for their technical
discussion.

Abbreviations:
MTS methyltrichlorosilane
C/C carbon-carbon
CVD chemical vapor deposition
lpm Litre (standard) Per Minute
Notations:
a order of pyrolysis reaction with respect to methyltrichlorosilane [-]
b order of pyrolysis reaction with respect to hydrogen [-]
di diameter of the reactor gas inlet [m]
e order of deposition reaction with respect to CH4 [-]
f order of deposition reaction with respect to SiCl2 [-]
g order of deposition reaction with respect to substrate surface area [-]
KP rate constant for pyrolysis reaction [s-1 (mol/cc)-0.1]
KD rate constant for deposition reaction [s-1]
T temperature [K]
CMTS concentration of methyltrichlorosilane in gas stream [mol/cc]
CH2 concentration of hydrogen in gas stream [mol/cc]
CCH4 concentration of methane in gas stream [mol/cc]
CSiCl2 concentration of dichlorosilylene in gas stream [mol/cc]
CSiC gas phase concentration of silicon carbide [mol/cc]
UT total mixed gas velocity [cm/s]
UAr velocity of argon in gas stream [cm/s]
UH2 velocity of hydrogen in gas stream [cm/s]
QT total mixed gas flow rate [lit/min.]
QAr argon flow rate [lit/min.]
QH2 hydrogen flow rate [lit/min.]
z characteristic height of the tubular reactor [cm]
h position of the sample along reactor height [cm]
S average surface area of all faces of sample [cm2]
Ep average activation energy for pyrolysis reaction [KJ mol-1]
Ed average activation energy for deposition reaction [KJ mol-1]
R gas constant [J mol-1 K-1]
rE rate of coating, experimental, [mg. cm-2 min-1]
rS rate of coating, simulated, [mg. cm-2 min-1]

References

[1] R. Venugopalan, J. Prakash, J. Nuwad, C. G. S. Pillai, A. K. Tyagi, D. Sathiyamoorthy, Int. J. Mater. Res.
103, (2012) 10 1251-1256.

[2] L.F. Cheng, Y.D. Xu, L.T. Zhang, Carbon 40 (12) (2002) 2229-2234.

[3] T. Kachi, Y. Kato, N. Ushigome, N. Arai, Energy Convers. Manage. 42 (15-17) (2001) 1963-1975.

[4] J.G. Lee, I.B. Culter, Am. Ceram. Soc. Bull. 54 (1975) 195-198.

[5] F. Qiangang, L. Hejun, S. Xiaohong, L. Kezhi, H. Zhibiao, W. Jian, Mater Lett. 59, (2005) 2593-2597.

[6] G. Sun, H. Li., Q. Fu, M. Huang, W. Cao, H. Wu, Comp. Mater. Sci. 46 (2009) 1002-1006.

[7] K. L. Choy, Prog. Mater. Sci. 48 (2) (2003) 57-170.

[8] Y.J. Lee, D.J. Choi, S.S. Kim H.L. Lee, H.D. Kim, Surf. Coat. Technol. 177-178 (2004) 415-419.

[9] J. Meziere, M. Ucar, E. Blanquet, M. Pons, P. Ferret, L. Di Cioccio, J. Cryst. Growth 267 (3-4) (2004) 436-
451.

[10] W. Ji, P.M. Lofgren, C. Hallin, C.Y. Gu, G. Zhou, J. Cryst. Growth 220 (2000) 560-571.

[11] P. K. Mollick , Ashok Kumar, J. Banerjee, J. Radhakrishna ,J. K. Chakravartty, J. of Nucl. Mat. 473 (2016)
229-236.
[12] Kaushik Sanyal, Sangita Dhara, S. Sanjay Kumar, N. L. Misra, P. K. Mollick, P. T. Rao, R. Venugopalan,
V. Rajesh Pai, N. Kumar, S. K. Mukerjee, J. K. Chakravartty, S. K. Aggarwal, J. Radioanal Nucl. Chem., DOI
10.1007/s10967-014-3593-0

[13] F. Liao, S. Park, J.M. Larson, M.R. Zachariah, S.L. Girshick, Materials Lett. 57 (2003) 1982–1986.

[14] S. H. Mousavipour, V. Saheb, S. Ramezani, J. Phys. Chem. A, 108(2004)1946-1952.

[15] Igor M. Kostjuhin, Stratis V. Sotirchos, Ind. Eng. Chem. Res. 40(2001) 2586-2596.

[16] B. Bahavar, R. J. McCluskey, M. I. Chaudhry, Znd. Eng. Chem. Res.34(1996) 1859-1867.

[17] P.T. Rao, U. Jain, P.K. Mollick, A. Agarwal, J.K. Chakravartty, Sanjay Kumar, Journal of Nuclear
Materials 456 (2015) 200–205.

[18] Zhigang Zang, Atsushi Nakamura, and Jiro Temmyo, Optics Express Vol. 21, No. 9 (2013)
DOI:10.1364/OE.21.011448 11449

[19] Tommi Kaplas and Polina Kuzhir, Nanoscale Research Letters (2017) 12:121, DOI 10.1186/s11671-017-
1896-0

[20] T. Kondo, H.Funakubo, K.Akiyama, H. Enta , Y. Seki, M. H.Wang, T. Uchida, Y. Sawad, Journal of
Crystal Growth 311 (2009) 642–646.

[21]Tommi Tynell, Takashi Aizawa, Isao Ohkubo, Katsumitsu Nakamura, Takao Mori, J. of Cryst. Growth 449
(2016),10–14.

[22] Tsutomu Kaneko, Noriaki Miyakawa, Hayato Sone, Makoto Iijirna, J. of Cryst. Growth 174 (1997) 658.

[23] Zhigang Zang , Atsushi Nakamura, Jiro Temmyo, Materials Letters 92 (2013) 188–191.

[24] A. Santoni,, J. Lancok, S. Loreti, I. Menicucci, C. Minarini, F. Fabbri, D. Della Sala, J. of Cryst.Growth
258 (2003) 272–276

[25] P. K. Mollick, D. Sathiyamoorthy , P. T. Rao and V. G. Rao, Ind. Eng. Chem. Res., 50 (2011),
13313-13321

[26] P. K. Mollick, R. Venugopalan, M. Roy, P. T. Rao, D. Sathiyamoorthy, P. Sengupta, G. Sharma,

C. B. Basak, J. K. Chakravartty, Chem. Eng. Sci., 128(2015)44–53
[27] D.Sathiyamoorthy, V.Govardhana Rao, P.T. Rao and P. K. Mollick , Indian Journal of
Engineering & Material Science, 17(2010), 349-352

[28] P. Mishra, N. Verma, Chem. Eng. Res. Des. 90 (2012 ) 2293–2301

[29] C. Lu, L. Cheng, C. Zhao, L. Zhang, Y. Xu, Apl. Surf. Sci., 255 (2009) 7495-7499.

[30] G.D. Sun, H.J. Li, S.Y. Zhang, Q.G. Fu, W. Cao, Y.Q. Jiao, Mater. Sci. Forum 610-613 (2009) 635-640.

[31] D. R. Alfonso, S. E. Ulloa, App. Phys. Lett, 74 (1999), 55-57

[32] A. Dollet, S. de Persis , M. Pons , M. Matecki, Surf. Coat. Technol. 177–178 (2004) 382–388

[33] J. Y. Ofori, S. V. Sotirchos, Ind. Eng. Chem. Res., 35 (1996), 1275-1287

[34] V. Hlavacek, J. A. Puszynski, Ind. Eng. Chem. Res. 35 (1996) 349-377

[35] L. J. Velenyi, Y. Song, J. C. Fagley, Ind. Eng. Chem. Res. 30 (1991), 1708-1712

[36] A. Ramos, W.O. Filtvedt, D. Lindholm, P.A. Ramachandran, A. Rodríguez, C. del Cañizo, J. Cryst. Growth
431 (2015) 1-9.

[37] Y. Ge, M. S. Gordon, F. Battaglia, R. O. Fox, J. Phys. Chem. A (2010), 114, 2384–2392

[38] S. Leone, O. Kordina, A. Henry, S. Nishizawa, O. Danielsson, E. Janzen, Cryst. Growth Des. 12 (2012)
1977-1984

[39] H.Cheng, R.Tu, S. Zhang, M. Han, T. Goto, L. Zhang, J. Euro. Cer. Soc. 37 (2017) 509–515

[40] A. Joseik, F. Langlais, J. Cryst. Growth, 160 (1996) 253-260.

[41] M. Jérôme, P. Michel, D. Jean-Marc, B. Elisabeth, F. Pierre, D. C. Léa, B. Thierry, Mater. Sci. Forum 433-
436 (2003) 141-144.

[42] G. Melnychuk, Y. Koshka, S. Yingquan, M. Mazzola, C.U. Pittman Jr., Mater. Sci. Forum 433-436 (2003)
177-180.

[43] C. Cavallotti, F. Rossi, S. Ravasio, and M. Masi, Ind. Eng. Chem. Res. 53, (2014) 9076−9087.
Fig. 1. Schematic of the SiC deposition process.
 Top surface

Peripheral

Bottom surface

Fig. 2 Schematic of experimental set up

 Substrate Substrate

(a) Total gas flow rate (QT) = 2 plm (b) Total gas flow rate (QT) = 6 plm

Fig. 3. Fluid velocity contour (m/s) and flow directional vector (normalized) below and above the

substrate plate.

(a) (b)
 (c) (d)

Total gas flow rate (QT) = 2 plm (b) Total gas flow rate (QT) = 6 plm

Temperature scale, (K)

Velocity magnitude, (m/s)

Substrate plate
Substrate plate

(e) (f)
Fig. 4. (a) & (b) Line graph of velocity magnitude along the axis of the reactor, (c) & (d) contour of

velocity field (3D) of radial componant (represented in height) along the height of the reactor

(sample position at 200 mm) comparing magnitude and direction of radial componant of velocity, (e)

3D contour of velocity profile near sample plate showing intencity of gas flow in bottom, side and

top surface, (f) proportional velocity vector showing the direction and intensity of flow near surfaces

coroured by tempereature.

Substrate plate

(a) Contour of Temeprature

(b) Line graph of temeprature along the axis of reactor

Fig. 5. Temperature profile inside CVD chamber (reactor) and around the susbstrate plate
 3.0
MTS: 0.5 ml/min. MTS: 1.0 ml/min. MTS: 2.0 ml/min.
2.5 Experiment Experiment Experiment
Simulation Simulation Simulation
Deposition rate, (mg/ min.-cm2)

2.0

1.5

1.0

0.5

0.0
0.4 0.6 0.8 1.0 1.2
Hydrogen flow rate, (lit./ min.)

Fig. 6 Comparison of experimental deposition rate with model predicted deposition rate and

(estimation of kinetic parameter with closest fitting) at total flow rate of gas at 2 lpm with reactor

wall temperature 1400 °C (1673 K).

(a) (b)

Fig. 7 (a) Temperature profile inside reactor and across the substrate plate for wall temperatre TW =

1050 °C (1323 K), (see Fug. 5 for comparison), (b) respective line graph for temperature along the

axis.
(a) (b)

Fig. 8 (a) Temperature profile inside reactor and across the substrate plate for variation gas

flow rate QT= 6 lpm (see Fug. 5 for comparison), (b) respective line graph for temperature along the

axis.
 1.4
Experimental Simulation
Deposition rate, (mg/ min.-cm2)

1.2

1.0

0.8

0.6

0.4

0.2
1000 1100 1200 1300 1400 1500
Temperature of the reactor wall, (°C)

Fig. 9 Variation of deposition rate with temperature of the reactor wall.

 0.9
Experiment Simulation
Deposition rate, (mg/ min.-cm2)

0.8

0.7

0.6

0.5

0.4

0.3

1 2 3 4 5 6 7
Flow rate of mixed gas (lit./ min.)

Fig. 10 Variation of deposition rate with total flow rate of mixed gas.
 6
Reactor dia. = 5 cm, Inlet dia. = 0.3 cm
Reactor dia. = 6 cm, Inlet dia. = 0.3 cm (dimension of present reactor)
Reactor dia. = 9 cm, Inlet dia. = 0.3 cm
5 Reactor dia. = 6 cm, Inlet dia. = 6.0 cm
Deposition rate, (mg/min.-cm )
2

0
0 5 10 15 20 25 30 35 40
Sample holding position at maximum heat zone (from inlet), (cm)

Fig. 11: Variation of coating rate with position of the sample along reactor height for various reactor

diameter and inlet nozzle size.

Fig. 12 Temperature profile inside CVD chamber for (a) reactor dia. 5 cm & inlet dia. 0.3 cm, (b)
reactor dia. 9 cm & inlet dia. 0.3 cm, (c) ) reactor dia. 6 cm & inlet dia. 6 cm for the total flow rate of
gas at 2 lpm with reactor wall temperature 1400 °C (1673 K).
 (a) Camera image of coated sample plate

(b) Optical Microscopic image of coated sample plate

(c): SEM image of coated sample plate

Fig. 13 Images of SiC coated C/C composite plate

 Table1- Constants in model equations

Parameter(s) Value Unit Remarks

/ constant(s)

KP0 1.0*1015 s-1 average reported value in literature [9, 12, 19, 22, 27-
(mol/cc)-0.1 33]*

KD0 1.2*10-6 s-1 assumed value to closely fit experimental coating rate

a 1.0 - assumed value to closely fit experimental coating rate

b 0.1 - assumed value to closely fit experimental coating rate

e 0.25 - assumed value to closely fit experimental coating rate

f 0.25 - assumed value to closely fit experimental coating rate

g 0.5 - assumed value to closely fit experimental coating rate

Ep 120 kJ mol-1 average reported value in literature [9, 12, 19, 22, 27-
33]*

Ed 50 kJ mol-1 assumed value to closely fit experimental coating rate

Tp Value from K max. temperature below substrate surface for pyrolysis

CFD
simulation

Td Value from K max. temperature over substrate surface for deposition

CFD
simulation

S 12 cm2 approx. available surface area for deposition

R 8.31 J mol-1 K-1 Ideal gas constant

*Average value from literature for starting trial guess and final value reached based on best

fitting with experimental data.

Table 2- Experimental parameters

Expt. QMTS QH2 QAr QTotal Tw

No. (cc/min.) (lit./min.) (lit./min.) (lit./min.) (at furnace
wall) (K)
1 2.0 0.4 1.6 2.0 1673

2 2.0 0.8 1.2 2.0 1673

3 2.0 1.2 0.8 2.0 1673

4 1.0 0.4 1.6 2.0 1673

5 1.0 0.8 1.2 2.0 1673

6 1.0 1.2 0.8 2.0 1673

7 0.5 0.4 1.6 2.0 1673

8 0.5 0.8 1.2 2.0 1673

9 0.5 1.2 0.8 2.0 1673

10 0.5 0.0 2.0 2.0 1673

11 0.5 0.8 3.2 4.0 1673

12 0.5 1.2 4.8 6.0 1673

13 1.0 1.2 0.8 2.0 1523
14 1.0 1.2 0.8 2.0 1323
Highlights:

• Synthesis of SiC crystal in hot wall tubular reactor is modelled and simulated
• CFD coupled kinetic modelling is performed for vertical CVD chamber
• CH4 & SiCl2 can reasonably be considered as major intermediates from MTS pyrolysis
• Model shows significant effect of hydrodynamics and heat transfer in SiC yield
• Model suggests reactor design aspects for better yield of product