Professional Documents
Culture Documents
“MASTER OF TECHNOLOGY”
in
Chemical Engineering
By
SHIVAM BHATIA
under my/our supervision and that this work has not been submitted elsewhere for a
degree.
It is further certified that the student has fulfilled all the requirements of the
Degree.
ii
DECLARATION BY THE CANDIDATE
I, SHIVAM BHATIA, certify that the work embodied in this thesis is my own bona fide
work and carried out by me under the supervision of Dr. Vijay Shinde from June 2020
in this thesis has not been submitted for the award of any other degree/diploma. I declare
that I have faithfully acknowledged and given credits to the research workers wherever
their works have been cited in my work in this thesis. I further declare that I have not
wilfully copied any other’s work, paragraphs, text, data, results, etc. , reported in journals,
books, magazines, reports, dissertations, thesis, etc. , or available at websites and have
not included them in this thesis and have not cited as my own work.
Date:
It is certified that the above statement made by the student is correct to the best of
my/our knowledge.
iii
ACKNOWLEDGEMENT
acknowledging those without whom this project would not have been a reality.
First, I would like to extend my heartfelt gratitude to the institution - IIT BHU for
imparting first class education. I would like to express my sincere gratitude to my project
guide, Dr. Vijay Shinde, Assistant professor, Department of Chemical Engineering and
Technology, for his continuous support and guidance in research and study. His
mentorship, insight and encouragement will continue to inspire me both personally and
professionally.
I am grateful to the Head of the Department, Dr. Vijay Laxmi Yadav, for her sterling
direction which inspires the department to keep up its pace and move on with graces.
I would also like to thanks Dr. Pramod Kumar Jain, Director, IIT BHU, for providing
I am also grateful to all other teaching and non-teaching staff members of the Department
of Chemical Engineering and Technology for their kind co-operation during the project.
iv
ABSTRACT
This thesis emphasises on computer simulation & optimization of a horizontal flow CVD
deposition of silicon carbide from gas mixtures containing MTS-H2-Ar was set up on
commercial CFD code ANSYS FLUENT 2020 R2. The geometry creation was done
using ANSYS SpaceClaim and meshing was done using ANSYS Meshing. A 2 step rate
mechanism that describes the deposition of silicon carbide was introduced. Analysis of
various operating parameters and its coupling effects on silicon carbide deposition rate
and uniformity of deposition was carried out using RSM. An accurate RSM model was
obtained based on the CFD model to predict deposition rate and uniformity of deposition
using numerous parameters such as pressure, temperature, mole ratio of H2-Ar &
flowrate.
Analysis of variance (ANOVA) was conducted on Design Expert 11 for determining the
regression model. Optimum set of operating parameters for the reactor taken into account
were susceptor and wall temperature of 850°C, operating pressure of 150mbar, H2:Ar
mole ratio of 1.5 and gas flowrate as 200lph. The maximum deviation between predicted
value and computational value of uniformity index was found to be 1.24% and that of
deposition to be 6.92%, which indicates results were very close to the CFD model.
obtained from a 2-dimensional model was proposed to describe mass transfer, heat
transfer and gas flow in a kinetically controlled chemical vapour deposition reactor.
v
TABLE OF CONTENT
PAGE NO.
CERTIFICATE ii
DECLARATION BY THE CANDIDATE iii
ACKNOWLEDGMENT iv
ABSTRACT v
TABLE OF CONTENTS vi
LIST OF FIGURES viii
LIST OF TABLES x
LIST OF SYMBOLS xi
vi
CHAPTER 4 Mathematical Modelling using Design Expert 47
4.1 Introduction 47
4.2 Analysis of Results 47
4.3 Uniformity of SiC film on substrate deposition 50
4.4 Deposition rate of SiC on substrate 54
4.5 Optimisation of Process Parameters 58
CONCLUSION 64
REFERENCES 65
vii
LIST OF FIGURES
FIGURE TITLE PAGE
NO. NO.
2.1 Schematic representation of series of steps involved in CVD 10
2.2 Graph of deposition vs velocity and temperature 11
3.1 2-Dimensional geometry of CVD reactor 22
3.2 Meshed geometry of 2-D CVD reactor 23
3.3 Graph of uniformity index vs number of elements 24
viii
4.2 Predicted vs actual value for response 1 53
3D surface model graph of uniformity vs temperature and
4.3 pressure 54
4.4 3D surface model graph of uniformity vs flowrate and mole ratio 54
4.5 Box cox plot for log transform for response 2 57
4.6 Predicted vs actual value for response 2 57
4.7 3D surface model graph of deposition vs temperature and pressure 58
4.8 3D surface model graph of deposition vs flowrate and mole ratio 58
4.9 3D surface model graph of desirability 59
4.10 Contours for 2-D CVD reactor operating at optimum conditions 60
4.11 Contours for 3-D CVD reactor operating at optimum conditions 63
ix
LIST OF TABLES
x
LIST OF SYMBOLS
xi
μ Dynamic viscosity
υ Stoichiometric Coefficient
°
ΔG𝑟𝑥𝑛 Gibbs free energy of chemical reaction
° Gibbs free energy of reactants
ΔG𝑓,𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠
° Gibbs free energy for products
ΔG𝑓,𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠
K Equilibrium constant
A Pre exponential factor
β Temperature coefficient
θ Fraction of occupied sites
EA Activation Energy
yi Mass fraction of species i
Kn Knudsen number
𝜆 Molecular mean free path
DC Length scale of substrate
V Volume
A Area
Τ∅ Diffusion coefficient for ϕ
u Velocity in x-direction
v Velocity in y-direction
Q Flowrate
M Mole ratio
xii
Chapter: 1
1.1 Introduction:
precursors which react in vapour phase to form a product which deposits on the wafer.
The gas moves through the reactor and distributes uniformly across the surface of the
wafer which begins to absorb them on their surface. The difference between chemical
vapour deposition and physical vapour deposition lies in the fact that in CVD, the
chemical structure of the deposited substance is different from that present in the gas
phase whereas PVD is generally a condensation process and therefore, the chemical
structure of deposited and gas phase species is the same. The main advantages of PVD
over CVD technique are purity of coating and availability of a broad range of precursors
which is compensated by low deposition rate and line of sight effects. Typical CVD
i. Pyrolysis
ii. Reduction
iii. Oxidation
iv. Nitration
Pyrolysis: In this bulk phase chemical reaction, a molecule is split into its element. Such
reactions are simple as they require only one precursor gas. Reactions could be halide
Reduction: In such type of bulk phase chemical reaction, the oxidation state of an
are as follows:
Oxidation: In this type, the oxidation state of an element is increased or in other words
an element loses an electron. Usual sources of oxygen are elemental oxygen, ozone and
2
-Carbidization and Nitridation: Carbide deposition is usually done by reacting
hydrocarbon with a halide; and nitride is generally based on ammonia which is preferred
over nitrogen since ammonia has a positive free energy of formation. A very common
Horizontal flow CVD reactor: In this type of reactor, flow direction of precursors and
carrier gas is parallel to the substrate resulting in the boundary layer formation which
leads to more deposition in the leading edge hence adversely affecting uniformity of the
film formed. Film deposition can be made uniform by inclining the substrate, diluting the
3
Vertical flow CVD reactor: In this type of reactor, precursors and carrier gas are
vapour in the vicinity of substrate; hence forming thicker film compared to horizontal
Cold Wall CVD reactor: In this type of reactor, the substrate is kept at the higher
temperature compared to the walls of the reactor (TS >> Tw). The advantage of using this
within the reactor, vapours start moving away from the substrate resulting in less
Hot Wall CVD reactor: In this type of reactor, the temperature of walls of the reactor
are kept at the same temperature as that of the substrate (TS ≈ Tw). The advantage of this
technique is that Soret Diffusion is not observed. The disadvantage of HWCVD reactor
is that it may result in homogeneous nucleation which would result in formation of flaky
film. In this process, deposition occurs not only on substrate but also on reactor walls
Atmospheric Pressure CVD reactor: Operating pressure of such reactors is kept close
to atmospheric pressure (1 atm). Advantage of such reactors is that they can be operated
in continuous as well as batch mode. Since no vacuum is needed, therefore the cost of
operation is low. The disadvantage of such reactors is that gas flow dynamics is difficult
to control. APCVD is generally operated as a cold wall reactor and is exposed to impurity.
Such reactors are used for films which do not have high sale value or rate of deposition
4
Low Pressure CVD reactor: LPCVD reactor is operated at a pressure lower than
atmospheric pressure (about 1/10th of atmospheric pressure). The low pressure CVD
reactors can be operated as both cold wall and hot wall reactors but the latter is preferred.
The disadvantage of low pressure hot walled CVD reactors is that nucleation can happen
anywhere in the reactor but by driving to sufficient low pressure one can minimise
homogeneous nucleation in hot walled low pressure CVD reactor. The advantage of such
good step coverage and uniformity compared to APCVD. LPCVD reactors are more
Plasma Enhanced CVD: PECVD reactors are generally operated at a pressure which is
about 1/100th of atmospheric pressure. In the PECVD reactor, most of the energy required
for the reaction is provided by a glow discharge plasma produced in precursors. The
colliding ions create a more active substrate surface which along with the produced
reactive species aids the deposition of the film on the substrate. The highly reactive
1. Reactor chamber
3. Vacuum system
4. Energy source
5
6. Exhaust system
The supply of precursors to the reactor chamber is achieved with the help of gas delivery
system. Reactor chamber is the space in which deposition occurs. The introduction and
removal of substrate is done using substrate loading mechanism. Energy required by the
precursors to react or decompose is provided by the energy source. Other gaseous species
excluding those needed for reaction or deposition are removed with the help of the
vacuum system. Withdrawal of volatile by-products from the reactor is achieved with the
help of exhaust system. Exhaust treatment system is used for treatment or conversion of
gases leaving the reactor to non-toxic compounds before releasing into the atmosphere.
Various process control equipment are needed to control process parameters such as
flowrate, temperature and pressure. In situ inspection system is used to check the quality
uniformity, purity, hardness, density, step coverage, refractive index etc. [3]
practicality of the film. The two main processes that determine the layer morphology are
nucleation and surface diffusion which depend on parameters such as partial pressure of
gaseous species, substrate temperature and total pressure of the reactor. Various analytical
methods implemented to determine layer morphology are AFM, SEM, XRD, etc. Three
6
i. Epitaxial layers: In epitaxial layers or crystalline film a fine layer of crystals is
to have a substrate free of contamination and defect. High temperature (> 700°C)
adsorbed precursor species. Equation 1.1 quantifies the diffusion coefficient for
E (1.1)
Di,adsorbate =Geometry factor (lattice dimension)2 exp D (Attempted jump frequency)
RT
For crystalline film, growth rates should be low as surface diffusion has to be fast in
comparison to the new growth initiating species reaching the surface. This permits
ii. Amorphous layers: In Amorphous layers, precursors reach the surface faster than
the surface diffusion of adsorbed species. Low substrate temperature and high
growth rates favour formation of amorphous films. Etching rate of amorphous films
sites, which leads to the formation of islands. These islands coalesce to form
7
1.6 Applications of CVD:
CVD is one of the most popular methods used for preparation of coatings and films. CVD
offers relatively high deposition rate, high purity along with wide range of precursors that
can be used for deposition. Some of the industrial applications of CVD are as follows:
1. Coatings: One of the early applications of CVD involves deposition of coatings for
and combination thereof. Certain nitrides and carbides provide wear resistance. CVD
coatings are also used to reduce friction. Most common application for CVD coating
films of the semiconductor material such as doped germanium, silicon, etc., insulating
components like solid state diode lasers are often grown by CVD. Ga-As &
amorphous polysilicon are used in some photovoltaic devices and integrated circuits.
Composites (CMC) are light weight, high temperature and corrosion resistant
materials that are substituting high nickel super alloys in certain applications like
shows great advantage over conventional wet methods for preparation of supported
catalyst.[5]
8
1.7 Objective of the work:
substrate.
optimum conditions obtained from a 2-dimensional model which describes gas flow,
9
Chapter: 2
Many steps involved in the CVD process are transport related. Although, the effect of
flowrate, reactants partial pressure and the reaction chemistry involved. These parameters
thermodynamic controlled. Figure 2.1 represents the series of steps involved in a CVD
process.
In a mass transport limited process, the diffusion or transport of reactant to the substrate
surface i.e. step 3 or step 8 are the slowest hence rate determining steps. In transport
10
transfer and fluid flow. In a kinetic limited step, surface processes are the slowest i.e.
steps 4-7 are not as fast as step 1 or 9 and 3 or 8. In this regime, fluid flow has a marginal
controlled regime, mass transfer towards and away from the substrate is much slower than
any other step i.e. rate controlling steps are the step 1 and step 9.
Temperature is the key parameter in kinetically controlled regime and flowrate is the key
curve or deposition vs velocity curve is plotted as shown in figure 2.2. Region on the left
side of the red line in figure 2.2 represents transport limited and that on the right side
11
In mass transport limited process, the controlling factors are diffusion of reactants across
the boundary layer and volatile by-products out of this layer. It mostly happens at high
pressure and temperature as decomposition of reactants occurs more rapidly and also
molecules reaching the surface react immediately. Hence, velocity is low making the
boundary layer thicker making which results in less reactants reaching the deposited
surface.
When the process is surface reaction controlled, the rate of the reaction depends upon the
concentration of reactant gas. This usually happens at low temperature and pressure, as
reaction occurs slowly and due to thin boundary layer reactants diffuse to the surface with
ease.
transport controlled. Damköhler number is the ratio of surface reaction rate to diffusion
rs
DaII = (2.1)
JD
For a typical CVD reactor Damköhler number varies from 10 -3 to 103. If DaII >>1, it
means the surface reaction is faster compared to diffusion and hence the process is
transport controlled and if DaII <<1, it means the process is kinetic controlled.
At a very low flowrate the reactants residence time is high enough to equilibrate with the
substrate. The deposition is thermodynamically limited and increases linearly with flow
independent of the total flowrate. Thus, in order to optimise the CVD reactor, a profound
knowledge of transport phenomena, gas and surface chemical reactions is needed. [6]
12
2.2 Governing Equations:
This part deals with the general conservation equations of energy, momentum, overall
and Eulerian control volume is used. The general form of these equations in differential
b
F BV (2.2)
t
⃗⃗ is
In the above equation b is field density (or conserved quantity per unit volume), 𝑭
conserved quantity flux (consist of both convective and diffusive terms) and BV is
Flux
Mass balance 𝜌 0 𝜌𝑢
⃗⃗ 0
Species 𝜌𝜔 𝑗𝑖 𝜌𝜔𝑢
⃗⃗ 𝑀𝑖 𝑅𝑖
balance
Momentum 𝜌𝑢
⃗⃗ 𝜋 𝜌𝑢
⃗⃗𝑢
⃗⃗ 𝜌𝑖 ⃗⃗⃗⃗
𝑔𝑖
13
Energy 𝜌𝑐𝑝 𝑇 𝑘∇𝑇 𝑐𝑝 𝜌𝑢
⃗⃗𝑇 𝑆
Continuity:
u 0 (2.3)
t
Chemical species:
i g
i u ji M i Rig 0
n
(2.4)
t j 1 ij
i 1
i 1 (2.6)
𝒑
In equation 2.5, 𝒋𝒊 represents the diffusive flux of ith species due to pressure gradient.
𝒈
Generally this flux is negligible in CVD reactors. The forced diffusion term 𝒋𝒊 is also
zero if there is no external force other than gravity. According to Wilke, the diffusional
ji Di i DT ,i ln T (2.7)
The first term in the equation 2.7 represents diffusion due to concentration gradient and
14
1 xi
Di (2.8)
n
x
i
j 1 Di , j
i j
The diffusion coefficient for a binary mixture 𝑫𝒊𝒋 is given by Chapman-Enskog formula.
In the equation 𝜺 𝒂𝒏𝒅 𝝈 are the characteristic energy and collision diameter of the
3 1 1 2 R3T 3
Dij (2.9)
16 M i M j N 2 K BT
ij
A
n
c2
DT ,i M i M j Dij KijT (2.10)
j i
Using diluted species in horizontal flow CVD (𝜔𝑖 → 0), makes the Wilke’s
approximation same as Stefan Maxwell equation which can be written as: [8]
N Xi X j T N X i X j DT , j DT ,i
j 1, j i Dij
(V j Vi ) di
T
j 1, j i
Dij j
i
(2.11)
Momentum:
u
n
uu i gi 0 (2.12)
t i 1
p (2.13)
2
u u k ( u )
T
(2.14)
3
15
Energy:
cPT p
N N SP NRe act
Sp
DiT n
Hi
c p uT K T RT ln i ji ij H i RJ u S
t t i 1 M i i 1 M i i 1 j 1
(2.15)
In case of CVD operations which deal with low Mach number (Ma < 0.3) and low viscous
flow, effect of change of pressure on temperature and viscous energy dissipation are
negligible. Dufour effect and inter diffusion results in little influence on energy flux at
T N N Re act
cP cP uT K T ij H i R j S (2.16)
t i 1 j 1
Chemical thermodynamics deals with the inter conversion of one form of energy to
another as well as transport of energy between chemical systems. For Chemical Vapour
Deposition, the transport of energy takes place when the gaseous reactants enter the
The first step in the CVD reactor taken into account is to determine the viability of the
chemical reaction. To determine the feasibility of reaction free energy is calculated. The
reaction of the CVD process is said to be thermodynamically favourable only when the
free energy is negative (ΔGrxn <0). Gibbs free energy is calculated by the formula:
Gibbs free energy of the species can be determined by the equation 2.18:
16
T T
cP
G f (T ) H f (298) c dT TS (298)
298
P
298
T
dT (2.18)
JANAF tables are used for determining the thermodynamic data such as Gibbs free
energy, enthalpy, entropy, and specific heat. Precise values of specific heat and entropy
The Gibbs free energy of formation depends on various factors like temperature, pressure,
K ai (2.21)
Out of all probable chemical reactions the one having highest negative free energy forms
the most stable products. Also free energy analysis helps in selecting the precursor for the
given CVD process. Along with this, precursors have to be stable enough to get
In this, the rapidity with which reaction proceeds is studied. For a reaction,
𝒂𝑨 + 𝒃𝑩 → 𝒄𝑪 + 𝒅𝑫
1 d [ A] 1 d [ B] 1 d [C ] 1 d [ D]
(2.22)
a dt b dt c dt d dt
17
For the reactions taking place at constant temperature and which are far away from
1 d A
k A B
P q
(2.23)
a dt
Where, p and q are order of reaction with respect to A and B respectively and p+q is the
overall order of reaction. k is called the reaction rate constant and it incorporates the
E
k A exp A (2.24)
RT
and in order to take this effect into account, modified Arrhenius equation has to be applied
E
k AT exp A (2.25)
RT
The detailed surface chemistry model includes the adsorption of reactants and
reaction by products. It is almost impossible to quantify a model that is effective for all
conditions as the process is very complex. Hence simplified semi-empirical models are
18
𝑵𝒈 𝑵𝑺 𝑵𝒃
𝑵𝒈 𝑵𝑺 𝑵𝒃
↔ ∑ 𝒂′′ ′′ ′′ ′′ ′′ ′′
𝒊 𝑨𝒊 (𝒈) + ∑ 𝒃𝒊 𝑩𝒊 (𝒔) + ∑ 𝒄𝒊 𝑪𝒊 (𝒃)
𝒊=𝟏 𝒊=𝟏 𝒊=𝟏
NG NS NG NS
ai' bi' ai'' bi''
rf k f Ai' Bi' kr Ai'' Bi'' (2.26)
i 1 i 1 i 1 i 1
Sometimes surface reaction rate also depends on the surface coverage and pressure of few
components. For such cases surface reaction rate are specified in terms of sticking
adsorption and desorption are not dependent on the fraction of occupied sites and number
E Edes
pi bi (1 ) exp ads b2 exp (2.27)
RT RT
api
(2.28)
1 bpi
19
CHAPTER: 3
3.1 Introduction:
Computer simulations which drastically reduce scale-up and experimental efforts are used
both surface phase and gas phase and heat transfer is important. The model incorporates
complexities of industrial reactors and the effect of the temperature, pressure, mole ratio
In the present study, a horizontal flow, low pressured, hot wall CVD reactor for growth
of silicon carbide was modelled. The process gases, methyl-trichloro-silane (MTS) and
hydrogen (H2) entered the reactor at 353K from two 9mm diameter inlet pipe at bottom.
These gases flown over hot cone shaped substrate depositing a thin layer of silicon carbide
on it in a uniform, repeatable manner. These gases flown in a laminar manner inside the
reactor and finally discharged out of the reactor through a long pipe of diameter 80mm.
The surface deposition on the heated surface was given by 2-step rate mechanism which
is as follows:
20
𝑺𝒊𝑪𝒍𝟐 + 𝑪𝑯𝟒 → 𝑺𝒊𝑪 + 𝟐𝑯𝑪𝒍 + 𝑯𝟐
The inlet gas was a mixture of methyl-trichloro-silane, hydrogen and argon. Flowrate of
MTS in the inlet feed is 600mg/hr and rest is H2 and Ar. The mole ratio of H2:Ar was
varied from 0.2 to 1.5. Flowrate of gas was also varied from 100lph to 200lph. Operating
pressure of the reactor was varied from 100mbar to 300mbar. The wall of the reactor and
susceptor both were kept at the same temperature and was varied from 850°C to 1400°C.
The operating parameters and their range is summarised in the table below:
Parameters Range
Temperature 850-1400 °C
0.2-1.5 H2:Ar
21
3.3 Geometry:
The 2-D reactor geometry used for running 32 simulations was created using ANSYS
SpaceClaim. The dimensions of the geometry are shown in the figure. The reactor has
an inlet nozzle of 9mm and two such nozzles were taken on the plane of the reactor. The
substrate is cone shaped with upper diameter as 100mm and bottom diameter as 250mm
and height of the cone is 400mm. The inner diameter of the outlet is taken as 80mm. The
inner diameter of the reactor is 540mm. Before exporting the model to ANSYS Meshing,
tools like fit curves, curve gaps, duplicate curves, and small curves were used to repair
the geometry. Metric type of units with millimetre as length was used for the geometry
creation.
3.4 Mesh:
A quadrilateral dominant method with free face mesh type as quad/tri was used for
meshing. Since the flow was laminar therefore there was no need for inflations at
boundaries. The size of the element was taken to be 4.5mm and element order was kept
22
linear. The total number of elements were 33122. The maximum skewness of the mesh
was 0.6 which was far away from the upper limit of 0.98. Similarly, the aspect ratio value
was 2.4738 which was also far from the upper value of 11. Meshed 2-D geometry is
shown in figure. Grid independency test was also carried out and a graph of uniformity
and number of elements is plotted. It was observed that variation of uniformity index with
33900 0.8571
54309 0.8637
74166 0.8562
106533 0.8635
23
1.0
0.9
0.8
Uniformity Index
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
40000 60000 80000 100000 120000
Number of elements
3.5 Setup:
3.5.1 Solver: A pressure based steady state solver was used along with planar spatial
definition in 2D space list. The direction of gravity was specified in y direction as -9.81
m/s2. None- explicit source chemistry solver was used for chemical reaction solver.
3.5.2 Model Specification: This problem solved energy, continuity, momentum, and
full multicomponent diffusion and thermal diffusion were enabled along with heat of
surface reaction and mass deposition on surface. Diffusion energy enables the effect of
24
diffusive fluxes for each constituent in the mixture with respect to the concentration
gradients. Thermal diffusion causes lighter components to diffuse more quickly towards
the heated surface compared to heavy molecules. The flow was modelled as laminar flow
argon, methane, silicon dichloride and hydrogen chloride were selected from ANSYS
Fluent database. Properties of the species are listed in the table 3.4. Chemical kinetics
was modelled as 2-step mechanism given by following reaction using the reaction
r1 k1 MTS H 2
1 0.1
(3.1)
In the above equation k1 and k2 are given by modified Arrhenius theory and are calculated
by,
E
k AT exp A (3.3)
RT
25
EA is activation energy (AE)
Number of Reactants 2 2
Coefficient
1.2e+08, TE = 0 5e+07, TE = 0
Number of Products 3 3
Coefficient
26
Table 3.4: Properties of species
trichloro-silane Chloride
conductivity
Weight
Enthalpy
Entropy
Parameter
Characteristic
4 2.92 4 4
length
27
Parameters CH4 SiCl2 SiC
Polynomial Polynomial
Conductivity
Enthalpy
Parameter
Characteristic
3.758 4 0
Length
28
For a gas mixture, the incompressible ideal gas was adopted. The density variation of the
flow primarily happens due to temperature variation and is given by following expression:
P
(3.4)
y
RT i
M W ,i
Specific heat of the mixture was calculated by the mixing law which was mathematically
given by,
cP yi c p ,i (3.5)
expression,
k yi ki (3.6)
yi i (3.7)
Mass diffusivity and thermal diffusion coefficient were calculated by kinetic theory.
3.5.4 Boundary Conditions: Mass flow inlet boundary condition was specified for both
the inlets as flowrate was specified in standard temperature and pressure conditions
(STP). Pressure –outlet type of boundary condition was specified to outlet and prevent
reverse flow condition was given as long pipe was used for the same. Outer wall, substrate
and rotor all were kept at the same temperature and no slip boundary condition was given
as Knudsen Number was less than 0.01. Zero reaction rates for each of the chemical
species was used for reactor walls and inner heating walls. Boundary conditions for
29
Kn (3.8)
DC
K BT
(3.9)
2 d 2 P
3.6 Solution:
30
Coupled scheme was used for pressure velocity coupling. Momentum, energy and species
were solved using second order upwind technique and pressure was solved using second
order scheme. Least square gradient spatial discretization was used for evaluation of
The pressure based coupled algorithm solved a coupled system of equations consisting of
momentum equations and pressure based continuity equations. Rest of the equations were
solved in the decoupled manner as solved in the segregated model. ANSYS FLUENT
uses a control volume based technique to convert a general scalar transport equation to
t
dV u d A d A S dV (3.10)
u S (3.11)
t
Where,
Equation ∅
Continuity 1
x-momentum u
y-momentum v
Energy h
31
The equation was applied to each control volume or element in the computational domain.
Since the reactor was operated at steady state therefore first term could be neglected from
u v
S (3.12)
x y x x y y
u v
n e n e n e n e n e
(3.13)
𝚻
Taking 𝜌u as F and 𝛿𝑥∅ as De we get,
Fee y Fwwy Fnn x Fss x e w y n s x S xy
y y s
x e x w n
(3.14)
The calculation of Φe, Φw, Φn and Φs were done by second order upwind scheme and for
𝝏∅ 𝝏∅ 𝝏∅ 𝝏∅
all schemes (𝝏𝒙) , (𝝏𝒙) , (𝝏𝒚) , (𝝏𝒚) were calculated using centre difference scheme.
𝒆 𝒘 𝒏 𝒔
e e E P (3.15)
x e x
e
De (3.16)
x
Similarly other gradients were also found and the equation reduces to
32
Where ∑Nb depicts sum of neighbouring cells. Under relaxation factors for momentum,
3.7 Results:
The net deposition on the substrate was the result of both surface reaction and mass
transfer process. Since both the steps were in series therefore slower of the two dominated
limiting step and diffusion was rate limiting at higher temperature. Therefore, it was
just by changing the temperature. In practice the crossover temperature and the
temperature range for different regimes depend on various factors, such as precursors,
activation energy of reaction, pressure and flow condition in the reactor. It can be seen
from figure 3.4 that the deposition was controlled by surface reaction kinetics in
temperature range of 575°C -1000°C and was transport controlled beyond 1000°C.
Temperature and its effect on deposition rate (both rotor and cone wall deposition) at
200mbar pressure, 150lph flowrate and 0.85 H2:Ar is summarised in table 3.7.
575 56.88
750 117
33
850 144
1000 158.76
1125 161.64
1400 162.72
1675 163.26
2.2
log Deposition
2.0
1.8
1.6
It was observed that the deposition on a substrate surface first increased and then
decreased with increase in temperature as more deposition occurred on the rotor wall than
34
on the substrate wall at high temperature. Figure 3.5 shows the deposition of silicon
1000°C is also shown. In fig. the top left photo represents surface deposition rate at 850°C
and top right represents surface deposition rate at 1000°C. The surface deposition rate
difference contour at bottom shows that deposition rate on rotor wall increases with
increase in temperature and that on substrate wall decreases. Table 3.8 summarises the
same.
80
Deposition on substrate (mg/hr)
70
60
50
40
30
20
10
0
0 500 1000 1500 2000
Temperature (K)
35
Figure 3.6: Case Comparison Study for surface deposition rate at 850°C and
1000°C
Surface Deposition
Temperature
Deposition on rotor wall Deposition on substrate wall
36
3.7.3 Effect of pressure on overall deposition:
rotor together should increase as reaction rate increases whereas in transport controlled
regime lowering pressure increases deposition rate as diffusivity varies inversely with
deposition. Effect of pressure on overall deposition in both kinetic controlled regime and
transport controlled regime is shown in figure 3.7. Figure 3.7 was drawn for flow rate of
150
100
100 mbar
50 200 mbar
300 mbar
increases. The same is shown in figure 3.8, whereas in transport controlled regime as
37
pressure increases deposition on substrate decreases as shown in figure 3.9. Both graphs
were drawn for flowrate of 200 standard litre per hour and mole ratio of 1.5.
80
Deposition (mg/hr)
60
40
20
0
100 200 300
Pressure (mbar)
38
75
60
Deposition (mg/hr)
45
30
15
0
100 200 300
Pressure (mbar)
regime
in the reactor and therefore less deposition of silicon carbide is observed. The same is
shown in figure 3.10, whereas as when flowrate increases in transport controlled regime
in the reactor. Therefore increasing flowrate in transport controlled regime favours the
deposition on substrate.
x
(3.18)
Re x
39
ux
Re x (3.19)
90
Substrate Deposition (mg/hr)
75
60
45
30
15
0
100 150 200 250
Flowrate (lph)
40
Substrate Deposition (mg/hr) 61
60
59
58
regime
deposition in kinetic controlled regime as MTS was the limiting reactant whereas
substrate deposition increased slightly with increase in mole ratio of H2:Ar. Increase in
discussed later. Figure 3.12 shows the increase in substrate deposition with increase in
H2:Ar ratio. This is because with increase in heavier gas concentration in carrier gas,
deposition rate increases along the axial distance up to a certain height and then starts
decreasing due to consumption of SiCl2 and CH4. The same is suggested by Teng Yin et
al. in their work [12]. Also surface deposition rate curves are shown for different mole ratio
in figure 3.14.
41
70
Substrate Deposition (mg/hr)
68
66
64
62
0.0 0.5 1.0 1.5
Mole Ratio (H2:Ar)
regime
140
Overall Deposition (mg/hr)
120
100
80
60
40
20
0
0.0 0.5 1.0 1.5
Mole Ratio (H2:Ar)
-
Figure 3.13: Graph of overall deposition vs mole ratio in kinetic controlled regime
42
Figure 3.14 (a): H2:Ar = 0.2
Figure 3.14: Graph of surface deposition rate for different mole ratio
43
180
140
120
100
0.0 0.5 1.0 1.5
Mole Ratio (H2:Ar)
regime
Substrate Deposition (mg/hr)
60
58
56
54
0.0 0.5 1.0 1.5
Mole Ratio (H2:Ar)
regime
44
3.7.7 Effect of mole ratio on uniformity:
film increased. Hydrogen improved the stability of the flow field and also the reactant
distribution was much more uniform in hydrogen as compared to argon. The same is
shown in the figures 3.17 and 3.18. The figures 3.17 and 3.18 were drawn for operating
pressure of 100mbar, substrate and wall temperature of 850°C and flowrate of 200lph.
98
96
Uniformity Index (%)
94
92
90 Substrate Deposition
Overall Deposition
88
0.0 0.5 1.0 1.5
Mole Ratio (H2:Ar)
Figure 3.17: Graph of uniformity index vs Mole ratio for kinetic controlled regime
45
Uniformity Index (%) 72
70
68
Substrate Deposition
Overall Deposition
Figure 3.18: Graph of uniformity index vs mole ratio for transport controlled
regime
regime in which the reactor was operating. The same is shown in figure 3.19. This was
due to decrease in precursor concentration along axial direction resulted from deposition
46
100
90
Uniformity Index (%)
80
70
60 Substrate Uniformity
Overall Uniformity
50
500 1000 1500
Temperature (C)
transport controlled regime. At high pressure reaction rate increases and concentration of
precursor decreases along the length of the reactor. Also, increase of pressure results in
more recirculation within the reactor hence adversely affecting the uniformity. The same
47
98
94
92
90
88 Substrate Uniformity
Overall Uniformity
86
100 200 300
Pressure (mbar)
75
Uniformity Index (%)
70
65
60
Substrate Deposition
Overall Deposition
55
100 200 300
Pressure (mbar)
48
3.7.10 Effect of flowrate on uniformity:
As flowrate increased, overall as well as substrate uniformity increased for the transport
limited regime whereas for kinetic controlled regime flowrate had insignificant effect on
uniformity of film. It was observed that uniformity index was more sensitive to flow rate
in transport controlled regime in comparison to kinetic regime. The same could be seen
73
Uniformity Index (%)
72
71
70
Substrate Uniformity
Overall Unifromity
69
100 150 200 250
Flowrate (lph)
controlled regime
49
Uniformity Index (%) 98
96
94
92
Substrate Uniformity
Overall Uniformity
90
100 150 200 250
Flowrate (lph)
controlled regime
50
CHAPTER: 4
4.1 Introduction:
An effective method for describing the combined effect of various factors is Response
used for process optimisation. Central Composite Design and Box-Behnken Design are
the two major factorial designs, used to assess response surface and for developing
models. The most important benefit of this technique is the reduction in the number of
experimental trials required to estimate the effect of various factors as well as the
processes. Ramadan Z. used the RSM technique in his research paper for optimization of
[13]
operating parameters in planetary CVD reactor. Alledin also used RSM to examine
and optimise the effects of operating parameters on the synthesis of Nano carbons in a
In the present work, Response Surface Methodology was employed to examine the impact
of operating parameters i.e. inlet flowrate, mole ratio of H2:Ar, operating temperature and
reactor pressure on silicon carbide deposition and uniformity. 30 runs were performed to
create the design of experiments as shown in table 4.1. Anova was employed to evaluate
contribution of process parameters and their interactions. The optimised parameters were
The statistical tests- F-value, P-value, R-squared value, predicted R-squared value and
adjusted R-squared value were used to indicate the goodness of fit of the model. The
51
model with P-value less than 0.05 was considered to be statistically significant. P-value
greater than 0.1 indicated model terms were not significant. The R-squared value ranges
from 0 to 1 and has better prediction when R-squared values are close or equal to 1. Also,
goodness of fit of the model. Plot of predicted response values and actual response values
were used to detect the values that were not easily predicted by the model.
Response Response
Factor 1 Factor 2 Factor 3 Factor 4
1 2
D:mole
Run A:Temperature B:Pressure C:Flowrate Uniformity Deposition
ratio
52
9 850 300 200 1.5 0.9172 75.132
53
26 850 300 200 0.2 0.889 71.28
Power transform with lambda value -1.92 was used to model the uniformity of SiC film.
Two factor interaction process order was suggested by the software. Sequential p-value
for the model was less than 0.0001 and predicted R2 of 0.9876 was in reasonable
agreement with adjusted R2 of 0.9928. High R-squared value of 0.9876 means only 1.24%
of the total variations were not explained by the model. The current lambda value lied
between low and high confidence interval (-2.41, -1.4). Fit Summary report is shown in
the table 4.2. Analysis of variance results of two factor interaction model for uniformity
are shown in table 4.3. Model F-value of 398.84 implied the model was statistically
significant. In this case A, B, C, D, AB, AC, AD were significant model terms. Adequate
precision ratio of 66.45 indicated an adequate signal and model could be used to navigate
in the space. The actual vs predicted value of uniformity is shown in figure 4.2. Variance
inflation factor for all factors is one indicating all factors were orthogonal to each other.
A response equation in terms of actual factors that can be used to predict uniformity of
SiC film is given by equation 4.1. 3D surface model graph of uniformity with various
54
𝑼𝒏𝒊𝒇𝒐𝒓𝒎𝒊𝒕𝒚−𝟏.𝟗𝟐 = −𝟎. 𝟓𝟏𝟎𝟖𝟗𝟐 + 𝟎. 𝟎𝟎𝟏𝟖𝟎𝟒 × 𝑻 − 𝟎. 𝟎𝟎𝟑𝟕𝟑𝟓𝟎 × 𝑷 +
𝑷 × 𝑴 + 𝟎. 𝟎𝟎𝟎𝟓𝟕𝟓 × 𝑸 × 𝑴 (4.1)
Sum of Mean
Source df F-value p-value
Squares Square
<
Model 10.87 10 1.09 398.84 significant
0.0001
A- <
8.50 1 8.50 3119.85
Temperature 0.0001
55
<
B-Pressure 1.45 1 1.45 533.75
0.0001
<
C-Flowrate 0.0709 1 0.0709 26.04
0.0001
<
D-mole ratio 0.1259 1 0.1259 46.21
0.0001
<
AB 0.4708 1 0.4708 172.81
0.0001
56
Figure 4.1: Box-Cox Plot for Power Transform for response 1
57
Figure 4.3: 3D surface model graph of uniformity vs temperature and pressure
Figure 4.4: 3D surface model graph of uniformity vs flowrate and mole ratio
Log transform with lambda value 0 was used to model the deposition rate of silicon
carbide. Two factor interaction model process order was suggested by the software.
Sequential p-value for the model was less than 0.0001and predicted R 2 value of 0.9308
was in reasonable agreement with adjusted R2 value of 0.9659. High R-squared value of
0.9308 means only 6.92% of the total variations were not explained by the model. The
58
current lambda value lied between low and high confidence interval (-0.58, 0.7). Fit
summary report is shown in the table 4.4. Analysis of variance results of two factor
interaction model for deposition rate are shown in table 4.5. Model F-value of 83.19
implied model was statistically significant. In this case A, B, D, AB, AC, AD and BD
were significant model terms. Adequate precision ratio of 33.951 indicates an adequate
signal and model can be used to navigate in the space. The actual vs predicted value of
deposition rate is shown in figure 4.6. Variance inflation factor for all factors was one
indicating all factors were orthogonal to each other. A response equation in terms of actual
factors that could be used to predict deposition rate of SiC film is given by equation 4.2.
3D surface model graph of deposition rate with various operating parameters are shown
(4.2)
59
Table 4.5: Analysis of variance for 2FI model for response 2
Sum of Mean F-
Source df p-value
Squares Square value
A-
1.09 1 1.09 518.17 < 0.0001
Temperature
60
Figure 4.5: Box-Cox plot for log transform for response 2
61
Figure 4.7: 3D surface model graph of deposition vs temperature and pressure
Figure 4.8: 3D surface model graph of deposition vs flowrate and mole ratio
Numerical Optimisation feature of design expert was used to optimise the process
parameters. The ultimate goal was to maximise the uniformity and deposition by keeping
process parameters in the specified range. The optimised operating conditions were
mbar, flowrate of 100lph and mole ratio of 1.5.The optimum operating parameters
62
obtained are summarised in table 4.6. Under these operating conditions deposition
uniformity was 0.9648 and deposition rate was 76.48 mg/hr. The contours of the 2D
reactor for the optimum conditions are shown in figure 4.10. Also 3D contours of
Temperature 850°C
63
Figure 4.10 (a): Contour of Figure 4.10 (b): Contour of mole
temperature for 2-D reactor fraction of hydrogen for 2-D reactor
Figure 4.10 (c): Contour of mole Figure 4.10 (d): Contour of mole
fraction of argon for 2-D reactor fraction of methane for 2-D reactor
64
Figure 4.10 (e): Contour of mole Figure 4.10 (f): Contour of
fraction of hydrogen chloride mole fraction of silicon
for 2-D reactor dichloride for 2-D reactor
Figure 4.10 (g): Contour of mole Figure 4.10 (h): Contour of surface
fraction of methyl trichloro deposition rate of silicon carbide for
silane for 2-D reactor 2-D reactor
65
Figure 4.10 (i): Contour of Figure 4.10 (j): Pathlines for 2-D
velocity for 2-D reactor reactor
Figure 4.10: Contours for 2-D CVD reactor operating at optimum condition
66
Figure 4.11(a): 3D Geometry of Figure 4.11(b): Meshed section
CVD Reactor view of CVD Reactor
67
CONCLUSION
It was observed that deposition rate first increases with increase in temperature, reaches
a maximum value and then starts decreasing with further increase of temperature.
Uniformity on the other hand showed linear decrease with increase of temperature.
regime but showed opposite behaviour in transport controlled regime whereas uniformity
deposition rate on substrate and uniformity of deposition increased with increase in mole
fraction of hydrogen in carrier gas irrespective of the operating regime. It was also
observed that with surface deposition rate increased with increase in mole fraction of
argon in carrier gas. As flowrate was increased in kinetic controlled regime deposition
rate decreased while increase of flowrate in transport controlled regime increased the
sensitive to flowrate in comparison to kinetic controlled regime. In both cases the increase
performed using design expert optimised parameters obtained for CVD reactor were
operating pressure of 100mbar, substrate and wall temperature of 850°C, mole ratio of
68
References
10. Örjan Danielsson, Matts Karlsson, Pitsiri Sukkaew, Henrik Pederson, Lars
Ojamäe, “A Systematic Method for Predictive in Silico Chemical Vapor
Deposition,” The Journal of Physical Chemistry, 2020.
12. Teng Yin, Bing-yan Jian, Zhe-an Su, Zhe-qiong Fan, Qi-zhong Huang, Numerical
Simulation of Carrier Gas effects on flow field, species concentration and
deposition rate of chemical vapour deposition of carbon, Science Direct, 2018.
69
13. Zaher Ramadan, Ik-Tae Im, Optimisation of Operating Parameters in a planetary
CVD Reactor using Response Surface Methodology, Springer, 2018.
14. Ghazaleh Allaedini, Siti Masrinda Tasirin, Payam Aminayi, Yield optimisation of
nanocarbons prepared via chemical vapor deposition of carbon dioxide using
response surface methodology, Elsevier, 2016.
70