1. A Numerical Model of Template based
Chemical Vapor Deposition Process for
Carbon Nanotube Manufacturing
1
Thesis Proposal
Thursday, July 31th, 2014
Yashar Seyed Vahedein
Thesis Committee Members:
Michael G. Schrlau, PhD;
Mechanical Engineering (Advisor)
Robert Parody, PhD; Statistician
Steven Day; PhD;
Mechanical Engineering
Patricia Taboada-Serrano, PhD;
Chemical and Biomedical Engineering
Agamemnon Crassidis, PhD;
Mechanical Engineering
2. Nano-Bio Convergence
Molecular Switch
DNA barcode
Molecular Imaging
Biochip / Biosensor
Nano-therapy /
Delivery
Bio-TechnologyNano-Technology
Bionano-machine /
Nano-Robot
Bio-inspired device and system
Development of tools and
methods
• More sensitive
• More specific
• Multiplexed
• More efficient and economic
Implementation:
Diagnosis and treatment of diseases
• Rapid and sensitive detection
(Biomarkers, Imaging)
• Targeted delivery of therapeutics
Drug development
• Understanding of life science
3. Analytical tools : Atomic force microscopy(AFM), Electron microscopy (EM)
Nano-sized materials
– Magnetic nanoparticles (Ferromagnetic, super paramagnetic)
– Gold or Carbon nanotubes
– Quantum dots (Semiconductor nanocrystals)
Carbon nanotubes:
– Properties:
High Thermal conductivity(3500
𝑊
𝑚𝐾
> diamond), strength(>100 Gpa), durability relative to
their small size, one dimensional transport.
– Applications in nanobiotechnology:
Intracellular electrochemistry, drug delivery and fluid injection (single cell analysis) and etc.
– Ideal for single cell analysis. Useful in many different fields, e.g electronics, optics and etc.
Example of Tools in Nano-biotechnology
4. Carbon Nano Tubes (CNTs) Development
Research groups What they did Impact
(L. V.Radushkevich, V.
M. Lukyanovich 1952)
Multiwall nano tubes First to find imperfect
CNTs
(Iijima 1991) Highly perfect
multiwall carbon nano
tubes,
used arc discharge
And (Thess et al. 1996)
used Laser ablation to
manufacture CNTs
(Iijima and Ichihashi
1993)
Single Wall CNTs Dragged attention of
Researchers
(Zhang and Li 2009) Review of some other
types
Bent, waved, helically
coiled, branched and
beaded CNTs
(Choy 2003) Review on those Used
Chemical Vapor
Deposition
Introduced as most
efficient way of making
CNTs
(Kyotani, Tsai, and
Tomita 1995), (Martin
1994), (Schrlau et al.
2008)
Template Based-CVD
(TB-CVD)
Can produce perfectly
aligned amorphous
CNTs
(Sarno et al. 2012)
(Ciambelli et al. 2011)
Structural analysis of
CNTs made by TB-CVD
Effect of deposition
time, temperature, gas
mixture on CNT
synthesis 4(M. Golshadi, J. Maita, D. Lanza, M.
Zeiger, V. Presser, M. G. Schrlau)
2.5 µm
Study effect of gas flow rate,
temperature of furnace and time
of process on CNT synthesis using
TB-CVD
5. Schematics of the TB-CVD Setup in NBIL1
1. NBIL: Nano Bio Interface Laboratory
5
Exhaust
Heated region - causing
deposition
Heater
Heater
Flow
meter
Precursorgas
Carriergas
Temperature knobs
Position of the
templates
Carbon
deposited in
a template
6. Template based
manufacturing
CVD Experiment
CVD Simulation
fabricating
CNTs
Need for template
based manufacturing
of CNTs
Output of the
processNBIL
Single
cell
analysis
Electrical,Bio
,nano,
Mechanical
app.
High
conductivity
& strength
Inlet
Gasses
AAO
Template
Furnace
Dimensions
Temperature
and Flow
rate
NBIL
6
Motivations:
• Control the process and the effect of the
parameters on deposition.
• save time and resources by simulating the
process
• Create a universal method to be used by
others for TB-CVD (not currently available in
literature)
Diagram of the Driving
Needs, Process and
Outcome of TB-CVD
7. Research Questions and Plan
7
How can temperature profile and
flow characteristics near the
templates be identified?
Is it possible to simulate the
deposition due to CVD process for
a single (>50nm) nano-pore in a
template and to predict the
deposition rate of carbon for
different flow rates and
temperatures?
Is it possible to create a more
flexible and comprehensive model
that can predict deposition in
different flow, temperature and
furnace conditions? Schematics of the CVD 𝐌𝐨𝐝𝐞𝐥 2
2. Model by Spear 1982
8. 8
Research groups What they did Impact
(Oberlin, Endo, and
Koyama 1976b),(Tibbetts,
Devour, and Rodda 1987),
(T. Kato, K. Haruta 1992)
& etc.
Understanding reactor
operation and product
morphology
Illustrated importance of
CFD on vapor grown
carbon fibers (VGCFs)
(Endo et al. 2004a),
(Kazunori Kuwana and
Saito 2005), (Kazunori
Kuwana, Li, and Saito
2006)
Predicted Carbon deposition
rate for catalytic
decomposition of xylene
Modeled catalytic CVD
process including
reactions using CFD
(He, Li, and Bai 2011)-
with 3 stage heating tube
furnace
Investigated non-uniform
nanotube growth in
horizontal CVD reactor and
suggested changes for
experimental setup
accordingly
2D model and Experiment
on space dependent
growth rate, temperature
and flow structure coupled
with pyrolysis kinetics for
samples
(Mishra and Verma 2012) 2D, CFD simulation on the
vertical furnace
Made modifications on
the CFD code to raise the
accuracy of the model
(Ibrahim and Paolucci
2011b), (Zhou and
Wolden 2003), (Cheng, Li,
and Huang 2008) & etc.
Information on mass,
momentum, concentration
and energy conservation in
porous media
Provided information on
how to model reactions in
porous media
Use of Computational Fluid Dynamics(CFD) on
CVD Simulation
9. The Gap in Simulations Conducted so Far and
Significance of This Work
• CNT synthesis using TB-CVD is controlled by parameters such as:
Temperature of the furnace, Flow rate and Time of the process. <resulted
from empirical study in NBIL>
• No simulations have been found on TB-CVD processes without catalyst
and using only temperature as the reaction activator (Raji and Sobhan
2013).
• Numerical models provided useful information about similar experimental
setups.
• Therefore a CFD simulation is suggested to provide insight on the
fundamentals of the TB-CVD process being run in NBIL and to predict the
carbon deposition rate.
9
10. 10
Schematics of The Problem
Pore size bigger than 50nm = continuum regime
Inlet flow
rates=20 to
300 sccm
11. Heated Walls
𝑇1
= 668.160
𝑐
Mass Flow Inlet =
3.93𝑒 − 07 to
5.9𝑒 − 06
𝑘𝑔
𝑠
Static pressure
𝑝𝑠 = 0
D=3.88 mm and
4.88 mm above
the bottom wall
𝑇1
= 688.160
𝑐
𝑇1
= 668.160
𝑐
𝑑𝑇
𝑑𝑥
= 0
5 × 304.8 𝑚𝑚
𝑑𝑄
𝑑𝑥
= 0 𝑑𝑄
𝑑𝑥
= 0
No slip condition on tube wall:
𝑑𝑢
𝑑𝑥
= 0
Boundary Conditions of the Furnace
11
12. Velocity and
temperature
profiles or
Deposition
rate as
Output
3D Steady
state
and/or
transient
Simulation
of the
Processes
Iterating
continuity,
momentum,
energy and
species
conservation eq.
Defining Species
Transport Model With
Reactions for gas
decomposition and
deposition of
substances
Initial Velocity,
Species
Concentration
Creating the
Meshed
Model
UDF 3
Development
Creating a
Numerical
Model for
Reactions
and/or velocity
in the model to
be able to code
them
Defining
Boundary
Conditions
Velocity,
Temperature,
Concentration
Field and
Reaction Rates
3. UDF: User Defined Function for implementing in FLUENT code
Repeating the process for different conditions of the furnace and parameters
Simulation Steps
12
13. Trend of the Model Development
13
v1 v2 v3 v4 v5 v6 v7 v8 v9 v10 v11 v12
2D
3D
Coarse mesh
Fine mesh
Adapted Mesh
Residuals 10e-3
Residuals 10e-4
Residuals 10e-6
SIMPLE Solver
SIMPLEC Solver
PISO Solver
COUPLED Solver
COUPLED Solver - Psuedo Transient
Laminar
Turbulent - standard k-epsilon
Constant Fluid Properties
Energy equation on
Ideal gas-temp dep Cp
Temp dep Visc, Thermal cond.
Shell Conduction
Tube Furnace
Just 60 SCCM
20, 40, 60, 80, 100, 150, 300 SCCM Flow Rate
20, 40, 60, 80, 100, 200, 300 Flow Rate
axisymmetric
scaling fixed
Temp fixed based on exp data
fixed inlet Diameter
Sample
Boat 2D
Boat 3D
VERSION
SPECS
talkabout whythis
semi-transientsolver
has beenusedand
whyit isnot needed
in 2D
meshstudy,
meshadaption
Solversettings
ResidualStudy
Odd V#
Even V#
Tried
parameters
Semi-
transient
solver is
needed for 3D
but not in 2D
Mesh study,
mesh
adaption
dimensionality
Mesh quality and
adaption
Residual
study
Solver study
for best
convergence
Flow regime
Properties,
Equations,
boundary
conditions
Different Flow
rates
Modification
on model
Sample and
boat model
14. Expected Flow Conditions Inside the Tube
14
(Fotiadis and Jensen 1990) - Smoke-test - Interference holography(Giling 1982)
• 𝑹𝒆 =
𝜌𝑉𝐷
𝜇
= 3968.4063 (In 60 sccm 4000>Reynolds at inlet>2300, therefore Laminar )
• Re inside tube =
𝜌𝑉𝐷
𝜇
= 220 States Laminar flow inside the tube
• 𝑮𝒓 =
𝑔𝛽 𝑇𝑠−𝑇∝ 𝐷3
𝜈2 = 1.212980888 𝑒8, 𝛽 = 0.00113, 𝑹𝒂 = 𝐺𝑟. 𝑃𝑟 = 4.75 × 106
> 3 × 105
Turbulent
(Chiu et al. 2000)
What we
would
expect to
see in
simulation
Cross
Flows
Recirculation
15. 2D Vs. 3D
15
Boltzmann number =16.875 so Radiation
is neglected
2D does not capture the
uniform Temperature region
correctly
3D model is able to
capture the cross
flows in YZ plane
16. Comparing Mesh and Geometry of Two
Modelled Cases w and w/out Boat (Holder)
16
Mesh Report - CASE 2 : BOAT AND TEMPLATES
Mesh Information
Domain Nodes Elements
gas 1711039 1367244
Mesh Report - CASE 1 : ONLY TEMPLATES
Mesh Information
Domain Nodes Elements
gas 516040 498575
Boat
Two solid circles
as templates
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0 500000 1000000 1500000 2000000
Velocitymagnitude(m/s)
Number of mesh elements
Finding Mesh Independent Solution
Boundary layer
meshing
17. 17
Temperature Data from Simulation showing
the same trend with Experimental Results
645
650
655
660
665
670
675
680
685
690
695
2 2.25 2.5 2.75 3
Temperature(C)
Length (ft)
Temperature vs Length (Flow Rate = 500 sccm experiment and simulation)
Radial Position = 0 (inch) Data from simulation 500 SCCM near wall
Radial Position = 1 (inch) Data from simulation 500 SCCM 1 Inch
Radial Position = 1.5 (inch) Data from simulation 500 SCCM 1.5 inch
Boat
0.47582 ft
Potential reasons for data disagreement
• piece-wise linear temperature dependent properties.
• Tube considered to be isolated from outside(in simulation).
• Thermal resistivity of the tube has been neglected and temperature considered to be uniform on each zone.
• Accuracy of the sensors may create mistakes.
• Shape of the sensor which will affect the flow has not been considered in the CFD model
18. Properties on cross-section along longitude:20 and
300sccm
• Recirculation is the main
cause for having a curve
shaped temperature
distribution along the
tube.
Recirculation
regions
Recirculation
regions
20. Temperature distribution on middle-cross section
961.550961.547961.544961.541961.538961.535961.532
Median
Mean
961.54300961.54275961.54250961.54225961.54200961.54175961.54150
1st Q uartile 961.54
Median 961.54
3rd Q uartile 961.55
Maximum 961.55
961.54 961.54
961.54 961.54
0.01 0.01
A -Squared 36.97
P-V alue < 0.005
Mean 961.54
StDev 0.01
V ariance 0.00
Skewness -0.32227
Kurtosis -1.06591
N 2327
Minimum 961.53
A nderson-Darling Normality Test
95% C onfidence Interv al for Mean
95% C onfidence Interv al for Median
95% C onfidence Interv al for StDev
95% Confidence Intervals
Summary for Temperature [K]-80sccm
21. Temperature distribution on middle-cross section
• Linear increase in
temperature range in
Laminar phase.
• Temperature range
change is (0.025 K)
and can be
considered constant
for CVD
• By getting to
turbulent phase,
Temperature range
becomes narrower in
200 sccm
22. Contours of Velocity Components (u[x],v[y],w[z])
20 sccm 300 sccm200 sccm100 sccm
∆𝑈=0.0089𝑚/𝑠
TurbulentLaminar
∆𝑉=0.0356𝑚/𝑠
velocity>00 velocity
∆𝑊=0.0247𝑚/𝑠
0.025m
scale
Highest velocity value
between 3 components
but not near the Templates
23. U-velocity distribution on middle-cross section
• Linear increase in
range and mean
velocity with
increasing flow rate.
• Does not depend on
change from Laminar
to turbulent.
• Compared to v and w
component, u
velocity has the
highest change by
increasing flow rate
(∆𝑚𝑒𝑎𝑛 =
0.001156 𝑚/𝑠)
24. V-velocity distribution on middle-cross section
• Change in mean value is
in scale of 0.000001 m/s
in Laminar phase.
• Outliers in box plot and
std of .004854 are the
outcome of cross flows
due to natural convection
& buoyancy effects
(sudden changes in
density)
• ∆𝑚𝑒𝑎𝑛=.000615 m/s
between laminar and
turbulent
• This drop is due to
having more disturbed
flow
25. W-velocity distribution on middle-cross section
• Very small decrease in
mean velocity by
changing from Laminar
phase to turbulent.
∆𝑚𝑒𝑎𝑛 = .000004 m/s
• Outliers in the boxplot,
demonstrate the
disturbance which exist
here duo to buoyancy
driven flow.
• By contour plots and
statistical data, flow
conditions around
sample can be extracted
26. Conclusions from preliminary work
• Buoyancy and flow regime v and w vel.
• Threshold of change to turbulent ≤ 200 sccm and the
deposition change observed in experiments can be caused
by this.
• Symmetric and nearly constant temperature.
• Tube furnace model with boat and template as final model.
• Using Peclet number, effect of small changes in velocity
components on diffusion and reaction can be tested.
• A statistical method for relating the data in different cross
sections, flow conditions and furnace temperatures is
required.
26
Diffusion
Characteristic
length
Characteristic
velocity
27. Research Questions and Plan
27
How can temperature profile and flow characteristics near the templates be
identified?
Is it possible to simulate the deposition due to CVD process for a single (>50nm)
nano-pore in a template and to predict the deposition rate of carbon for different
flow rates and temperatures?
Study reaction kinetics and reaction-diffusion systems
Study how to simulate species transport, reaction and porous media
Develop the user defined functions (UDF) for reactions
Create model for carbon deposition in one pore
Compare carbon deposition rate with experiment to modify the model
Is it possible to create a more flexible and comprehensive model that can predict
deposition in different flow, temperature and furnace conditions?
Check requirements to achieve a general model applicable to different furnaces
Modify the numerical model to match requirements
28. Introducing diffusion-reaction system to
FLUENT
28
Generalized Source term
(constant and linear)
Generalized diffusion
coefficient
Generalized
transport variable
By taking divergence 𝛻 of these
two, they can be transformed into
volume integrals.
𝜕𝑐 𝐴
𝜕𝑡
+ 𝛻. 𝑐 𝐴 𝑉∗
= −𝑐𝛻. 𝐷𝐴𝐵 𝛻𝑥 𝐴 + 𝑛 𝐴
Rate of increase of
mole of the species 𝑖
Net rate of additions of
mole of the 𝑖 𝑡ℎ
species per
unit volume by convection
Molar-averaged
velocity
Net rate of mole of the 𝑖 𝑡ℎ
species per unit
volume by diffusion in a binary system of
components, otherwise 𝐷𝐴𝐵 𝛻xa is replaced by 𝑗𝑖
∗
The molar rate of production
of species 𝑖 by chemical
reaction. 𝑛
𝑛𝑖 = 𝑚𝑖 =
𝑗=1
𝑁 𝑐
𝑎𝑖𝑗 𝑀𝑖 𝑅𝑗
If the number of chemical reactions
taking place in the system is 𝑁𝑐, the
mass production rate is:
stoichiometric coefficient
Difference between the forward and backward reactions
29. Diffusion
AAO
Membrane
Deposited
carbon
Outlet boundary
condition from
macro-scale model
and gas spectrometry
Reaction zone
Diffusion of
remaining gas
and by-
products of
reactions
Control Volume for model
Schematics of the Micro-Scale Model
29
Inlet boundary
condition from
Macro-scale
model
• Dehydrogenization
or Coking?
By trying both and
comparing the results
with experimental data
for carbon deposition
We anticipate the
results to be presented
1. Concentration [vol
ppm] plot for each
substance versus
position along axis
of Tube
2. Carbon deposition
rate[mgc m-2h-1]
plot for different
flow rates versus
position[X]
After conducting gas
chromatography, the
substances that exist here will
be revealed
31. Required Facilities
31
• High-end computer with these specs is required for decreasing the
simulation time:
Processor: Intel(R) Xeon(R) CPU E5-2620 v2 @ 2.10GHz (2 processors)
Enabled Processor Count: 12
Total Memory: 16 GB
Local Storage: 931.51 GB (1 drives)
Graphics Card & Driver: FirePro W7000
• 3 stage ZTF CARBOLITE furnace. - Already Exist.
• AAO membranes (Whatman Anodisc 13, nominal pore diameter: 200 nm,
nominal thickness: 60 µm), Ethylene-helium gas mixture and Argon, in
already set CVD setup. - Already Exist.
• Gas chromatography - Cooperating with chemical engineering department
of RIT
• High performance research computing – Cooperating with research
computing Section of RIT
32. Acknowledgments
• I would like to thank my advisor, Professor Michael G. Schrlau and
PHD candidate Masoud Golshadi,, Dr. Taboada-Serrano, Dr. Robert
Parody Dr. Steven Day and Dr. Agamemnon Crassidis for their valued
advice and support of this work at Rochester Institute of
Technology.
• Furthermore, I would like to thank Ms. Brenda Mastrangelo and Mr.
Thomas Allston for their helps on conducting gas chromatography
and Mr. William Finch for his kind cooperation in the process of
buying the required hardware.
• Ayomipo Ayowosola, Ryan Dunn Karen De Souza Martins and all the
other lab members who helped and supported my work.
32
My noite: Chemical Vapour Deposition (CVD) involves the dissociation and/or chemical
reactions of gaseous reactants in a activated (heat, light, plasma) environment, followed
by the formation of a stable solid product. The deposition involves homogeneous
gas phase reactions, which occur in the gas phase, and/or heterogeneous
chemical reactions which occur on/near the vicinity of a heated surface leading to
the formation of powders or films, respectively.
V and w velocity are governed by buoyancy effects and flow being laminar or turbulent