AN ABSTRACT OF THE DISSERTATION OF
Yu-Wei Su for the degree of Doctor of Philosophy in Chemical Engineering
presented on June 8, 2011.
Title:
CdS
Nanocrystalline
Thin
Films
Deposited
by the
Continuous
Microreactor-Assisted Solution Deposition (MASD) Process: Growth Mechanisms
and Film Characterizations
Abstract approved:
__________________________________________________________________
Chih-Hung Chang
The continuous microreactor-assisted solution deposition (MASD) process
was used for the deposition of CdS thin films on fluorine-doped tin oxide (FTO)
glass. The MASD system, including a T-junction micromixer and a microchannel
heat exchanger is capable of isolating the homogeneous particle precipitation from
the heterogeneous surface reaction. The results show a dense nanocrystallite CdS
thin films with a preferred orientation at (111) plane. Focused-ion-beam was used
for TEM specimen preparation to characterize the interfacial microstructure of CdS
and FTO layers. The band gap of the microreactor-assisted deposited CdS film was
determined at 2.44 eV. X-ray Photon Spectroscopy show the bindings energies of
Cd 3d3/2, Cd 3d5/2, S 2p3/2 and S 2p1/2 at 411.7 eV, 404.8 eV, 162.1 eV, and 163.4 eV,
respectively.
The film growth kinetics was studied by measuring the film thickness
deposited from 1 minute to 15 minutes in physical (FIB-TEM) and optical
(reflectance spectroscopy) approaches. A growth model that accounts for the
residence time in the microchannel using empirical factor (η) obtained from
previous reported experimental data. Applying this factor in the proposed modified
growth model gives a surface reaction rate of 1.61*106 cm4mole-1s-1, which is
considerable higher than the surface reaction rates obtained from the batch CBD
process. With the feature of separating homogeneous and heterogeneous surface
reaction, the MASD process provides the capability to tailor the surface film
growth rate and avoid the saturation growth regime in the batch process.
An in-situ spectroscopy technique was used to measure the UV-Vis
absorption spectra of CdS nanoparticles formed within the continuous flow
microreactor. The spectra were analyzed by fitting with the sum of three Gaussian
functions and one exponential function in order to calculate the nanoparticle size.
This deconvolution analysis shows the formation of CdS nanoparticles range from
1.13 nm to 1.26 nm using a residence time from 0.26 s to 3.96 s. Barrier-controlled
coalescence mechanism seems to be a reasonable model to explain the
experimental UV-Vis data obtained from the continuous flow microreactor, with
a rate constant k’ value of 2.872 s-1. Using CFD, low skewness value of the RTD
curve at high flow rate (short τ) suggests good radial mixing at high flow rate is
responsible for the formation of smaller CdS nanoparticles with a narrower size
distribution.
The combination of CdS nanoparticle solution with MASD process
resulted in the hindrance of CdS thin film deposition. It is hypothesized that the
pre-existing sulfide (S2-) ions and CdS nanoparticles changes the chemical species
equilibrium of thiourea hydrolysis reaction. Consequently, the lack of thiourea
slows down the heterogeneous surface reaction.
To test the scalability of the MASD process, a flow cell and reel-to-reel
(R2R)-MASD system were setup and demonstrated for the deposition of CdS films
on the FTO glass (6ʺ x 6ʺ) substrate. The film deposition kinetics was found to be
sensitive to the flow conditions within the heat exchanger and the substrate flow
cell. The growth kinetics of the CdS films deposited by R2R-MASD process was
investigated by with a deposition time of 2.5 min, 6.3 min, and 9 min. In
comparison with the continuous MASD process, the growth rate in R2R-MASD is
higher, however more difficult to obtain a linear relationship with the deposition
time.
© Copyright by Yu-Wei Su
June 8, 2011
All Rights Reserved
CdS Nanocrystalline Thin Films Deposited by the Continuous
Microreactor-Assisted Solution Deposition (MASD) Process: Growth Mechanisms
and Film Characterizations
by
Yu-Wei Su
A DISSERTATION
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Doctor of Philosophy
Presented June 8, 2011
Commencement June 2012
Doctor of Philosophy dissertation of Yu-Wei Su presented on June 8, 2011.
APPROVED:
__________________________________________________________________
Major Professor, representing Chemical Engineering
__________________________________________________________________
Head of the School of Chemical, Biological and Environmental Engineering
__________________________________________________________________
Dean of the Graduate School
I understand that my dissertation will become part of the permanent collection of
Oregon State University libraries. My signature below authorizes release of my
dissertation to any reader upon request.
__________________________________________________________________
Yu-Wei Su, Author
ACKNOWLEDGEMENTS
I would like to sincerely express my truthful gratitude and appreciation to my
advisor Prof. Chih-Hung Chang for his enormous support and guidance during the
past seven years. I greatly appreciate that he gave me the opportunities to work on
the GAP project of the Nanobits Inc., the NSC funded project with National
Taiwan University, and the ITP project funded by the US Department of Energy. I
would also like to extend my appreciation to my committee members: Dr. Daniel
Palo, Prof. Goran Jovanovic, Prof. Brian K. Paul, Prof. Vinod Narayanan, and Prof.
Michael Penner.
Special thanks are given to the ITP team co-workers, Dr. Sudhir Ramprasad,
Mr. Clayton Hires, Mrs. Supriya Pawar, Mr. Don Higgins, and Dr. Jair Lizarazo
Adarme for their assistance in the flow cell design, experimental operation and
technical support. Without them, I could not have produced good samples for
characterization and the kinetics study could not have been accomplished. I would
also like to thank Dr. Yi Liu for his assistance on focused-ion-beam and TEM
operations.
Thanks to every group member in Chang’s group for their help and discussion
on my research. Especially, I appreciate Mrs. Kathy Han for her writing correction
on this dissertation. Thanks my lovely little buddy, Ryan Chang (1.5 year-old) and
his parents (Ean Chang, and Nancy Kao) leading me into a photography world. We
almost stepped and took pictures on every corner in Oregon. These pictures
brought me a sweet and unforgettable memory in my PhD life in USA. The last,
my deepest gratitude goes to my family for their spiritual support, and to my
girlfriend, Winnie Lai, for her encouragement from Taiwan every day. I am so
touchable that she spends her 7-year colorful life on long-distance love with me.
Finally, I want to thank all my brothers and sisters in CCCC (Corvallis
Chinese Christian Church). We did fellowship on bible study and worshiped God
every Sunday. In the past 7-year life, knowing God and then becoming a follower
of Jesus Crist are much worth than getting the PhD degree.
TABLE OF CONTENTS
Page
Chapter 1
Introduction .......................................................................................... 1
Chapter 2
Dense CdS Thin Films on Florine-doped Tin Oxide Surface by
High-rate Microreactor-Assisted Solution Deposition ........................ 6
2.1
Introduction .......................................................................................... 7
2.2
Experimental ........................................................................................ 8
2.3
Film Characterization ........................................................................ 10
2.3.1
XRD ............................................................................................ 10
2.3.2
TEM ............................................................................................ 12
2.3.3
Surface Property ......................................................................... 12
2.3.4
Optical Property .......................................................................... 15
2.4
Conclusion ......................................................................................... 15
Chapter 3
Study of Growth Kinetics for the CdS Thin Films by Continuous
Microreactor-Assisted Solution Deposition (MASD) on
Fluorine-doped Tin Oxide Surface .................................................... 24
3.1
Introduction ........................................................................................ 25
3.2
Experimental ...................................................................................... 32
3.3
Characterization ................................................................................. 32
3.3.1
Thickness .................................................................................... 32
3.3.2
Micro Structure ........................................................................... 35
3.4
Growth Kinetics ................................................................................. 37
3.5
Conclusion ......................................................................................... 40
TABLE OF CONTENTS (Continued)
Page
Chapter 4
Investigation of CdS Nanoparticle Formation and Deposition using
Continuous Microreactors.................................................................. 52
4.1
Introduction ........................................................................................ 52
4.2
Experimental ...................................................................................... 55
4.3
Results and Discussion ...................................................................... 56
4.3.1
Part 1: Nanoparticle Formation .................................................. 56
4.3.2
Part 2: Thin Film Deposition ...................................................... 60
4.4
Simulation of Residence Time Distribution (RTD) ........................... 61
4.5
Conclusion ......................................................................................... 64
Chapter 5
Influence of Flow Conditions on CdS Thin Film Growth Kinetics by
Continuous Microreactor-Assisted Solution Deposition (MASD) .... 74
5.1
Introduction ........................................................................................ 74
5.2
Experimental ...................................................................................... 75
5.3
Computational Fluid Dynamics (CFD).............................................. 76
5.3.1
Stacked Heat Exchanger ............................................................. 76
5.3.2
Flow Cell .................................................................................... 77
5.4
Chapter 6
Result and Discussion ........................................................................ 79
Analysis of CdS Thin Film by Reel-to-Reel Microreactor-Assisted
Solution Deposition (R2R-MASD) ................................................... 88
6.1
Introduction ........................................................................................ 88
6.2
Experimental ...................................................................................... 89
6.3
Film Characterization ........................................................................ 90
TABLE OF CONTENTS (Continued)
Page
6.3.1
XRD ............................................................................................ 90
6.3.2
TEM ............................................................................................ 90
6.3.3
Surface Property ......................................................................... 91
6.3.4
Optical Property .......................................................................... 92
6.4
Result and Discussion ........................................................................ 92
Chapter 7
Conclusion and Future Work ........................................................... 104
Chapter 8
Bibliography .................................................................................... 106
Appendix A: Harmonic Oscillator Approximation ............................................. 118
Appendix B: Optical and Physical Measurement Results ................................... 119
Appendix C: De-convolution Fitting Results ...................................................... 121
Appendix D: Thickness Data Points of Large CdS Films (6ʺ x 6ʺ)..................... 122
Appendix E: Time-dependent Outlet Tracer Concentration ............................... 125
LIST OF FIGURES
Figure
Page
2.1 Schematic diagram of microreactor-assisted solution deposition
(MASD) process……………………………………………………... 17
2.2
2.3
XRD spectrum of (a) bare FTO glass substrate, (b) CdS layer on
FTO after one-pass deposition and (c) two-pass deposition, (d) SnO2
(#411445, Tetragonal), (e) CdS (#750581, Cubic), and (f) CdS
(#653414, Hexagonal)………………………………………………..
18
(a) Low magnification image of CdS/FTO structure (magnification:
75,000X) (b) HRTEM images of the CdS/FTO boundary
(magnification: 620,000X). (c) Fast Fourier transformed diffraction
pattern of CdS layer (region 1). (d) Fast fourier transformed
diffraction pattern of FTO layer (region 2)…………………………..
19
2.4
Raman spectra of CdS layer deposited on FTO substrate…………....
20
2.5
AFM images of (a) bare FTO substrate (RMS = 8.15 nm), (b)
2.6
2.7
3.1
one-pass deposited CdS film (RMS = 11.32 nm), and (c) two-pass
deposited CdS film (RMS = 8.11 nm) (insert: 3D morphological
images)….………………………………………………………….....
21
XPS spectra (dashed line: as received, solid line: after etching, red
dotted line: Gaussian fitting) of the CdS films deposited by
continuous MASD (a) O 1s, (b) Cd 3d, (c) C 1s, (d) S
2p……………………………………………………………………...
22
Plot of (αhν)2 versus hν showing the band gap energy of 251.72 nm
CdS films by continuous MASD process…………………………….
23
Schematic diagram of continuous MASD process for CdS films
growth kinetic study…………………………………………………..
42
LIST OF FIGURES (Continued)
Figure
Page
3.2 Measured reflectance (solid) and calculated reflectance (dotted) of the
continous MASD deposited CdS films with a deposition time of (a) 1
min, (b) 2 min, (c) 3 min, (d) 4 min, (e) 5 min, (f) 8 min, (g) 10 min,
and (h) 15 min…………………………………………………………. 43
3.3
Schematic diagrams (a) ~ (e) of TEM specimen prepartion by FIB
process…………………………………………………………….........
3.4
3.5
3.6
3.7
3.8
3.9
4.1
44
TEM cross sectional images of the continuous MASD deposited CdS
films with a deposition time of (a) 1 min, (b) 2 min, (c) 3 min, (d) 4
min, (e) 5 min, (f) 8 min, (g) 10 min, and (h) 15min…………………..
45
GIXRD of the continuous MASD deposited CdS films with a
deposition time of 0 to 15 min…………………………………………
46
Instrumental calibration curve of CdS films by MASD process during
various deposition times (Correlation factor is 0.9104)………………..
47
Growth kinetics of CdS thin film deposited by batch process at 60 °C,
70 °C, and continuous MASD process. The series data of batch
process are from Dona et al. [31]. Solid fitted lines are based on
Kostoglue’s model (3-27) with non-fixed kH (a), and fixed kH (b)…...
48
The CdS film growth rate on SiO2/Si substrate at various surface
temperature versus residence time, and the fitting curve (solid line)
based on equation (3-29)……………………………………………….
50
Growth kinetics of CdS thin film deposited by continuous MASD
process. The fitting line is based on the modified model (3-30). The kH
is fixed at 0.0263 cm3mole-1s-1…………………………………............
51
Schematic diagrams of in-situ spectroscopic flow cell measurement
and thin film deposition by combining MASD process………………..
65
LIST OF FIGURES (Continued)
Figure
Page
4.2 Absorbance spectra of CdS nanoparticles formed at various flow
rates……………………………………………………………………. 66
4.3
Absorbance spectra (black solid line) of CdS nanoparticles formed at
various flow rates are fitted through 1st stage calculation by using
(4-9) (red dotted line), including Gaussian and exponential functions
(black dashed line)……………………………………………………..
4.4
67
Absorbance spectra (black solid line) of CdS nanoparticles formed at
various flow rates are fitted through 2nd stage calculation by using
(4-9) (red dotted line), including Gaussian and exponential functions
(black dashed line)…………………………………………………….. 68
4.5
Energies difference of the fitted three absorption bands plotted as a
function of CdS (Eg = 2.48 eV) particle size………………………….. 69
4.6
Dependence of the optical size of CdS nanopartices at various
residence times after the 1st (red squares) and 2nd (black dots) stage
analysis. The second analysis result was fitted by the
barrier-controlled coalescence model (solid line)…………………….
69
4.7
4.8
4.9
Photograph of CdS films deposited by MASD process with and
without nanoparticle precipitation……………………………………..
70
Schematic diagram of a T-junction micromixer with extended tubes
used for simulating RTD by COMSOL 4.2…………………………..
71
The plot of average time-dependent concentration profile (a) and the
normalized RTD curve (b) in different inflow velocities……………... 72
4.10 The plot of coefficient of variation (σθ) and skewness (s) as a function
of residence time (τ)……………………………………….................... 73
LIST OF FIGURES (Continued)
Figure
Page
5.1 Schematic diagram of continuous MASD system…………………… 80
5.2
5.3
Photograph (top) and cross-section diagram (bottom) of parallel
flow cell and deflected flow cell……………………………………..
80
2-D velocity and temperature profiles in the silicon heat exchanger
with various inlet velocities………………………………………......
81
5.4
Velocity contour maps of the parallel and deflected flow cell [97]….
84
5.5
The map of 36 measurement points on a 6-inch substrate using
MASD process with a flow cell……………………………………...
85
2-D CdS film thickness line profiles (row A to row F) along the flow
direction of reactants in the parallel (a) and deflected (b) flow
cell.…………………………………………………………………...
86
3-D thickness surface profiles of CdS films deposited by continous
MASD with (a) parallel flow cell and (b) deflected flow cell……….
87
6.1
Working principle of the reel-to-reel coating machine [99]………….
94
6.2
Schematic diagram of the reel-to-reel continuous MASD…………...
94
6.3
GIXRD of the reel-to-reel continuous MASD deposited CdS films
on FTO substrate……………………………………………………..
95
TEM cross-sectional images of the reel-to-reel continuous MASD
deposited CdS films on FTO substrate. (Magnification:
125,000kX)…………………………………………………………...
96
5.6
5.7
6.4
LIST OF FIGURES (Continued)
Figure
Page
6.5 AFM images of (a) bare FTO substrate (RMS = 8.15 nm), (b)
reel-to-reel continuous MASD deposited CdS film (RMS = 11.35
nm) at 6.5 min deposition time. (Insert: 3D morphological images)...
97
6.6
XPS spectra (dashed line: as received, solid line: after etching, red
dotted line: Gaussian fitting) of the CdS films deposited by
R2R-MASD (a) O 1s, (b) Cd 3d, (c) C 1s, (d) S 2p………………….
6.7
98
Plot of (αhν)2 versus hν showing the band gap energy of the
reel-to-reel continuous MASD deposited CdS films…………………
99
6.8
The location map of 36 measurement points on a 6-inch substrate
using R2R-MASD process…………………………………………... 100
6.9
2-D thickness line profiles of CdS films deposited by continuous
R2R-MASD during (a) 2.5 min, (b) 6.3 min, and (c) 9 min………… 101
6.10 3-D thickness surface proflies of CdS films deposited by continuous
R2R-MASD during (a) 2.5 min, (b) 6.3 min, and (c) 9 min….……... 102
.
6.11 Growth kinetics of CdS thin film deposited by batch process at (a)
70°C, (b) 60°C, (c) continuous MASD process, and (d) R2R-MASD
process. The series data of (a) and (b) are from Dona et al. [31].
Solid lines are fitting curves of Kostoglue’s model (3-27) and the
modified model (3-30)………………………………………………. 103
LIST OF TABLES
Table
Page
3.1 Comparison of film thickness measurement techniques……………..
42
4.1
Fit parameters in the de-convolution analysis procedure…………….
66
4.2
Kinetic constants of early stage CdS nanoparticles fitted by three
different growth models………………………………………………………..
70
Comparison of the residence time, coefficient of variation, and
skewness at different flow rates……………………………………...
73
4.3
B.1
Four optimized parameters of harmonic oscillator approximated
equation for bare FTO layer…………………………………………. 119
B.2
Optical CdS thickness at different locations measured by
spectroscopic reflectance…………………………………………….. 119
B.3
Physical CdS thickness at different locations measured by TEM…… 120
C.1
First and second stage de-convolution fitting results of CdS
nanoparticles by continuous MASD………………………………… 121
D.1
CdS films deposited by continuous MASD in the parallel flow cell...
D.2
CdS films deposited by continuous MASD in the deflected flow cell. 122
D.3
CdS films deposited by continuous R2R-MASD at 2.5min…………. 123
D.4
CdS films deposited by continuous R2R-MASD at 6.3min…………. 123
D.5
CdS films deposited by continuous R2R-MASD at 9min…………… 124
122
CdS Nanocrystalline Thin Films Deposited by the Continuous
Microreactor-Assisted Solution Deposition (MASD) Process:
Growth Mechanisms and Film Characterizations
Chapter 1 Introduction
In the past 40 years, Cadmium sulfide (CdS) is regarded as one of the most
popular semiconductor materials due to its potential application in photo-detectors
[1], thin-film transistors (TFTs) [2-4], CdTe and CIGS thin film solar cells [5].
CdS is usually fabricated as a thin or thick film for various applications. For
photo-detectors, the thickness should be at least 1µm to absorb enough light and
have electrons excite to higher energy level. For TFT and thin film solar cells, the
thickness is from 50nm to 150nm in order to show p-n diode function instead of
conductive phenomena. CdS thin films can be deposited by gas and liquid process.
Physical vapor deposition (PVD) and chemical vapor deposition (CVD) are belong
to gas processes, which usually need expensive vacuum equipment to obtain the
longer mean free path of the molecular. For a PVD process, H.S. Kowk et al. [6]
reported using the ArF laser fluence (1~5 J/cm2) in laser-evaporation deposition
(LEDE) in 1988. The target was made by high purity CdS powders (99.999%). The
CdS film was deposited on a quartz substrate, which was heated to 350°C and
located at 3cm above the target. In addition to ArF laser, Nd:YAG laser [7] at
355nm (UV) and 1064nm (IR) with 10Hz pulses are also performed in thin film
deposition. Meanwhile, thermal co-evaporation [8, 9] was applied to make CdS
thin films doped with various impurities, such as In, Mn, Se, and Sb. High purity
CdS powders and trace amounts of dopants were co-evaporated at high
temperature to the substrate. An Rf-sputtering process was also applied for CdS
2
thin film deposition in the early 1980’s. Substrate temperature (60 ~ 300 °C),
power density (0.24 ~ 2 W/cm2), and gas pressure (2~10 mTorr) are experimental
parameters explored in the process to study the growth rate [10] and electrical
properties [11] of polycrystalline CdS thin films. The film thickness was increased
with the deposition time, however the bandgap showed decreasing. This result was
expected due to a thicker film caused by larger grain size. Recently, Tomakin et al.
[12] reported how low substrate temperature (-173 ~ 27 °C) and post-deposition
annealing (200 ~ 400 °C) affect the grain size and dark resistivity. Also, the
substrate rotation was combined to reduce surface roughness and obtain the fast
vapor deposition process [13].
For a CVD process, organic metal precursors are carried by nitrogen or
hydrogen gasses to a low pressure chamber for reaction. Boone et al. [14] and
Berry et al. [15] used dimethylcadmium (DMC) and hydrogen sulfide (H2S)
diluted in helium as reagent sources. The CdS film was grown on (111) silicon
substrate at 270 ~ 550 °C in the pressure of 565 ~ 2660 Pa (5 ~ 20 Torr). Uda et al.
[16] used diethyl sulfide (DES) instead of H2S for the sulfide source. Ajayi et al.
[17] and Fainer et al. [18] respectively used bis-(morpholinodithioato-S,S’)
cadmium and cadmium dithiocarbamates as the single-source precursor instead of
the conventional two precursors process. The single-source precursor was carried
by nitrogen to the chamber, vaporized at 200 ~ 300 °C and then thermally
decomposed to CdS on the substrate (quartz glass). The reported advantages are
the simpler reactor design than conventional MOCVD and the potential to deposit
large-area devices. Until recently, only more structurally complicated single-source
precursors were used in CVD processes [19].
3
The only liquid process, chemical bath deposition (CBD) has advantages of
high quality film production and low cost fabrication. During 1960’s to 1970’s,
Russian scientists Kitaev et al. and Betenekov et al. [20-22] started the CBD
research from the quantum mechanical calculations and radiochemical experiments
of the possible dissociation mechanism in thiourea and the reaction with
Cd(NH)3)42+ ions. In 1980’s, Mexico scientists P. K. Nair and his colleague studied
the photoconductive property [1, 23-25] of CdS films, and also the growth kinetics
and material utilization yield of CBD process [26, 27]. Until the 1990’s, Dona et al
[28] reported applying an electrochemical open-circuit potential change (EOCPC)
method for in-situ measurement to understand the growth kinetics of CdS films by
CBD process. Meanwhile, Lincot et al [29] also reported a novel measurement
technique of in-situ quartz crystal microbalance (QCM) combined with
electrochemical impedance spectroscopy (EIS) to study the CdS thin film layers
during CBD. Ortega-Borges et al. [30] initially established a heterogeneous surface
reaction mechanism by controlling different variables in the CBD reaction, such as
concentrations of cadmium salt, thiourea, aqueous ammonia, ammonium salt, and
reaction temperature. In 1997, Dona et al. [31] modified Ortega-Borges’s
mechanism by replacing the original adsorbed complex intermediates (Cd(OH)2) to
the complex (Cd(OH)2(NH3)2). Oladeji et al. [32] found the heterogeneous surface
reaction competes with the homogeneous reaction, which limited the CdS film
growth rate. Optimization of CBD was studied to obtain the high thin film growth
rate. Kostoglou et al. [33-37] developed a detail comprehensive mathematical
model to study the temporal evolution of reagent concentrations in bulk fluid. The
complete numerical solutions are obtained by solving the mass balance equations
4
of each species.
So far, all CBD researches encounter the co-existing of homogeneous
reaction and heterogeneous reaction in the same space. A modified CBD process,
continuous microreator-assisted solution deposition (MASD), was developed to
spatially separate the homogeneous particle formation and heterogeneous surface
nucleation. Microreactor technology has advantages of accelerating heat and mass
transfer by high surface area-to-volume ratios within the microscale space. Our
MASD process provides the benefit of impinging a constant flux of solution to the
substrate, and efficiently controls the quantity of reagents. This continuous MASD
process has been applied successfully for depositing various thin films, such as
CdS [3, 38, 39], CuInS2 [40], CuInSe2 [41], CuSe [42], ZnO [43-45] and ZnS [46].
In addition to thin film deposition, the microreactor-assisted system was also
applied in synthesizing polyamide dendrimers [47, 48], gold nanoparticles [49],
SnTe nanorods [50] and ZnO nanowires [51, 52] for various applications.
In this research, CdS films was deposited on the fluorine-doped tin oxide
(FTO) coated glass by continue MASD process. Various characterization
techniques (chapter 2) were performed to understand the thin film microstructure;
also the growth kinetics (chapter 3) was studied by applying previous
mathematical model. The early nucleation growth of CdS nanoparticles (chapter 4)
in a microreactor was investigated by using UV-vis spectroscopy to realize the
temporal particle size. For the industrial application, MASD process was
incorporated with a closed space flow cell for large area (6 by 6 inches) thin film
deposition. The influence of flow condition for thin film uniformity was elucidated
in chapter 5. The further modified reel-to-reel (R2R) MASD process (chapter 6) is
5
similar as the industrial automatic production line with advantages of fast loading
and unloading of the substrate.
6
Chapter 2 Dense CdS Thin Films on Florine-doped Tin Oxide
Surface by High-rate Microreactor-Assisted Solution Deposition
Yu-Wei Su1, Sudhir Ramprasad2, Seung-Yeol Han1, Wei Wang1, S. O. Ryu3,
Daniel R. Palo2, Brian K. Paul4,5 and Chih-hung Chang1,5*
1
School of Chemical, Biological & Environmental Engineering, Oregon State
University, Corvallis, OR 97330, USA
2
Energy Processes and Materials Division, Pacific Northwest National Laboratory,
Corvallis, OR 9730, USA
3
School of Display and Chemical Engineering, Yeungnam University, 214-1
Dae-dong, Gyeonsan, Gyeongbuk 712-749, Republic of Korea
4
School of Mechanical, Industrial & Manufacturing Engineering, Oregon State
University, Corvallis, OR 97330, USA
5
Oregon Process Innovation Center for Sustainable Solar Cell Manufacturing, Corvallis,
OR 9730, USA
[*]
E-mail: chih-hung.chang@oregonstate.edu
Abstract
Continuous
microreactor-assisted
solution
deposition
(MASD)
is
demonstrated for depositing CdS thin film on fluorine-doped tin oxide (FTO) glass.
The continuous flow system consists of a microscale T-junction micromixer with
the co-axial water circulation heat exchanger to isolate the homogeneous particle
precipitation from the heterogeneous surface reaction. The result shows dense
nanocrystallite CdS thin films with a preferred orientation at (111) plane and a
cubic structure. Focused-ion-beam was used for TEM specimen preparation to
characterize the interfacial microstructure of CdS and FTO layers. Surface
structural property was characterized by AFM. The band gap was confirmed at
2.44 eV by UV-vis absorption spectroscopy.
7
2.1 Introduction
Cadmium sulfide (CdS) is a one of the most popular semiconductor
materials due to its potential application in photo-detectors [1], thin-film transistors
(TFTs) [2, 3], and most importantly CdS/Cu(In,Ga)Se2, and CdS/CdTe
hetero-junction solar cells [5]. In the early 1990’s, the CdTe/CdS solar cell
efficiency reached 15% [53] and the module product was commercialized by First
Solar Inc. within 10 years. Many studies [53-59] have reported that chemical bath
deposition (CBD) is a low-cost and stable method for the deposition of CdS thin
films on transparent conducting (TCO) glass. However, the conventional CBD
process has nanoparticle growth and film growth simultaneously happened in the
same reactor confinement. This phenomenon causes the CdS films have high
roughness and low area coverage. The proposed microreactor-assisted solution
deposition (MASD) process can perform better surface coverage and uniformity of
CdS films on SiO2/Si
substrate in comparison with the conventional CBD
process [38]. In this paper, we reported high quality CdS thin films on
fluorine-doped tin oxide (FTO) coated glass substrate deposited by MASD process.
Also, focus-ion-beam milling process with lift-off TEM sample preparation
technique was used for observing the interface structure between CdS and FTO
layers.
8
2.2 Experimental
Our microreactor setup make use of a micromixer for efficient mixing of
two reactant streams, and confine the homogeneous reaction in the heat
exchanger before impinging on the substrate. The MASD process (Figure 2.1)
in this study consists of a microprocessor-controlled peristaltic pump (Ismatec
REGLO Digital) for pumping each reactant stream through a 1.22 mm ID
Tygon ST tube (Upchurch Scientific). The T-junction micromixer (Upchurch
Scientific) was used for mixing these two streams. Stream A was prepared by
dissolving 0.073 g CdCl2 (Mw = 183.31 g/mole), 0.214 g NH4Cl (Mw = 53.49
g/mole), and 4.16 ml of 14.82 M (28 wt%) NH4OH in DI water. The total
volumn was added up to 50 mL. Stream B was prepared by dissolving 0.305 g
thiourea (Mw = 76.12 g/mole) in 50 mL DI water. With this recipei, the stream
A stock solution contains 0.008 M CdCl2, 0.08 M NH4Cl, and 1.23 M NH4OH;
the stream B stock solution contains 0.08M thiourea. After mixing by equal
volumn ratio, the concentration of each components in the PEEK tube was
reduced to half. The polycaryl-ether etherketone (PEEK) tube was enclosed
co-axially in a tygon-tube, which serves as a shell and tube heat exchanger with
hot water circulation maintained at 80-85 °C by a constant temperature bath.
The flow rate of each stream was controlled at 0.434mL/min to impinge the
solution onto the FTO glass substrate (Pilkington TEC 8), which was taped on a
hotplate to keep the surface temperature at 80-90 °C. The residence time of the
solution was controlled at 35 seconds. Once the process was completed, the
substrate was rinsed by DI water and dried under nitrogen gas. To better
understand the CdS crystal growth structure, a second CdS deposition process
9
was performed on the as-deposited CdS/FTO/glass. The first and second
depostion process were both performed using a deposition time of 5 minutes.
XRD (Bruker D8 Discover, CuKα=1.54056Å) was applied to characteize the
crystal structure of the CdS films after one-pass and two-pass depositions.
Focused ion beam (FEI Helios Dual Beam Microscopy) was used to prepare
cross-sectional samples for TEM analysis. Platinum was deposited on a rectangular
area of 10 μm (length) x 2 μm (width) with a 0.5 μm thick by electron beam (1.4 nA),
which was cleaned by an ion-beam with a current of 0.2 nA. Two areas with 11 μm x
6 μm, close to the platinum deposited area, were milled down in a 3 μm depth by
ion-beam (2.8 nA). The cross-section of the vertical surface was cleaned by ion-beam
with a current of 0.93 nA. Finally, small pieces of specimen were mounted on an
omniprobe and then transferred to a supported grid. The thickness of the specimen was
again milled down to less than 0.1 μm. Surface analysis of morphology, roughness,
and chemical bonding were performed by AFM (Bruker, Innova Scanning Probe
Microscope), Raman (Witec, alpha 500), and XPS (ESCALAB 250). The optical
property was characterized by UV-vis absorption spectroscopy using an Ocean Optics
USB2000, spectrometer with a halogen lamp as light source.
10
2.3 Film Characterization
2.3.1
XRD
Figure 2.2 shows the XRD diffractograms of bare FTO glass substrate (a)
and CdS films by one-pass (b) and two-pass deposition (c). Compared to the
tetraganol SnO2 (JCPDS #411445, d), the three major peaks at a 2θ value of 26.37°,
37.66°, and 51.42° of FTO substrate (a) shifted 0.2°~0.4° degree lower. The reason
could be attributed to the larger d-spacing of fluorine doped SnO2 than the pure
SnO2. All peaks showed in the FTO glass substrate are much close to SnO2, and the
orientation planes can be indexed to tetraganol structure. The XRD diffractograms
still resembled the diffractogram of bare FTO substrate after one-pass and two-pass
deposition except the peaks around 26°. It can be observed that one-pass and
two-pass deposition caused the peak at 26.39° shift to 26.63° and 26.71°,
respectively (inset figure). The inset figure magnified the 2θ range from 25.5° to
27.5° and shows a small transition peak at 26.41° and a following main peak at
26.63° for one-pass deposited CdS films (b). For thicker CdS films made by
two-pass deposition (c), the main peak shifted to 26.71° also the intensity increased
in comparison with the previous thinner films (b). Mazon-Montijo et al. [60]
reported that the underneath substrate peaks can still be seen when the top film is
not sufficiently thick. Therefore, the small transition peak (26.41°) of one-pass
deposited CdS films (b) matches with the 26.39° of FTO substrate, which has
orientation of T(110) plan. The peak at 26.63° and 26.71° from the one-pass and
two-pass deposited films could be assigned to either C(111) plane of cubic CdS
(JCPDS #760581, e) or H(002) plane of hexaganol CdS (JCPDS #653414, f).
Similar finding has also been reported by Chu et al [61]. Additional other peaks at
11
43.8° and 52° that could be assigned as C(220)/H(110) and C(311)/H(112).
However, the peak at 43.8° was not observed in these samples. The other peak at
52° of FTO substrate is index to T(211) plan of tetraganol SnO2 (JCPDS #411445,
d). According to our previous grazing incidence X-ray diffraction pattern, the
C(111) plane in CdS films can be differentiated from the FTO-coated glass
substrate [62]. Therefore, grazing incidence X-ray source could be used to identify
any possible overlap peaks. Other peaks at 33.60°, 37.68°, 51.38°, 61.42°, and
65.40° from one-pass and two-pass deposited CdS layers could be assigned to the
underneath FTO glass substrate with respective orientations of T(101), T(200),
T(211), T(310), and T(301) planes. Ikhmayies et al. [63] ever reported the shifting
T(200) plane of CdS:In/FTO at 480°C annealing was due to the formation of solid
solution CdS1-xSnx. The non-shift T(200) plan of the as-deposited CdS films can
explain no solid solution CdS1-xSnx was formed without thermal annealing.
12
2.3.2
TEM
Figure 2.3(a) and (b) show TEM images of the interfacial micro structure
between the CdS and FTO layers at low magnification (75,000X) and high
magnification (620,000X) respectively. The thickness of the two-pass deposited
CdS layer is about 251.72nm on the FTO layer with 344.83nm. The average
growth rate could be calculated as 25.2 nm/min. This rate is significantly higher
than batch CBD process. For example, Kokotov et al. [64] reported highly-textured,
columnar CdS films with an average deposition rate of 2.5 nm/min from a CdSO4
and EDA bath and 1.67 nm/min from CdCl2 and ammonia, respectively. In figure
2.3(b), the FTO layer exhibits large single crystalline structure. At the interface, the
deposited CdS layer consists of several nano-size grains and shows columnar
structure [64]. A few grains are attached closely to FTO boundary. Kim et al. [65]
reported that CdS nanocrystalline structure is cubic due to strong correlations with
the FTO structure. It implies that lattice mismatch is reduced a lot between
cubic-CdS and tetragonal FTO. The figure 2.3(c) shows the selected area
diffraction of CdS (region 1) with a fractured ring pattern, which shows the
nanocrystalline structure of the CdS film. The FTO substrate (figure 2.3(d): region
2) presents a highly oriented dot pattern, which presents highly single crystalline
material.
2.3.3
Surface Property
The Raman scattering measurement was performed at room temperature by
using an Argon ion laser with excitation wavelength at 514.5nm. Figure 2.4 shows
the observed Raman spectrum of the deposited CdS films on FTO coated glass
substrate. Two phonon peaks at 298.5 cm-1, and 592.5 cm-1 represent 1LO
13
(longitudinal optical) mode and 2LO mode. The full with at half maximum
(FWHM) of the 1LO is 24.1 cm-1. The phonon peaks position agreed well with the
reported result in earlier studies [66, 67]. Compared to bulk CdS (1LO: 305 cm-1),
it shows the down-frequency shift of the 1LO Raman peak. The reason may come
from the grain-size effect. However, none TO (transverse optical) vibration mode
was found in CdS/FTO sample. Tong et al. [68] explained that lattice mismatch
and thermal expansion coefficient mismatch between CdS film and different
substrates could cause the deformation potential.
Figure 2.5 (a) ~ (c) show AFM scanned 2D and 3D (insert) images of the
bare FTO glass (RMS = 8.15 nm), one-pass deposited films (RMS = 11.32 nm) and
two-pass deposited films (RMS = 8.11 nm). The RMS roughness of the one-pass
deposited CdS films is 11.32 nm, which is higher than 7nm RMS for the CdS film
deposited on ITO glass. This difference comes from the FTO glass substrate with
higher roughness than the ITO glass surface. Kim et al. [69] also reported the RMS
roughness between 7 and 15 nm depending on the thiourea concentration in
chemical solution deposition. The results show that a lower RMS value is obtained
by a higher thiourea mole concentration. The other finding in this research is that
the RMS value of the two-pass deposited film is lower than the one-pass deposited
film because of the voids being filled out after two-pass deposition.
XPS was performed to obtain the chemical binding information of the CdS
films. Figure 2.6 shows the presence of oxygen (O), cadmium (Cd), carbon (C),
and sulfur (S) from CdS layer. Figure 2.6 (a) shows the oxygen on the surface at
532.2 eV due to the formation of hydroxide (Cd(OH)2) layer. The data obtained
were corrected by taking the specimen charging and referring to C 1s at 284.9 eV
14
(figure 2.6 (c)), which was originated in the atmospheric contamination. For the
as-received condition, the binding energies of Cd 3d3/2 and Cd 3d5/2 shows at 411.7
eV and 404.8 eV respectively. These spectrums of Cd and S after etching can be
fitted by Gaussian function (red dot lines). For Cd 3d energy level, the binding
energy of 3d3/2 and 3d5/2 orbital are 412.7 eV and 405.9 eV respectively, with the
splitting energy of 6.8eV. For S 2p energy level, the binding energies of S 2p3/2 and
S 2p1/2 orbitals are 162.2 eV and 163.4eV respectively, with the splitting energy of
1.2eV. The obtained result shows exactly agreement with previous researches [70,
71].
15
2.3.4
Optical Property
The optical band gap (Eg) of CdS thin film was characterized by absorption
spectroscopy and determined from the formula
𝛼ℎ𝜐 = 𝐴(ℎ𝜈 − 𝐸𝑔 )𝑛
(2-1)
where hυ is the incident photon energy, A is a constant, α(cm-1) is the absorption
coefficient, and the exponent n depends on the type of transition, n = 1/2 and 2 for
direct and indirect transition, respectively. By using equation (2-2), the absorption
coefficient α can be obtained by the measured wavelength-dependent transmittance
(T), and the film thickness (t) was determined by TEM observation.
𝑇 = exp⁡(−𝛼𝑡)
(2-2)
Figure 2.7 shows the Tauc plot of the 10 minutes deposited CdS thin film (251.72
nm). The optical band gap is determined by extrapolating the curve at around
2.44eV.
2.4 Conclusion
The continuous MASD process was successfully demonstrated to deposit
dense CdS thin films on FTO substrate. The average growth rate at 25.2 nm/min is
significantly higher than batch CBD process. The CdS film shows a preferred
orientation in cubic-(111), which become more significant than the tetragonal-(110)
from FTO substrate with increasing the film thickness. The nanocrystalline
structure was observed by HRTEM through FIB specimen preparation. Two-pass
deposition can fill out the voids created by the one-pass deposition. The roughness
of two-pass deposited film (RMS = 8.11nm) shows equivalent as the bare FTO
substrate (RMS = 8.15nm).
16
Acknowledgement
The work was funded by the US Department of Energy, Industrial
Technologies
Program
through
award
#NT08847,
under
contract
DE-AC-05-RL01830 to PNNL. We are thankful to Dr. Yi Liu in Oregon State
University Microscope Facility for his assistance on FIB sample preparation and
TEM operation.
17
Figure 2.1 Schematic diagram of microreactor-assisted solution deposition
(MASD) process.
18
Figure 2.2 XRD spectrum of (a) bare FTO glass substrate, (b) CdS layer on FTO
after one-pass deposition and (c) two-pass deposition, (d) SnO2 (#411445,
Tetragonal), (e) CdS (#750581, Cubic), and (f) CdS (#653414, Hexagonal).
19
Figure 2.3 (a) Low magnification image of CdS/FTO structure (magnification:
75,000X) (b) HRTEM images of the CdS/FTO boundary (magnification:
620,000X). (c) Fast Fourier transformed diffraction pattern of CdS layer (region 1).
(d) Fast fourier transformed diffraction pattern of FTO layer (region 2).
20
Figure 2.4 Raman spectra of CdS layer deposited on FTO substrate
21
Figure 2.5 AFM images of (a) bare FTO substrate (RMS = 8.15 nm), (b) one-pass
deposited CdS film (RMS = 11.32 nm), and (c) two-pass deposited CdS film (RMS
= 8.11 nm) (insert: 3D morphological images).
22
Figure 2.6 XPS spectra (dashed line: as received, solid line: after etching, red dot
line: Gaussian fitting) of the CdS films deposited by continuous MASD (a) O 1s,
(b) Cd 3d, (c) C 1s, (d) S 2p
23
Figure 2.7 Plot of (αhν)2 versus hν showing the band gap energy of 251.72 nm
CdS films by continuous MASD process.
24
Chapter 3 Study of Growth Kinetics for the CdS Thin Films by
Continuous Microreactor-Assisted Solution Deposition (MASD) on
Fluorine-doped Tin Oxide Surface
Yu-Wei Su1, Sudhir Ramprasad2, Daniel R. Palo2, Brian K. Paul3,4 and Chih-hung
Chang1,4*
1
Oregon State University, School of Chemical, Biological & Environmental Engineering,
Corvallis, OR 97330, USA
2
Energy Processes & Material Division, Pacific Northwest National Laboratory, Corvallis,
OR 97330, USA
3
School of Mechanical, Industrial & Manufacturing Engineering, Oregon State University,
Corvallis, OR 97330, USA
4
Oregon Process Innovation Center for Sustainable Solar Cell Manufacturing, Corvallis,
OR 9730, USA
[*]
E-mail: chih-hung.chang@oregonstate.edu
Abstract
Microreactor-assisted solution deposition (MASD) process was used to
deposit CdS thin films on fluorine doped tin oxide (FTO) glass. The film growth
kinetics was studied by measuring the film thickness deposited from 1 minute to
15 minutes in physical (FIB-TEM) and optical (reflectance spectroscopy)
approaches. The crystalline structure was identified by grazing-incidence XRD
(GIXRD) and presented a preferred orientation at cubic (111) plan. A new
proposed modified model incorporated the emperical residence time factor (η)
obtaines the surface reaction rate (k0) of 1.61*106 cm4 mole-1 s-1 and R-square value
of 0.9695. Comparing to the saturation growth region in the batch process, MASD
process can prevent the free particle formation by separating homogeneous
reaction from heterogeneous surface reaction. This unique property of MASD
process alters the growth kinetics from within saturation growth to only a linear
growth.
25
3.1 Introduction
CdS is regarded as an excellent heterojunction partner for p-type CdTe or
as a buffer layer in p-CuInSe2 solar cells. In these applications, only a thin layer
around 50 to 1000 nm is required. In addition to solar cells, other applications
including photochemical cells, light meters, and image intensifiers require a
working thickness of CdS above 1μm which is much more challenging to achieve
by a batch CBD process. For example, Oladeji [32] reported that thicker CdS film
(400 ~ 500 nm) can be obtained by a batch CBD process. The entire deposition
time took between 2 hours to 4 hours. Microreactor-assisted solution deposition
(MASD) offers a solution to overcome this challenge by continuously supply
optimum reacting chemical solution to the surface [38]. High quality, high-rate
deposition of thick CdS thin films have been demonstrated by MASD approach
[39]. In this paper, we use a laboratory scale MASD system to investigate the
growth kinetics of CdS thin films.
During the mid-1960’s to 1970’s, Kitaev et al. and Betenekov et al. [20-22]
applied thermodynamic equilibrium foundation to study the solubility of cadmium
hydroxide (Cd(OH)2) and cadmium-ammonia complex ions (Cd(NH3)42+) in
alkaline solution. They suggested a solid phase precipitate Cd(OH)2 in the solution
provides the catalytic surface for decomposing thiourea and releasing the sulfur ion
(S2-) for CdS films formation. Aside from the thermodynamic study, quantum
mechanical calculations and radiochemical experiments were also performed to
explore the possible dissociation mechanism of thiourea and the reaction with
Cd(NH3)42+ ions. Kitaev and Betenekov’s efforts have made significant
contribution to better understand the solution chemistry of chemical bath
26
deposition.
In the 1990’s, the growth mechanisms of CdS thin film were studied more
thoroughly using in-situ measurement techniques. J. M. Dona et al [28] applied an
in-situ electrochemical open-circuit potential change (EOCPC) method to study the
CBD CdS growth kinetics in combination with scanning electron microscopy
(SEM) observation of thin film morphology. Lincot et al [29] reported the first
in-situ quartz crystal microbalance (QCM) and electrochemical impedance
spectroscopy (EIS) study of CBD CdS thin film deposition. The principle of QCM
for measuring film thickness is to detect the frequency changes of a quartz sensor
based on different amount of material adsorbed on its surface. The frequency
change data is then converted to film thickness. The other method, EIS, uses two
parallel capacitors and a resistor. Equivalent film thickness data can be converted
from the capacitance values. An important conclusion from these studies is to
identify three major growth regimes in batch CBD CdS deposition. The first
regime, induction/coalescence, and the second regime, compact layer growth, are
composed of an inner compact layer. The third regime is the outer porous layer
growth, which occurs at a longer reaction time. Later, Ortega-Borges et al. [30]
reported the use of in-situ QCM to monitor the CdS film growth kinetics. The
kinetic growth mechanism was established by investigating different variables in
the CBD reaction, including concentrations of cadmium salt, thiourea, aqueous
ammonia, ammonium salt, and reaction temperature. According to their
mechanism, free Cd2+ ions forms due to the dissociation of cadmium salt precursor
CdSO4 → Cd2+ + SO2−
4
(3-1)
In the presence of ammonium hydroxide, the equilibrium of ammonia and
27
ammonium ions (as buffer reagent) follows the relationship.
OH − + NH4+ ⇄ NH3 + H2 O
(3-2)
Then the free cadmium ion binds with ammonia molecular to form a complex with
a coordination number of four.
Cd2+ + 4NH3 ⇄ Cd(NH3 )2+
4
(3-3)
Thiourea (CS(NH2)2) dissociates in a basic environment to slowly release HS- ions
and then undergoes a second reaction to form S2- ions.
CS(NH2 )2 + OH − → HS − + CH2 N2 + H2 O
(3-4)
HS − + OH − ⇄ S 2− + H2 O
(3-5)
When the ionic product exceeds the solubility limit of CdS (~10-25), the
homogeneous nucleation and particle formation of CdS begins by the reaction of
free cadmium ions and sulfide ions.
Cd2+ + S 2− → CdS(s)
(3-6)
Based on Ortega-Borges’s reported experimental data, the growth mechanism is no
longer considered atom-by-atom growth but growth by complex adsorption and
decomposition mechanism. The CdS formation model is proposed as following.
(I). Reversible adsorption of cadmium mechanism hydroxide species
k
1

 Cd(OH)
−
Cd(NH3 )2+

2,ads + 4NH3
4 + 2OH + site 
k
(3-7)
-1
(II). Formation of a surface complex with thiourea
k
2
 [CdSC(NH2 )2 (OH)2,ads ]
Cd(OH)2,ads + SC(NH2 )2 
∗
(3-8)
(III). Formation of CdS with site regeneration
∗
k
3
 CdS + CH2 N2 + 2H2 O + site
[CdSC(NH2 )2 (OH)2,ads ] 
(3-9)
28
The growth rate is derived from above reactions
𝑟=
− 2
𝑘1 𝑘2 𝐶𝑠 [𝐶𝑑(𝑁𝐻3 )2+
4 ][𝑆𝐶(𝑁𝐻2 )2 ][𝑂𝐻 ]
𝑘1 𝑘2
[𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
𝑘−1 [𝑁𝐻3 ]4 +𝑘1 [𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
4 ]+𝑘2 [𝑆𝐶(𝑁𝐻2 )2 ]+
4 ][𝑆𝐶(𝑁𝐻2 )2 ]
𝑘3
(3-10)
Dona et al. [31] proposed a modified heterogeneous surface reaction mechanism
supported by SEM images and XPS data. The adsorbed complex intermediates
Cd(OH)2 proposed by Kitav and Ortega-Borges with hydroxy ions are modified to
become a dihydroxy-diammino-cadimiumn complex Cd(OH)2(NH3)2. Previous
reactions proposed by Ortega-Borges were re-written in (3-11) to (3-13). The
reasons behind this proposing change is that the transition metal easily exists in the
form of aqueous-ammonia complexes with coordination number of four and the
hydroxy (OH-) ions have a high tendency to be adsorbed on the glass surface.
Reaction (3-7) is modified to
k
1

 [Cd(NH ) (OH)
−
Cd(NH3 )2+

3 2
2,ads ] + 2NH3
4 + 2OH + site 
k
-1
(3-11)
Reaction (3-8) is modified to
k
2
 [Cd(NH3 )2 SC(NH2 )2 (OH)2,ads ]
[Cd(NH3 )2 (OH)2,ads ] + SC(NH2 )2 
∗
(3-12)
Reaction (3-9) is modified to
∗
k
3
 CdS + CH5 N3 + NH3 + 2H2 O + site
[Cd(NH3 )2 SC(NH2 )2 (OH)2,ads ] 
(3-13)
The growth rate is derived from above reactions
𝑟=
− 2
𝑘1 𝑘2 𝐶𝑠 [𝐶𝑑(𝑁𝐻3 )2+
4 ][𝑆𝐶(𝑁𝐻2 )2 ][𝑂𝐻 ]
𝑘1 𝑘2
[𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
𝑘−1 [𝑁𝐻3 ]2 +𝑘1 [𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
4 ]+𝑘2 [𝑆𝐶(𝑁𝐻2 )2 ]+
4 ][𝑆𝐶(𝑁𝐻2 )2 ]
𝑘3
(3-14)
29
So far, the study of the CdS thin films growth model was focused on
heterogeneous
surface
reaction
by
introducing
the
assumed
cadmium
complex-thiourea intermediate. Considering the role of homogeneous reaction in
the bulk solution phase, Kostoglou et al. [33-36, 72] developed a complete
comprehensive mathematical model to study temporal evolution of reagent
concentrations and the rate of nucleation, particle growth, coagulation and
particular deposition. In CBD process, the mass balance equations of Cd are given
as
𝑑[𝐶𝑑]
𝑑𝑡
=−
𝐴𝑟
𝑉
𝜌
∞
𝜌𝛼
− 𝛿 𝑚 ∫0 𝐺(𝑥)𝑓(𝑥, 𝑡)𝑑𝑥 − 𝑚 𝐻(𝑆)
𝑤
(3-15)
𝑤
[𝐶𝑑] = [𝐶𝑑 2+ ] + [𝐶𝑑(𝑁𝐻3 )2+
4 ]
(3-16)
The nomenclature of all symbols is listed in table 3.1. The term on the left-hand
side of equation (3-15) represents the temporal evolution of the total cadmium ions
concentration, including the cadmium-ammonia complex ion. The terms on the
right-hand side (from left to right) represent the Cd ions consumed for film growth,
particle growth in the bulk, and generation of new nuclei. The mass balance
equation of thiourea is given as
𝑑[𝑆𝐶(𝑁𝐻2 )2 ]
𝑑𝑡
=−
𝐴𝑟
𝑉
𝜌
∞
− 𝑘𝐻 [𝑆𝐶(𝑁𝐻2 )2 ][𝑂𝐻 − ] − (1 − 𝛿) 𝑚 ∫0 𝐺(𝑥)𝑓(𝑥, 𝑡)𝑑𝑥 (3-17)
𝑤
The terms on the right-hand side of equation (3-17) represent (from left to right)
consumption of thiourea for film growth, hydrolysis, and particle growth in the
bulk. The mass balance equation of sulfide ions are given as
𝑑[𝑆]
𝑑𝑡
𝜌
∞
𝜌𝛼
= 𝑘𝐻 [𝑆𝐶(𝑁𝐻2 )2 ][𝑂𝐻 − ] − 𝛿 𝑚 ∫0 𝐺(𝑥)𝑓(𝑥, 𝑡)𝑑𝑥 − 𝑚 𝐻(𝑆)
[𝑆] = [𝑆 2+ ] + [𝐻𝑆 − ]
𝑤
𝑤
(3-18)
(3-19)
The term on the left-hand side of equation (3-18) represents the time evolution of
30
total sulfide ions concentration, including hydrosulfide ions. The terms in the
right-hand side of equation (3-18) represent (from left to right) sulfide ions
produced by thiourea hydrolysis, sulfide ions consumed for particle growth in the
bulk, and sulfide ions consumed for generation of new nuclei. The particle
population balance is given as
𝜕𝑓(𝑥,𝑡)
𝜕𝑡
1
𝜕𝐺(𝑥)𝑓(𝑥,𝑡)
𝜕𝑥
∞
∞
= 2 ∫0 𝐵(𝑦, 𝑥 − 𝑦)𝑓(𝑦, 𝑡)𝑓(𝑥 − 𝑦, 𝑡)𝑑𝑦 − 𝑓(𝑥, 𝑡) ∫0 𝐵(𝑥, 𝑦)𝑓(𝑦, 𝑡)𝑑𝑦 −
− 𝐷(𝑥)𝑓(𝑥, 𝑡) + 𝐻(𝑆)𝛿(𝑥 − 𝛼)
(3-20)
The rate of nucleation is expressed as
𝐻=
𝐷𝑛
5/3
𝑉𝑚
𝑒𝑥𝑝 (
−∆𝐺𝑐𝑟𝑖𝑡
𝑘𝐵 𝑇
)
(3-21)
where Dn is the diffusivity of the nucleus, Vm is the molecular volume of CdS,
kB is the Boltzmann constant, and ∆Gcrit is the maximize free energy change at
critical nucleus size, which can be expressed as
∆𝐺𝑐𝑟𝑖𝑡 =
16𝜋
2
𝜎3 𝑉𝑚
(3-22)
3 (𝑘𝐵 𝑇 𝑙𝑛 𝑆⁡)2
where σ the value of is interfacial surface energy. From equation (3-6), S is given
as the supersaturation ratio.
𝑆=(
[𝐶𝑑2+ ][𝑆 2− ]
𝐾𝑠𝑝
1/2
)
(3-23)
Assuming all particles are spherical shape, the particle growth rate is expressed as
𝐺(𝑥) = (36𝜋)1/3 𝑥 2/3
𝑚𝑤 𝑟
𝜌
(3-24)
where G(x) is a function of particle volume “x”. Mw is molecular weight of CdS, ρ
is the density, and r is the surface reaction rate. The coagulation rate is used for
describing particle coagulation in colloidal solution based on Brownian motion
theory. The Brownian motion rate is given as
31
𝑥 1/3
2𝑘 𝑇
𝐵
𝐵(𝑥) = 3𝜇𝑊
[2 + (𝑦)
𝑐
𝑦 1/3
+ (𝑥 )
]
(3-25)
where (x,y) is a pair of particles with volume “x” and “y”, μis the viscosity of the
liquid. Wc is a constant. The last part is the particulate deposition rate, which is
given as
2/3 1/2
0.3𝐷𝑝 𝑢∞
𝐷(𝑥) = 𝑊
𝑑𝑣
1/6 𝐿 1/2
(3-26)
where Dp is the diffusivity of particles of volume x, ν is the kinetic viscosity of the
solution, 𝑢∞ ⁡is a characteristic velocity of the fluid, L the characteristic length of
the submerged surface, and Wd the stability ratio for particulate deposition. The
computational solved results provide abundant information on temporal evolution
of reagent (Cd, thiourea, and sulfide) concentrations, film thickness evolution for
various ammonia concentration, and final film thickness versus ammonia
concentration for various pH values. A simplified model (3-27) was proposed to
predict CdS film thickness and demonstrate the capabilities of solving the
complete model.
ℎ=
𝑚𝑤 𝑘0 [𝐶𝑑]0 [𝑆𝐶(𝑁𝐻2 )2 ]0 [𝑂𝐻 − ]
(1
𝑘𝐻 𝜌[𝑁𝐻3 ]2
− 𝑒𝑥𝑝(−𝑘𝐻 [𝑂𝐻 − ]𝑡))
(3-27)
where kH is the hydrolysis constant from equation (3-4), and k0 is the overall
reaction rate constant. Two years later, Kostoglou et al. [37] reported using SEM to
observe CdS film surface morphology on tin-oxide-coated (TOC) glass. SEM
pictures showed that the discrete CdS particles on the surface tend to coalesce with
neighboring ones to form a continuous surface with time. A model was developed
to examine the temporal film thickness evolution of instantaneous surface
nucleation and constant surface nucleation.
32
3.2 Experimental
A commercial fluorine doped tin oxide (FTO) glass (Pilkington TEC-15)
was employed as the substrate for CdS thin film deposition. Positive displacement
pumps (Acuflow Series III) were used to pump reagents at a constant flow rate of
31 ml/min. Cadmium source reagent consisted of cadmium chloride (0.008 M),
ammonium chloride (0.08 M), and ammonium hydroxide (0.08 M) in water. Sulfur
source reagent consisted of thiourea (0.08 M) in water. The reagents from the two
streams were mixed in a T-junction micromixer before entering the heat exchanger.
The mixed reagent was heated to 85 °C in the heat exchanger and impinged to the
FTO glass substrate, which was held at a surface temperature at 80 °C. The
experimental set-up is illustrated in Figure 3.1. The deposition time was varied
from 1 minute to 15 minutes for kinetics study. After deposition, the film was
rinsed with DI water to remove any particulates and by-products from the substrate.
The thin film characterization was performed on the as-deposited films.
3.3 Characterization
3.3.1
Thickness
The film thickness measurement attracts great interests due to the effects of
thickness on optical and electrical properties. Many techniques have been
developed so far to accurately measure the thickness of different materials.
Choosing a proper technique is very important in thin film metrology. All
measurement techniques can be classified as either mechanical or optical
approaches. Mechanical approaches, such as AFM and surface profiler, use a tiny
33
stylus dragged on the surface, and the result is presented as a 3D surface plot or 2D
line plot. Stylus surface profiler could not provide reliable measurement on these
samples due to the significant roughness of FTO surface. The other approach is
using microscopy, which relies on FIB (focus-ion-beam) to etch a cross-section
surface for observation. This process needs to been performed in the chamber of
scanning electron microscopy (SEM). The optical approach, spectroscopic
ellipsometry, measures the phase shift and intensity of electromagnetic wave
propagating in different media. Reflectance spectroscopy measures the intensity of
reflected light from the surface. Both of these two optical approaches do not need
surface contact and specimen preparation in vacuum, but refractive index (n) and
extinction coefficient (k) values for thickness fitting is required. The comparison of
all above approaches is showed in table 3.1. Therefore, a proper measurement
approach can obtain the accurate film thickness. In this research, reflectance
spectroscopy was applied to measure the film thickness. Direct observation of film
thickness was done by TEM to calibrate the reflectance spectroscopy
measurement.
A film stack model, FTO/Soda lime glass, was initially built to fit the
reflectance spectrum of a bare FTO glass substrate according to the harmonic
oscillator model, which was expressed as following.
m
𝜀(𝐸) = 1 +

j 0
1
𝐴𝑗 𝑒 −𝑖𝜃𝑗 (𝐸+𝐸
𝑗 +𝑖Γ𝑗
1
− 𝐸−𝐸
𝑗 +𝑖Γ𝑗
)
(3-28)
The (E) represent the complex dielectric constant as a function of electron energy
(E). Four parameters, Aj, Ej, j, and j for j = 0, 1, 2 were obtained by getting a
minimize MSE (mean square error) value. The reflectance (R) can be expressed in
34
terms of n and k values (Appendix A). The calculated thickness of FTO layer was
370 nm, which match well with TEM observation. A combination layer of
CdS/void/FTO was added on top of the FTO layer with fixed parameters to
improve the spectrum fitting [73] and obtain the desired CdS film thickness. Figure
3.2 (a) ~ (h) show the measured and calculated reflectance spectrum of various
CdS film thickness deposited from 1 to 15 minutes. It can be observed that the
fitting improves with the increased film thickness.
35
3.3.2
Micro Structure
The FIB lift-out process was commonly used for TEM specimen
preparation. Figure 3.3 (a) ~ (e) illustrate the FIB process for the TEM specimen
preparation. First, Platinum was deposited in a rectangular area of 10 μm by 2 μm
with a 0.5 μm thick by electron beam on CdS surface (Figure 3.3 (a)). The
platinum coating is to protect the specimen from damage during ion milling. Then,
two areas near the Platinum deposited area were milled down to a 3 μm depth by
gallium ion bombardment (figure 3.3 (b)). The cross-section of the vertical surface
was formed and cleaned by the ion-beam. After that, figure 3.3 (c) shows the
omniprobe with a sharp tip that was inserted to approach the vertical surface and
then welded with the surface by depositing Platinum ions. Then ion milling was
used to cut the edges connecting to the substrate. Once the specimen was separated
from the substrate, the omniprobe with specimen attached was carefully lifted
(Figure 3.3 (d)) and placed onto a specimen support. The last step was to weld the
specimen on a support and then detach the omniprobe (Figure 3.3 (e)). Finally, the
specimen requires ion-bombardment to mill the width from 2 μm to 100 nm in
order to get high quality TEM images. Figure 3.4 (a) ~ (h) show the TEM images
of the CdS film deposited from 1 min to 15 min. The thickness of each specimen is
averaged by 5 different locations (red double arrows). Many of the large grains
beneath the CdS layer are crystalline structured FTO layer.
Figure 3.5 shows the grazing-incidence (0.5⁰) X-ray diffractograms of
various CdS film thicknesses. Three peaks of bare FTO layer are index to T(110),
T(101), and T(200) of tetragonal SnO2 (JCPDS-411445). A comparison between
36
the CdS/FTO and the bare FTO layer reveals that the first peak appears obviously
in the thicker CdS films. The first peak (26.45⁰ ~ 27.0⁰) agrees well with C(111) of
cubic CdS (JCPDS-750581) and H(002) of hexagonal CdS (JCPDS-653414) [74].
Due to the absence of H(100) and H(101) plans, the as-deposited CdS films is
considered to have cubic structure.
37
3.4 Growth Kinetics
The growth kinetics was studied by fitting the thickness data points to a
theoretical model. All spectroscopic reflectance measurement results are shown in
Appendix B. Before studying the growth kinetics, calibration work was performed
first to obtain the correlation of optical thickness and physical thickness. Figure 3.6
shows the correlation factor of 0.9104 (ideal value is 1) and R-square value of
0.9803, which indicate the optical measurement is quite reliable.
For the conventional batch process, Kostoglue et al. [34] reported a model
(3-27) to fit the experimetal results given by Dona et al. [31].
ℎ=
𝑚𝑤 𝑘0 [𝐶𝑑]0 [𝑆𝐶(𝑁𝐻2 )2 ]0 [𝑂𝐻 − ]
(1
𝑘𝐻 𝜌[𝑁𝐻3 ]2
− 𝑒𝑥𝑝(−𝑘𝐻 [𝑂𝐻 − ]𝑡))
(3-27)
The h is film thickness, and each chemical species is substituted by [Cd]0 =
0.025*10-3 mole/cm3, [SC(NH2)2]0 = 0.035*10-3 mole/cm3, [OH-] = 0.005*10-3
mole/cm3, [NH3] = 1.68*10-3 mole/cm3, and k0 is the surface reaction rate obtained
by curve fitting. The molecular weight (MW) of CdS is 144.6 g/mole, and density
(ρ) is 4.82 g/cm3. Marcotrigiano et al. [75] in 1972 reported the thiourea hydrolysis
rate (kH) is about 0.0263 cm3 mole-1 s-1 at 80 °C. The kH at 60 °C, 70 °C is 0.0058
cm3 mole-1 s-1 and 0.0127 cm3 mole-1 s-1 respectively. These two values can be
calculated by Arrhenius equation with the activation energy of 17867 cal/mole.
Figure 3.7 shows the fitting results along with the experimental growth data
from the batch procss at 60 °C, 70 °C and MASD based on equation (3-27). In this
fitting analysis, k0 and kH were set as target fitting parameters in figure 3.7 (a), and
k0 as only target fitting parameter in figure 3.7 (b). The adjusted R-square of these
three series data all show good fitting results ( > 0.95). The fitting results were
38
listed in Table 3.2. The obtained hydrolysis rate is 306.8 cm3 mol-1 s-1, 716.8 cm3
mol-1 s-1 and 421.2 cm3 mol-1 s-1, respectively for 60oC batch, 70oC batch and
MASD data. These values are significantly higher than the hydrolysis rate reported
by Marcotrigiano et al. One could possibly attributed this higher thiourea
hydrolysis rate to the catalytic surface provided by the Cd(OH)2 precipitaes. This
explanation would also result in significantly higher homogeneous particle
formation.
The feature of the continous MASD process is to separate the residence
time (τ) of homogenious reaction in a microreactor and the deposition time (t) of
heterogenous reaction on substrate surface. From the previous result in figure 3.8
[39, 76], the growth rate of CdS films deposited on SiO2/Si substrate at various
surface temperature (60 ~ 80 °C) shows an increasing trend with the residence time,
and decreasing at 70 s. This observation arise a hypothesis that much longer
residence time can not enhance the film growth rate because the forming of CdS
nanoparticles consume the thiourea concentration. Considering the residence time
effect, Yu-Jen Chang’s result [76] was used to obtain the emperical residence time
factor for the continous MASD process instead of Kostoglou’s model. The
modified model originated from Dona’s derived equation (3-14).
𝑟=
− 2
𝑘1 𝑘2 𝐶𝑠 [𝐶𝑑(𝑁𝐻3 )2+
4 ][𝑆𝐶(𝑁𝐻2 )2 ][𝑂𝐻 ]
𝑘1 𝑘2
[𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
𝑘−1 [𝑁𝐻3 ]2 +𝑘1 [𝑂𝐻 − ]2 [𝐶𝑑(𝑁𝐻3 )2+
4 ]+𝑘2 [𝑆𝐶(𝑁𝐻2 )2 ]+
4 ][𝑆𝐶(𝑁𝐻2 )2 ]
𝑘3
(3-14)
Compared to k-1[NH3]2, the other terms in denominator can be neglected due to the
small values [30]. The equation (3-14) is rewritten in (3-28).
𝑟=𝑘
[𝐶𝑑]0 [𝑆𝐶(𝑁𝐻2 )2 ]0 [𝑂𝐻 − ]2
[𝑁𝐻3 ]2
(3-28)
The left term r (reaction rate of CdS) is rearranged and expressed in the thickness
39
growth rate. Also, the initial thiourea concentration ([SC(NH2)2] 0) should be
substitute to the effective thiourea concentration by multiplying an empirical factor
(η), which is a function of residence time (τ). Then, (3-28) is modified to (3-29).
dh
dt
=
mw k0 [Cd]0 [OH− ]2 [SC(NH2 )2 ]0
(exp(Aτ)
ρ[NH3 ]2
− exp(Bτ))
(3-29)
Equation (3-29) is used for fitting the CdS film growth rate on SiO2/Si substrate at
the surface temperature from 60°C to 80 °C. The average constant of A and B are
-0.003682 and -0.3106, respectively. Each chemical specie is substituted by [Cd]0 =
0.004*10-3 mole/cm3, [SC(NH2)2]0 = 0.04*10-3 mole/cm3, [OH-] = 0.001*10-3, and
[NH3] = 0.655*10-3 mole/cm3. Then, a derived equation (3-30) is used for fitting
the data from continous MASD process in Figure 3.9.
ℎ = 1.11 × 10−7 𝑘0 𝑡(𝑒𝑥𝑝(−0.003682𝜏) − 𝑒𝑥𝑝(−0.3106𝜏))
(3-30)
The thickness value of MASD process in figure 3.9 is obtained by multiplying the
correlation factor (Figure 3.6) from optical thickness. The final fitting result shows
R-square value and k0 are 0.9695 and 1.61*106 cm4mole-1s-1, respectively.
The Kostoglue’s model (3-27) presents ideal fitting only in the initial linear
region of batch process, when kH is set 0.0263 cm3mole-1s-1. Comparing to the
batch process, the growth kinetics of the CdS films by MASD process presents a
linear growth rate and without saturation region. It is concluded that the MASD
process can improve free homogenous particle formation by continuously
impinging fresh reagent to the surface.
40
3.5 Conclusion
The CdS films by continuous MASD process have been characterized and
calibrated by FIB-TEM and reflectance spectroscopy. GIXRD was successfully
applied to identify the CdS films with a preferred orientation on the C(111) plan.
The thiourea hydrolysis rate (kH) is already considered in the empirical factor (η)
based on previous experimental result. Applying this factor in the proposed
modified model gives a surface reaction rate of 1.61*106 cm4mole-1s-1. With the
feature of separating homogeneous and heterogeneous reaction, the MASD process
can alter the behavior of the saturation growth region in the bath process to linear
growth.
Acknowledgement
The work was funded by the US Department of Energy, Industrial
Technologies
Program
through
award
#NT08847,
under
contract
DE-AC-05-RL01830 to PNNL. Additional matching funds were received from the
Oregon Nanoscience and Microtechnologies Institute (ONAMI) under a matching
grant to Oregon State University. We are grateful to Mr. Don Higgins for his
assistance in the experimental setup and operation and to Dr. Jair Lizarazo-Adarme
for his help with data acquisition and control. Special thanks to Dr. Leo Asinovski
from Semiconsoft, Inc. for his help on model fitting of reflectance spectrum. We
are also thankful to the staffs in Oregon State University Microscope Facility and
the Center for Advanced Materials Characterization in Oregon (CAMCOR) their
assistance on FIB sample preparation and TEM operation.
41
List of Symbols
A
B(x)
D(x)
f(x,t)
G(x)
H(x)
h
kH
Substrate area
Coagulation rate
Particulate deposition rate
Particle number density population
Particle growth rate
Nucleation rate
Film thickness
Thiourea hydrolysis rate constant
k1
k-1
Forward rate constant of the adsorption reaction for complex cadmium and
hydroxide
Reverse rate constant of the dissociation reaction for complex cadmium
k2
k3
Mw
r
S
t
hydroxyl ions
Rate constant of the formation of a surface complex with thiourea
Rate constant of the formation of CdS with site regeneration
Molecular weight (g/mole)
Surface reaction rate
Supersaturated ratio
time
V
x
y
α
ρ
δ
Reactor volume
Particle volume
Particle volume
Volume of nucles
Density of solid CdS
=0 (completely film growth)
=1 (completely particle growth)
42
Figure 3.1 Schematic diagram of continuous MASD process for CdS films
growth kinetic study.
Vacuum
Advantage
Disadvantage
Requirement
Surface
No
profilometry
Output
Information
Easy and quick
Mask required
to make a scan
Thickness v.s
horizontal scan
length
AFM
No
Precise surface
Mask required
morphology
SEM
Yes
(FIB lift-out)
3D or 2D
images
Physical and
Difficult
Real surface
accurate
specimen
morphology
thickness is
preparation
achieved
Spectroscopic No
Non-destructive
Collecting data
 and  v.s
Ellipsometry
measurement
and model
wavelength
fitting both take
time.
Reflectance
spectroscopy
No
Non-destructive
Requires film
Reflectance v.s
and quick
stack model for
wavelength
measurement
curve fitting.
Table 3.1 Comparison of various film thickness measurement techniques.
43
Figure 3.2 Measured reflectance (solid) and calculated reflectance (dotted) of the
continous MASD deposited CdS films with a deposition time of (a) 1 min, (b) 2
min, (c) 3 min, (d) 4 min, (e) 5 min, (f) 8 min, (g) 10 min, and (h) 15min.
44
Figure 3.3 Schematic diagrams (a) ~ (e) of TEM specimen prepartion by FIB
process.
45
Figure 3.4 TEM cross sectional images of the continous MASD deposited CdS
films with a deposition time of (a) 1 min, (b) 2 min, (c) 3 min, (d) 4 min, (e) 5 min,
(f) 8 min, (g) 10 min, and (h) 15min.
46
Figure 3.5 GIXRD of the continuous MASD deposited CdS films with a deposition
time of 0 to 15 min.
47
Figure 3.6 Instrumental calibration curve of CdS films by MASD process during
various deposition times (Correlation factor is 0.9104).
48
Figure 3.7 Growth kinetics of CdS thin film deposited by batch process at 60 °C,
70 °C, and continuous MASD process. The series data of batch process are from
Dona et al. [31]. Solid fitted lines are based on Kostoglue’s model (3-27) with
non-fixed kH (a), and fixed kH (b).
49
Batch
Batch
MASD
at 70°C
at 60°C
at 80°C
(Dona et al.)
(Dona et al.)
Two fit parameters
k0 = 2.9*105
kH = 716.8
k0 = 1.2*105
kH = 306.8
k0 = 7.8*105
kH = 421.2
Fit k0 (cm4 mole-1 s-1)
k0 = 1.49*105
k0 = 6.55*104
k0 = 6.8*105
Fixed kH (cm3 mole-1 s-1)
kH = 0.0127*
kH = 0.0058*
kH = 0.0263
Fit k0 (cm4 mole-1 s-1)
Fit kH (cm3 mole-1 s-1)
One fit parameter
Table 3.2 Fitting results of Kostoglue’s redudecd model for batch and
microreator-assisted processes.
−𝐸
Calculated by 𝑘𝐻 = 𝐴𝑒𝑥𝑝 ( 𝑅𝑇𝑎 ),
*
Ea = 17867 cal/mole,
50
Figure 3.8 The CdS film growth rate on SiO2/Si substrate at various surface
temperature versus residence time, and the fitting curve (solid line) based on
equation (3-29).
51
Figure 3.9 Growth kinetics of CdS thin film deposited by continuous MASD
process. The fitting line is based on the modified model (3-30). The kH is fixed at
0.0263 cm3mole-1s-1.
52
Chapter 4 Investigation of CdS Nanoparticle Formation and
Deposition using Continuous Microreactors
4.1 Introduction
In 1996, Alivisatos published an article [77], “Semiconductor clusters,
nanocrystals and quantum dots” in the journal “Science”. A semiconductor cluster
is defined as a fragment of semiconductor consisting of hundred to many
thousands of atoms. Generally speaking, any material with a dimension of less than
100 nm should be referred to as a nanoparticle. The term “nanocrystal” or
“nanocrystalline” can be used to describe nanoparticles with a single crystalline
structure. A quantum dot combines all of these concepts as a semiconductor matter
whose excitons are confined in all three spatial dimensions. The physical
characteristic of the spatial confinement effects can cause quantum dots to have a
higher band gap than the bulk material [78, 79]. So far, most science and
engineering researchers regarded the nanocrystals with strongly size-dependent
optical and electrical property as the quantum dots. Many researchers have
reported that II-VI semiconductor nanocrystals, such as CdSe [80], CdS [81, 82],
and ZnS [83, 84] can be synthesized by a continuous flow microreactor. It is
well-known that the size-dependent optical property occurs when the size is
smaller than the bulk Bohr radius. Edel et al. [81] first reported using a continuous
flow microreactor to produce CdS nanoparticles. Sodium polyphosphate ((NaPO3)n)
was added to the cadmium nitrate solution as a surfactant to stabilize the
nanoparticles, and then the solution was mixed with sodium sulfide in the
microchannel. Each inlet stream was split into two substreams and each substream
was split again into the other two substreams. The entire micromixer part was
composed of totally 16 micro-scale substreams for each inlet. The flow rate was
53
controlled at 0.001 ~ 0.3 mL/min per each inlet and total mixing volume in the
micromixer was less than 0.6 μL. Comparing to the bulk CdS band gap of 2.48 ~
2.58 eV, the produced band gaps of the nanoparticles at room temperature were
estimated at 3.6 eV (3.2 nm particle) and 2.4 eV (12 nm particle).
After several years, Hung et al. [85] reported an alternative droplet
generation device that can make CdS nanocrystals with the sizes from 4.2 nm (Eg =
2.9 eV) to 8.2 nm (Eg = 2.6 eV). This device has advantages of rapid and highly
efficient mixing in the micro-channel. Tiemann et al. [83, 84] reported a kinetics
growth study of ZnS nanoparticles by in-situ stopped-flow UV/Vis absorption
spectroscopy. The temporal evolution of the particle size was estimated by
deconvoluting the Uv-vis absorption spectrums.
Recently, Shayeganfar et al. [82] reported effective mixing by turbulence
dispersion to control CdS nanoparticles. Absorbance spectroscopy and TEM were
applied to characterize the optical size and image size respectively. Mullaugh et al.
[86, 87] also set CdS as a model to study the size-dependent spectroscopic
properties of long-tern stable nanoparticles. In this study, the spectroscopy
associated with a flow cell is utilized to take an in-situ measurement of the CdS
nanoparticles precipitated in a commercial T-junction micromixer. The feature of
this method is optical in-situ analysis instead of ex-situ TEM analysis. Nosaka’s
finite potential model [79] and Tiemann’s [83] spectrum analysis approaches were
introduced to study the effect of residence time on sulfide nanoparticle size.
The growth kinetics can be obtained by fitting experimental data to three
previous reported models. The first model is derived from the Ostwald ripening
mechanism, which describes how small crystals re-dissolve while larger crystals
54
grow by consumption of the solute species. This mechanism can be described by
Lifshitz-Slyozov-Wagner (LSW) model.
𝑑 − 𝑑0 = 𝑡1/𝑛
(4-1)
where d is the particle diameter, d0 is the mean particle diameter at t = 0, a is the
material constant, and n is expected to be between 2 and 4. The value of n
represents the ripening controlled by surface diffusion at solid/liquid interface (n =
2), volume diffusion in the liquid medium (n = 3), and dissolution kinetics of initial
species (n = 4). The second mechanism is called the orientated attachment
mechanism. Ribeiro et al. [88] developed a simple kinetic model to describe the
crystal growth by orientated attachment. This model is referred to as the barrierless
coalescence model due to assuming no activation energy needed for coalescence.
The particle size is described in (4-2)
2𝑘[𝐴] 𝑡+1 1/3
𝑑 = 𝑑0 ( 𝑘[𝐴] 0𝑡+1 )
0
(4-2)
where [A]0 and d0 are the concentration of particle and the mean diameter when t =
0; k is the rate constant and approximated by
𝑅𝑇
𝑘=
(4-3)
where R is the universal gas constant, T is the temperature, and η is the liquid
viscosity. Huang et al. [89] proposed the third model (4-4), which is referred as
barrier-controlled reorientation.
3
𝑑 = 𝑑0
( √2𝑘 𝑡+1)
(𝑘 𝑡+1)
The rate constant k’ is related to activation energy.
(4-4)
55
4.2 Experimental
This experimental section includes the CdS nanoparticles formation in a
T-junction micromixer (part I) and the integration with thin film deposition by
MASD process (part II). Figure 4.1 (part I) shows the nanoparticle in-situ
measurement system, composed of a peristaltic pump, a flow cell and a UV/Vis
spectrophotometer. CdS nanoparticles were precipitated by mixing metal salt
solution (0.0004M Cd(NO3)2*4H2O, 0.0004M sodium polyphosphate) and sodium
sulfide solution (0.0004M Na2S*9H2O) through a T-junction micromixer. The
spectrum was captured when the flow approaches the steady state conditions. To
study the effect of residence time on particle size, each spectrum was captured by
varying the flow rate of each stream from 0.5 mL/min to 9 mL/min. The estimated
residence time could be calculated by dividing the constant internal tube volume
by the real total flow rate.
The purpose of the part II experiment is to investigate the effect of CdS
nanoparticles in thin film deposition. The precipitation of nanoparticles was
formed by pre-mixing metal salt and sodium sulfide solution, which are the same
as part I (with sodium polyphosphate). Then, the nanoparticles are mixed with the
reagents, which is composed of cadmium chloride solution (0.008M CdCl2, 0.08M
NH4Cl, and 1.23M NH4OH) and thiourea solution (0.08M). The flow rates are set
at 0.1 mL/min and 0.434 mL/min for pump 1 and pump 2, respectively.
56
4.3 Results and Discussion
4.3.1
Part 1: Nanoparticle Formation
In order to calculate the optical particle size, the de-convolution method
was derived from a finite potential model. The band gap difference of a
nanoparticle and bulk status is described as in (4-5).
𝐸𝑙,𝑛 − 𝐸𝑔 = 𝐸𝑙,𝑛 + 𝐸𝑙,𝑛 −
2
1.
(4-5)
𝑟
The  ( = r0 = 5.04*10-11) is the actual permittivity, and the subscript notation (l,
n) represent (0,0), (0,0)ˊ, (1,0) for the first excitonic transition, the first splittling
transition and the second excitonic transition. The kinetic energies of an electron
and a hole can be expressed as (4-6).
𝑖
𝐸0,0
=
0
0,0
[
+
0,0
2
𝑚∗𝑖
(𝑟 √𝑉0 +𝑐0,0 )
𝑚
⁡⁡ = 𝑒, ℎ
(4-6)
]
Numerical values of V0 (potential depth), me* (effective electron mass), mh*
(effective hole mass), and me (electron mass) are available from previous studies.
According to Nosaka et al [79], this finite potential model can be applied for CdS
nanoparticles. These parameters of CdS potential model are V0 = 3.6 eV, me* = 0.19
me, mh* = 0.8 me and a0,0 = 0.0025, b0,0 = 0.325, c0,0 = 0.4 for an electron; a0,0 = 0,
b0,0 = 0.4, c0,0 = 0.28 for a hole. Substituting all above parameters into (4-6), then
the finite potential model of CdS can be derived and approximated as
1.711
𝐸0,0 − 𝐸𝑔 = 0.00 + (𝑟+0.4
4)2
0.5
+ (𝑟+0.165)2
1.554
(𝑟+0.2514)2
(4-7)
And the nanoparticle size (r: nm) is
1.554
𝑟 = √𝐸
0,0 −𝐸𝑔
− 0.2514
(4-8)
57
The absorbance spectrum data was approximated by the sum of three Gaussian
functions and one exponential function.
3
𝐴=

i 0
𝑎𝑖
𝑒𝑥𝑝 [
𝜋
𝑤𝑖 √
−2(𝐸−𝐸𝑖 )2
𝑤𝑖2
2
𝐸
] + 𝑐1 𝑒𝑥𝑝 (𝑐 ⁡)
2
(4-9)
where the subscript notation “1”, ”2” and “3” represent the first excitonic (E0,0),
splitting (E0,0’), and second transition (E1,0) respectively.
Figure 4.2 shows how the absorption spectra of CdS nanoparticles changes
steeper with the increased volumetric flow rate. This result matches well with the
previous reported results [81, 83]. As can be seen clearly, the higher volumetric
flow rates result smaller sizes with a shorter residence time. Also, the spectra of
CdS nanoparticles have a specific wavelength at which the same absorbance for
different flow rates. This point is called the isosbestic point, which is used as
reference points in the study of reaction rates. The isosbestic point of CdS
nanoparticles located between 387 nm ~ 393 nm (3.16 ~ 3.20 eV), which is very
closed to 3.12 eV [81].
For the least-square fitting analysis, the original 11 parameters in table 4.1
were reduced to 9 unconstrained parameters for the first stage analysis by
constraining a2 = 0.4a1, and w2 = w1. The fitting process was performed by the GUI
function (curve fitting tool) in MATLAB. After the first stage analysis, the
preliminary particle size was calculated based on the energy difference of the first
excitonic transition (E1). Then the obtained first splitting excitonic transition (E2)
and the second excitonic transition (E3) were put into the (4-9) for the next
calculation. The final results needed to be constrained by (4-10)
𝐸2⁡
𝑟⁡3
𝐴
− 𝐸𝑔 = (𝑟+
)2
(4-10)
58
Also the parameters of a3 and w3 should be constrained by the linear relationship
with a1 and w1 in the second stage analysis. If E2 and E3 could not be constrained
by (4-10), then the calculation process need to be repeated from the first stage
analysis by guessing new 9 values of parameters for fitting until obtaining the
optimized results. Finally, the values of E2 and E3 calculated from (4-10), and a3 =
1.848a1, w3 = 1.56w1 were substituted into (4-9) for the second stage analysis. Then
the fitting accuracy was increased by reducing the unconstrained parameters from
9 to 5. Figure 4.3 and 4.4 show the de-convolution results of each spectrum at the
first and second stage analysis, respectively. The peak position (Ei), intensity (ai),
and full width at half maximum (wi) values and the constants (c1, c2) of the
exponential function are showed in Appendix C. Figure 4.5 shows the plot of the
energy difference at three excitonic transitions versus the particle size, which is
calculated based on (4-8). The real lines represent equation (4-10) with different
values of A and B for E2 and E3.
Figure 4.6 shows the final fitting results of CdS namoparticle size growing
with the residence time. The first (red square) and second (black circle) analysis
shows similar particle sizes from the 1st and 2nd stage calculation. This indicates
that the fitting process is stable and reliable. In order to understand the growth
mechanism, the size data of the second analysis are fitted by three proposed
models,
Ostwald
ripening
(4-1),
barrierless
coalescence
(4-2),
and
barrier-controlled reorientation (4-4). All the fitted parameters of these three
models are listed in table 4.2. The obtained value from the Ostwald ripening model
shows n ≥ 20, beyond the reasonable range (4 ≥ n ≥ 2). Hence, the Ostwald
ripening model seems highly unlikely to explain the mechanism of precipitated
59
CdS nanoparticles in the flowing status. The barrier less coalescence model shows
the value of [A]0 is 10-9 mole/L, which is much lower than the initial concentration
(5*10-4 mole/L) of reagent solution. The low value would mean that the most
precursor ions are still in solution before ripening [90]. Therefore this mechanism
is still unable to describe the real situation. The solid line in figure 4.6 shows the
fitting result by the barrier-controlled coalescence model proposed by Hung et
al.[89]. The modified rate constant k’ is 2.872 s-1 much higher than 0.002 s-1 [87].
The possible reason for the variance could be the reported time scale of this
long-term observation was lasting for several thousand seconds.
60
4.3.2
Part 2: Thin Film Deposition
Figure 4.7 shows the photograph of the CdS films deposited by MASD
process with and without combining precipitated CdS nanoparticles. It can be
visually observed that the combination of CdS nanoparticles does not produce a
visible yellow CdS films on the substrate. That indicates the combination of CdS
nanoparticles hinder the CdS thin film growth rate of the original MASD process
(Figure 2.1). A possible mechanism is that the CdS nanoparticles from
precipitation (3-6) in stream E (figure 4.1: part II) acts as seeds after mixing with
stream F. The already existing sulfide (S2-) ions from sodium sulfide solution will
also precipitate with cadmium ions (3-6) from stream F. The consuming of sulfide
ions changes the chemical species equilibrium and forwards the reaction (3-5) and
(3-4) going to right hand side.
CS(NH2 )2 + OH − → HS − + CH2 N2 + H2 O
(3-4)
HS − + OH − ⇄ S 2− + H2 O
(3-5)
Cd2+ + S 2− → CdS(s)
(3-6)
Therefore, the amount of thiourea adsorbed on the substrate surface to form
thiourea-hydroxyl complex was decreased, and reaction (3-8) and (3-9) were
reduced significantly.
k
2
 [CdSC(NH2 )2 (OH)2,ads ]
Cd(OH)2,ads + SC(NH2 )2 
∗
k
∗
3
 CdS + CH2 N2 + 2H2 O + site
[CdSC(NH2 )2 (OH)2,ads ] 
(3-8)
(3-9)
61
4.4 Simulation of Residence Time Distribution (RTD)
Residence time distribution (RTD) was calculated in this study to better
understand the impacts of flow rate on the CdS nanoparticle synthesis within the
continuous flow system. Adeosum and Lawal [91-94] compared the RTD function
by computational fluidic dynamic (CFD) with experimental data of a
multi-laminated micromixer and a T-junction micromixer. COMSOL 4.2 is used to
simulate the flow distribution in the microchannel of the in-situ spectroscopic flow
cell measurement system. Figure 4.8 shows the geometry of the flow system
including a T-junction micromixer, an extended PEEK tube and a connection tube.
The module of laminar flow and transport of diluted species are added in a
time-dependent setting and solved by incompressible Navier-Stokes equations. The
boundary condition “Normal inflow velocity” was used for two inlets, “zero
pressure” for outlet, and “No slip” for other boundaries set as wall. Different
inflow velocities were set at 0.042, 0.085, 0.161, 0.242, 0.323, 0.402, and 0.721
m/s to observe the velocity effect on the RTD functions. The velocity is calculated
by dividing the inflow volume rate (1.0, 2.0, 3.8, 5.7, 7.6, 9.5 and 17.0 mL/min) by
the cross-section area of inlets. Finally, this computational velocity field is used as
the initial condition for the convection term in the module of transport of diluted
species. According to the experimental precursor concentration in section 4.2, the
dilute tracer specie concentration in inlets was set at a value of 0.4 mole/m3 with an
injection time of 0.2 seconds, and “outflow” was set in the outlet as the boundary
condition. Also, the diffusion coefficient was set at a value of 1*10-9 m2/s in this
system.
62
The computational time-dependent outlet tracer concentration profiles
given in Appendix E show the outlet concentration decreased with the decreasing
inflow velocity, since the enhanced convection in axial direction is caused by
longer residence time. Therefore, the same amount tracer is diluted and lower
concentration is detected in the outlet. The RTD function E(t) is mathematically
expressed as
𝐸(𝑡) =
𝐶(𝑡)
∞
∫0 𝐶(𝑡)𝑑𝑡
𝐶(𝑡𝑖 )
∞
∑𝑖=0 𝐶(𝑡𝑖 )∆𝑡𝑖
⁡
(4-11)
where ∆ti is the time step for the measurement. Figure 4.9 (a) shows the C(t) is
the outlet time-dependent tracer concentration. Each data point in a C(t) curve
represent the mean concentration at a certain time step. A normalized E(θ)
function in figure 4.9 (b) is used instead of E(t) when two different flow
conditions need comparison. The relationship of E(t) and E(θ) RTD functions are
expressed by (4-12) and (4-13).
𝐸(𝜃) = 𝜏𝐸(𝑡)
(4-12)
𝑡
𝜃=𝜏
(4-13)
∞
The residence time (τ) is defined in (4-14), since ∫0 𝐸(𝑡)𝑑𝑡 = 1.
∞
𝜏=
∫0 𝑡𝐸(𝑡)𝑑𝑡
∞
∫0 𝐸(𝑡)𝑑𝑡
∞
= ∫0 𝑡𝐸(𝑡)𝑑𝑡
(4-14)
The variation (σ2), and coefficient of variation (σθ) are dimensionless characteristic
parameters for quantifying the mixing performance of the flow system. They can
be derived from (4-15) and (4-16).
∞
𝜎 2 = ∫0 (𝑡 − 𝜏)2 𝐸(𝑡)𝑑𝑡
𝜎𝜃 ⁡ =
𝜎
𝜏


i 0
(𝑡𝑖 − 𝜏)2 𝐸𝑖 (𝑡)∆𝑡𝑖
(4-15)
(4-16)
63
The variation is a parameter to measure the width of a RTD curve. Lower value
means the RTD curve a narrower width. There is another useful parameter, s,
which is used for identifying the skewness (4-17) of a RTD curve.
∞
𝑠=
∫0 (𝑡−𝑡𝑚 )2 𝐸(𝑡)𝑑𝑡
𝜎 1.5
𝜏
( )
(4-17)
Figure 4.10 shows the CFD results of variation (σθ) and skewness (s) as a function
of residence time (τ). The corresponded flow rate and inflow velocity are also
listed in table 4.3 for comparison. The skewness measures the degree of symmetric
distribution. When good radial mixing is achieved at high flow rate, the RTD curve
is highly symmetric with low skewness [95]. Comparing to the particle size growth
in figure 4.6, a large number of small CdS nanoparticles are formed by good radial
mixing at high flow rate. On the contrary, poor mixing at slow flow rate causes few
amounts of large particles within the longer residence time. However, the
coefficient of variation does not present absolutely decreasing with the residence
time in our T-junction micromixer system. This finding was ever reported by
Adeosum and Lawal [94] but no further discussion on the relationship of
coefficient of variation and residence time was addressed. Table 4.3 also shows the
comparison of the residence time value obtained by estimation (τe) and RTD cure
(τ). The τe is given by dividing the microreactor internal volume (0.00393 cm3) to
the flow rate. This estimation method shows a good approach for a reactor with
simple geometry. Experimental RTD approach is still essential and more accurate
for the complicated geometric system.
64
4.5 Conclusion
In this study a novel in-situ spectroscopy technique was developed to
capture the UV absorption spectra of CdS nanoparticles precipitated through a
T-junction micromixer. The spectra were successfully fitted by the sum of three
Gaussian functions and one exponential function in order to obtain the nanoparticle
size. This deconvolution analysis shows the size between 1.13 nm to 1.26 nm
under the residence time from 0.26 s to 3.96 s. Barrier-controlled coalescence is a
reasonable model for the precipitated CdS nanoparticles growth, and the modified
rate constant k’ is 2.872 s-1. By using CFD, low skewness value of the RTD curve
at high flow rate (short τ) can explain good radial mixing at high flow rate, also a
large number of small CdS nanoparticles are formed in this condition.
The precipitated reaction combined with MASD process for thin film
deposition was found a very low surface reaction rate. The pre-existing sulfide (S2-)
ions and CdS nanoparticles changes the chemical species equilibrium of thiourea
hydrolysis reaction. Consequently, the lack of thiourea adsorbed to the surface for
cadmium-ammonia-thiourea complex interrupts the heterogeneous surface
reaction.
65
Figure 4.1 Schematic diagrams of in-situ spectroscopic flow cell measurement
and thin film deposition by combining MASD process.
66
Figure 4.2 Absorbance spectra of CdS nanoparticles formed at various flow rates.
Fit parameters in equation (4-9)
Peak 1
1st stage analysis
2nd stage Analysis
E1
Unconstrained
Unconstrained
a1
Unconstrained
Unconstrained
w1
Unconstrained
Unconstrained
E2
Unconstrained
a2
a2 = 0.4a1
a2 = 0.4a1
w2
w2 = w1
w2 = w1
𝐸2 − 𝐸𝑔 =
3.626
(𝑟 + 0.64 1)2
Peak 2
Unconstrained
E3
𝐸3 − 𝐸𝑔 =
34.06
(𝑟 + 3.6360)2
Peak 3
exponential
a3
Unconstrained
a3 = 1.848a1
w3
Unconstrained
w3 = 1.56w1
c1
Unconstrained
Unconstrained
c2
Unconstrained
Unconstrained
Table 4.1 Fit parameters in the de-convolution analysis procedure.
67
Figure 4.3 Absorbance spectra (black solid line) of CdS nanoparticles formed at
various flow rates are fitted through 1st stage calculation by using (4-9) (red dotted
line), including Gaussian and exponential functions (black dashed line).
68
Figure 4.4 Absorbance spectra (black solid line) of CdS nanoparticles formed at
various flow rates are fitted through 2nd stage calculation by using (4-9) (red
dotted line), including Gaussian and exponential functions (black dashed line).
69
Figure 4.5 Energies difference of the fitted three absorption bands plotted as a
function of CdS (Eg = 2.48 eV) particle size.
Figure 4.6 Dependence of the optical size of CdS nanopartices at various
residence times after the 1st (red squares) and 2nd (black dots) stage analysis. The
second analysis result was fitted by the barrier-controlled coalescence model (solid
line).
70
Ostwald Ripening
Barrier less
Barrier-controlled
(4-1)
Coalescence
Coalescence
(4-2)
(4-4)
d0 = 0.1997 nm
d0 = 1.014 nm
d0 = 1.014 nm
-1
a = 0.9995
k[A]0 = 2.268 s
n = 21.59 (not reasonable)
k = 2.47*109 L mole-1 s-1
adjusted R2 = 0.9887
[A]0 = 0.92*10-9 mole L-1
kˊ = 2.872 s-1
adjusted R2 = 0.9580
adjusted R2 = 0.9598
Table 4.2 Kinetic constants of early stage CdS nanoparticles fitted by three
different growth models.
Figure 4.7 Photograph of CdS films deposited by MASD process with and
without nanoparticle precipitation.
71
Figure 4.8 Schematic diagram of a T-junction micromixer with extended tubes
used for simulating RTD by COMSOL 4.2
72
(a)
(b)
Figure 4.9 The plot of average time-dependent concentration profile (a) and the
normalized RTD curve (b) in different inflow velocities.
73
Figure 4.10 The plot of coefficient of variation (σθ) and skewness (s) as a function
of residence time (τ).
Flow rate
Velocity
Residence Time
(mL/min)
(m/s)
(seconds)
σ2
Coefficient of Skewness, s
variation, σθ
τe
τ
1
0.042
3.960
2.995
2.687
0.547
6.635
2
0.085
1.980
1.392
0.499
0.508
1.381
3.8
0.161
1.042
1.028
0.441
0.646
0.850
5.7
0.242
0.695
0.729
0.255
0.693
0.442
7.6
0.322
0.521
0.621
0.179
0.682
0.318
9.5
0.402
0.417
0.505
0.102
0.632
0.203
17
0.721
0.233
0.332
0.049
0.664
0.090
Table 4.3 Comparison of the residence time, coefficient of variation, and skewness
at different flow rates.
74
Chapter 5 Influence of Flow Conditions on CdS Thin Film
Growth Kinetics by Continuous Microreactor-Assisted Solution
Deposition (MASD)
5.1 Introduction
The conventional chemical solution reaction in a batch reactor has low
thin-film production yield, which is determined by the mass ratio of final film
thickness and the original reagents quantity. The reason is that the formation of
homogeneous CdS nanoparticles competes with heterogenous surface reaction by
consuming cadmium salt and thiourea. In order to increase the yield in a batch
reactor, a small spacing distance between two substrates was performed to trap
reagents via the surface tension [26, 27]. Lee et al. [96] prepared ZnO seed layer in
poly-dimethylsiloxane (PDMS) microfluidic channels, and then grew ZnO
nanowires by chemical vapor deposition (CVD). We presented the continuous
MASD process composed of microfluidic channels and a deposition chamber to
separate the homogeneous particles formation and heterogeneous surface
nucleation. This MASD process has been applied in growing ZnO [43, 45] and
CdS [3, 38, 39] thin films. McPeak et al. [51] designed the continuous flow reactor
to make reagents contact the substrate surface in a closed space for growing ZnO
nanowires. Compared to the bath reactor, this continuous flow reactor gives higher
deposition rate and yield (30% ~ 50%). The result showed that higher flow rate can
accelerate the nanowire growth rate, and the growth rate decreased with the
downstream position from the inlet [52]. However, previous researches were all
focused on the small dimensional size (less than 3 cm).
In this research, the industrial pilot-scale production was firstly performed
on depositing CdS films on the 6-inch squared FTO glass substrate by our
75
continuous MASD process. In reality, the original design, parallel flow cell, has
parabolic flow distribution in the channel, and caused the final CdS films have
large variation in thickness. A modified version, deflected flow cell, was designed
to compensate the lateral flow variation across a high aspect ratio channel. Both
parallel and deflected flow cells were investigated to understand how the different
flow conditions affected the final CdS film thickness.
5.2 Experimental
Figure 5.1 illustrates the continuous MASD system. One reactant stream
was composed of 0.008 M CdCl2, 0.08 M NH4Cl, and 1.27 M NH4Cl aqueous
solution. The other reactant stream was composed of 0.08 M thiourea aqueous
solution. Both streams were pumped through a Tee mixer and into a stacked heat
exchanger with a total flow rate of 24 mL/min. After that, the solution was guided
into the flow cell section to allow the surface reaction of CdS film growth on the
FTO substrate (Pilkington TEC-15). The residence time in the heat exchanger can
be determined by stacking different numbers of copper plates.
Figure 5.2 shows the designs of parallel and deflected flow cells. Two main
features were designed for deflected flow cell. The first feature was the silicon
gasket changed from the polygon shape to the curvature shape. The second feature
was one more polycarbonate sheet added between the substrate and the top cover
plate. Several poles with screws were installed into the top cover plate, which can
push the beneath polycarbonate sheet to manipulate a deflected flow cell. This
modification altered the parabolic flow profile to an even flow profile, solving the
non-uniform thin film issue caused by flow pattern.
76
5.3 Computational Fluid Dynamics (CFD)
5.3.1
Stacked Heat Exchanger
The momentum and heat transport of reactant flow in stacked heat
exchanger are solved numerically by using finite elemental analysis CFD software
(COMSOL 4.2). Navier-Stokes equation (5-1) can be expressed as velocity at xand y- coordination (5-2a, b) for the velocity profiles of incompressible flow.
𝐷
2
=− 𝑝+
𝐷𝑡
𝜕𝑣
𝜕𝑣
( 𝜕𝑡 +
𝜕𝑥
𝜕𝑣
𝜕𝑣
( 𝜕𝑡 +
𝜕𝑥
+
+
+
(5-1)
𝜕𝑣
𝜕2 𝑣
𝜕
) = − 𝜕𝑥 + ( 𝜕𝑥 2 +
𝜕𝑦
𝜕𝑣
𝜕2 𝑣
𝜕
) = − 𝜕𝑦 + ( 𝜕𝑥 2 +
𝜕𝑦
𝜕2 𝑣
𝜕𝑦 2
𝜕2 𝑣
𝜕𝑦 2
)+
(5-2a)
)+
(5-2b)
For heat transfer, the analogue form of the Navier-Stokes equation can be
expressed as (5-3). Substituting in velocity vectors, it can be extended as the
complete form (5-4). The left and right hand side of the equal sign represent
respectively the convection and conduction in fluids.
𝑐
𝜕𝑇
𝜕𝑡
𝐷𝑇
𝐷𝑡
+
=𝑘
𝜕𝑇
𝑥 𝜕𝑥
+
2
𝑇
𝜕𝑇
𝑦 𝜕𝑦
(5-3)
𝑘
𝜕2 𝑇
𝜕2 𝑇
= 𝜌𝑐 (𝜕𝑥 2 + 𝜕𝑦 2 )
𝑝
(5-4)
After drawing the geometry, the boundary conditions of incompressible Newtonian
fluid module were added to simulate the velocity profile in the heat-exchanger.
Figure 5.3 (a), (c), and (e) show the velocity profiles for the setting inlet velocities
at 0.04 m/s, 0.008 m/s, and 0.001 m/s. The color mapping shows the dead volume
in the entrance square region that is formed at high inlet velocities. Then the
boundary conditions of heat conduction and convection were added and
incorporated with the simulated velocity profiles. The temperature profiles
corresponding to the various inlet velocities were solved and are shown in figure
77
5.3 (b), (d), and (f). With the increased fluid velocity, heat convection in the liquid
becomes more significant than the heat conduction from the walls. Therefore, an
obvious temperature gradient can be seen in the heat-exchanger channels. It
implies that longer residence time is needed to let the high velocity fluid approach
to the desire temperature (85 °C).
5.3.2
Flow Cell
Figure 5.4 shows the velocity contour map of the parallel and deflected
flow cell. The CFD analysis was performed by FLUENT 6.3 with an inlet velocity
of 0.089 m/s. Hires [97] proposed the Hagen-Poiseuille equation (5-5) on
designing the deflected flow cell. The original parallel flow cell (Figure 5.4 (a))
has three stream paths (L1 > L2 > L3) from the inlet to outlet and equivalent pressure
drop (ΔP1 = ΔP2 = ΔP3). The hydraulic diameter, DH (5-6), can be thought as an
imaginary rectangular long channel with a distance of width (W) and height (H).
The other constant, µ, is dynamic viscosity of fluid. The relationship of hydraulic
diameter and stream path is expressed in (5-7) due to the constant DH. Therefore,
the velocity (v) of these three stream paths shows v1 < v2 < v3 to fulfill the
equivalent pressure drop. Figure 5.4 (a) shows the parabolic velocity distribution in
parallel flow with the highest velocity (v3) in middle stream path than the other paths
(v1 and v2).
32𝜇𝐿
2
𝐻)
∆ = (𝐷
(5-5)
4𝑊𝐻
𝐷𝐻 = 2(𝑊+𝐻)
𝐿1
(𝐷𝐻1 )
2
𝐿2
(𝐷𝐻1 )
(5-6)
2
𝐿3
(𝐷𝐻1 )
2
(5-7)
78
To produce the uniform velocity profile (v1 = v2 = v3), the (5-7) is modified to (5-8),
and also the DH in three paths should follow the relationship of DH1 > DH2 > DH3 to
fulfill the equivalent pressure drop.
𝐿1
(𝐷𝐻1 )
2
=
𝐿2
(𝐷𝐻1 )
2
=
𝐿3
(𝐷𝐻1 )
2
(5-8)
According to (5-6), the height (H) is the only manipulated parameter in this flow
cell system. Hires [97] provided detail data of gap distance between the bottom of
cover plate and the substrate surface on different locations. Figure 5.4 (b) shows
the velocity profile in the deflected flow cell is improved to more uniform from the
original parabolic shape. This theoretical background gives a well support to
compare the film uniformity from the parallel and deflected flow cell.
79
5.4 Result and Discussion
There are totally 36 points (Figure 5.5) are measured by reflectance
spectroscopy to determine the thickness profiles of the large scale CdS films
deposited by using the parallel and deflected flow cell (Appendix D: Table D.1 and
D.2). Figure 5.6 shows the 2-D thickness line profiles (row A to row F) of the CdS
films made by the parallel (a) and deflected flow (b) cell system. The line profile is
plotted parallel to the flow direction. It is observed that the deflected flow cell did
improve the film uniformity and has a much lower variance than the film made by
the parallel flow cell. Figure 5.7 (a, b) show the 3-D thickness surface profiles
based on the line profiles in figure 5.6. The overall average thickness and standard
deviation are 39.85 nm ± 22.54 nm and 33.36 nm ± 4.51 nm for parallel flow cell
and deflected flow cell, respectively.
For the parallel flow cell, the center part (row D) has thicker film, and also
apparent thickness gradient along the flow direction. The thickness gradient is
caused by the non-uniform flow patterned in the flow cell. Blasius’s solution for
the laminar boundary layer on a flat plate is expressed as equation (5-9)
𝛿 = 5√𝑣
𝑥
∞
(5-9)
The boundary thickness ( δ) is inversely proportional to the square root of
external flow velocity (v∞). The high velocity flow in the centre part leads the
thinner boundary layer, and then shortens the diffusion length in perpendicular to
the flow direction. Therefore, the reactant in solution is easily transported to the
surface and forms the thicker films in the centre part than edge parts.
Based on this result, it is concluded that the modified deflected flow cell
can obtain uniform large area CdS films by controlling the same velocity of the
80
internal flow in each psition.
Figure 5.1 Schematic diagram of continuous MASD system.
Figure 5.2 Photograph (top) and cross-section diagram (bottom) of parallel flow
cell and deflected flow cell.
81
(a) Velocity profile of V = 0.04 m/s
(b) Temperature profile of V = 0.04 m/s
82
(c) Velocity profile of V = 0.008 m/s
(d) Temperature profile of V = 0.008 m/s
83
(e) Velocity profile of V = 0.001 m/s
(f) Temperature profile of V = 0.001 m/s
Figure 5.3 Velocity and temperature profiles in the silicon heat exchanger with
various inlet velocities.
84
Figure 5.4 Velocity contour maps of the parallel and deflected flow cell [97].
85
Figure 5.5 The map of 36 measurement points on a 6-inch substrate using MASD
process with a flow cell.
86
Figure 5.6 2-D CdS film thickness line profiles (row A to row F) along the flow
direction of reactants in the parallel (a) and deflected (b) flow cell.
87
Figure 5.7 3-D CdS film thickness surface profiles by continous MASD process
with the parallel (a) and deflected (b) flow cell.
88
Chapter 6 Analysis of CdS Thin Film by Reel-to-Reel
Microreactor-Assisted Solution Deposition (R2R-MASD)
6.1 Introduction
Solar cell is the most promising renewable energy technology in current
solar energy technology. The best research-cell efficiency was reported to be
43.5% in III-V group three-junction tendon cell, 20.3% in Cu(In,Ga)Se2 cell,
17.3% in CdTe/CdS cell, and 8.3% in organic cell. The development of solar cell
production technology plays an important role in PV product market. The first
generation crystalline silicon solar cell has achieved mature production technology,
and the commercial product still dominated the market share. Due to the high
manufacturing cost in crystalline silicon, new generation low-cost CdTe and
Cu(In,Ga)Se2 thin film solar cells have attracted more attention and have the
biggest market share for the non-silicon solar cells. Continuous reel-to-reel process
has a potential to further lower down the cost. Winkler et al. [98] reported a
reel-to-reel (R2R) process for fabricating CuInS2 solar cells on the flexible Cu-tape
in a non-vacuum ambient. The average efficiency of the small module reached
7.1%. Blankenburg et al. [99] fabricated P3HT/PCBM organic solar cells by using
a R2R coating machine shown in figure 6.1. Low viscous polymer solution was
dispensed on the moving foil substrate, and then the solvent was dried out through
hot air convection. Romeo et al. [100] reported the fabrication of CdTe/CdS thin
film cells by an in-line process. The whole process is composed of three deposition
sections, including the ITO layer, p-n junction layers, and the back contact layer
[100]. In this chapter, analysis of CdS thin films by R2R-MASD process was
reported.
89
6.2 Experimental
Figure 6.2 illustrates the schematic diagram of the R2R-MASD, which
includes the solution reaction region and the film deposition region. In solution
reaction region, the reagent of stream A was composed of 0.008 M CdCl2, 0.08 M
NH4Cl, and 1.27 M NH4OH aqueous solution. The stream B was composed of 0.08
M thiourea aqueous solution. Both streams were pumped through a Tee mixer and
into a stacked heat exchanger at a total flow rate of 45 mL/min. After that, the
solution was guided into the nozzle to be dispensed on the substrate.
In the film deposition region, the substrate was installed on a stage carried
by a conveyor. In step 1, the substrate stayed at the original point and waited for
pre-heating. In step 2, the substrate moved to the IR lamp pre-heating zone for
maintaining the surface temperature to 80 ~ 90 °C. Then the pre-heated substrate
moved to the spray zone in step 3. The reagent running through a heat exchanger
was dispensed on a rotating rod, which enhance the coverage of solution on the
substrate surface. After dispensing the proper amounts of reagents, the substrate
moved into the heating zone for growing the CdS film in step 4. To improve the
film quality, the substrate moved back and forward between the spraying and
heating zones several times. Finally, the substrate moved to the vacuum zone to
clean the excess residual solution on surface in step 5. All steps were controlled
automatically and tuned to the optimum conditions.
90
6.3 Film Characterization
6.3.1
XRD
Smaller-pieces of the sample (1ʺ x 1ʺ) were taken from the as-deposited
larger size CdS/FTO (6ʺ x 6ʺ) sample for various characterizations. The CdS films
were deposited by the R2R-MASD process with a deposition time of 6.5 minutes.
The GIXRD (Bruker D8 Discover, CuKα=1.54056Å) scan was collected with a
grazing incidence angle of 0.5°. The detector was moving to receive the signals.
Figure 6.3 shows the CdS films with four reflections at 26.26°, 30.47°, 43.84°, and
52.12°, which can be indexed as C(111), C(200), C(220), and C(311) of cubic CdS
(JCPDS-750581). The bare FTO substrate shows the reflection at 26.61°, 33.39°,
37.60°, and 51.55°, which can be indexed as T(110), T(101), T(200), T(211), and
T(220) of tetragonal SnO2 (JCPDS-411445). A comparison between the FTO
substrate with and without CdS films reveals that T(101), T(200), T(211) and
T(220) can’t be observed in CdS/FTO. The other finding is that the CdS films
contribute to an intense C(111) peak that block the T(110) peak from bare FTO
substrate. The GIXRD result clearly identifies the formation of CdS films.
6.3.2
TEM
Figure 6.4 shows the TEM (Philip CM-12) cross-sectional image of
CdS/FTO structure at low magnification (125,000X). A platinum layer was
deposited for the protection of samples during ion-milling, and the underneath
carbon layer function as a conducting material during FIB operation. The average
physical thickness of the CdS layers was determined to be around 80.66 nm and
the optical thickness was determined to be 102 nm obtained by the reflectance
spectroscopy.
91
6.3.3
Surface Property
Surface morphology and the surface roughness were measured by AFM
(Innova Scanning Probe Microscope, Bruker). The sample was cut to a small size
(1ʺ x 0.5ʺ) in order to be loaded onto the sample stage. Figure 6.5 (a), (b) show
AFM tapping mode images of the bare FTO substrate and the CdS films deposited
at 6.5 min respectively. The RMS roughness of bare FTO substrate is 8.15 nm,
which is higher than the reported value of 6.7 nm from an ITO substrate [101]. The
RMS roughness of the CdS films on FTO is 11.35 nm, which is much higher than
the reported value of 7 nm for CdS film on ITO with equivalent deposition time
[101]. FTO surface used for PV application normally has higher roughness than
typical ITO surface. Therefore, it is not surprise to see a higher surface roughness
from our CdS/FTO samples. Kim et al. [69] also reported the RMS roughness
around 7 ~ 15 nm depending on the thiourea molar concentration from 0.1 M to 0.4
M. The low RMS value was obtained from a batch with higher thiourea
concentration. Our RMS value (11.35 nm) is reasonable in light of Kim’s reported
range (7 ~ 15 nm). Mazon-Montijo et al. [60] reported the RMS value of 3.6 nm in
CdS films on ITO and 2.7 nm for the bare ITO substrate. The deposited film layer
has higher roughness compared to the bare substrate because of particle formation
during the solution deposition process.
XPS was performed to obtain the chemical binding information of the
CdS film. Figure 6.6 shows the presence of oxygen (O), cadmium (Cd), carbon (C),
and sulfur (S) from CdS layer. The data obtained were corrected by taking the
specimen charging and referring to C 1s at 284.9 eV (figure 6.6 (c)), which was
originated in the atmospheric contamination. The oxygen (figure 6.6 (a)) on the
92
surface at 532.4 eV appears to be due to hydroxide (Cd(OH)2) relative surface
chemisorbed oxygen. For the as-received condition, the binding energies of Cd
3d3/2 and Cd 3d5/2 shows at 412 eV and 405.3 eV respectively. These spectrums of
Cd and S after etching can be fitted by Gaussian function (red dotted lines). For Cd
3d energy level, the binding energy of 3d3/2 and 3d5/2 orbital are 411.7eV and 405
eV respectively, with the splitting energy of 6.8 eV. For S 2p energy level, the
binding energies of S 2p3/2 and S 2p1/2 orbitals are 161.1 and 162.3 eV respectively,
with the splitting energy of 1.2 eV. The obtained result shows excellent agreement
with previous researches [70, 71].
6.3.4
Optical Property
The optical property was characterized by UV-vis absorption spectroscopy
(Ocean Optics USB2000, Halogen lamp as light source). Figure 6.7 shows the plot
of (αhν)2 versus hν (band gap energy) of the CdS film with a thickness of 80.66 nm.
The optical band gap is determined by extrapolating the curve across the x-axis at
2.38 eV, which is quite consistent with previous reported results.
6.4 Result and Discussion
The growth kinetics was studied by analyzing the film thickness. Figure 6.8
shows a 6-inch sample with 36 measurement points, which were labeled by
column 1 to column 6 and row A to row F. The film thickness was determined by
the spectroscopic reflectance, which was elucidated in Chapter 3. Figure 6.9 shows
the 2-D thickness line profiles from row A to row F at the deposition time of 2.5
min, 6.3 min, and 9 min. Data points are showed in table D.3 ~ D.5 in Appendix D.
The flat line profiles show the lower standard deviations (σ) at 2.5 min deposition
93
time. Longer deposition time change the line profiles from flat to a curvature shape,
which represents uneven film obtained. Figure 6.10 (a) ~ (c) show the 3-D
thickness surface profiles based on the above line profiles. The standard deviation
of all 36 points is 3.94 nm, 5.18 nm, and 31.34 nm for 2.5 min, 6.3 min, and 9 min
respectively. It is observed that the center region has significantly thicker films
than the films around the edge for samples deposited at longer deposition time. The
non-uniform film growth rates most likely come from the non-uniform distribution
of the dispensed reagents on the substrate.
The average thickness of all 36 points is 72.58 nm, 94.78 nm, and 232.28
nm for 2.5 min, 6.3 min, and 9 min respectively. These three points are plotted in
figure 6.11 (d) and compared with the results obtained from the lab-scale
continuous MASD (c), batch processes at 60 °C (a) and 70 °C (b). The equation
(3-30) derived from the revised model (3-29) was applied to analyze the
experimetal results.
𝑑
𝑑𝑡
=
𝑚𝑤 𝑘0 [𝐶𝑑]0 [𝑂𝐻 − ]2 [𝑆𝐶(𝑁𝐻2 )2 ]0
(𝑒𝑥𝑝(𝐴𝜏)
𝜌[𝑁𝐻3 ]2
− 𝑒𝑥𝑝(𝐵𝜏))
ℎ = 1.11 × 10−7 𝑘0 𝑡(𝑒𝑥𝑝(−0.003682𝜏) − 𝑒𝑥𝑝(−0.3106𝜏))
(3-29)
(3-30)
For continuous R2R-MASD process, the concentrations of each species are the
same as ones previously used in Chapter 3. They are [Cd]0 =0.004*10-3 mole/cm3,
[SC(NH2)2] = 0.04*10-3 mole/cm3, [OH-] = 0.001*10-3 mole/cm3, and [NH3] =
0.655*10-3 mole/cm3. The molecular weight (MW) of CdS is 144.6 g/mole, and
density (ρ) is 4.82 g/cm3. The fitting result shows the k0 = 3.567*106 cm4mole-1s-1,
and the R-square is 0.777. This growth rate is higher than the lab-scale continous
flow MASD process.
94
Figure 6.1 Working principle of the reel-to-reel coating machine [99].
Figure 6.2 Schematic diagram of the reel-to-reel continuous MASD.
95
Figure 6.3 GIXRD of the reel-to-reel continuous MASD deposited CdS films on
FTO substrate.
96
Figure 6.4 TEM cross-sectional images of the reel-to-reel continuous MASD
deposited CdS films on FTO substrate. (Magnification: 125,000X)
97
Figure 6.5 AFM images of (a) bare FTO substrate (RMS = 8.15 nm), (b)
reel-to-reel continuous MASD deposited CdS film (RMS = 11.35 nm) at 6.5 min
deposition time. (Insert: 3D morphological images).
98
Figure 6.6 XPS spectra (dashed line: as received, solid line: after etching, red
dotted line: Gaussian fitting) of the CdS films deposited by R2R-MASD (a) O 1s,
(b) Cd 3d, (c) C 1s, (d) S 2p
99
Figure 6.7 Plot of (αhν)2 versus hν showing the band gap energy of the
reel-to-reel continuous MASD deposited CdS films.
100
Figure 6.8 The location map of 36 measurement points on a 6-inch substrate
using R2R-MASD process.
101
Figure 6.9 2-D thickness line profiles of CdS films deposited by continuous
R2R-MASD during (a) 2.5 min, (b) 6.3 min, and (c) 9 min.
102
Figure 6.10 3-D thickness surface proflies of CdS films deposited by continuous
R2R-MASD during (a) 2.5 min, (b) 6.3 min, and (c) 9 min.
103
Figure 6.11 Growth kinetics of CdS thin film deposited by batch process at (a)
70°C, (b) 60°C, (c) continuous MASD process, and (d) R2R-MASD process. The
series data of (a) and (b) are from Dona et al. [31]. Solid lines are fitting curves of
Kostoglue’s model (3-27) and the modified model (3-30).
104
Chapter 7 Conclusion and Future Work
The continuous microreactor-assisted solution deposition (MASD) process
was used for the deposition of CdS thin films on fluorine-doped tin oxide (FTO)
glass. The MASD system, including a T-junction micromixer and a microchannel
heat exchanger, is capable of isolating the homogeneous particle precipitation from
the heterogeneous surface reaction. The results show a dense nanocrystallite CdS
thin films with a preferred orientation at (111) plane.
The film growth kinetics was studied and a growth model that accounts for
the residence time in the microchannel using empirical factor (η) obtained from
previous reported experimental data. Applying this factor in the proposed modified
growth model gives a surface reaction rate of 1.61*106 cm4mole-1s-1, which is
considerable higher than the surface reaction rates obtained from the batch CBD
process. With the feature of separating homogeneous and heterogeneous surface
reaction, the MASD process provides the capability to tailor the surface film
growth rate and avoid the saturation growth regime in the batch process
An in-situ spectroscopy technique was used to measure the UV-Vis
absorption spectra of CdS nanoparticles formed within the continuous flow
microreactor. The data shows the formation of CdS nanoparticles range from 1.13
nm to 1.26 nm using a residence time from 0.26 s to 3.96 s. Barrier-controlled
coalescence mechanism seem to be a reasonable model to explain the experimental
Uv-Vis data obtained from the continuous flow microreactor, with a rate constant k’
value of 2.872 s-1. Using CFD, low skewness value of the RTD curve at high flow
rate (short τ) suggests good radical mixing at high flow rate is responsible for the
formation of smaller CdS nanoparticles with a narrower size distribution. The
105
combination of CdS nanoparticle solution with MASD process resulted in the
hindrance of CdS thin film deposition. It is hypothesized that the pre-existing
sulfide (S2-) ions and CdS nanoparticles changes the chemical species equilibrium
of thiourea hydrolysis reaction. Consequently, the lack of thiourea slows down the
heterogeneous surface reaction.
A flow cell and a reel-to-reel (R2R)-MASD systems were setup to test the
scalability via the deposition of CdS films on the FTO glass (6ʺ x 6ʺ) substrate.
The film deposition kinetics was found to be sensitive to the flow conditions
within the heat exchanger and the substrate flow cell. The growth kinetics of the
CdS films deposited by R2R-MASD process was investigated. In comparison with
the continuous MASD process, the growth rate in R2R-MASD is higher, however
more difficult to obtain a linear relationship with the deposition time.
Two major areas for future work are listed here:

Investigate the homogeneous nanoparticle formation, particle sticking, and
surface reaction on the particle surface and substrate.

Fabricate CdTe and CIGS thin film solar cells using MASD process and
optimize the process for higher material utilization and cell performance.
106
Chapter 8 Bibliography
[1] P.K. Nair, J. Campos, M.T.S. Nair, Opto-Electronic Characteristics of
Chemically Deposited Cadmium-Sulfide Thin-Films, Semicond Sci Tech, 3 (1988)
134-145.
[2] G. Arreola-Jardón, L.A. González, L.A. García-Cerda, B. Gnade, M.A.
Quevedo-López, R. Ramírez-Bon, Ammonia-free chemically deposited CdS films
as active layers in thin film transistors, Thin Solid Films, 519 (2010) 517-520.
[3] Y.J. Chang, P.H. Mugdur, S.Y. Han, A.A. Morrone, S.O. Ryu, T.J. Lee, C.H.
Chang, Nanocrystalline CdS MISFETs fabricated by a novel continuous flow
microreactor, Electrochem Solid St, 9 (2006) G174-G177.
[4] F.Y. Gan, I. Shih, Preparation of thin-film transistors with chemical bath
deposited CdSe and US thin films, Ieee T Electron Dev, 49 (2002) 15-18.
[5] R.W. Birkmire, B.E. McCandless, CdTe thin film technology: Leading thin film
PV into the future, Curr Opin Solid St M, 14 (2010) 139-142.
[6] H.S. Kwok, J.P. Zheng, S. Witanachchi, P. Mattocks, L. Shi, Q.Y. Ying, X.W.
Wang, D.T. Shaw, Growth of highly oriented CdS thin films by laser-evaporation
deposition, Appl Phys Lett, 52 (1988) 1095.
[7] B. Ullrich, H. Sakai, Y. Segawa, Optoelectronic properties of thin film CdS
formed by ultraviolet and infrared pulsed-laser deposition, Thin Solid Films, 385
(2001) 220-224.
[8] T. Hayashi, T. Nishikura, T. Suzuki, Y. Ema, Formation and Properties of
in-Doped High-Conductivity Cds Film, J Appl Phys, 64 (1988) 3542-3550.
[9] F. Iacomi, M. Purica, E. Budianu, P. Prepelita, D. Macovei, Structural studies
on some doped CdS thin films deposited by thermal evaporation, Thin Solid Films,
515 (2007) 6080-6084.
107
[10] I. Martil, G. González-Díaz, F. Sanchez-Quesada, M. Rodriguez-Vidal,
Deposition dependence of r.f.-sputtered CdS films, Thin Solid Films, 90 (1982)
253-257.
[11] I. Martil, G. Gonzalezdiaz, F. Sanchezquesada, Electrical Transport-Properties
of Polycrystalline Rf Sputtered Cds Thin-Films, Journal of Vacuum Science &
Technology a-Vacuum Surfaces and Films, 2 (1984) 1491-1494.
[12] M. Tomakin, M. Altunbas, E. Bacaksiz, I. Polat, Preparation and
characterization of new window material CdS thin films at low substrate
temperature (< 300 K) with vacuum deposition, Mat Sci Semicon Proc, 14 (2011)
120-127.
[13] R. Castro-Rodriguez, J. Mendez-Gamboa, I. Perez-Quintana, R.
Medina-Ezquivel, CdS thin films growth by fast evaporation with substrate
rotation, Appl Surf Sci, 257 (2011) 9480-9484.
[14] J.L. Boone, S.A. Howard, D.D. Martin, G. Cantwell, Mocvd of Cds onto
(111)Silicon Substrates, Thin Solid Films, 176 (1989) 143-150.
[15] A.K. Berry, P.M. Amirtharaj, J.T. Du, J.L. Boone, D.D. Martin,
Photoluminescence and Raman Studies of Cds Films Grown by Metal
Organic-Chemical Vapor-Deposition on Si(111) Substrates, Thin Solid Films, 219
(1992) 153-156.
[16] H. Uda, H. Yonezawa, Y. Ohtsubo, M. Kosaka, H. Sonomura, Thin CdS films
prepared by metalorganic chemical vapor deposition, Sol Energ Mat Sol C, 75
(2003) 219-226.
[17] O.B. Ajayi, O.K. Osuntola, I.A. Ojo, C. Jeynes, Preparation and
Characterization of Mocvd Thin-Films of Cadmium-Sulfide, Thin Solid Films, 248
(1994) 57-62.
[18] N.I. Fainer, M.L. Kosinova, Y.M. Rumyantsev, E.G. Salman, F.A. Kuznetsov,
Growth of PbS and CdS thin films by low-pressure chemical vapour deposition
using dithiocarbamates, Thin Solid Films, 280 (1996) 16-19.
108
[19] D. Barreca, A. Gasparotto, C. Maragno, E. Tondello, CVD of nanosized ZnS
and CdS thin films from single-source precursors, J Electrochem Soc, 151 (2004)
G428-G435.
[20] G.A. Kitaev, Uritskarya, A. A.and Morkrushin, S. G. , Conditions for the
chemical deposition of thin films of cadmiumn sulfide on a solid surface, Russ. J.
Phys. Chem., 39 (1965) 1101.
[21] G.A. Kitaev, Makurin, Y. N. and Dvoinin, V. I. , Quantum-chemical
calculation of the electronicstructures of thiourea and its derivatives and
comparison of the reactivity indices, Russ. J. Phys. Chem., 50 (1976).
[22] N.D. Betenekov, Medvedev, V. P., Zhukovskaya, A. S.and Kitaev, G. A.,
Radiochemical investigation of chalcogenide films. II. Kinetics of the growth of
cadmium sulfide films on the surface of glass, Sov Radiochem+, 20 (1979).
[23] P.K. Nair, M.T.S. Nair, J. Campos, L.E. Sansores, A Critical Discussion of the
Very High Photoconductivity in Chemically Deposited Cadmium-Sulfide
Thin-Films - Implications for Solar-Cell Technology, Sol Cells, 22 (1987) 211-227.
[24] M.T.S. Nair, P.K. Nair, J. Campos, Effect of Bath Temperature on the
Optoelectronic Characteristics of Chemically Deposited Cds Thin-Films, Thin
Solid Films, 161 (1988) 21-34.
[25] M.T.S. Nair, P.K. Nair, R.A. Zingaro, E.A. Meyers, Conversion of Chemically
Deposited Photosensitive Cds Thin-Films to N-Type by Air Annealing and
Ion-Exchange Reaction, J Appl Phys, 75 (1994) 1557-1564.
[26] A.A.C. Readigos, V.M. Garcia, O. Gomezdaza, J. Campos, M.T.S. Nair, P.K.
Nair, Substrate spacing and thin-film yield in chemical bath deposition of
semiconductor thin films, Semicond Sci Tech, 15 (2000) 1022-1029.
[27] P.K. Nair, V.M. Garcia, O. Gomez-Daza, M.T.S. Nair, High thin-film yield
achieved at small substrate separation in chemical bath deposition of
semiconductor thin films, Semicond Sci Tech, 16 (2001) 855-863.
109
[28] J.M. Dona, J. Herrero, Chemical Bath Deposition of Cds Thin-Films Electrochemical Insitu Kinetic-Studies, J Electrochem Soc, 139 (1992) 2810-2814.
[29] D. Lincot, R.O. Borges, Chemical Bath Deposition of Cadmium-Sulfide
Thin-Films - Insitu Growth and Structural Studies by Combined Quartz Crystal
Microbalance and Electrochemical Impedance Techniques, J Electrochem Soc, 139
(1992) 1880-1889.
[30] R. Ortegaborges, D. Lincot, Mechanism of Chemical Bath Deposition of
Cadmium-Sulfide Thin-Films in the Ammonia-Thiourea System - in-Situ
Kinetic-Study and Modelization, J Electrochem Soc, 140 (1993) 3464-3473.
[31] J.M. Dona, J. Herrero, Chemical bath deposition of CdS thin films: An
approach to the chemical mechanism through Study of the film microstructure, J
Electrochem Soc, 144 (1997) 4081-4091.
[32] I.O. Oladeji, L. Chow, Optimization of chemical bath deposited cadmium
sulfide thin films, J Electrochem Soc, 144 (1997) 2342-2346.
[33] M. Kostoglou, N. Andritsos, A.J. Karabelas, Flow of supersaturated solutions
in pipes. Modeling bulk precipitation and scale formation, Chem Eng Commun,
133 (1995) 107-131.
[34] M. Kostoglou, A.J. Karabelas, Comprehensive modeling of precipitation and
fouling in turbulent pipe flow, Ind Eng Chem Res, 37 (1998) 1536-1550.
[35] M. Kostoglou, N. Andritsos, A.J. Karabelas, Modeling thin film CdS
development in a chemical bath deposition process, Ind Eng Chem Res, 39 (2000)
3272-3283.
[36] M. Kostoglou, N. Andritsos, A.J. Karabelas, Progress towards modelling the
CdS chemical bath deposition process, Thin Solid Films, 387 (2001) 115-117.
[37] M. Kostoglou, N. Andritsos, A.J. Karabelas, Incipient CdS thin film formation,
J Colloid Interf Sci, 263 (2003) 177-189.
110
[38] P.H. Mugdur, Y.J. Chang, S.Y. Han, Y.W. Su, A.A. Morrone, S.O. Ryu, T.J.
Lee, C.H. Chang, A comparison of chemical bath deposition of CdS from a batch
reactor and a continuous-flow microreactor, J Electrochem Soc, 154 (2007)
D482-D488.
[39] Y.J. Chang, Y.W. Su, D.H. Lee, S.O. Ryu, C.H. Chang, Investigate the
Reacting Flux of Chemical Bath Deposition by a Continuous Flow Microreactor,
Electrochem Solid St, 12 (2009) H244-H247.
[40] M.S. Park, S.Y. Han, E.J. Bae, T.J. Lee, C.H. Chang, S.O. Ryu, Synthesis and
characterization of polycrystalline CuInS(2) thin films for solar cell devices at low
temperature processing conditions, Curr Appl Phys, 10 (2010) S379-S382.
[41] C.R. Kim, S.Y. Han, C.H. Chang, T.J. Lee, S.O. Ryu, Synthesis and
characterization of CuInSe2 thin films for photovoltaic cells by a solution-based
deposition method, Curr Appl Phys, 10 (2010) S383-S386.
[42] C.R. Kim, S.Y. Han, C.H. Chang, T.J. Lee, S.O. Ryu, A Study on Copper
Selenide Thin Films for Photovoltaics by a Continuous Flow Microreactor, Mol
Cryst Liq Cryst, 532 (2010) 455-463.
[43] S.Y. Han, Y.J. Chang, D.H. Lee, S.O. Ryu, T.J. Lee, C.H. Chang, Chemical
nanoparticle deposition of transparent ZnO thin films, Electrochem Solid St, 10
(2007) K1-K5.
[44] T.J. Lee, Y.J. Lee, N.K. Park, G.B. Han, S.O. Ryu, C.H. Chang, The
preparation and desulfurization of nano-size ZnO by a matrix-assisted method for
the removal of low concentration of sulfur compounds, Curr Appl Phys, 8 (2008)
746-751.
[45] J.Y. Jung, N.K. Park, S.Y. Han, G.B. Han, T.J. Lee, S.O. Ryu, C.H. Chang,
The growth of the flower-like ZnO structure using a continuous flow microreactor,
Curr Appl Phys, 8 (2008) 720-724.
111
[46] D.H. Lee, J.Y. Jung, E.J. Bae, T.J. Lee, S.O. Ryu, C.H. Chang, Highly uniform
ZnS thin film through the continuous flow reaction process, J Korean Phys Soc, 53
(2008) 102-105.
[47] S. Liu, C.H. Chang, High Rate Convergent Synthesis and Deposition of
Polyamide Dendrimers using a Continuous-Flow Microreactor, Chem Eng Technol,
30 (2007) 334-340.
[48] S. Liu, C.-h. Chang, B.K. Paul, V.T. Remcho, Convergent synthesis of
polyamide dendrimer using a continuous flow microreactor, Chem Eng J, 135
(2008) S333-S337.
[49] H.D. Jin, A. Garrison, T. Tseng, B.K. Paul, C.H. Chang, High-rate synthesis of
phosphine-stabilized undecagold nanoclusters using a multilayered micromixer,
Nanotechnology, 21 (2010) -.
[50] H.D. Jin, C.-H. Chang, Continuous synthesis of SnTe nanorods, J Mater Chem,
21 (2011) 12218-12220.
[51] K.M. McPeak, J.B. Baxter, Microreactor for High-Yield Chemical Bath
Deposition of Semiconductor Nanowires: ZnO Nanowire Case Study, Ind Eng
Chem Res, 48 (2009) 5954-5961.
[52] K.M. McPeak, J.B. Baxter, ZnO Nanowires Grown by Chemical Bath
Deposition in a Continuous Flow Microreactor, Cryst Growth Des, 9 (2009)
4538-4545.
[53] C. Ferekides, J. Britt, Cdte Solar-Cells with Efficiencies over 15-Percent, Sol
Energ Mat Sol C, 35 (1994) 255-262.
[54] T.L. Chu, S.S. Chu, C. Ferekides, C.Q. Wu, J. Britt, C. Wang, 13.4-Percent
Efficient Thin-Film Cds/Cdte Solar-Cells, J Appl Phys, 70 (1991) 7608-7612.
[55] J. Britt, C. Ferekides, Thin-Film Cds/Cdte Solar-Cell with 15.8-Percent
Efficiency, Appl Phys Lett, 62 (1993) 2851-2852.
112
[56] M.E. Calixto, M. Tufino-Velazquez, G. Contreras-Puente, O. Vigil-Galan, M.
Jimenez-Escamilla, R. Mendoza-Perez, J. Sastre-Hernandez, A. Morales-Acevedo,
Study of chemical bath deposited CdS bi-layers and their performance in
CdS/CdTe solar cell applications, Thin Solid Films, 516 (2008) 7004-7007.
[57] S. Kohli, V. Manivannan, J.N. Hilfiker, P.R. McCurdy, R.A. Enzenroth, K.L.
Barth, W.P. Smith, R. Luebs, W.S. Sampath, Effect of Chemical Treatment on the
Optical Properties of a Cadmium Telluride Photovoltaic Device Investigated by
Spectroscopic Ellipsometry, J Sol Energ-T Asme, 131 (2009).
[58] J.F. Han, C. Liao, T. Jiang, G.H. Fu, V. Krishnakumar, C. Spanheimer, G.
Haindl, K. Zhao, A. Klein, W. Jaegermann, Annealing effects on the chemical
deposited CdS films and the electrical properties of CdS/CdTe solar cells, Mater
Res Bull, 46 (2011) 194-198.
[59] J.F. Han, C. Spanheimer, G. Haindl, G.H. Fu, V. Krishnakumar, J. Schaffner,
C.J. Fan, K. Zhao, A. Klein, W. Jaegermann, Optimized chemical bath deposited
CdS layers for the improvement of CdTe solar cells, Sol Energ Mat Sol C, 95
(2011) 816-820.
[60] D.A. Mazon-Montijo, M. Sotelo-Lerma, L. Rodriguez-Fernandez, L. Huerta,
AFM, XPS and RBS studies of the growth process of CdS thin films on ITO/glass
substrates deposited using an ammonia-free chemical process, Appl Surf Sci, 256
(2010) 4280-4287.
[61] T.L. Chu, S.S. Chu, N. Schultz, C. Wang, C.Q. Wu, Solution-Grown
Cadmium-Sulfide Films for Photovoltaic Devices, J Electrochem Soc, 139 (1992)
2443-2446.
[62] S. Ramprasad, Y.-W. Su, C.-H. Chang, B.K. Paul, D.R. Palo, Cadmium sulfide
thin film deposition: A parametric study using microreactor-assisted chemical
solution deposition, Sol Energ Mat Sol C, 96 (2012) 77-85.
[63] S.J. Ikhmayies, R.N. Ahmad-Bitar, Characterization of the SnO2:F/CdS:In
structures prepared by the spray pyrolysis technique, Sol Energ Mat Sol C, 94
(2010) 878-883.
113
[64] M. Kokotov, Y. Feldman, A. Avishai, M. DeGuire, G. Hodes, Chemical bath
deposition of CdS highly-textured, columnar films, Thin Solid Films, 519 (2011)
6388-6393.
[65] M. Kim, B.K. Min, C.D. Kim, S. Lee, H.T. Kim, S.K. Jung, S. Sohn, Study of
the physical property of the cadmium sulfide thin film depending on the process
condition, Curr Appl Phys, 10 (2010) S455-S458.
[66] C. Li, X.G. Yang, B.J. Yang, Y. Yan, Y.T. Qian, Growth of microtubular
complexes as precursors to synthesize nanocrystalline ZnS and CdS, J Cryst
Growth, 291 (2006) 45-51.
[67] Z.R. Khan, M. Zulfequar, M.S. Khan, Optical and structural properties of
thermally evaporated cadmium sulphide thin films on silicon (1 0 0) wafers,
Materials Science and Engineering: B, 174 (2010) 145-149.
[68] X.L. Tong, D.S. Jiang, W.B. Hu, Z.M. Liu, M.Z. Luo, The comparison
between CdS thin films grown on Si(111) substrate and quartz substrate by
femtosecond pulsed laser deposition, Appl Phys a-Mater, 84 (2006) 143-148.
[69] M.J. Kim, H.T. Kim, J.K. Kang, D.H. Kim, D.H. Lee, S.H. Lee, S.H. Sohn,
Effects of the Surface Roughness on Optical Properties of CdS Thin Films, Mol
Cryst Liq Cryst, 532 (2010) 437-444.
[70] H. Khallaf, G. Chai, O. Lupan, L. Chow, S. Park, A. Schulte, Characterization
of gallium-doped CdS thin films grown by chemical bath deposition, Appl Surf Sci,
255 (2009) 4129-4134.
[71] L. Wan, Z.Z. Bai, Z.R. Hou, D.L. Wang, H. Sun, L.M. Xiong, Effect of CdCl2
annealing treatment on thin CdS films prepared by chemical bath deposition, Thin
Solid Films, 518 (2010) 6858-6865.
[72] M.a.K. Kostoglou, A. J., Evaluation of numerical methods for simulating an
evolving particle size distribution in growth processes Chem Eng Commun, 136
(1995).
114
[73] S. Kohli, V. Manivannan, J.N. Hilfiker, P.R. McCurdy, R.A. Enzenroth, K.L.
Barth, W.P. Smith, R. Luebs, W.S. Sampath, Effect of Chemical Treatment on the
Optical Properties of a Cadmium Telluride Photovoltaic Device Investigated by
Spectroscopic Ellipsometry, J Sol Energ-T Asme, 131 (2009) -.
[74] R. Castro-Rodriguez, A.I. Oliva, V. Sosa, F. Caballero-Briones, J.L. Pena,
Effect of indium tin oxide substrate roughness on the morphology, structural and
optical properties of CdS thin films, Appl Surf Sci, 161 (2000) 340-346.
[75] G. Marcotrigiano, G. Peyronel, R. Battistuzzi, Kinetics of the desulphuration
of 35S-labelled thiourea in sodium hydroxide studied by chromatographic methods,
Journal of the Chemical Society, Perkin Transactions 2, (1972) 1539-1541.
[76] Y.J. Chang, Investigation of Low Temperature Solution-based Deposition
Process for Flexible Electronics, in: Chem Eng-New York, Oregon State
University, Corvallis, 2007.
[77] A.P. Alivisatos, Semiconductor clusters, nanocrystals, and quantum dots,
Science, 271 (1996) 933-937.
[78] T. Trindade, P. O'Brien, N.L. Pickett, Nanocrystalline semiconductors:
Synthesis, properties, and perspectives, Chem Mater, 13 (2001) 3843-3858.
[79] Y. Nosaka, Finite Depth Spherical Well Model for Excited-States of
Ultrasmall Semiconductor Particles - an Application, J Phys Chem-Us, 95 (1991)
5054-5058.
[80] S. Krishnadasan, J. Tovilla, R. Vilar, A.J. deMello, J.C. deMello, On-line
analysis of CdSe nanoparticle formation in a continuous flow chip-based
microreactor, J Mater Chem, 14 (2004) 2655-2660.
[81] J.B. Edel, R. Fortt, J.C. deMello, A.J. deMello, Microfluidic routes to the
controlled production of nanoparticles, Chem Commun, (2002) 1136-1137.
115
[82] F. Shayeganfar, L. Javidpour, N. Taghavinia, M.R.R. Tabar, M. Sahimi, F.
Bagheri-Tar, Controlled nucleation and growth of CdS nanoparticles by turbulent
dispersion, Phys Rev E, 81 (2010) 026304.
[83] M. Tiemann, Ö. Weiß, J. Hartikainen, F. Marlow, M. Lindén, Early Stages of
ZnS Nanoparticle Growth Studied by In-Situ Stopped-Flow UV Absorption
Spectroscopy, Chemphyschem, 6 (2005) 2113-2119.
[84] M. Tiemann, F. Marlow, F. Brieler, M. Linden, Early stages of ZnS growth
studied by stopped-flow UV absorption spectroscopy: Effects of educt
concentrations on the nanoparticle formation, J Phys Chem B, 110 (2006)
23142-23147.
[85] L.H. Hung, K.M. Choi, W.Y. Tseng, Y.C. Tan, K.J. Shea, A.P. Lee, Alternating
droplet generation and controlled dynamic droplet fusion in microfluidic device for
CdS nanoparticle synthesis, Lab Chip, 6 (2006) 174-178.
[86] K.M. Mullaugh, G.W. Luther, Spectroscopic determination of the size of
cadmium sulfide nanoparticles formed under environmentally relevant conditions,
J Environ Monitor, 12 (2010) 890-897.
[87] K.M. Mullaugh, G.W. Luther, Growth kinetics and long-term stability of CdS
nanoparticles in aqueous solution under ambient conditions, J Nanopart Res, 13
(2011) 393-404.
[88] C. Ribeiro, E.J.H. Lee, E. Longo, E.R. Leite, A kinetic model to describe
nanocrystal growth by the oriented attachment mechanism, Chemphyschem, 6
(2005) 690-696.
[89] F. Huang, H.Z. Zhang, J.F. Banfield, Two-stage crystal-growth kinetics
observed during hydrothermal coarsening of nanocrystalline ZnS, Nano Lett, 3
(2003) 373-378.
[90] M. Tiemann, F. Marlow, J. Hartikainen, O. Weiss, M. Linden, Ripening
Effects in ZnS Nanoparticle Growth, The Journal of Physical Chemistry C, 112
(2008) 1463-1467.
116
[91] J.T. Adeosun, A. Lawal, Mass transfer enhancement in microchannel reactors
by reorientation of fluid interfaces and stretching, Sensor Actuat B-Chem, 110
(2005) 101-111.
[92] J.T. Adeosun, A. Lawal, Numerical and experimental mixing studies in a
MEMS-based multilaminated/elongational flow micromixer, Sensor Actuat
B-Chem, 139 (2009) 637-647.
[93] J.T. Adeosun, A. Lawal, Numerical and experimental studies of mixing
characteristics in a T-junction microchannel using residence-time distribution,
Chem Eng Sci, 64 (2009) 2422-2432.
[94] J.T. Adeosun, A. Lawal, Residence-time distribution as a measure of mixing in
T-junction and multilaminated/elongational flow micromixers, Chem Eng Sci, 65
(2010) 1865-1874.
[95] D. Boskovic, S. Loebbecke, Modelling of the residence time distribution in
micromixers, Chem Eng J, 135 (2008) S138-S146.
[96] S.H. Lee, H.J. Lee, D. Oh, S.W. Lee, H. Goto, R. Buckmaster, T. Yasukawa, T.
Matsue, S.K. Hong, H. Ko, M.W. Cho, T.F. Yao, Control of the ZnO nanowires
nucleation site using microfluictic channels, J Phys Chem B, 110 (2006)
3856-3859.
[97] C.L. Hires, Uniform Residence Time in Microreactor-Assisted Solution
Deposition of CdS Thin-Films for CIGS Photovoltaic Cells, in: Industrial
Engineering, Oregon State University, Corvallis, 2011.
[98] M. Winkler, J. Griesche, I. Konovalov, J. Penndorf, J. Wienke, O. Tober,
CISCuT - solar cells and modules on the basis of CuInS2 on Cu-tape, Sol Energy,
77 (2004) 705-716.
[99] L. Blankenburg, K. Schultheis, H. Schache, S. Sensfuss, M. Schrodner,
Reel-to-reel wet coating as an efficient up-scaling technique for the production of
bulk-heterojunction polymer solar cells, Sol Energ Mat Sol C, 93 (2009) 476-483.
117
[100] N. Romeo, A. Bosio, A. Romeo, An innovative process suitable to produce
high-efficiency CdTe/CdS thin-film modules, Sol Energ Mat Sol C, 94 (2010) 2-7.
[101] A.I. Oliva, R. Castro-Rodriguez, O. Ceh, P. Bartolo-Perez, F.
Caballero-Briones, V. Sosa, First stages of growth of CdS films on different
substrates, Appl Surf Sci, 148 (1999) 42-49.
118
Appendix A: Harmonic Oscillator Approximation
According to the relationship (A-1) of dielectric constants and refractive index, the
real (ε1) and imaginary part (ε2) of dielectric constant are converted to n and k,
expressed in (A-2a) and (A-2b). Due to the dielectric constant being wavelength
dependent, the harmonic oscillator approximated equation (2-3) can be used to fit
the dielectric constant by adjusting all parameters.
ε(E) = ε1 + iε2 = (n + ik)2
m
𝜀(𝐸) = 1 +
2

j 0
1
𝐸+𝐸𝑗 +𝑖Γ𝑗
−
1
𝐸−𝐸𝑗 +𝑖Γ𝑗
)
(2-3)
2
√ε1 +ε2 +ε1
n=√
2
√ε21 +ε22 −ε1
k=√
𝐴𝑗 𝑒 −𝑖𝜃𝑗 (
(A-1)
2
(A-2a)
(A-2b)
Here, the real part of the refractive index (n) indicates the phase speed, while the
imaginary part (k) indicates the amount of absorption loss when the
electromagnetic wave propagates through the material. When the incident angle
and refracted angle are zero, the reflection coefficient also called reflectance is
defined in n, k and also expressed in ε1, ε2 by substituting (A-2a) and (A-2b) into
the Fresnel equation. The final calculated reflectance equation (A-3) is used for
fitting the measured reflectance spectrum.
( −1)2 +k2
=(
+1)2 +k2
√ε21 +ε22 +1−√2(√ε21 +ε22 +ε1 )
=
(A-3)
√ε21 +ε22 +1+√2(√ε21 +ε22 +ε1 )
119
Appendix B: Optical and Physical Measurement Results
Parameters
Aj
Ej
j
j
0
9.323
-0.056
-4.431
-2.203
1
-21.762
1.874
11.198
-0.258
2
0.484
3.693
0.567
0.785
j
Table B.1
Four optimized parameters of harmonic oscillator approximated
equation for bare FTO layer.
Position
Average
Deposition
Time (min)
1
2
3
4
5
thickness
6
(nm)
Standard
Deviation
15
144.27 145.46 150.92 177.25 160.6
158.00
156.08
12.25
10
108.87 105.89
110.5 106.54 112.71
112.64
109.53
2.94
8
98.67
104.15 100.82 102.59 106.29
102.34
102.48
2.63
5
68.42
67.43
71.35 66.37 72.98
76.29
70.47
3.77
4
37.83
38.34
36.95 35.40 36.38
32.74
36.27
2.02
3
29.49
32.27
29.08 30.34 30.18
29.49
30.141
1.15
2
25.36
25.20
25.29 16.59 23.99
24.48
23.49
3.42
1
~0
~0
~0
~0
~0
Table B.2 Optical CdS
spectroscopic reflectance.
~0
thickness
~0
~0
at
different
locations
measured
by
120
Location
Deposition
Time
(min)
1
2
3
4
5
Average
(nm)
Standard
Deviation
15
153.06 170.68 147.65 136.82 131.39
147.92
15.34
10
88.05
107.7
87.55
117.62
88.34
97.852
13.97
8
80.46
81.42
93.03
83.34
82.93
84.236
5.05
5
69.35
59.62
62.97
52.27
54.22
59.686
6.87
4
44.84
40.45
39.38
47.38
46.63
43.736
3.63
3
26.68
36.47
38.34
50.39
33.46
37.07
8.67
2
11.19
15.1
16.98
17.18
16.3
15.35
2.46
1
N/A
N/A
N/A
N/A
N/A
~0
Table B.3
N/A
Physical CdS thickness at different locations measured by TEM.
121
Appendix C: De-convolution Fitting Results
3.172
0.09
0.09
0.4625
0.4566
0.5195
3.516
3.582
3.488
3.450
0.0397
0.036
0.0406
0.036
0.036
0.4154
0.4263
0.4414
0.4625
0.4566
0.5195
3.892
3.940
3.965
3.923
3.965
3.908
3.883
a3
0.1752
0.1748
0.1730
0.1663
0.1882
0.1663
0.1736
=1.848*a1
a3
0.1817
0.6669
0.6257
0.6074
0.6480
0.6558
0.6886
0.6844
0.7123
0.6778
=1.56*w1
w3
w3
1.0E-9
2.06E-9
1.0E-9
1.0E-9
1.01E-9
2.2E-9
2.18E-7
2.43E-9
3.89E-8
1E-9
0.2226
0.2187
0.2271
0.2181
0.2159
0.2175
0.2273
0.2988
0.2286
0.2677
0.2165
0.9953
0.9972
0.9977
0.9980
0.9981
0.9974
0.9988
0.9989
0.9992
0.9992
0.9950
0.9976
Adjust
3.16
0.1016
0.4414
3.590
0.0378
0.4032
3.942
0.1781
0.6363
1.45E-9
0.2198
0.9962
a2 =
3.196
0.09
0.4263
3.548
0.036
0.4011
3.969
0.1776
0.6436
1.0E-9
0.2208
0.9927
Flow rate
3.16
0.09935
0.4154
3.463
0.0393
0.3971
3.955
0.1735
0.6243
1.09E-9
0.2217
1st
2nd
1st
2nd
1st
2nd
1st
ed R2
3.224
0.09458
0.4032
3.551
0.0386
0.4079
3.959
0.1827
0.6502
1.0E-9
C2
3.18
0.09
0.4011
3.597
0.0384
0.375
3.961
0.1700
0.6148
C1
3.227
0.09832
0.3971
3.577
0.0395
0.4002
3.968
0.1723
E3
3.215
0.09659
0.4079
3.600
0.0395
0.3569
3.973
w2 = w1
1st
3.25
0.09612
0.375
3.589
0.036
0.3941
E2
2nd
3.249
0.09877
0.4002
3.600
0.0373
w1
1st
3.261
0.09885
0.3569
3.613
a1
2nd
3.263
0.09
0.3941
E1
1st
3.283
0.09324
2nd
0.4*a1
2nd
3.289
((mL/min)
1.0
2.0
3.8
5.7
7.6
9.5
17.0
Table C.1 First and second stage de-convolution fitting results of CdS
nanoparticles by continuous MASD.
122
Appendix D: Thickness Data Points of Large CdS Films (6ʺ x 6ʺ)
All the thickness data are expressed as the unit of nm.
σ
Avg.
1
2
3
4
5
6
29.23
21.01
6.87
0.00
0.21
2.01
54.77
62.20
A
11.58
36.47
33.66
30.37
29.72
32.57
32.60
59.91
B
14.36
49.46
25.40
43.02
47.99
64.23
62.70
53.42
C
5.31
67.83
68.83
60.27
63.55
67.64
72.01
74.68
D
24.20
32.79
39.00
46.48
49.00
57.31
1.92
3.04
E
10.75
31.51
28.99
29.74
39.35
40.67
38.26
12.08
F
Table D.1 CdS films deposited by continuous MASD in the parallel flow cell.
(Flow direction: from right to left)
std
avg
1
2
3
4
5
6
5.79
29.94
30.32
34.28
32.44
35.90
26.55
20.17
A
6.51
30.57
34.91
26.70
38.75
35.26
24.09
23.69
B
4.06
35.07
36.06
40.17
37.98
34.42
28.46
33.35
C
1.78
33.75
31.10
35.16
33.22
35.17
32.42
35.44
D
2.17
35.39
33.58
38.13
34.12
35.28
37.95
33.26
E
2.17
35.45
35.94
36.62
37.73
36.51
31.71
34.20
F
Table D.2 CdS films deposited by continuous MASD in the deflected flow cell.
(Flow direction: from right to left)
123
σ
Avg.
6
5
4
3
2
1
2.04
71.99
68.80
73.38
73.80
73.33
72.50
70.12
A
2.09
75.34
71.94
78.33
75.96
75.73
74.46
75.63
B
1.11
74.28
73.24
74.85
72.95
74.54
74.09
76.00
C
1.76
74.14
74.43
76.80
73.41
74.34
74.46
71.38
D
4.46
71.77
72.28
76.93
73.37
73.76
70.54
63.74
E
5.79
67.94
70.53
72.30
71.56
65.80
70.41
57.07
F
Table D.3
CdS films deposited by continuous R2R-MASD at 2.5min. (Moving
direction: from left to right)
σ
Avg.
6
5
4
3
2
1
5.68
91.02
93.80
99.46
93.88
88.25
86.44
84.29
A
2.29
96.93
97.58
99.12
99.99
94.96
95.02
94.90
B
5.08
97.87
99.59
102.51
99.26
101.42
95.86
88.60
C
2.71
94.81
97.91
98.50
93.67
92.53
93.85
92.38
D
6.81
96.33
97.67
104.25
99.81
100.30
88.52
87.45
E
4.76
91.71
94.46
97.74
94.54
90.96
85.40
87.15
F
Table D.4 CdS films deposited by continuous R2R-MASD at 6.3min. (Moving
direction: from left to right)
124
σ
Avg.
6
5
4
3
2
1
24.81
188.46
173.94
198.39
215.68
208.64
185.84
148.28
A
26.91
237.46
211.33
253.17
264.58
259.17
237.59
198.93
B
18.92
252.16
236.25
266.69
271.84
263.63
250.79
223.78
C
16.71
259.22
239.76
279.90
272.90
266.77
254.86
241.12
D
20.07
239.58
223.04
274.24
247.46
235.85
238.77
218.09
E
21.50
216.78
239.77
247.40
209.90
200.63
207.35
195.59
F
Table D.5 CdS films deposited by continuous R2R-MASD at 9min. (Moving
direction: from left to right)
125
Appendix E: Time-dependent Outlet Tracer Concentration
V = 0.042 m/s
Q = 1.0 mL/min
V = 0.085 m/s
Q = 2.0 mL/min
126
V = 0.161 m/s
Q = 3.8 mL/min
V = 0.242 m/s
Q = 5.7 mL/min
127
V = 0.322 m/s
Q = 7.6 mL/min
V = 0.402 m/s
Q = 9.5 mL/min
128
V = 0.721 m/s
Q = 17.0 mL/min