Defect Engineering of Cuprous Oxide Thin-Films
for Photovoltaic Applications
by
YUN SEOG LEE
ARCHNES
B.S., Mechanical and Aerospace Engineering,
Seoul National University, Korea (2006)
APR 15)
M.S., Mechanical Engineering,
Stanford University, CA, USA (2007)
LLRAIE
SUBMITTED TO THE DEPARTMENT OF MECHANICAL EINGINEERING
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
FEBRUARY 2013
@ 2013 Massachusetts Institute of Technology
All rights reserved
........
Yun Seog Lee
Department of Mechanical Engineering
January 25q-2Q13
S ignature of A uthor.......................................................,
............
Certified by........................................................
/,fess9f
Tonio Buonassisi
ineering
ofjegyic
is Supervisor
...........................
David E. Hardt
Professor of Mechanical Engineering
Chairman, Committee for Graduate Studies
Accepted by...............................................
Defect Engineering of Cuprous Oxide Thin-Films
for Photovoltaic Applications
by
Yun Seog Lee
Submitted to the Department of Mechanical Engineering
on January 25, 2013 in partial fulfillment of the
requirements for the degree of Doctor of Philosophy in
Mechanical Engineering
ABSTRACT
Thin-film solar cells are promising for renewable-energy applications due to their
low material usage and inexpensive manufacturing potential, making them
compatible with terawatts-level deployment. Cuprous oxide (Cu 2O) is an earthabundant semiconductor with desirable properties for light-absorbing layers.
However, power conversion efficiencies of solar cells comprising this absorber
material remain significantly below the theoretical limit. In this thesis, I utilize
novel materials and device geometries to engineer defects in Cu 2 O thin-films and
overcome the low power-conversion-efficiency of Cu 20-based solar cells.
First, nitrogen doping is proposed as an effective p-type doping method to control
optical and electrical properties of Cu 2 O thin-films. The film's p-type conductivity
is elucidated by temperature-dependent Hall effect measurements and a
compensated semiconductor model. Secondly, an atomic-layer-deposited
amorphous zinc-tin-oxide buffer layer is developed to mitigate non-ideal band
alignment and interfacial defect-assisted recombination in Cu 2 0 - zinc oxide
(ZnO) heterojunction devices. Reduced interfacial recombination is demonstrated
by incorporating a 5-nm-thick buffer layer in the device. Finally, I propose a
spatially controlled vertical ZnO nanowire array to overcome the short minority
carrier diffusion length in Cu 2 0. A scalable fabrication process is developed
using colloidal lithography and hydrothermal growth of ZnO nanowires. Optical
simulations are also conducted to investigate the effect of nanostructured device
geometry on light-absorption properties.
Thesis Supervisor: Tonio Buonassisi
Title: Associate Professor of Mechanical Engineering
2
Table of Contents
Table of Contents ........................................................................................................
3
List of Figures ...................................................................................................................
6
List of Tables ..................................................................................................................
13
Acknowledgements .....................................................................................................
14
Citations to Published W ork......................................................................................
15
CHAPTER 1. Introduction.........................................................................................
16
1.1. M otiv atio n .............................................................................................................
17
1.2. Thesis overview .................................................................................................
23
CHAPTER 2. High-Mobility Cu 2O Thin-Film Deposition by Reactive Sputtering .25
2 .1. Introdu ction ...........................................................................................................
26
2.2. Structural properties of Cu 20 thin films ..........................................................
27
2.3. Electrical properties of Cu 2 0 thin films.............................................................
30
2 .4 . C onclu sion s ...........................................................................................................
36
CHAPTER 3. Interface Engineering of Cu 2 0-ZnO Heterojunction Solar Cells by a
Buffer Layer........................................................................................
37
3 .1. Introdu ction ...........................................................................................................
38
3.2. Amorphous zinc-tin-oxide buffer layer for Cu 2 0-based solar cells..................
41
3.3. Device characterization and analysis .................................................................
47
3.3.1. Current density - voltage characteristics under illumination....................
47
3.3.2. Band-alignment characterization...............................................................
49
3.3.3. Capacitance-frequency (C-]) characteristics ............................................
55
3.3.4. Current density - voltage characteristics under dark condition .................
57
3
3.4. Optical Simulation ............................................................................................
59
3.5. Experim ental details...........................................................................................
63
3.5.1. Atomic layer deposition of a-ZTO and AZO thin-films ............................
63
3.5.2. Thin-film solar cell fabrication .....................................................................
64
3.5.3. Characterization ........................................................................................
64
3.6. Conclusions...........................................................................................................
66
CHAPTER 4. P-Type Doping of Cu 2O Thin-Films by Nitrogen for a HoleTransporting Layer .............................................................................
67
4.1. Introduction...........................................................................................................
68
4.2. Structural and chem ical properties....................................................................
71
4.3. Electrical and optical properties........................................................................
76
4.4. Cu 2 0 :N film s as a hole-transporting layer in solar cells....................................
82
4.5. Experim ental details...........................................................................................
87
4.5.1. Cu 20 :N thin-film deposition .........................................................................
87
4.5.2. Thin-film solar cell fabrication .....................................................................
87
4.5.3. Characterization ........................................................................................
88
4.6. Conclusions...........................................................................................................
89
CHAPTER 5. Spatially-Controlled ZnO Nanowire Array for Cu 20 Thin-Film Solar
Cells......................................................................................................
90
5.1. Introduction...........................................................................................................
91
5.2. Spatially-controlled ZnO nanowire array...........................................................
93
5.2.1. Fabrication process....................................................................................
93
5.2.2. Colloidal lithography for patterning a ZnO nanowire array....................... 95
5.2.3. Seed layer deposition for vertical ZnO nanowire growth ..........................
98
5.3. Device fabrication and characterization..............................................................
101
5.4. Optical simulation...............................................................................................
102
5.5. Experim ental details............................................................................................
104
5.5.1. Cu 20 thin-film deposition...........................................................................
4
104
5.5.2. Characterization ..........................................................................................
5.6. Conclusions.........................................................................................................
104
105
CHAPTER 6. Summary and Future Directions.........................................................106
References
...................................................................
5
111
List of Figures
Figure 1.1. Forecast of global energy consumption rate. Data from [1].......................
17
Figure 1.2. Abundance of elements normalized to the abundance of silicon in the Earth's
crust. Blue-colored elements could be main constituents of photovoltaic devices.
A fter [4 ]. ...................................................................................................................
18
Figure 1.3. Annual electricity production potential for known semiconductor materials for
PV applications. A fter [3].....................................................................................
19
Figure 1.4. Upper limit of the efficiency of p-i-n/p-i-n type tandem solar cells as a
function of the bandgaps of the top layer (Eg,top) and the bottom layer
(Eg,bottom).
A fter [7 ]....................................................................................................................2
0
Figure 1.5. Improvement in band gap prediction accuracy. Calculated band gaps are
plotted against the experimentally measured values for 111 compounds. Red circles
are obtained from standard DFT (mean error ± 1.0 eV) and blue crosses are from a
newly developed method taking into account the dielectric screening (mean error ±
0.3 eV). Courtesy of Dr. M . K. Chan....................................................................
21
Figure 2.1. Upper limits of solar cell efficiency as a function of absorber material's
bandgap and carrier mobility. Only a radiative recombination path is assumed. After
[10 ]............................................................................................................................
26
Figure 2.2. The zone model of grain-size growth in sputtered thin films by Movchan and
D em chishin. A fter [18, 19] ...................................................................................
28
Figure 2.3. SEM images of Cu 2 0 films with growth temperatures of (a) 300 K, (b) 600 K,
and (c) 1070 K. All images to same scale; all size bars represent 1 pm............... 28
6
Figure 2.4. X-ray diffraction patterns of the samples with varying growth temperatures.
The patterns are normalized to the same maximum height. Dotted lines represent the
reference peaks of Cu2 0 (ICDD PDF# 01-071-3645)..........................................
Figure 2.5. Temperature-dependent
carrier density of Cu 2 0
29
films with growth
temperatures (a) 600 K (blue circle) and (b) 1070 K (red square). Lines represent the
exact solution from the theoretical model with given acceptor densities............. 31
Figure 2.6. Temperature-dependent Hall mobility of Cu 2 0 films with growth temperature
600 K (blue circles) and 1070 K (red squares). Open symbols represent
monocrystalline Cu 20 from various references; Ref. [33] (triangle-up), Ref. [34]
(triangle-down), Ref. [31] (hexagon), Ref. [30] (diamond), and Ref. [24] (circle).
Lines represent theoretical limits by: LO phonon scattering [32] (black dash),
ionized center scattering for the Tgroth = 600 K sample with NA= 2.2 x 1018 cm-3
(blue dash-dot), and for the Tgroth = 1070 K sample with NA= 2.7 x 1017 cm-3 (red
dash -dot-dot).............................................................................................................
34
Figure 3.1. Band structure at the junction and back contact (inset) of a HIT solar cell with
ap-type Si w afer. A fter [43].................................................................................
39
Figure 3.2. Schematic energy-band alignment diagram of a CIGS solar cell when the
conduction band of window layer is (a) above and (b) below that of CIGS layer.
A fter [4 7 ]. .................................................................................................................
40
Figure 3.3. (a) A schematic structure of the substrate-type Cu 20-based solar cells with aZTO buffer layer. (b) Cross-sectional SEM image of the device, exhibiting a highly
textured top surface stemming from (111) preferred growth of Cu 2 0. Scale bar, 1
pm. (c) Magnified SEM image near the junction interface. Conformal coating of an
ALD-oxide is demonstrated. Scale bar, 500 nm. (d) HR-TEM image near the
junction interface. Separated amorphous ZTO layer from crystalline Cu 2 0 and AZO
layers is show n. Scale bar, 5 nm ..........................................................................
42
Figure 3.4. (a) RBS spectra of the a-ZTO films with three different Zn to Sn ALD cycle
ratios. The carbon signal comes from the glassy carbon substrates used for the
7
measurements. (b) GAXRD spectra of the ZTO films. ZnO is included as a
reference. All ZTO films are amorphous...............................................................
44
Figure 3.5. (a) Optical absorption coefficient of atomic layer deposited a-ZTO and ZnO
thin-films. Increasing Sn contents in a-ZTO films reduces high-energy photon
absorption. (b) Bandgap estimation from the plot of (ahv)"2 as a function of photon
en erg y ........................................................................................................................
46
Figure 3.6. Current-voltage characteristics under 1-sun illumination (AM1.5G spectrum)
of the devices with different buffer layers. ...........................................................
48
Figure 3.7. XPS spectra of electrochemically deposited Cu 2 O thin-films. Cu 2P3/2 core
level spectra show only Cu2
states with a relative energy level of 932.47 eV
reference to V B E ....................................................................................................
50
Figure 3.8. XPS spectra of (a) Zn core level and (b) valence band for atomic layer
deposited a-ZTO and ZnO thin-films. Zn 2P3/2 core level peaks of ZnO and a-ZTO
with a Zn:Sn ratio of 1:0.27, 1:0.59 and 1:1.8 exhibit relative energy levels of
1019.02, 1019.44, 1019.57, 1019.47 eV reference to their VBEs, respectively. In this
analysis, small humps near valence band edges of a-ZTO are attributed to tail states
in bandgap [62], which originate from increased SnO 2 contents. The hump near
VBE of SnO 2 is often observed from photoelectron spectroscopy when a high
energy photon source is used [63], while the states are not expected from theoretical
calculations [64, 65]...............................................................................................
51
Figure 3.9. UPS spectra of the valence band for atomic layer deposited a-ZTO and ZnO
thin-films measured using a He-I photon source (hv = 21.2 eV). The inteinsities of
small humps near valence band edges of a-ZTO are smaller than the spectra from
X P S...........................................................................................................................
52
Figure 3.10. XPS spectra of Zn and Cu core levels from 2-nm-thick ZnO and a-ZTO
film s on Cu 2 0 thin-film s........................................................................................
8
53
Figure 3.11. Relative alignments of conduction band (CB) and valence band (VB) for aZTO and ZnO overlayers to Cu 20 thin-films investigated in this work. The values
were m easured by XPS technique........................................................................
54
Figure 3.12. Capacitance - frequency characteristics of the devices with a-ZTO buffer
layers at room tem perature....................................................................................
56
Figure 3.13. Effect of a-ZTO (Zn:Sn = 1:0.27) buffer layer on dark current-voltage
characteristics of the devices. Inset schematics at top and bottom show electronicband structures of the devices with ZnO and a-ZTO buffer layers, respectively. Grey
areas indicate defect-rich interfaces with deep-trap states. Red arrow lines represent
interfacial recombination paths for electrons (filled circles) from the ZnO layer. The
a-ZTO buffer layer impedes electron movement to the interface where holes (open
circles) are provided from the Cu 20 layer, thus reducing J. ................................
58
Figure 3.14. Simulated optical absorption profile in the Cu 2 0 layer for 500 nm
wavelength light with the intensity of that in the AMI.5G spectrum. Scale bar
(white), 1 pm . Courtesy of J. P. M ailoa...............................................................
59
Figure 3.15. Modeled collection probability profiles for photo-generated carriers in the
C u 2 0 lay er.................................................................................................................6
1
Figure 3.16. Effect of the minority carrier diffusion length of Cu 2 0 on EQE. Colored
lines represent calculated EQE from optical simulation with a carrier-collection
probability profile with increasing Lc and an ideal 100 % collection case. Dotted
black line represents measured EQE of the a-ZTO (Zn:Sn = 1:0.27) buffer layer
incorporated device...............................................................................................
62
Figure 4.1. Valence and conduction band edge positions of various oxides with their
doping limits, showing dopable and undopable cases. After [85]........................
69
Figure 4.2. Film growth temperature effects on (a) electrical resistivity, (b) carrier density,
and (c) mobility of 0.6-pm-thick Cu 20:N films measured by Hall effect
9
measurements at room temperature. Dotted lines are to guide the eye. N2 flow rate
was maintained at 1 secm. All samples exhibited p-type conductivity. ...............
72
Figure 4.3. Nitrogen concentrations in Cu 2 0:N films measured by SIMS. A typical
calibration error range is up to ± 30 %. ...............................................................
73
Figure 4.4. XRD spectra of Cu 2 0 and Cu 20:N films, indicating two main peaks from
Cu 2 0 (111) and (200). Inset SEM images of the Cu 20 (bottom) and Cu 2O:N (top)
films. The both scale bars represent 300 nm........................................................
74
Figure 4.5. XPS spectra of (a) copper and (b) nitrogen core levels of Cu 2 0 and Cu 20:N
films. A grey arrow indicates a small nitrogen peak at 397.1 ± 0.2 eV................ 75
Figure 4.6. Effects of nitrogen doping on electrical resistivity of Cu 2 0:N films measured
by four-point-probe...............................................................................................
78
Figure 4.7. Temperature-dependent Hall mobility of Cu 2 0 and Cu 2 O:N films. .......... 78
Figure 4.8. Effects of nitrogen doping on carrier densities of Cu 2 0:N films: Temperaturedependence of carrier (hole) density determined by Hall effect measurements....... 79
Figure 4.9. Carrier (hole) activation energies of Cu 2 0:N films estimated from the
measured carrier density in the sample temperature range of 200 - 330 K.......... 79
Figure 4.10. Effects of nitrogen doping on the optical absorption of Cu 2 0:N films
measured by a spectrophotometer with an integrating sphere...............................
81
Figure 4.11. A cross-sectional SEM image of Cu 2 O-based thin-film solar cell with a
Cu 2 0:N hole-transporting layer. Scale bar, 1 pm.................................................
83
Figure 4.12. Dark J-V characteristics of a Cu 20:N-incorporated device and a control
dev ice . .......................................................................................................................
84
Figure 4.13. J-V characteristics of a Cu 2 0:N-incorporated device and a control device
under 1-sun illumination (AM1.5G, 100 mW-cm )............................................
10
84
Figure 4.14. External quantum efficiency of of a Cu 2 0:N-incorporated device and a
control device at zero bias voltage........................................................................
86
Figure 4.15. Schematic diagram of energy band alignments in Cu 2 0-based thin-film solar
cells: (a) the control device and (b) Cu 2 0:N hole-transporting layer incorporated
dev ice . .......................................................................................................................
86
Figure 5.1. Schematic diagram of the spatially-controlled ZnO nanowire array growth
process, consisting of (a) (0001)-textured ZnO:Al seed layer deposition on SiO 2
substrate, (b) colloidal PS sphere assembly, (c) 5-nm-thick TiO 2 mask layer by
atomic layer deposition, (d) PS-sphere removal, (e) ZnO nanowire growth by the
hydrothermal method, and (f) Cu 20 and Au electrode deposition. Courtesy of Dr. J.
J00 .............................................................................................................................
94
Figure 5.2. A closely packed 500 nm diameter PS-sphere array transferred to a substrate.
The scale bar is 1 pm ............................................................................................
95
Figure 5.3. A SEM image of a ZnO nanowire array grown on a single crystal ZnO
substrate (300 tilted view). The scale bar is 1 pm. ................................................
97
Figure 5.4. A SEM image of a ZnO nanowire array grown on a single crystal ZnO
substrate (top view). The scale bar is 1 m. .........................................................
97
Figure 5.5. XRD spectra of the sputtered ZnO:Al films and grown ZnO nanowires....... 99
Figure 5.6. A side view of the vertically grown ZnO nanowire array on the textured
ZnO:Al seed layer patterned by 1 jim spheres. The scale bar is 1 pm. ................
99
Figure 5.7. ZnO nanowires grown on a polycrystalline ZnO:Al film without a mask layer.
The scale bar is 1 pm . .............................................................................................
100
Figure 5.8. ZnO nanowires grown on a polycrystalline ZnO:Al film with a patterned TiO 2
mask layer, exhibiting larger diameter than nanowires without patterning. The scale
bar is I I m . .............................................................................................................
11
10 0
Figure 5.9. J-V characteristics of ZnO nanowire array incorporated devices and a control
device under 1-sun illum ination condition. ............................................................
102
Figure 5.10. 2-D optical simulations of the optical absorption profile for photons with a
wavelength of 500 nm in Cu 20 layer (middle area) between ZnO nanowires with
periods of (a) 350 nm, (b) 500 nm, (c) 700 nm, (d) 1000 nm, and (e) a planar
structure. The incident photon flux is based on the AMI.5G solar spectrum. White
lines indicate interfaces between ZnO and Cu 2 0. All scale bars (bottom) represent
200 nm . Courtesy of J. P. Mailoa............................................................................
103
Figure 6.1. Band alignment in ZnO/GaN, Cu 20/GaN and Cu 20/ZnO heterojunctions.
A fter [1 18]..............................................................................................................
12
10 9
List of Tables
Table 1.1. Candidate earth-abundant semiconductor compounds for thin-film solar cells.
...................................................................................................................................
22
Table 3.1. Photovoltaic characteristics under 1-sun illumination.................................
48
Table 4.1. Photovoltaic characteristics under 1-sun illumination.................................
85
Table 5.1. Photovoltaic characteristics under 1-sun illumination...................................
13
101
Acknowledgements
I would like to express my sincere thanks to a number of people for their support
throughout my years at MIT.
First of all, I would like to thank Prof. Tonio Buonassisi for his tremendous support and
guidance on my research with patience. I am very fortunate to have him as my advisor.
His enthusiasm, optimism, and wisdom always encouraged me through my graduate
study. It has truly been a fruitful and enjoyable experience to work with him.
I appreciate kind guidance and advice from my thesis committee members, Prof. Joseph
Jacobson and Prof. Gerbrand Ceder. Their insightful feedback strengthened my
dissertation.
I thank all the members of the Photovoltaic Laboratory for their kind support. I have great
memories of working together with our talented team members, including Mark T.
Winkler, Sin Cheng Siah, Riley E. Brandt, Jonathan P. Mailoa, Prof. Mariana I. Bertoni,
Jim Serdy, and Michael Lloyd.
I also would like to thank Dr. Inna Kozinsky, Prof. Roy G. Gordon, Prof. Jaeyeong Heo,
Meng-Ju Sher, Dr. Sangwoon Lee, Dr. Jaebum Joo, Dr. Kimin Jun, Prashant Patil, Shin
Young Kang, Dr. Maria K. Chan, Youngsoo Joung, Dr. Ching-Mei Hsu, Shuang Wang,
Kurt Broderick, Dr. Scott Speakman, Libby Shaw, Ed Macomber, Mac Hathaway, Dr.
Jongbae Park, Byungchul Son, and Dr. Alan Wan for providing invaluable discussions
and experimental supports on my research.
I would like to thank Pamela A. and Arunas A. Chesonis and Doug and Barbara Spreng
for their generous support through my research. I also thank Bosch for providing an
internship opportunity at the Bosch Research and Technology Center in Palo Alto as well
as the strong support to my research.
Finally, I would like to express heartwarming gratitude to my family and friends for their
love and support.
14
Citations to Published Work
Parts of this dissertation cover research reported in the following articles:
[1] Y. S. Lee, M. Bertoni, M. K. Chan, G. Ceder and T. Buonassisi, "Earth Abundant
Materials for High Efficiency Heterojunction Thin Film Solar Cells," Proc. 34th
IEEE PhotovoltaicSpecialists Conference (PVSC), pp. 2375-2377, 2009
[2] Y. S. Lee, M. T. Winkler, S. C. Siah, R. Brandt and T. Buonassisi, "Hall Mobility of
Cu 2 0 Thin Films Deposited by Reactive DC Magnetron Sputtering," Applied Physics
Letters, vol. 98, p. 192115, 2011.
[3] S. C. Siah, Y. S. Lee, Y. Segal and T. Buonassisi, "Low Contact Resistivity of Metals
on Nitrogen-Doped Cuprous Oxide (Cu 2 O) Thin-Films," Journalof Applied Physics,
vol. 112, p. 084508, 2012.
[4] K. Jun, Y. S. Lee, T. Buonassisi and J. M. Jacobson, "High Photocurrent in Silicon
Photoanodes Catalyzed by Iron-Oxide Thin Films for Water Splitting," Angewandte
Chemie InternationalEdition, vol. 51, pp. 423-427, 2012.
[5] Y. S. Lee, J. Heo, S. C. Siah, J. P. Mailoa, R. E. Brandt, R. G. Gordon and T.
Buonassisi, "Ultrathin Amorphous Zinc-Tin-Oxide Buffer Layer for Improving
Heterojunction Interface Quality in Metal-Oxide Solar Cells," submitted, 2012.
[6] Y. S. Lee, J. Heo, M. T. Winkler, S. C. Siah, R. G. Gordon and T. Buonassisi,
"Nitrogen-doped Cuprous Oxide for P-Type Hole Transporting Layer in Metal Oxide
Solar Cells," in preparation.
[7] Y. S. Lee, J. Joo, J. P. Mailoa, J. M. Jacobdson and T. Buonassisi, "Scalable
Fabrication of Spatially Controlled Vertical Zinc Oxide Nanowire Array for Efficient
Light Absorption in Thin-Film Solar Cells," in preparation.
15
CHAPTER 1.
Introduction
16
1.1. Motivation
Photovoltaics (PV) is a promising renewable energy source that can meet the
terawatt-level global energy demand shown in Figure
1.1, while keeping low
concentrations of CO 2 in the atmosphere. To become a major energy source, module
costs should be reduced significantly and energy conversion efficiencies should be
increased. In terms of reducing module cost, thin-film solar cells represent a viable option
due to their lower material usage and inexpensive manufacturing potential. Comparing to
wafer-based crystalline silicon solar cells, approximately a hundred times less material is
needed for the light-absorbing layer in thin film technologies, since direct-bandgap
semiconductors are more efficient at absorbing light. The lower materials usage can
reduce costs during module production and reduce energy-payback time. Moreover, the
flexible nature of thin-film devices can also allow for new types of applications and
inexpensive manufacturing processes (e.g. roll-to-roll processing).
25
20
0
15
0D CU
10
> 0
tC
OE
0
1980
1990
2000
2010
2020
2030
Year
Figure 1.1. Forecast of global energy consumption rate. Data from [1].
17
Conventional materials for thin-film solar cells such as cadmium telluride (CdTe)
and copper indium gallium (di)selenide (CIGS) suffer from concerns over resource
scarcity (e.g., tellurium and indium) and toxicity (e.g., cadmium) and are believed to be
limited to sub-terawatts deployment (Figure 1.2 and Figure 1.3) [2, 3]. From a materials
abundance and toxicity perspective, a very promising thin-film solar cell device type is
the so-called micromorph Si solar cell. This device consists of a bottom layer of microcrystalline silicon (ic-Si) with a top layer of amorphous silicon (a-Si). Since each layer
preferentially absorbs a certain region of the solar spectrum, the tandem structure device
can theoretically achieve higher power conversion efficiency than a single-junction Si
device. Given the bandgaps of pic-Si (1.1 eV) and a-Si (1.7 eV), the maximum theoretical
efficiency of a tandem p-i-n/p-i-n type device is expected to be 35 % (Figure 1.3).
102
10
-
~
SI
( -
Eio'-
-100
H
Al
Na
~
K
Ca
Mg
1-2
10
Ti
F p
LiS
o E10-4
0~ 0
Fe
Sr
N
BeO
10-6
CI
Ba
I:.
V
Zr
VCrr Cu Zn Rb
Rb
Co Ni Ga
Nb
Sc CO
YT
AssSn
Ge
Br
Pb
Cs
IW
Cd
Ag
10~7
T
Hf Ta+
U*
Rare earth
b eelements
10-8
B1
8Pt
10-9-
Te
Au
10-10 -20
40
60
RU number
Atomic
Rh
80
Os
Rel Ir
Figure 1.2. Abundance of elements normalized to the abundance of silicon in the Earth's
crust. Blue-colored elements could be main constituents of photovoltaic devices. After
[4].
18
1.00E+11
MAnnual Electricity FromKnownEconomicReserves
(TWhl
1.00E+10
X AnnualElectricity FromAnnualProduction(TWh)
1.00E+09
1.OOE+08
1.00E+07
1.00E+06
M
1.OOE+05
....
...
...
...... ... ... .
.. ..
.. ... ... *
17,000 TW h
1.OOE+04
4,0I~
1.OOE+03
-
1.OOE+02
1.OOE+01
-
1.00E+00
) .
CL
0
Figure 1.3. Annual electricity production potential for known semiconductor materials for
PV applications. After [3].
However, power conversion efficiencies for large-scale micromorph Si cells are
only around 10 % [5]. This is mostly due to the poor carrier mobility of a-Si in the topcell and its degradation by the Staebler-Wronski effect [6]. While the device structure has
a great potential for high efficiencies with low manufacturing costs, the a-Si layer is a
bottleneck for higher device performance. Thus it would be desirable to keep the inherent
advantages of thin-film materials, while replacing the top layer of a-Si with another
material. This material needs to fulfill the following constraints for a scalable device with
a power conversion efficiency over 20 %: (a) abundant in nature, (b) inexpensive, (c)
proper electrical and optical properties, (c) non-toxic, (d) manufacturable, (e) defect
tolerant.
19
One of the most important material properties limiting solar cell power
conversion efficiency is the bandgap, which determines the rate of photon absorption. A
bandgap between 1.6 and 2.0 eV is required to complement a silicon-based bottom layer
in a tandem device structure theoretically capable of supporting efficiencies higher than
30 % (Figure 1.4).
1.6
Iv
1
1.2
1.4
1.6
1.8
2
2.2
ES top [eV)
Figure 1.4. Upper limit of the efficiency of p-i-n/p-i-n type tandem solar cells as a
function of the bandgaps of the top layer (Eg,top) and the bottom layer
(Eg,bottom).
After [7].
To find good candidate materials efficiently, computational methods has been
developed to predict the bandgap of a given material. The bandgap prediction can be
carried out to various levels of accuracy by using different computational methods, with
more computationally intensive methods generally producing more accurate results. The
most efficient method is the calculation of the bandgap using standard density functional
theory (DFT). Although DFT typically underestimates the bandgap compared to
20
experimental values, it is still useful as a first screening tool since existing accurate
methods typically require 100 - 1000 times the computational effort of DFT (Figure 1.5).
To meet the terawatts level deployment demand, a list of low-cost elements based
on their abundance in Earth's crust and manufacturing production capacity was derived.
From these elements, 190 semiconductor compounds were found from a combinatorial
search. Large-scale bandgap calculations and literature reviews were performed, which
reduced the list to tens of candidate semiconductor compounds. Finally, Cu 2O is
investigated as the first selection for thin-film solar cells materials.
~44
0
-
2
-
3
4
Measured Band gaps (eV)
Figure 1.5. Improvement in band gap prediction accuracy. Calculated band gaps are
plotted against the experimentally measured values for 111 compounds. Red circles are
obtained from standard DFT (mean error ± 1.0 eV) and blue crosses are from a newly
developed method taking into account the dielectric screening (mean error ± 0.3 eV).
Courtesy of Dr. M. K. Chan.
21
Table 1.1. Candidate earth-abundant semiconductor compounds for thin-film solar cells.
compound
bandgap
aluminum dodecarboride (AIB 12)
1.9
calcium silicide (Ca 2 Si)
1.9
cuprous oxide (Cu 2 0)
1.9
-
2.1
copper nitride (Cu 3 N)
1.2
-
1.9
silicon diphosphide (SiP 2)
1.9
zinc diphosphide (ZnP 2)
1.7 - 2.2
zinc phosphide (Zn 3 P2)
1.4 - 2.2
zirconium sulfide (ZrS 2)
1.7
Cuprous oxide (Cu2O), a compound semiconductor with a direct bandgap of 1.9 2.1 eV,[8] is a promising material for thin-film photovoltaic applications due to its
elemental abundance in the Earth's crust[3] and non-toxicity. Although the ShockleyQueisser efficiency limit for Cu 2 0 is about 20%, the maximum efficiency realized using
oxidized Cu metal foils is significantly below the limit [9]. This low record efficiency
stems from a variety of factors that remain poorly understood in Cu 20, including poor
collection probability of photo-excited carriers, un-optimized band structure in the device
structure, and high surface recombination.
22
1.2. Thesis overview
This study aims to overcome the principal performance-limiting defects currently
restricting cell efficiencies. The main goal in this research is to develop Cu 2 0-based thinfilm solar cells that can be used for the top cell in a tandem device structure. Following
the introduction in this chapter, I apply novel device geometries and materials to reduce
major energy loss mechanisms and improve solar cell performance.
In Chapter 2, polycrystalline Cu 2 0 thin-films are deposited by reactive sputtering.
To improve grain structure and carrier mobility of the Cu 2 0 films, the substrate
temperature of the film is increased up to 1070 K during the film growth. The Cu 2 0 films
deposited at higher temperature shows a significant enhancement in Hall mobility
comparable to single crystal Cu 2 0 mobility at above 250 K. Detailed analysis with a
compensated semiconductor model is conducted to elucidate mobility-limiting factors.
In Chapter 3, I develop Cu 20-based thin film solar cells comprising a ZnO-Cu 2 0
heterojunction structure and discuss the effect of band alignment on device performance.
To mitigate non-ideal band alignment between Cu 2 0 and ZnO and reduce interfacial
recombination problems, an amorphous zinc tin oxide buffer layer is developed. Using
the atomic layer deposition technique, highly conformal films with precisely tunable
properties can be deposited. By adjusting the film's band alignment and electronic
properties, the buffer layer enhances the device open circuit voltage.
In Chapter 4, I develop a p-type hole transporting layer by nitrogen-doped Cu 2 0
films to minimize the back-contact energy barrier in the device. By controlling the
nitrogen content in the film, carrier density and optical transmittance are tuned to create a
23
low contact resistance interface with minimal optical absorption. By increasing the carrier
density of the Cu 2 0:N layer, a tunnel junction is formed between the Ag back electrode
and Cu 20 to reduce the contact resistance. I demonstrate that the addition of a Cu 2 0:N
layer results in a sizeable fill-factor (FF) and power conversion efficiency enhancement
relative to control samples without the layer.
In Chapter 5, I develop a scalable fabrication method of a spatially controlled
vertical ZnO nanowire array to overcome the short carrier-collection length problem and
enhance photo-generated carrier-collection
probability. Colloidal lithography with
polystyrene nano- and micro-spheres is used to enable hexagonal alignment of ZnO
nanowire array. Using the ZnO nanowire array, Cu 2 0-based heterojunction thin-film
solar cells are developed. The strong enhancement of photo-generated carrier collection
by incorporating ZnO nanowire array in the device is demonstrated. In addition, optical
simulations using the three-dimensional FDTD method are used to further analyze optical
effects of the ZnO nanowire array in the device.
24
CHAPTER 2.
High-Mobility Cu 2 O Thin-Film
Deposition by Reactive Sputtering
25
2.1. Introduction
In this chapter, polycrystalline Cu 2O thin-films are deposited by sputtering to
enhance their carrier mobility. High mobility of photo-generated carriers in the Cu 2 0 film
can increase the carrier's diffusion length and the solar cell power conversion efficiency
(Figure 2.1) [10].
Polycrystalline Cu2 0 thin films have been deposited by various
methods, such as sputtering [11, 12], pulsed laser deposition [13], molecular beam
epitaxy [14], chemical vapor deposition [15], and electro-chemical deposition [16].
Among these deposition methods, reactive direct-current (DC) magnetron sputtering is a
relatively cost-effective process that can be used for large-area device fabrication. The
sputtered Cu 2 0 films have sufficiently large grain size for thin-film photovoltaic
applications. Additionally, temperature-dependent Hall effect measurements are carried
out to identify the dominant mechanism limiting carrier mobility, and determined that
Cu 2 0 films grown via sputtering exhibit majority carrier mobilities sufficiently high for
thin-film photovoltaic applications.
35
-
-
S6 iit
10
30
25
.
20
=1510
1.0 1.2 1.4
0.8
Band Gap Energy E9 [eV]
Figure 2.1. Upper limits of solar cell efficiency as a function of absorber material's
bandgap and carrier mobility. Only a radiative recombination path is assumed. After [10].
26
2.2. Structural properties of Cu 20 thin films
Cu 2 0 thin films were deposited on GE-124 fused quartz glass substrates by
reactive DC magnetron sputtering using an ATC-2200 (AJA International) in an argon
and oxygen atmosphere. The substrate temperature was controlled using quartz lamps. A
constant power (DC 50 W) was applied to a metallic copper target (2 inch dia., 99.999%
pure, Kurt J. Lesker Company). The base and working pressures of the chamber were
1.3 x 10- Pa and 0.53 Pa respectively. The phase purity of Cu 2O was controlled by
varying the flow rate ratio of argon and oxygen between 1:0.35 and 1:0.39. The average
deposition rate was ~3.4 nm/min.
For thin-film photovoltaic applications, columnar grain structure with a grain size
larger than the film's thickness is desired [17]. To control morphology, the substrate
temperature during film growth was varied. By adopting the Zone Model proposed by
Movchan and Demchish (Figure 2.2) [18] for sputtered films, the temperatures were
chosen to be 300 K (0.2 Tm, where Tm= 1508 K is the Cu 2 0 melting temperature), 600 K
(0.4- Tm) and 1070 K (0.7- Tm). These choices represent each regime proposed in the
model. Film morphology was studied using a Zeiss ULTRA55 field-effect scanning
electron microscope (SEM). The SEM micrographs in Figure 2.3 show a change from
fiber-like grains to columnar grains, as well as an increase in grain size, as the substrate
temperature increases. Digital image processing was used to estimate average grain sizes:
79
+
17 nm, 228 ± 57 nm and 884 ± 373 nm for the samples grown at 300 K, 600 K and
1070 K, respectively.
27
0.3
0.5
SUBSTRATE
TEMPERATURE (T/Tm)
Figure 2.2. The zone model of grain-size growth in sputtered thin films by Movchan and
Demchishin. After [18, 19].
a plane view
b
c
cross-section
nennw
Figure 2.3. SEM images of Cu 2 0 films with growth temperatures of (a) 300 K, (b) 600 K,
and (c) 1070 K. All images to same scale; all size bars represent 1 tm.
28
The phase and crystal structure were characterized by X-ray diffraction (XRD)
using PANalytical X'Pert Pro diffractometer with Cu-Ka radiation. XRD confirmed that
higher substrate temperature results in films with better crystallinity. As observed from
the Bragg-Brentano scans in Figure 2.4, the diffraction peaks of all samples are well
matched to the reference pattern of Cu 2 O and the peaks of other phases (e.g., Cu and
CuO) were not detected. All samples also exhibited (200) out-of-plane preferred
orientation. Samples grown at higher substrate temperatures showed narrower diffraction
peaks due to the increase of grain size. The smallest full-width at half maximum
(FWHM) of the (200) peak was 0.145' with a substrate temperature of 1070 K.
JFWHM
0.6010
Cd
T
-
iI
011
1 1"
kk,
=300K:
600KL
.
0.432*
600K:,
0.1450
1070K
20
30
40
50
60
70
80
90
2-0 (0)
Figure 2.4. X-ray diffraction patterns of the samples with varying growth temperatures.
The patterns are normalized to the same maximum height. Dotted lines represent the
reference peaks of Cu 2 0 (ICDD PDF# 01-071-3645)
29
2.3. Electrical properties of Cu 20 thin films
The temperature-dependent Hall effect was measured by using the van der Pauw
configuration. Ohmic Au contacts were deposited on the corners of 1 x 1 cm2 Cu2 0 film
samples using electron beam evaporation. Samples were placed in a closed-cycle He
cryostat on a copper cold finger in a near-vacuum environment (P < 0.1 Pa); a resistive
heater was used for temperature control. Measurement temperatures were kept below 400
K to prevent bulk phase change and persistent photo-conductivity decay [20]. All
samples exhibited p-type conductivity only and strong temperature dependence. It was
unable to measure reproducible Hall voltages from the sample grown at 300 K due to its
low mobility (< 1 cm 2/V s); the remaining samples exhibited stable Hall voltages. Hall
voltages VH were measured using magnetic field B = 0.65 T and excitation current I.
Carrier density p was calculated using the relationship,
(1
p = IB(edVH) ',
where I is the current, B is the magnetic field, e is an electron charge, d is the film
thickness, and VH is the Hall voltage.
Figure 2.5 shows the temperature dependences of carrier density for samples
grown at 600 K and 1070 K. The low-temperature portion of the data was fitted by using
the low-temperature approximation [11, 21] (p < N
-
N1 ) for carrier density in a
compensated semiconductor,
p =(21rm*kT/h2)
3 2
/
/kT),
-[(N A/N)-1]-exp(-E
30
(2)
where the effective mass m* can be taken as 0.58. mo [22], k is the Boltzmann constant,
h is the Planck constant, NA is the acceptor density, N is the donor density, and E is
the activation energy. This model assumes only one type of singly-charged acceptor is
present, and that all donors are ionized (
ND
= ND ). Using this model, EA were
measured to be 0.23 eV and 0.19 eV for samples grown at 600 K and 1070 K,
respectively. These activation energies are in the range of previously reported
experimental values, between 0.16 eV and 0.42 eV [11, 23, 24].
b
a
S18
I017
- -
E=(
+-NA
10
10
-
1o~17
10
16
9 eV
0.86
10
10
101-------
-101
10
10
T
10
.101"+-N(3)
2
=600
-
10,
1000O/T (K-4)
1ooo/T (K-')
Figure 2.5. Temperature-dependent
carrier density of Cu 2 0
films with growth
temperatures (a) 600 K (blue circle) and (b) 1070 K (red square). Lines represent the
exact solution from the theoretical model with given acceptor densities.
31
Fitting the low-temperature portion of our data with Eq. (2) provides estimates of
both
EA
and the compensation ratio (N) / NA). The fits yield the ratios of 0.20 and 0.86
for samples grown at 600 K and 1070 K, respectively. The acceptor density was
estimated by using the exact form[25] of Eq. (1) - which will saturate at high temperature
when acceptors are completely ionized - and extending our fits to include all of our data.
By calculating the net carrier density for various values of
and
ND /NA
(using the values of
NA
EA
provided by the low temperature fits), samples grown at 600 K and 1070 K
are expected to have acceptor densities of at least 2.2
x 1018
cm~3 and 2.7
x
1017 cm-3,
respectively. However, since the carrier densities of these samples did not saturate at the
experiment's maximum temperature, these values represent lower bounds on the acceptor
densities.
For the sample grown at 600 K, a change of the slope in Figure 2.5 is observed as
measurement temperature increases. This behavior cannot be fit using the single-acceptor
model, and could be caused by multiple types of acceptors with different energy levels.
Previous experimental work[26, 27] and ab initio calculations [28, 29] have both
suggested the possibility of multiple acceptor levels. A substantially improved fit can be
generated using a two-acceptor model, which provides acceptor level energies of 0.20 eV
and 0.37 eV. Interestingly, the low energy acceptor level is close to the acceptor level
(0.19 eV) of the sample grown at 1070 K. However, due to the number of parameters, I
have less confidence in this fit. In addition, the lower bound on
NA
provided by the two-
acceptor model is consistent with the one-acceptor model; thus, the results from the oneacceptor model are used in the subsequent analysis for simplicity.
32
In Figure 2.6, the temperature-dependence of our samples' Hall mobilities are
compared to theoretical and experimental values of monocrystalline Cu 20 [24, 30-34].
Shimada and Masumi [32] modeled the Hall mobility of Cu 20 when limited by
longitudinal-optical (LO) phonon scattering with 220 K and 960 K modes; the origin of
the discrepancy between theoretical and experimental mobilities at temperatures above
200 K is currently unresolved. The Hall mobilities of both sputtered samples are
comparable to that of monocrystalline Cu 2 0 at temperature above 250 K. The Hall
mobility of the Tgrowth = 1070 K sample was 62 cm 2 /V s at room temperature (293 K) and
43 cm 2 /V -s at a typical solar cell operating temperature of 60 'C (333 K).
33
10
-
- -j
(10 0K
600K)
10~
...--
-00K-
-
100
101
0
-
T
0
11
2
3
4
=1070 K
=600K
5
6
1OOrrT (KT)
Figure 2.6. Temperature-dependent Hall mobility of Cu 2 o films with growth temperature
600 K (blue circles) and 1070 K (red squares). Open symbols represent monocrystalline
Cu 2 O from various references; Ref. [33] (triangle-up), Ref. [34] (triangle-down), Ref. [31]
(hexagon), Ref. [30] (diamond), and Ref. [24] (circle). Lines represent theoretical limits
by: LO phonon scattering [32] (black dash), ionized center scattering for the Tgrowth
K sample with NA
=
2.2
x
600
1018 cm~3 (blue dash-dot), and for the Tgwth = 1070 K sample
with NA = 2.7 x 1017 cm-3 (red dash-dot-dot).
34
The mobilities of our samples appear to be limited by different factors than
monocrystalline Cu 2 0 at temperatures below 250 K. Because our films exhibit higher
carrier concentrations relative to monocrystalline Cu 2 O, there will be a higher density of
ionized centers, likely native defects. The ionized impurity-limited mobility p,, is
calcultated by using [35]:
p
HN
12 (E0 2--2
)kT
128,[2-7r
V N,Z2e+b
where b = 24m*kT/h
2
p,j2
,
r1 ,
ln(1+b)
b
,
(3)
is the Hall coefficient for ionized impurity scattering
(equal to 1.93), c is the relative dielectric constant of Cu 2 0, E( is the dielectric constant
of vacuum, Z is the charge of the scattering center, e is the electron charge, and f, is the
inverse screening length. For a p-type semiconductor compensated by singly-charged
donors, the ionized impurity density N, is equal to p + 2N
. The estimated lower
bounds of donor density were used to calculate the impurity density. Therefore, our
calculation of u,, is actually an upper bound, and could be shifted downwards in Figure
2.6. Other factors for polycrystalline Cu 2 0 films such as scattering due to grain
boundaries and dislocations were considered, but they could not accurately model the
measured Hall mobility. Thus, it is concluded that the scattering from ionized centers is
the likely limiting mechanism in these samples at lower temperatures.
35
2.4. Conclusions
In summary, it is shown that reactive DC magnetron sputtering can deposit Cu 2 0
thin-films with controllable structural and electrical properties. High-quality Cu 2 0 films
can be deposited at high substrate temperature, resulting in suitable properties for thinfilm photovoltaic applications. Temperature-dependent Hall measurements reveal that the
sputtered films exhibit high Hall mobility, comparable to that of monocrystalline Cu 20 at
temperature above 250 K. Lastly, it is deduced that the Hall mobility is limited by the
scattering from ionized-centers at low temperature.
36
CHAPTER 3.
Interface Engineering of Cu 2 0-
ZnO Heterojunction Solar Cells by a Buffer
Layer
37
3.1. Introduction
Due to difficulties in doping Cu 2 0 to n-type, the most common approach to make
PV devices is using a Cu 20-ZnO heterojunction structure. However, record efficiencies
with the devices remain low, with wafer-based Cu 20 devices reaching 4.1% [36] and
thin-film Cu 2 0 devices reaching 1.3% [37], exhibiting strong dependence on the junction
interface [38, 39]. Thin-film devices generally have significant manufacturing cost
advantages over wafer-based approaches, but generally have a higher concentration of
lifetime-limiting bulk defects and are more susceptible to poor surface/interface quality
[40, 41].
In this chapter, an approach to reduce the impact of interface defects on the
performance of Earth-abundant thin-film devices is demonstrated. This approach could
be applicable to a wide range of Earth-abundant materials. I apply an approach
successfully implemented for crystalline silicon PV technology: Introducing a thin (~5
nm) amorphous electron blocking layer (i.e., "buffer layer") between the absorber and the
transparent conducting oxide. Such an approach has been successfully employed by the
silicon-based heterojunction with intrinsic thin layer (HIT) devices (Figure 3.1),
demonstrating a high open circuit voltage (Voc) [42]. The layer can be thin enough to
avoid significant optical absorption (current loss) or band perturbation of the absorber
(voltage loss).
38
a-Si:H(n)
1H
a-Si:H(i)
-1-ITO
I0.-
W -2W-3-3-
250.03
5 67
0.1
2
250.06
250.09
Position [pmn]
3
4 5 67
Position [pm]
Figure 3.1. Band structure at the junction and back contact (inset) of a HIT solar cell with
a p-type Si wafer. After [43].
Previous studies of electron blocking layers in more established materials indicate
that it is essential to "tune" the conduction band offset (CBO), to avoid current losses
(stemming from too high CBO) or voltage losses (stemming from a negative CBO)
(Figure 3.2) [44]. Such tunability has been shown possible using ternary compounds,
whereby the ratio of two cations (anions) typically modifies the conduction (valence)
band position [45, 46]. Judicious composition selection of this ternary compound allows
one to "tune" the conduction band energy of the buffer layer, thus serving as an effective
blocking layer for electron diffusion from the transparent conductive layer into the
absorber layer, thus reducing interfacial recombination and increasing Voc.
39
CIGS
-
Window
Layer
Defect Level
-
E
Ev
Injection Hole
~~~w
OVC no
V
Defect Level
(b) below
(a) above
Figure 3.2. Schematic energy-band alignment diagram of a CIGS solar cell when the
conduction band of window layer is (a) above and (b) below that of CIGS layer. After
[47].
In this chapter, an ultrathin amorphous zinc tin oxide (a-ZTO) film is introduced
as an electron blocking layer to inhibit recombination at the Cu 20-ZnO interface.
Amorphous metal oxides are a new class of semiconductors recently highlighted in
transparent electronics due to their superior electronic transport properties with high
optical transparency [48, 49]. Specifically, a-ZTO is a non-toxic and scalable material
due to elemental abundance, and has shown a field-effect mobility up to 13 cm2. V-1-swith good thermal stability [50]. Recently, ZTO has shown potential as a Cd-free buffer
layer to CIGS solar cells exhibiting a comparable performance to devices with CdS
buffer layers [51]. In this work, the potential utilization of atomic layer deposited a-ZTO
films in all-metal-oxide thin-film solar cells is investigated. By controlling the atomic
composition of the a-ZTO films, optical and electrical properties of the film and its band
alignment with Cu 20 and ZnO layers can be tuned precisely. A sizable enhancement of
Voc and fill-factor (FF) is demonstrated by inserting the ultrathin a-ZTO buffer layer,
resulting in a power conversion efficiency of 2.65 %.
40
3.2. Amorphous zinc-tin-oxide buffer layer for Cu 2 0-based
solar cells
Cu 2 0-based thin-film solar cells are fabricated in the substrate configuration with
ultrathin a-ZTO buffer layers. A schematic structure and electron microscopy images of
the devices are shown in Figure 3.3. 2.5-pim-thick Cu 2O films were deposited by an
electrochemical method on patterned Au electrodes. Grains in the Cu 2 O films have (111)
preferred orientation when deposited by lactate solution [52], which resulted in highly
textured top surface morphology (Figure 3.3(b)) exhibiting reduced optical reflection
desirable to photovoltaic applications. 5-nm-thick a-ZTO buffer layer and 80-nm-thick
Al-doped ZnO (AZO) layer were sequentially deposited by ALD at 120 'C. Crosssectional images of the solar cells taken with high-resolution transmission electron
microscopy (HR-TEM) show the individual layers clearly (Figure 3.3(d)).
The
amorphous nature of the ZTO buffer layer is observed in contrast to crystalline Cu 20 and
AZO layers. It is confirmed that the ALD process can enable highly conformal coverage
of a-ZTO and AZO films on the textured Cu 2 0 surfaces without pin-holes (Figure 3.3(c)
and (d)). The AZO films exhibited an electrical resistivity of 5.9 x 10-3 0 cm. Al grids
were deposited for top-side electrodes. As a control, a baseline device containing an
undoped ZnO buffer layer grown by ALD at 120 'C with the same thickness was also
fabricated.
41
b
Al
AZO (80 nm)
a-ZTO (5 nm)
CU0(2.5
-
pm)
Au
Substrate
c
d
-AZO
-a-ZTO
-Cu 20
Figure 3.3. (a) A schematic structure of the substrate-type Cu 2O-based solar cells with aZTO buffer layer. (b) Cross-sectional SEM image of the device, exhibiting a highly
textured top surface stemming from (111) preferred growth of Cu 2 0. Scale bar, 1 pLm. (c)
Magnified SEM image near the junction interface. Conformal coating of an ALD-oxide is
demonstrated. Scale bar, 500 nm. (d) HR-TEM image near the junction interface.
Separated amorphous ZTO layer from crystalline Cu 20 and AZO layers is shown. Scale
bar, 5 nm.
42
A newly developed cyclic amide of tin (1,3-bis(1,1-dimethylethyl)-4,5-dimethyl(4R,5R)-1,3,2-diazastannolidin-2-ylidene)Sn(II)) as a Sn precursor [53] and diethylzinc
as a Zn precursor enabled the low temperature ALD of a-ZTO thin-films [50]. Atomic
composition of the a-ZTO buffer layer was varied by choosing the ratio of ALD subcycles for ZnO and SnO 2 deposition to be 3:1, 1:1 or 1:3. Atomic concentrations of Zn,
Sn, and 0 in the films were measured by Rutherford backscattering spectroscopy (RBS).
Figure 3.4 (a) shows the RBS spectra for three different a-ZTO films with different ZnO
to SnO 2 sub-cycle ratios, grown on glassy carbon substrates pre-treated by UV-ozone. No
obvious peaks except for Zn, Sn, and 0 from the films, and C from substrates were
observed. The Zn to Sn ratio was measured to be 1:0.27, 1:0.59, and 1:1.8 for the films
with ZnO to SnO2 sub-cycle ratios of 3:1, 1:1, and 1:3, respectively. Oxygen
concentrations were measured to be 13 - 19 % higher than the values calculated
assuming stoichiometric ZnO and SnO2. Similar oxygen-rich composition was observed
for pure Sn0
2
ALD using hydrogen peroxide (H2 0 2 ) as an oxidant [50]. The
microstructure of the films was evaluated by glancing-angle X-ray diffraction (GAXRD)
as shown in Figure 3.4 (b). Diffraction peaks from pure undoped ZnO film indicate
crystalline hexagonal ZnO. On the other hand, all ZTO films were amorphous, which is
consistent with TEM results. Broad peaks at around 340 are observed for the three ZTO
films, which is characteristic of a-ZTO [54]. Formation of amorphous films resulted from
the different crystal structures of each pure binary oxide; the crystal structure of ZnO and
Sn0 2 are wurtzite and rutile, respectively [55, 56].
43
energy (MeV)
a
0
C
0
500
1000
channel
1500
2000
b
C
20(o)
50
Figure 3.4. (a) RBS spectra of the a-ZTO films with three different Zn to Sn ALD cycle
ratios. The carbon signal comes from the glassy carbon substrates used for the
measurements. (b) GAXRD spectra of the ZTO films. ZnO is included as a reference. All
ZTO films are amorphous.
44
Optical and electrical properties of the a-ZTO films were tuned by controlling the
Zn to Sn ratio. All films exhibit high transmittance in the visible wavelength range and
higher transmittance than crystalline ZnO in the UV range (Figure 3.5 (a)). The Tauc
model is normally used to find fundamental bandgap values for amorphous metal-oxide
semiconductors [57, 58]. The bandgaps of a-ZTO films are determined from the plot of
(ahv)12 as a function of photon energy shown in Figure 3.5 (c). As the Zn to Sn ratio of
the film decreases from 1:0.27 to 1:1.8, the bandgap of a-ZTO films gradually increases
from 3.12 to 3.37 eV. Hall measurements revealed an n-type conductivity with electron
density of 3.5 x 1016 cm-3 and mobility of 3.7 cm2. V-1 s-I for the a-ZTO film with a Zn to
Sn ratio of 1:0.27, while the other a-ZTO films with higher Sn contents exhibited
resistivity higher than 2 x 103 Q-cm, with Hall voltages too small to measure. Low
carrier concentrations of the grown films appear to be due to the usage of the strong
oxidant H2 0 2 , reducing oxygen vacancy concentrations in the films. Similar behavior was
also reported in sputtered a-ZTO layers [59, 60]. Undoped ZnO as a control buffer layer
exhibited a resistivity of 2.0
x
10-2 Q-cm, with an electron density of 1.8 x 1019 cm-3 and
a mobility of 1.7 x 101 cm2-V- -s-1.
45
a
101
5
105
a-ZTO
- 1:0.27 (Zn:Sn)
1:0.59
1:1.8
-- ZnO
o 10
C
2-4
1012
3
4
5
photon energy (eV)
b
5
- 1:0.27 (Zn:Sn)
4 - 1:0.59
- 1:1.8
-
3 -
0
2
3
4
photon energy (eV)
5
Figure 3.5. (a) Optical absorption coefficient of atomic layer deposited a-ZTO and ZnO
thin-films. Increasing Sn contents in a-ZTO films reduces high-energy photon absorption.
(b) Bandgap estimation from the plot of (ahv)m as a function of photon energy.
46
3.3. Device characterization and analysis
3.3.1. Current density - voltage characteristics under illumination
Incorporating an a-ZTO buffer layer into the solar cells led to a strong
enhancement of the power conversion efficiency. Current-voltage characteristics of the
solar cells with a-ZTO buffer layers and a baseline device under AMI.5G illumination
(100 mW-cm~2 ) are shown in Figure 3.6. Characteristic device performance properties are
summarized in Table 1. These devices show a strong dependence of Voc on the
incorporated buffer layer material, while short-circuit current densities (Jsc) remain 7.3 7.5 mA-cm-2.2 The Jsc matches the integrated value of measured external quantum
efficiency (EQE) data with the AMI.5G spectrum. The a-ZTO buffer layer with a Zn to
Sn ratio of 1:0.27 exhibited the highest Voc of 0.553 V with a fill-factor of 65.0 %,
resulting in a power conversion efficiency of 2.65 %. As the Zn to Sn ratio decreases to
1:1.8, the Voc decreases to 0.406 V which is lower than that of the baseline device.
Buffer layers with Zn to Sn ratios higher than 1:0.27 didn't further increase power
conversion efficiencies.
47
0
-2
-4
a)
-6
I
-8
0.0
0.2
0.4
0.6
bias (V)
Figure 3.6. Current-voltage characteristics under 1-sun illumination (AMI.5G spectrum)
of the devices with different buffer layers.
Table 3.1. Photovoltaic characteristics under 1-sun illumination
buffer layer
Zn:Sn
Voc
[mV]
Jsc
[mA cm-2]
[%]
efficiency
[%]
FF
ZnO
-
425
7.35
58.3
1.82
a-ZTO
1:0.27
553
7.37
65.0
2.65
a-ZTO
1:0.59
497
7.50
64.0
2.39
a-ZTO
1:1.8
406
7.26
58.2
1.72
48
3.3.2. Band-alignment characterization
To investigate the effect of the a-ZTO buffer layer, its band alignment to the
Cu 2 0 layer was characterized by photoelectron spectroscopy following the procedure
outlined by Waldrop et al. [61]. An X-ray (Al-Ka) photon source was used to measure
the binding energies Cu and Zn core levels with respect to the valence band maximum
energy of Cu 2 0 and a-ZTO bulk film samples. Two-layer samples that consist of 2.5-pmthick Cu 2 0 films covered by ~2-nm-thick a-ZTO films were prepared to measure their
valence band level alignments. Tauc gap values were added to estimate conduction band
positions. The valence band offset (AE,,) in the heterojunction was calculated by using
the following equation:
AEB
= (E" 2 -y)(2-
-
-
Ea-
-
(EU 2 P - EfB"),
(4)
where (E("-Z7"(IU2( - Ea- ucu2o) is the binding energy difference between the Cu-2p and
Zn-2p core levels measured at the Cu 20/a-ZTO interface. (E a-ZT0
(E((,
-
-
E,"4'") and
Ev"20 ) are the positions of the core level peaks referenced to the valence band
maximum (VBM). Subsequently, the conduction band offset (AEcaB)
can be determined
by using the following relation:
AE
where E'
"20
=A
-(Ea-ZTO
(5)
- E( 2'),
and E"-Z"0 are the bandgaps of Cu 2 0 and a-ZTO, respectively.
49
Cu 2p1 /2
C
932.47 eV
VBM
960
950
940
930
6
4
2
-2
binding energy (eV)
Figure 3.7. XPS spectra of electrochemically deposited Cu 20 thin-films. Cu 2P3/2 core
level spectra show only Cu2+ states with a relative energy level of 932.47 eV reference to
VBE.
50
ab
a
Cd
ZnOZn
Zn:Sn
.2
A-
:1
Zn:S
1:0.27
1:0.27-
1:0.59
1:0.59_
1:1.81:1.8
1050
1030
1010
8
6
4
2
0
binding energy (eV)
Figure 3.8. XPS spectra of (a) Zn core level and (b) valence band for atomic layer
deposited a-ZTO and ZnO thin-films. Zn 2P3/2 core level peaks of ZnO and a-ZTO with a
Zn:Sn ratio of 1:0.27, 1:0.59 and 1:1.8 exhibit relative energy levels of 1019.02, 1019.44,
1019.57, 1019.47 eV reference to their VBEs, respectively. In this analysis, small humps
near valence band edges of a-ZTO are attributed to tail states in bandgap [62], which
originate from increased SnO2 contents. The hump near VBE of SnO 2 is often observed
from photoelectron spectroscopy when a high energy photon source is used [63], while
the states are not expected from theoretical calculations [64, 65].
51
CdZnO
-
Zn:Sn
r>
1:0.27-
.2
1:0.59
8
6
4
2
0
binding energy (eV)
Figure 3.9. UPS spectra of the valence band for atomic layer deposited a-ZTO and ZnO
thin-films measured using a He-I photon source (hv = 21.2 eV). The inteinsities of small
humps near valence band edges of a-ZTO are smaller than the spectra from XPS.
52
Cu 2p.
28 eV
_~Zn 2p
89.
24 eVC
189.
('3
p,
89. 16 eV
C
1:0.59
1:1.8
1050
1040
-189.
1030
89. 26 eV
1020
960
950
940
930
binding energy (eV)
Figure 3.10. XPS spectra of Zn and Cu core levels from 2-nm-thick ZnO and a-ZTO
films on Cu 2O thin-films.
Figure 3.11 shows the band alignments of a-ZTO overlayers on Cu 2 0 films. The
Cu 2 O/ZnO interface yields a cliff-type conduction band offset (AECB) of -1.47 ± 0.2 eV,
comparable to previous reports [66, 67]. The a-ZTO layers create a barrier in conduction
band which prevents electrons in the AZO layer from recombining with holes in the
Cu 2 0 layer, while allowing photo-generated electrons in the Cu 2 0 layer to be collected in
the ZnO layer. This improvement in conduction band alignment is proposed to cause the
observed increase in Voc of the a-ZTO incorporated devices.
53
SE
-0.96± 0.2 -0.86
zw (ev)
=f-
0.2 -1.22±0.2
-1.52
0.2_
LU
______(1:1.8)
Cu20
(20.19
(1:0.27)
a-ZTO (Zn:Sn)
ZnO
Figure 3.11. Relative alignments of conduction band (CB) and valence band (VB) for aZTO and ZnO overlayers to Cu 2O thin-films investigated in this work. The values were
measured by XPS technique.
54
3.3.3. Capacitance-frequency (C-f) characteristics
Capacitance-frequency (C-]) characteristics at room temperature are measured to
investigate the decrease in the Voc of the Sn-rich a-ZTO buffer layer (Figure 3.12). The
frequency dependence on diode capacitance is used to measure the density and energy
level of trap states present in the depletion region [68]. At high frequencies, the device
behaves like an insulator due to dielectric freeze-out and exhibits a geometric capacitance
(Cg). The Cg was calculated using the relationship of Cg = eAlt, where E is the dielectric
constant, A is the device area and t is the thickness of Cu 20 layer. At frequencies near 1
MHz, capacitances of all devices converge to 4.5 - 4.7 nF cm- 2 , close to Cg of the device
(~2.7 nF-cm-2). At low frequencies, the capacitance of all devices plateau to a depletion
capacitance (C), which is affected by charging and discharging of interfacial and bulk
defect levels present in the depletion region [69]. Due to identical geometry and
fabrication processes across the devices except for 5-nm-thick buffer layer materials with
high resistivity, the relative change in Cd is attributed to the defects originated from the
buffer layers. The highest efficiency device exhibits the lowest Cd, indicative of lower
defect densities. On the other hand, the lowest Voc of the device with Sn-rich (1:1.8) aZTO buffer layer exhibits the highest Cd possibly due to higher densities of subgap states.
55
,,60
Zn:Sn
Cd
U-
+1:0.27
-1:0.59 -
c40
-1:1.8
.
20
C',
08
--
0
-
- -
103
-
-
- -
-
-
--
-
-
--
-
-
105
i04
--
-
-
--
-
-
10,
frequency (Hz)
Figure 3.12. Capacitance - frequency characteristics of the devices with a-ZTO buffer
layers at room temperature.
56
3.3.4. Current density - voltage characteristics under dark condition
The Voc of a heterojunction solar cell is strongly affected by the density and
energy levels of interface states, which increase the dark saturation current by promoting
interfacial recombination [70]. The Voc can be simplified as:
'J
-nkT
Inr w +1,
q
JO
(6)
V1c ~k
where n is the diode ideality factor, k is the Boltzmann constant, T is temperature, q is the
electron charge and Jo is the dark saturation current density. Recombination of carriers
through interfacial traps increases Jo, thereby reducing the Voc. Current-voltage
characteristics of two devices with ZnO and a-ZTO (Zn:Sn = 1:0.27) measured in the
dark condition are plotted on a semi-log scale, as shown in Figure 3.13. Addition of the
5-nm-thick a-ZTO buffer layer reduces J significantly, resulting in lower dark saturation
current densities than the control device by a factor of ~ 40 under forward bias, while
both devices show similar ideality factors of 1.5. The a-ZTO buffer layer introduces an
effective interfacial recombination barrier that impedes movement of electrons to a
defect-rich interface as illustrated in the inset of Figure 3.13. Only 5 nm is a sufficient
thickness for the buffer layer to block tunnelling while minimizing additional series
resistance.
57
10~2
CN
0U0ZZnO
0
a)
a-ZTO
U8
o
e*AE
0.3 eV
a-ZTO
100
-0.4
-0.2
0.0
0.2
0.4
0.6
bias (V)
Figure 3.13. Effect of a-ZTO (Zn:Sn = 1:0.27) buffer layer on dark current-voltage
characteristics of the devices. Inset schematics at top and bottom show electronic-band
structures of the devices with ZnO and a-ZTO buffer layers, respectively. Grey areas
indicate defect-rich interfaces with deep-trap states. Red arrow lines represent interfacial
recombination paths for electrons (filled circles) from the ZnO layer. The a-ZTO buffer
layer impedes electron movement to the interface where holes (open circles) are provided
from the Cu 2 O layer, thus reducing Jo.
58
3.4. Optical Simulation
Possible further Voc improvements by enhanced Jsc are studied. Due to the low
carrier density of electrochemically deposited Cu 20 thin-films (p
=
1013
-
1014 cm-3), a
fully depleted Cu 20 layer is expected in the device [71]. However, the photo-generated
carrier collection probability profile is empirically modeled to have a drift-dominated
region near the junction (depth (w) of 0.27 im) and a diffusion dominated area where the
collection probability decays exponentially with a minority carrier diffusion length (LD)
of 0.16 pim [72]. Optical absorption in the device is simulated by using the finite
difference time domain (FDTD) method. The device was modeled as a two-dimensional
geometry, determined by atomic force microscopy, under incident light with a transverse
magnetic mode polarization. The calculated spatial absorption profile for 500 nm
wavelength light is shown in Figure 3.14.
photon absorption (cm-3 s- nm-4)
Figure 3.14. Simulated optical absorption profile in the Cu 2 0 layer for 500 nm
wavelength light with the intensity of that in the AM1.5G spectrum. Scale bar (white), 1
pm. Courtesy of J. P. Mailoa.
59
EQE curves are calculated by weighting the generated minority carriers with the
spatial collection probability functions with varying LD (Figure 3.15). In the FDTD
simulation, the device was modeled as a textured two-dimensional layer stack. The Cu 20
texture was characterized using the AFM and had a feature size of approximately 1 tm.
To properly incorporate the randomness of this texture, the FDTD simulation area was
chosen to be 10 ptm wide. Incident light with TM polarization was then used to simulate
the light propagation with 300 - 650 nm wavelength range within this structure. FDTD
Solutions software (Ver. 7.5, Lumerical, Inc.) was used to calculate the electromagnetic
field profile inside the Cu 20 as a function of space and frequency. Using this
electromagnetic field profile, the divergence of the Poynting vector was calculated to
obtain the power absorption per unit volume as a function of space and frequency inside
the Cu 20. The EQE of the device was further simulated by using the optical absorption
profile. A collection probability function for photo-generated carriers was modeled as
shown in Figure 3.15. EQE was calculated using the equation:
EQE (A)
f
s (x,y, A) CP(x,y) dxdy
E(OTAL
((7)
where PABS(x, y, X) is the spatial profile for power absorption per unit volume for a
specific wavelength, CP(x, y) is the spatial profile for carrier collection, and PTOTALQ) is
the total incident power per unit length for a specific wavelength.
60
100
0
100% collection
80 -60L2m2000 nm
C-
40-
20-
0
0
30010
500
1000
1500
2000
2500
3000
depth from junction (nm)
Figure 3.15. Modeled collection probability profiles for photo-generated carriers in the
Cu 20 layer.
Using this method, the EQE curves as a function of carrier collection length (Lc =
LD +
w) were generated and measured EQE of the best performing device was plotted, as
shown in Figure 3.16. Lc of the a-ZTO device is estimated to be 500 nm approximately,
by comparing the EQE curves from the FDTD simulation and the actual measurement.
Improving carrier collection with longer LD and light trapping will be one of the next
routes to achieve devices with higher efficiencies.
61
,-%100
*
R-
100% Col.
'",-
.
e
80 -/
\
2000 nm
-1000 nm
500 nm
300 nm
E E6-to
measured
40-
E 20
U)
0
300
400
500
600
700
wavelength (nm)
Figure 3.16. Effect of the minority carrier diffusion length of Cu 20 on EQE. Colored
lines represent calculated EQE from optical simulation with a carrier-collection
probability profile with increasing Lc and an ideal 100 % collection case. Dotted black
line represents measured EQE of the a-ZTO (Zn:Sn = 1:0.27) buffer layer incorporated
device.
62
3.5. Experimental details
3.5.1. Atomic layer deposition of a-ZTO and AZO thin-films
Amorphous-ZTO, AZO, and undoped ZnO thin-films were synthesized by ALD
using a custom-built hot-wall reactor with a chamber volume of 0.627 L at 120 'C.
Cyclic
amide
of
tin
(1,3-bis(1,1-dimethylethyl)-4,5-dimethyl-(4R,5R)-1,3,2-
diazastannolidin-2-ylidene)Sn(II)) and diethylzinc (Sigma Aldrich) were used as Sn and
Zn precursors, respectively [50, 53]. A 50 wt.% hydrogen peroxide (H2 0 2 ) solution was
used as a common oxidant for a-ZTO growth. To deposit a-ZTO films, an ALD
supercycle scheme with ZnO and Sn0 2 sub-cycles was employed. Three different
ZnO:SnO2 sub-cycle ratios of 3:1, 1:1, 1:3 were used to find the optimized composition
as a buffer layer in the Cu 20-ZnO heterojunction. Their nominal growth per one
supercycle were 3.1, 1.6 and 4.1
thin
films
were
also
A, respectively,
synthesized
at
120
measured by X-ray reflectivity. AZO
'C
by
ALD
using
diethylzinc,
trimethylaluminum, de-ionized water as Zn, Al precursors and an oxidant, respectively.
Optimized A12 0 3 doping ratio to ZnO (1 cycle of A12 0 3 after every 19 cycles of ZnO)
was selected to obtain the lowest resistivity (5.9 x 10-3 Q-cm) of the films. The nominal
film growth per one supercycle, which consists of 19-ZnO and 1-Al 2 0 3 sub-cycles, was
32 A. Undoped ZnO as a control buffer layer was grown at 120 'C without Al doping
with de-ionized water as an oxidant.
63
3.5.2. Thin-film solar cell fabrication
An Au bottom electrode (200-nm-thick, 3.2 cm 2 area) with a 5-nm-thick Ti
adhesion layer was deposited on a 1 x 1 inch 2 fused silica by e-beam evaporation. A 2.5pim-thick Cu 2 0 film was deposited on the Au film selectively at 40 'C by the
galvanostatic electrochemical method [73]. A lactate-stabilized copper sulphate aqueous
solution was prepared with 3 M lactic acid (Sigma Aldrich), 0.2 M cupric sulfate
pentahydrate (CuO 4 S -5H2 0, Sigma Aldrich) and de-ionized water (18.3 MQ cm, Ricca
Chemical) was prepared and 2 M sodium hydroxide (NaOH, Sigma Aldrich) aqueous
solution was added to adjust the pH of the solution to 12.5. All reagent grade chemicals
were used and the solution was filtered and stirred thoroughly. A constant current density
of 0.23 mA-cm-2 was applied by a Keithley 2400 sourcemeter with a Pt counter electrode
to grow Cu 20 films. The Cu 2 0 film surfaces were rinsed with de-ionized water.
Amorphous-ZTO and AZO layers were deposited by ALD, followed by Al electrode
deposition. The Al electrodes (300-nm-thick) were deposited by e-beam evaporation with
a grid spacing of 1 mm defined by a shadow mask. 3 x 5 mm 2 cell area was defined by
photolithography and wet etching with nitric acid solution.
3.5.3. Characterization
The microstructures of the films were characterized by JEM-2100 HR-TEM
(JEOL) and GAXRD using a PANalytical X'Pert Pro diffractometer with Cu-Ka
radiation (o> = 0.90). Surface morphologies of the grown films were analyzed by Ultra 55
FESEM (Zeiss). Optical properties of the films were measured by a U-4100 UV-VIS
spectrophotometer (Hitachi) and a VASE spectroscopic ellipsometer (J.A. Woollam Co.,
64
Inc.). Electrical properties of the films were characterized by Hall effect measurement
using the van der Pauw configuration and magnetic field 0.75 T. The XPS was measured
by PHI VersaProbe II (Physical Electronics) and AXIS Nova (Kratos analytical, in KBSI
Jeonju). Samples were etched by Ar ion beam to remove carbon contaminants on the
surface. The current-voltage and the capacitance-frequency characteristics of the device
were measured by Agilent 4156C and Keithley 4200 semiconductor characterization
systems. The standard 1-sun illumination was generated by a Newport Oriel 91194 solar
simulator with a 1300 W Xe-lamp with AM1.5G filter and a Newport Oriel 68951 flux
controller calibrated by an NREL-certified Si reference cell equipped with a BG-39
window. EQE of the device was measured by QEX7 (PV Measurements, Inc.) calibrated
by a NIST-certified Si photodiode. The top surface morphology of the device was
measured by atomic force microscopy using MFP-3D SA (Asylum Research).
65
3.6. Conclusions
In summary, a strong enhancement of power conversion efficiency of Cu 2 0-based
all-metal-oxide thin-film solar cells is successfully demonstrated by introducing an
ultrathin a-ZTO buffer layer. Tunable electrical and optical properties of a-ZTO films are
enabled by an ALD process, controlling the zinc-to-tin ratio. Detailed measurements
reveal that the a-ZTO buffer can reduce interfacial recombination by acting as an electron
blocking barrier and decreasing the density of interface defects. Due to highly conformal
coverage and precise thickness control by ALD, only a 5-nm-thick buffer layer is
necessary to improve junction quality without sacrificing carrier transport and optical
transmission.
66
CHAPTER 4.
P-Type Doping of Cu 2O Thin-
Films by Nitrogen for a Hole-Transporting
Layer
67
4.1. Introduction
To develop high-efficiency photovoltaic
devices, proper band alignment
throughout the entire device structure is necessary to minimize energy losses from nonideal interfaces. Most materials for light-absorbing layers are p-type, possess low carrier
density, and often have deep valence band edge positions, which can result in
unfavorable band alignment and a highly resistive Schottky barrier formation between the
absorber layer and metal or transparent conducting oxide electrode [74, 75]. To mitigate
absorber-electrode interface problems, hole-transporting layers have been introduced at
the back contact of various device types [76-79]. The layers are required to have high
carrier density and proper band position to reduce contact resistance and create an
"electron-reflecting" back surface field to promote carrier collection. Various transitionmetal oxides including molybdenum oxide (MoO 3 ) and vanadium oxide (V 2 0 5 ) have
shown promising properties as hole transporting layers due to their high work functions
(<) [78, 80, 81]. However, their n-type natures and deep-lying states inhibit holeselective and electron-blocking functions [82].
P-type doping of metal oxides is generally challenging, due to self-compensation
and dopant solubility limits [83, 84]. However, theoretical calculations predict that a few
intrinsically p-type oxides can be extrinsically doped p*-type (Figure 4.1) [85]. Cuprous
oxide (Cu 2 O) is one such intrinsically p-type semiconductor with conduction band edge
(Ec) and valence band edge (Ev) positions at 3.2 and 5.2 eV from the vacuum level,
respectively [86]. With the addition of extrinsic acceptors, the Fermi level (EF) can be
further shifted toward Ev, without lowering the formation energies of compensating
defects to negative values [85]. Various dopants including N, Si, Ge, and some transition
68
metals have been tested to increase the p-type conductivity of Cu 2O thin-films [87-90].
Among those elements, nitrogen is expected to substitute oxygen with a high solubility,
given their similar ionic radii. Previously reported nitrogen doping has been shown to
reduce electrical resistivity of Cu 2O by a factor of ~100 [90]. Thus, I hypothesize that it is
possible to tune the electrical and optical properties of a nitrogen-doped Cu 2O (Cu 2 0:N)
layer and to incorporate it into a thin-film photovoltaic device as a hole-transporting layer
to reduce back-contact resistance. In principle, such a layer could widen the range of
possible back contact materials suitable for solar-cell device fabrication, by reducing the
severity of the so-called "roll-over" (a current saturation at high forward bias) [91].
0
-1
-2
-3
-4
21
0)
C
-5
wU-6
-7
-8
-9
Figure 4.1. Valence and conduction band edge positions of various oxides with their
doping limits, showing dopable and undopable cases. After [85].
69
In this chapter, the potential of a nanometer-scale Cu 20:N film as a p-type hole
transporting layer for photovoltaic devices is demonstrated. By controlling the nitrogen
content in the film, carrier density and optical transmittance are tuned to create a low
contact resistance interface with minimal optical absorption. Cu 20-based metal oxide
heterojunction thin-film solar cells are fabricated and a 20-nm-thick Cu 2 0:N hole
transporting layer is inserted between a silver back contact and a Cu 2 0 absorber layer.
Silver is chosen as a back electrode, because its high reflectance around 550 nm enhances
light trapping near the bandgap of Cu 2 0 and the peak of the solar spectrum, yet its
relatively small work function creates an inherently resistive back contact with intrinsic
Cu 2 0. By increasing the carrier density of the Cu 20:N layer, a tunnel junction is formed
that reduces the contact resistance. I demonstrate that the addition of a Cu 2 0:N layer
results in a sizeable fill-factor (FF) and power conversion efficiency enhancement
relative to control samples without the layer.
70
4.2. Structural and chemical properties
Reactive magnetron sputtering was used to incorporate nitrogen into Cu 2 0 thinfilms. Reactive sputtering of metallic copper using 02 and N2 as reactive gases has been
widely used for Cu 20 and copper nitride (Cu 3N) thin-film depositions [92, 93]. A range
of substrate temperatures during film growth ranging from 130 to 390
C was
investigated; 230 C provided the lowest electrical resistivity of Cu 2 0:N films (Figure
4.2). Substrate temperature higher than 230 C resulted in higher resistivities due to lower
hole densities. To control nitrogen content, the N 2 flow rate was varied from 0 to 6 sccm,
while the Ar and 02 flow rates were fixed to 40 and 11 sccm for high-purity Cu 2 0
deposition. Figure 4.3 shows the nitrogen content in the 0.6 pIm-thick Cu 2 0 films
measured by secondary ion mass spectrometry (SIMS) with depth profiling. A controlled
amount of nitrogen was implanted into an undoped Cu 20 thin-film sample and a Si wafer
for calibrating the measured data. As the N2 flow rate increased to 6 sccm, the nitrogen
concentration ([N]) increased in an approximately linear fashion to 1.7 at.% (1.3 x 1021
cm- 3). If all the nitrogen doping were substitutional at oxygen sites, this doping level
would correspond to replacing about 5 % of the oxygen atoms by nitrogen atoms.
71
a
60
U
E
c 40
.-U
20
0
b
,
CV?
100
300
200
growth temperature (0C)
400
100
300
200
growth temperature (OC)
400
10
E
Q)
o16
10
10
C
20
C4)
10
..- '
0
100
300
200
growth temperature (C)
400
Figure 4.2. Film growth temperature effects on (a) electrical resistivity, (b) carrier density,
and (c) mobility of 0.6-pm-thick Cu 2 0:N films measured by Hall effect measurements at
room temperature. Dotted lines are to guide the eye. N 2 flow rate was maintained at 1
sccm. All samples exhibited p-type conductivity.
72
2 (at.%)
nitrogen concentration, [N]
11
(
c-1.
1.0
-0.5
0
0
2
4
N2 flow rate (sccm)
6
0.0
Figure 4.3. Nitrogen concentrations in Cu 2 0:N films measured by SIMS. A typical
calibration error range is up to ± 30 %.
Structural and chemical properties of the host material (Cu 2 0) were maintained
for all nitrogen doses in our study. X-ray diffraction (XRD) measurement shows (111)
and (200) peaks from the cubic structure of Cu2 0 for undoped and doped ([N] = 1.7 at.%)
films (Figure 4.4). The Cu 2 0:N film exhibited reduced peak intensities as [N] increased
in the film, indicating lower crystallinity than the Cu 2 0 film. Peaks from metallic copper,
cupric oxide (CuO), and Cu3N were not detected in the measurement. X-ray
photoelectron spectroscopy (XPS) was used to investigate chemical states of elements in
the films. Cu 2P3/2 and 2pv/2 peaks at 932.8 ± 0.2 and 952.7 ± 0.2 eV shown in Figure 4.5
(a) indicate the same Cu oxidation state (Cue) for Cu 20 and Cu 20:N films. CuO peaks
from Cu2+ states in 940 - 945 eV [94] were not detected. Due to the low sensitivity of
73
XPS to light elements, the signal from nitrogen was difficult to resolve. A small peak
position from the nitrogen core level in the Cu 2 0:N film ([N] = 1.7 at.%) (Figure 4.5(b))
was measured at 397.1 ± 0.2 eV, close to nitrogen signal from Cu3 N (397.4 eV) [95]. The
peak position suggests that a considerable fraction of nitrogen in the Cu 2 0:N film bonds
with Cue, as a substitution for oxygen.
i
'
I
'
I
'
|
1 1
I
-0
N]= 1.7 at.%
U)
4-'
I
20
30
40
50
60
I
70
I
80
0
90
20(*)
Figure 4.4. XRD spectra of Cu 2O and Cu 2 0:N films, indicating two main peaks from
Cu 2 0 (111) and (200). Inset SEM images of the Cu 2 O (bottom) and Cu 2 0:N (top) films.
The both scale bars represent 300 nm.
74
a
960
950
940
binding energy (eV)
930
b
Nis
on
.
0ov
Mundoped
U%
soI
410
405
400
395
binding energy (eV)
390
Figure 4.5. XPS spectra of (a) copper and (b) nitrogen core levels of Cu 20 and Cu 2 0:N
films. A grey arrow indicates a small nitrogen peak at 397.1 ± 0.2 eV.
75
4.3. Electrical and optical properties
The effects of nitrogen-doping on electrical and optical properties are
characterized. In Figure 4.6, the measured resistivity exhibits a strong reduction as more
nitrogen is doped into the Cu 2 0 films. While undoped Cu 20 film exhibits a resistivity of
1.4
x
102
Q.cm,
the highest nitrogen dose (1.7 at.%) reduces the resistivity down to 1.8
x
10-1 Q -cm, which is the lowest value among reported Cu 20:N thin films to the knowledge
of the authors. Temperature-dependent Hall measurements were carried out to further
elucidate the carrier properties of these films. The films with [N] up to 1.2 at.% show ptype conductivities; the film with [N] = 1.7 at.% could not be measured as the Hall
voltage was very small due to its low mobility. Cu 2 0:N films exhibited reduced mobility
as [N] increased in the films (Figure 4.7), due to increased ionized-scattering-centers [92].
Figure 4.8 shows hole densities (p) increase to 1.1
x
1019 cm 3 at 300 K as the nitrogen
content increases to 0.62 at.%. The hole densities show strong dependence on
temperature. The carrier activation energies (EA) were estimated by fitting the data to a
compensated
semiconductor
model
with
a
low-temperature
approximation
(p<N -N)):
p =(2;m*kT / h2 )3/2 [(N/N))-1]exp(-EA/kT),
(8)
where m* is the effective mass, k is Boltzmann's constant, h is Plank's constant, NA is the
acceptor density,
ND
is the donor density [21, 92]. Cu 2O:N films with [N] higher than
0.06 at.% exhibit decreased EA of 0.12 eV (Figure 4.9). On the other hand, undoped
Cu 2 0 and lightly doped Cu 20:N ([N]= 0.06 at.%) films exhibited EA of 0.18 - 0.19 eV,
76
indicating that the dominating acceptor state moves close to the valence band at higher [N]
[88, 92].
77
103 .1
102
U
101
.5
.4-'
U
U
U
C,)
100
U
U
U
-
10111
0.0
0.5
1.0
1.5
2.0
[N] (at.%)
Figure 4.6. Effects of nitrogen doping on electrical resistivity of Cu 2 0:N films measured
by four-point-probe.
10
8C,)
[N]/at.%
0.06
018
undoped -
*0.40
0.56U
6 - . 0.62
* 0.73
.
=
4 -- 1.22
2
0 -owa&
100
N
200
300
temperature (K)
400
Figure 4.7. Temperature-dependent Hall mobility of Cu 2 0 and Cu 2 0:N films.
78
1020
[N](at.%)
0.73 , ,1.2
0.62 .
-
MENNEN
0.56
108
0.40
*
MNo
0.18
0.06"
10
2
3
4
5
1
1000/T (K )
6
7
Figure 4.8. Effects of nitrogen doping on carrier densities of Cu 2 0:N films: Temperaturedependence of carrier (hole) density determined by Hall effect measurements.
0.2
U
U
C
U
a)
C
0
U
U
0.1
0.0
I
0.0
I
I
0.5
[N] (at.%)
1.0
1.5
Figure 4.9. Carrier (hole) activation energies of Cu 2 0:N films estimated from the
measured carrier density in the sample temperature range of 200 - 330 K.
79
As a result of the increased carrier density, carrier transport between Cu 20:N and
the Ag electrode can be enhanced. Metals normally exhibit unfavorable Schottky barriers
at metal/Cu 2 0 interfaces except high work-function metals (e.g. Au and Pt) [96, 97]. If
the hole density in the Cu 2 0 film exceeds 1019 cm-3 , the barrier's depletion width in Cu 20
will be reduced sufficiently to form a tunnel junction [98]. Temperature-dependent
specific contact resistance (pc) between Ag and Cu 2 0:N films was measured using the
circular transmission line model [99]. The film exhibited reduced pc of 2.9 x 10-4 Q cm 2
at room temperature as [N] increased to 1.2 at.%, while an undoped Cu 2 0 film exhibited
a pc of 1.9 x 10-2 Q-cm 2 . The temperature dependence of pc became weaker for the films
with [N] = 1.2 at.%, indicating that field-emission rather than thermionic-emission is
dominating the carrier transport at the interface [75].
To utilize Cu 20:N film as a hole-transporting layer, optical absorption by the
layer needs to be minimized by tuning its absorption properties and layer thickness. The
effect of nitrogen doping on the optical absorption of Cu 2 0:N films was examined. As
shown in Figure 4.10, the Cu 2 0:N film exhibited a higher optical absorption coefficient
(a) than the undoped Cu 20 film. In particular, a significant increase of a in the subgap
wavelength region (k > 600 nm) was observed for Cu 2 O:N films. Zhao et al. suggested
substitutional nitrogen doping could increase the absorption near 1.7 eV by firstprinciples calculations [100]. However, the Cu 20:N films in this study exhibited a broad
absorption peak shoulder near 0.6 eV, and the peak intensity increased with film nitrogen
content. Similar absorption characteristics were reported from Cu 2 0:N and Cu 3 N thinfilms and were attributed to the nitrogen-induced subgap states and free-carrier
absorption contributions, respectively [90, 101]. Structural defects from the lower
80
crystallinity of the Cu 20:N films could also increase a over all wavelengths. Due to the
increases in a, a 20-nm-thick Cu 2 0:N ([N] = 1.2 at.%) layer would be proper for a hole
transporting layer with a sub-gap absorption lower than 5 % (0.55 ptm < k < 2 tm).
1.0
)
C:
A)
0.5
0
0.0
0.5
1.0
1.5
wavelength (pm)
2.0
Figure 4.10. Effects of nitrogen doping on the optical absorption of Cu 2 0:N films
measured by a spectrophotometer with an integrating sphere.
81
4.4. Cu 2 0:N films as a hole-transporting layer in solar cells
Using a 20-nm-thick Cu 2 0:N layer ([N] = 1.2 at.%) as an interlayer between the
Ag electrode and a Cu 2 0 absorber layer, metal-oxide thin-film solar cells were fabricated.
Figure 4.11 shows a cross-sectional scanning electron microscopy (SEM) image of the
heterojunction solar cell. Cu 2 0, amorphous zinc-tin-oxide (a-ZTO), and Al-doped ZnO
(ZnO:Al) layers were used as a p-type light absorber, an n-type buffer, and an n-type
transparent conducting oxide layers, respectively. A 1.2-pm-thick Cu 2 0 layer was
deposited by the electrochemical deposition technique, as this produces a highly
roughened surface morphology with anti-reflection properties and improved photogenerated carrier collection. Conformal depositions of 5-nm-thick a-ZTO and 80-nmthick ZnO:Al layers were followed by atomic layer deposition. The Zn-to-Sn ratio in the
a-ZTO layer was adjusted to 1:0.27 to reduce interfacial recombination between Cu 20
and ZnO:Al layers (Chapter 3). As a comparison, a control device using the same
geometry, but without the Cu 20:N layer, was also fabricated.
To investigate the effect of a Cu 20:N layer in the device, current density - bias
voltage (J-V) characteristics were measured. Figure 4.12 shows J- V curves of two devices
under dark condition. The device with a Cu 2 0:N layer exhibited a normal rectifying
behavior. However, the control device exhibited a suppressed current density at bias
voltages greater than 0.5 V, while the current density below 0.5 V was comparable with
the Cu 2 0:N device. Such flat J-V characteristics under dark condition are often observed
in CdTe thin-film solar cells with a back-contact barrier. Those devices have been
explained by a two-diode model where a p-n junction and a leaky back-contact Schottky
junction are connected in series with opposite directions [74, 103]. The low current
82
density of the control device is expected to be limited by a back-contact barrier
originating from the low work function of the polycrystalline Ag electrode
(<Ag
= 4.3 eV
[104]). By inserting the Cu 2 0:N layer at the Ag/Cu 20 interface, a narrow back-contact
barrier can be formed in the Cu 2 0:N layer, allowing a high tunneling current through the
junction. The effects of a Cu 20:N layer on the illuminated J-V characteristics are shown
in Figure 4.13, and their parameters are summarized in Table 5.1. A strong enhancement
in FF and open circuit voltage (Voc) were observed, resulting in a power conversion
efficiency of 2.56 %.
ZnO:AI
a-ZTO
CU20
ICu20:N
Figure 4.11. A cross-sectional SEM image of Cu 2 0-based thin-film solar cell with a
Cu 2 0:N hole-transporting layer. Scale bar, 1 pm.
83
1.0
0.8
0.6
0.4
-D
(D
"D
0.2
0.01
-0.2
0.0
0.2
0.4
0.6
bias (V)
Figure 4.12. Dark J-V characteristics of a Cu 2 O:N-incorporated device and a control
device.
0
(N
-2
U)
C
U)
V
C
U)
L..
I-
-4
-6
0
-8L
0.0
0.2
0.4
0.6
bias (V)
Figure 4.13. J-V characteristics of a Cu 20:N-incorporated device and a control device
under 1-sun illumination (AMI.5G, 100 mW-cm 2 ).
84
Table 4.1. Photovoltaic characteristics under 1-sun illumination.
device
Voc
[mV]
[mA-cm 2
Jsc
FF
[%]
efficiency
[%]
with Cu 20:N layer
557
7.30
61.0
2.56
control
485
6.87
46.7
1.56
A small improvement in the short circuit current density (Jsc) was also observed,
which was further investigated by external quantum efficiency (EQE) measurement in
Figure 4.14. The integrated values of the EQE data with the AMI.5G spectrum match
well with Jsc of the devices. Both devices exhibited a strong drop-off in EQE above ~490
nm, due to a sharp decrease in the optical absorption coefficient of Cu 2 0. Due to the
limited minority carrier diffusion length of the Cu 20 layer, the photo-generated carriers
far from the junction are only partially collected [72]. The device with the Cu 20:N layer
exhibited higher EQE relative to the control device when wavelength was increased from
300 nm to 600 nm. The enhancement in EQE can be explained by the electric field in the
absorber layer. Since the back-contact barrier is formed within the Cu2 0:N layer, the
absorber layer contains a larger electric field over a wider collection region, thereby
collecting more photo-generated carriers (Figure 4.15).
85
100-
80
60
E
4-0
C
40
1...
20
(U
0300
400
500
600
wavelength (nm)
700
Figure 4.14. External quantum efficiency of of a Cu 2 O:N-incorporated device and a
control device at zero bias voltage.
a
b
Cu 20:N
a-ZTO
AZO
Figure 4.15. Schematic diagram of energy band alignments in Cu 2 O-based thin-film solar
cells: (a) the control device and (b) Cu 2 O:N hole-transporting layer incorporated device.
86
4.5. Experimental details
4.5.1. Cu 2 0:N thin-film deposition
Cu 2 0 and Cu 2 0:N thin-films were deposited on GE-124 quartz glass substrates by
reactive magnetron sputtering using an PVD- 174 sputtering system (PVD Products, Inc.).
Substrate temperature was controlled using SiC heating elements. A constant power of 30
W (direct-current) was applied to a 99.999% pure metallic copper target of 2-inch-dia.
(K. J. Lesker Co.) to deposit 0.6-pm-thick films with a rate of 5 nm/min. The base
pressure and working pressure were 1.3 x10- Pa and 1.7x 10-' Pa, respectively.
4.5.2. Thin-film solar cell fabrication
A 200-nm-thick Ag bottom electrode with an underlying 5-nm-thick Ti adhesion
layer was deposited on a 1x 1 inch 2 Si0
2
substrate by e-beam evaporation. 20-nm-thick
Cu 2 0:N film was deposited as a hole transporting layer by sputtering. To prevent any
nitrogen decomposition during the subsequent electrochemical deposition process, an
additional 10-nm-thick Cu 20 film was sputtered. A 1.2-pim-thick intrinsic Cu 20 film as a
light absorbing layer was deposited at 40
C by the galvanostatic electrochemical
deposition method [73]. A copper sulfate aqueous solution was prepared by mixing 3 M
lactic acid (Sigma Aldrich), 0.2 M cupric sulfate pentahydrate (Sigma Aldrich) in deionized water (Ricca Chemical). Its pH level was adjusted to 12.5 by adding 2 M sodium
hydroxide (Sigma Aldrich) aqueous solution. A constant current density of 0.23 mA cm-2
was applied by a Keithley 2400 sourcemeter with a Pt counter electrode. 5-nm-thick
amorphous ZTO and 80-nm-thick Al-doped ZnO films were deposited by atomic layer
deposition using cyclic amide of tin (1,3-bis(1,1-dimethylethyl)-4,5-dimethyl-(4R,5R)87
1,3,2-diazastannolidin-2-ylidene)Sn(II)),
diethylzine
(Sigma
Aldrich),
and
trimethylaluminum (Sigma Aldrich) as Sn, Zn, and Al precursors, respectively [50]. 50
wt.% hydrogen peroxide (Sigma Aldrich) and de-ionized water was used as oxidant for aZTO and Al-doped ZnO deposition, respectively. Growth temperature was 120 C. 300nm-thick Al electrodes with a grid spacing of 1 mm were deposited by e-beam
evaporation. The cell area was defined to 3x5 mm 2 by photolithography and nitric acid
solution wet etching.
4.5.3. Characterization
Nitrogen concentrations in the films were measured by SIMS (EAGLAB PCORSIMSsm) with a depth profiling. The crystal structures of the films were characterized by
XRD using a PANalytical X'Pert Pro diffractometer with Cu-Ka radiation. Surface and
cross-sectional images of the films and the device were taken by Zeiss Ultra 55 FESEM.
The XPS was measured by an AXIS Nova (Kratos Analytical, in KBSI). Samples were
etched by an Ar ion beam to remove surface contaminants. The binding-energy scale was
calibrated by adjusting C Is peaks to 284.8 eV. Electrical properties of the films were
characterized by temperature-dependent Hall effect measurements using the van der
Pauw configuration and a magnetic field 0.75 T. Ohmic Au contacts of lxI mm 2 were
deposited on the corners of 1x 1 cm2 Cu 20:N film samples. A closed-cycle He cryostat
and a resistive heater were used to control measurement temperature. Specific contact
resistance was measured by preparing circular transmission line patterns with ring
spacings of 4 - 12 pm. Film optical properties of the films were measured by a Lambda
950 UV-VIS-NIR spectrophotometer (PerkinElmer Inc.). The J-V characteristics of the
device were measured by a Keithley 2400 sourcemeter. The standard I-sun illumination
88
was generated by a Newport Oriel 91194 solar simulator with a 1300 W Xe-lamp with an
AMI.5G filter and a Newport Oriel 68951 flux controller calibrated by an NRELcertified Si reference cell equipped with a BG-39 window. EQE of the device was
measured by QEX-7 (PV measurements, Inc.) calibrated by a NIST-certified Si
photodiode.
4.6. Conclusions
In summary, the potential of Cu 2 O:N films to serve as a hole-transporting
(electron-blocking) layer in a Cu 20-based solar cell is successfully demonstrated. The
nitrogen content in these films can be controlled by varying the nitrogen dose during film
deposition. Nitrogen doping can reduce film electrical resistivity down to 1.8 x 10-1 Q -cm
by increasing the hole concentration. A 20-nm-thick Cu 2 0:N film can create a tunnel
junction to the bottom electrode in a solar cell without absorbing much light in the
subgap wavelength range. The FF and power conversion efficiency of the device show
significant enhancements relative to the control device.
89
CHAPTER 5.
Spatially-Controlled ZnO
Nanowire Array for Cu 2 O Thin-Film Solar Cells
90
5.1. Introduction
Metal-oxide nanowires have demonstrated great potential for high performance in
various optoelectronic devices including solar cells, light-emitting diodes and watersplitting devices [105-107]. In particular, nanowire arrays in solar cells have shown
strong enhancement in photon absorption and photo-generated carrier collection
efficiencies [108, 109]. Specifically, properly controlled geometry of the nanostructure
allows tailoring of light absorption profiles beyond traditional limits [110]. Threedimensional device structures using vertical nanowire arrays (e.g. radial junctions) can
also decouple device geometry design constraints from the light absorption properties and
photo-generated carrier diffusion lengths in a light absorbing material. These advantages
over planar devices improve the prospects for many candidate materials for low-cost
solar cells, which often suffer from non-ideal light absorption and poor carrier transport
properties. However, creating vertically aligned ZnO nanowire array with a controlled
periodicity over a large area is challenging, which normally results in random device
structures. To realize high-performance nano-structured solar cells with potential for a
large-scale deployment, it is highly desirable to develop a high-throughput and scalable
method for fabricating a nanowire array with a controlled geometry.
Among various approaches to create metal-oxide nanowires, hydrothermal growth
of single-crystal ZnO nanowires from supersaturated aqueous solutions has been of great
interest due to the low processing temperature and scalability [111]. The morphology
(e.g. growth direction and aspect ratio) of ZnO nanowires can be controlled by adjusting
the growth conditions including pH-level and auxiliary agents in the solutions [112]. To
align a nanowire array with sub-micron periodicity, various methods including e-beam
91
lithography [113], interference lithography [114], and colloidal lithography [115] have
been used to define seeding areas for nanowire growth. In particular, colloidal
lithography offers a great potential of high-throughput and large-scale processing if it is
combined with the Langmuir-Blodgett process to transfer a self-assembled monolayer of
nanospheres onto the substrate surface [116].
In this chapter, I develop a scalable fabrication method to spatially control a
vertical ZnO nanowire array for photovoltaic applications. Colloidal lithography with PS
nano- and micro-spheres is used to achieve hexagonal alignment of the ZnO nanowire
array. ZnO nanowires can be grown by a parallel hydrothermal reaction at the nucleation
sites defined by self-assembled spheres. The spacing between nanowires and their
geometry are controlled by varying the sphere diameter. Using the ZnO nanowire array,
Cu 20-based heterojunction thin-film solar cells are developed. The strong enhancement
of photo-generated carrier collection by incorporating the ZnO nanowire array in the
device is demonstrated. In addition, optical simulations using the three-dimensional
FDTD method are used to further analyze optical effects of the ZnO nanowire array in
the device. It is found that the heterojunction-based system provides an additional benefit
on amplified light absorption at specific regions by the photonic crystal effect, due to
refractive index differences between ZnO and Cu 20.
92
5.2. Spatially-controlled ZnO nanowire array
5.2.1. Fabrication process
Figure 5.1 depicts the nanowire-array-incorporated solar cell fabrication process,
consisting of seed layer growth, colloidal lithography, ZnO nanowire array growth, Cu 2O
layer deposition, and Au electrode deposition. A textured Al:ZnO film with a crystal
orientation of c-axis (0001) is deposited on a SiO 2 substrate to grow ZnO nanowires
vertically and to create an n*-type transparent conducting oxide (TCO) layer in the solar
cells. A monolayer of closely packed PS spheres is coated on the seed layer for patterning
the ZnO nanowire array. The distance between nanowires can be controlled by varying
the PS sphere diameter. To define areas for nanowire growth, a conformal 5-nm-thick
TiO 2 film as a mask layer is deposited by atomic layer deposition (ALD). The TiO 2 layer
is used as a mask layer due to its chemical stability in the ZnO nanowire growth solution
and the crystal structure incompatible with ZnO, resulting in a high energy barrier for
ZnO nucleation. The TiO 2 -coated PS spheres are removed by ultrasonication in toluene,
exposing the ZnO:Al seed layer where the spheres were in contact with the substrate. A
vertically aligned ZnO nanowire array is grown by the hydrothermal method selectively
on the exposed area. An additional 20-nm-thick ZnO film is deposited by ALD to cover
the TiO 2 film and to enable uniform deposition of the Cu 2 0 film. As a light absorbing
layer, a 2.5-pm-thick Cu2 0 film is deposited on the ZnO nanowire array by
electrochemical deposition. Finally, an Au electrode is deposited by e-beam evaporation.
93
a
ab
C
d
e
f
Figure 5.1. Schematic diagram of the spatially-controlled ZnO nanowire array growth
process, consisting of (a) (0001)-textured ZnO:Al seed layer deposition on SiO 2 substrate,
(b) colloidal PS sphere assembly, (c) 5-nm-thick TiO 2 mask layer by atomic layer
deposition, (d) PS-sphere removal, (e) ZnO nanowire growth by the hydrothermal
method, and (f) Cu 2O and Au electrode deposition. Courtesy of Dr. J. Joo.
94
5.2.2. Colloidal lithography for patterning a ZnO nanowire array
To demonstrate the proposed method for vertical ZnO nanowire array alignment,
a ZnO nanowire array was grown on a (0001)-orientated single crystal ZnO wafer.
Sulfate-functionalized PS spheres with a diameter of 0.5 or 1 pm (Polysciences, Inc.)
were dispersed in an 1:1 mixture of ethanol and de-ionized water, with a solid-liquid ratio
of 1.2 wt.%. Aided by a tilted SiO 2 piece, the solution was applied gently onto de-ionized
water in a Langmuir-Blodgett trough to float the spheres on the surface while the
substrate is immersed in the water horizontally. A barrier position was adjusted to form a
closely packed monolayer of PS spheres on the surface. The monolayer is then
transferred to the substrate surface by draining water from the trough (Figure 5.2).
Figure 5.2. A closely packed 500 nm diameter PS-sphere array transferred to a substrate.
The scale bar is 1 pm.
95
A conformal 5-nm-thick ALD TiO 2 layer was deposited on the sphere-coated
ZnO wafer at 80 'C using a Savannah ALD (Cambridge NanoTech, Inc.) with tetrakisdimethylamino titanium (TDMAT) as a Ti precursor and H2 0 as an oxidant [117]. The
deposition rate was measured to be 0.95 A/cycle. After removing the TiO 2 -coated PS
spheres from the wafer, ZnO nanowires were hydrothermally grown in a 100 mL aqueous
solution at 50 - 60 'C for 4 hours. The solution was prepared by 0.01 M of zinc sulfate
heptahydrate (ZnSO47H 2 0, Sigma-Aldrich) and 0.3 M of ammonium chloride (NH 4 Cl,
Sigma-Aldrich) in de-ionized water. The pH of the solution was adjusted to 11.0 by
adding sodium hydroxide (NaOH, Fluka). Figure 5.3 and Figure 5.4 show micrographs of
the fabricated ZnO nanowire array using 500 nm diameter PS spheres. The 2-tm-long
ZnO nanowires with an aspect ratio of ~10 were grown vertically, maintaining the
periodic pattern from the colloidal lithography. The selectivity of ZnO nanowire growth
resulted from homogeneous nucleation on the ZnO surface, a more favorable reaction
than heterogeneous nucleation on the TiO 2 surface.
96
Figure 5.3. A SEM image of a ZnO nanowire array grown on a single crystal ZnO
substrate (30 tilted view). The scale bar is 1 pm.
Figure 5.4. A SEM image of a ZnO nanowire array grown on a single crystal ZnO
substrate (top view). The scale bar is 1 pm.
97
5.2.3. Seed layer deposition for vertical ZnO nanowire growth
I also demonstrate the growth of vertically aligned ZnO nanowire arrays on
ZnO:Al thin-films, for large-scale photovoltaic applications. A textured 1-pm-thick
ZnO:Al film was deposited as both a seed layer of ZnO nanowire growth and an n*-type
transparent conducting oxide (TCO) layer in the solar cells. The deposition parameters
are adjusted to achieve a preferred crystal orientation of c-axis (0001) of ZnO and a sheet
resistance lower than 10 Q/o. 1-pm-thick polycrystalline ZnO:Al films were deposited at
230 'C by an Orion 5 magnetron sputtering system (AJA International). A 2-inch
diameter ceramic target composed of 98 wt.% ZnO and 2 wt% A12 0 3 (K. J. Lesker) was
sputtered by Ar with a constant RF power of 150 W. The base and working pressures
were maintained to 1.3 x 10-6 and 4.Ox 10-3 Pa, respectively. Figure 5.5 shows XRD spectra
of the sputtered ZnO:Al films and grown ZnO nanowires. Both samples exhibit strong
(0001) peaks denoting ZnO crystal structure, indicating the seed layer is highly textured
and proper to grow vertical ZnO nanowires.
Figure 5.6 shows a side view of the vertically grown ZnO nanowire array on a
textured ZnO:Al seed layer patterned by 1 ptm spheres. Due to the polycrystalline nature
of the seed layer, a bundle of multiple ZnO nanowires could grow from the same
patterned area. Figure 5.7 and Figure 5.8 show that ZnO nanowires from the patterned
seed layer exhibit larger diameters than nanowires without patterning.
98
(0001)
(0002)
ZnO-NW
C2
ZnO:AJ seed layer
0
30
40
50 690
20(0)
Figure 5.5. XRD spectra of the sputtered ZnO:Al films and grown ZnO nanowires.
Figure 5.6. A side view of the vertically grown ZnO nanowire array on the textured
ZnO:Al seed layer patterned by 1 pm spheres. The scale bar is 1 pim.
99
Figure 5.7. ZnO nanowires grown on a polycrystalline ZnO:Al film without a mask layer.
The scale bar is 1 im.
Figure 5.8. ZnO nanowires grown on a polycrystalline ZnO:Al film with a patterned TiO 2
mask layer, exhibiting larger diameter than nanowires without patterning. The scale bar is
1 pm.
100
5.3. Device fabrication and characterization
Using the ZnO nanowire array, Cu 2 0-based thin-films solar cells are fabricated to
test its photo-generated carrier collection efficiency. 2-p.m-thick Cu 2 0 layers were
electrochemically deposited on the ZnO nanowire arrays with spacings of 0.5 and 1 tm.
A planar structure device and a device with dense ZnO nanowires without patterning
were also fabricated as control devices to study the effects of the periodic ZnO nanowire
array. SEM images of the fabricated devices show the trenches between ZnO nanowires
are filled with Cu 20. The performance of the fabricated devices were characterized by
current density - voltage (J-V) measurements under 1-sun (AMI.5G, 100 mW-cm-2)
illumination (Figure 5.9 and Table 5.1). The devices exhibited sizable enhancements in
short-circuit current densities (Jsc) and open-circuit voltages (Voc) by incorporating
periodic ZnO nanowires arrays, resulting in power conversion efficiencies (PCE) up to
0.88 % for a nanowire spacing of 0.5 pm. However, the device with dense ZnO
nanowires exhibited a PEC of 0.47 %, with a relatively small increase compared to the
planar device (PCE = 0.35 %).
Table 5.1. Photovoltaic characteristics under 1-sun illumination.
FF
efficiency
(mA-cm 2 )
(mV)
(%)
(%)
planar
4.62
187
40.4
0.35
ZnO nanowire (P = 500 nm)
7.32
267
45.0
0.88
ZnO nanowire (P =1000 nm)
6.08
279
46.5
0.79
device
101
VOC
VoC
0
C
a)
-8-
_0
0.0
0.1
0.2
0.3
bias (V)
Figure 5.9. J-V characteristics of ZnO nanowire array incorporated devices and a control
device under 1-sun illumination condition.
5.4. Optical simulation
The enhancements in Jsc were further analyzed by optical simulation using FDTD
method. FDTD Solutions software (Ver. 7.5, Lumerical, Inc.) was used to calculate the
electromagnetic field and light absorption profile inside the device as a function of space
and frequency. The power absorption profile was obtained by calculating the divergence
of the Poynting vector. Incident light with a transverse magnetic (TM) polarization was
then used to simulate the light propagation with 400 - 650 nm wavelength (I) range
within this structure. Figure 5.10 shows calculated absorption profiles for k = 500 nm by
102
varying a distance between ZnO nanowires in a 2-D model. Due to a large refractive
index (n) difference between Cu 2 0 (ncUoo
3) and ZnO (nzno ~ 1.8), a strong waveguide
effect was calculated; the cladding by ZnO nanowires resulted in locally amplified
absorption in Cu20. Calculations in a 3-D model also suggest a similar waveguiding
effect. As the spacing between ZnO nanowires is decreased, stronger absorption is
expected near the ZnO-Cu 2 0 interface where photo-generated carriers can be collected
more efficiently.
a
b
C
e
d
3 x 1019
0
(s-1- nm-1 cm-3 )
Figure 5.10. 2-D optical simulations of the optical absorption profile for photons with a
wavelength of 500 nm in Cu 20 layer (middle area) between ZnO nanowires with periods
of (a) 350 nm, (b) 500 nm, (c) 700 nm, (d) 1000 nm, and (e) a planar structure. The
incident photon flux is based on the AMI.5G solar spectrum. White lines indicate
interfaces between ZnO and Cu 2 0. All scale bars (bottom) represent 200 nm. Courtesy of
J. P. Mailoa.
103
5.5. Experimental details
5.5.1. Cu 2 0 thin-film deposition
2.5-pm-thick Cu 2 0 films were deposited at 40 'C by the galvanostatic
electrochemical method. A lactate-stabilized copper sulphate aqueous solution was
prepared with 3 M lactic acid (Sigma-Aldrich), 0.2 M cupric sulfate pentahydrate
(CuO 4 S -5H2 0, Sigma-Aldrich) and de-ionized water (18.3 MQ cm, Ricca Chemical),
and a 2 M sodium hydroxide (NaOH, Sigma Aldrich) aqueous solution was added to
adjust the pH of the solution to 12.5. All reagent-grade chemicals were used and the
solution was filtered and stirred thoroughly. A constant current density of 0.2 mAcm-2
was applied by a Keithley 2400 sourcemeter with a Pt counter electrode to grow Cu 2 0
films. The Cu 2 0 film surfaces were rinsed with de-ionized water after deposition.
5.5.2. Characterization
The microstructures of the ZnO nanowires and solar cells were analyzed by an
Ultra 55 FE-SEM (Zeiss) and XRD using an X'Pert Pro diffractometer (PANalytical)
with Cu-Ka radiation. The film thickness was measured by a VASE spectroscopic
ellipsometer (J.A. Woollam Co., Inc.). The electrical resistivities of the ZnO:Al films
were measured by four-point probe. The current-voltage and the capacitance-frequency
characteristics of the devices were measured by a Keithley 2400 sourcemeter. The
standard 1-sun illumination was generated by a Newport Oriel 91194 solar simulator with
a 1300 W Xe-lamp with AM1.5G filter and a Newport Oriel 68951 flux controller
calibrated by an NREL-certified Si reference cell equipped with a BG-39 window. EQE
104
of the device was measured by QEX7 (PV Measurements, Inc.) calibrated by a NISTcertified Si photodiode.
5.6. Conclusions
In summary, a scalable fabrication method is developed for spatially-controlled
vertical zinc oxide (ZnO) nanowire array growth. Colloidal lithography using PS nanospheres and a Langmuir-Blodgett trough enables the precise control of nanowire spacing.
Using a ZnO nanowire array, a sizable performance enhancement of Cu 2 0-based thinfilm solar cells is demonstrated. The ZnO nanowire incorporated devices exhibit power
conversion efficiencies 50 % higher than a planar ZnO-Cu 2 0 device, by increased Jsc
and Voc. FDTD optical simulations are carried out to further investigate optical
absorption enhancements in the devices. A properly controlled nanowire periodicity
enables amplified light absorption near the junction area. This work demonstrates the
strong potential for nanostructured metal-oxide materials to be incormorated into various
opto-electronics and energy harvesting applications.
105
CHAPTER 6.
Summary and Future Directions
106
This study develops Cu 2 0-based thin-film solar cells that can be used as a top cell
in a tandem structure. In particular, I investigate three principal energy loss mechanisms
that restrict power-conversion efficiencies in Cu 20-based solar cells: (a) a low efficiency
of photo-generated carrier collection originating from a high density of defects in the
material, (b) a low built-in voltage and high density of interface recombination due to
non-ideal energy band alignment in the p-n junction, and (c) a back electrode barrier
creating an unfavorable electric field in the Cu 20 layer of the device.
To enhance photo-generated carrier collection efficiency, the photo-generated
carrier mobility of Cu 20 thin-films needs to be increased. In Chapter 2, polycrystalline
Cu 2 0
thin-films were deposited by reactive sputtering with elevated
substrate
temperature. It was shown that high temperature enhances grain structure in the film and
increases Hall mobility to values comparable with single crystalline Cu 2 0
at
measurement temperatures above 250 K. Temperature-dependent Hall measurements
revealed that a lower defect density could further enhance the mobility and increase the
photo-generated carrier diffusion length.
The incorporation of an a-ZTO buffer layer was demonstrated to mitigate the nonideal band alignment in the ZnO-Cu 2 0 heterojunction. In Chapter 3, by introducing an
ultrathin a-ZTO buffer layer, I demonstrated a sizable enhancement of power conversion
efficiency of Cu 2 0-based thin-film solar cells. The a-ZTO buffer layers with precisely
tuned electrical and optical properties could reduce interfacial recombination by acting as
electron-blocking barriers and decreasing defect densities. The 5-nm-thick buffer layer
improved the junction quality without sacrificing
107
carrier transport and optical
transmission, resulting in an open-circuit voltage close to the built-in potential of the
ZnO-Cu 20 junction.
Chapter 4 discussed a back-contact barrier between a Cu 20 layer and a backelectrode in the solar cell. I proposed doping of nitrogen as an effective method to
increase carrier density in Cu 20 films and to create a low-resistance contact with the
metal electrode. The potential of Cu 2 0:N films as a hole-transporting layer in the Cu 20based solar cell was successfully demonstrated. The nitrogen doping reduced the
electrical resistivity of the films down to 1.8 x 101 Q-cm by increasing the hole density.
A 20-nm-thick Cu 2 0:N film was inserted as a p-type hole transporting layer, and the
layer created a tunnel junction to a silver bottom electrode in the solar cell without
sacrificing optical absorption in the Cu 20 layer.
To further enhance photo-generated carrier collection efficiency and optical
absorption in the device, a ZnO nanowire array was developed. In Chapter 5, a scalable
fabrication method for spatially-controlled vertical ZnO nanowire array growth was
developed. Colloidal lithography using the Langmuir-Blodgett method controlled the
nanowire array periodicity precisely. Incorporating the ZnO nanowire array into Cu 20based solar cells enhanced the performance of Cu 2 0-based thin-film solar cells relative to
planar devices without buffer layers. Optical simulations showed a locally amplified light
absorption in Cu 2 0, due to a waveguide effect from the ZnO nanowire array.
However, the solar cells studied in this thesis exhibited open-circuit voltages
lower than 0.6 V. The voltages can be further increased by improved band-alignment,
which can reduce the interfacial recombination current density and form a built-in
108
potential higher than that of a ZnO-Cu 20 heterojunction. The ideal material for a high
built-in potential will be a wide-gap n-type semiconductor with a conduction band edge
position close to Cu 2 0, but not too high to block electron transport. Several n-type
semiconductors including GaN and ZnS possess conduction band edge positions higher
than the conduction band edge of ZnO (Figure 6.1). If such materials can have n-type
conductivities with high carrier densities while maintaining low densities of deep-trap
states, they may replace the ZnO layer or be inserted between ZnO and Cu 20 layer to
increase the open-circuit voltage.
_______0.24
CBM
eV
0.73 eV
1.47 eV
VBM
ZnO
0.70eV
GaN
Cu 2O
Figure 6.1. Band alignment in ZnO/GaN, Cu 20/GaN and Cu 2 O/ZnO heterojunctions.
After [118].
The Cu 20 devices studied in this work exhibited low quantum efficiencies in the
wavelength range of 490 - 630 nm due to the limited carrier collection length of -500 nm,
which is much shorter than the Cu 20 layer thickness. The quantum efficiency near the
band-edge can be further enhanced by (a) improving the minority carrier diffusion length
109
of Cu 2 0 and (b) making the Cu 2 0 layer thickness less than the diffusion length. To
improve the minority carrier diffusion length, Cu 20 thin-films with low defect-densities
and high degrees of crystallinity are required to increase carrier mobility and lifetime.
Also, the device geometry needs to be optimized for the Cu 2 0 film properties; the entire
Cu 2 0 layer should have high carrier-collection efficiency and absorb photons in the
wavelength range effectively.
The universality of these approaches to improve device efficiencies can be
extended to other photovoltaic material systems. Also, the materials develped in this
study may be useful to other types of materials and photovoltaic devices. The a-ZTO
buffer layer can be used in various heterojunction solar cells to replace cadmiumcontaining buffer layers and increase open-circuit voltages. Nitrogen-doped Cu 20 may be
useful in other photovoltaic materials systems, improving carrier transport properties in
metal-semiconductor interfaces. Nanostructured metal-oxide materials can be utilized to
enhance the performance of various opto-electronic and energy harvesting devices.
110
References
[1]
International
Energy
Outlook
2008
http://www.eia.gov/forecasts/archive/ieo08/index.html
[Online].
Available:
[2]
M. A. Green, "Estimates of Te and In prices from direct mining of known ores,"
Progressin Photovoltaics: Research and Applications, vol. 17, pp. 347-3 59, 2009.
[3]
C. Wadia, et al., "Materials Availability Expands the Opportunity for Large-Scale
Photovoltaics Deployment," Environmental Science & Technology, vol. 43, pp.
2072-2077, 2009.
[4]
T. Unold and H. Schock, "Nonconventional (non-silicon-based) photovoltaic
materials," Annual Review of MaterialsResearch, vol. 41, pp. 297-321, 2011.
[5]
J. Meier, et al., "From R&D to Mass Production of Micromorph Thin Film Silicon
PV," Energy Procedia,vol. 15, pp. 179-188, 2012.
[6]
D. L. Staebler and C. R. Wronski, "Reversible conductivity changes in dischargeproduced amorphous Si," Applied Physics Letters, vol. 31, pp. 292-294, 1977.
[7]
F. Meillaud, et al., "Efficiency limits for single-junction and tandem solar cells,"
Solar Energy Materialsand Solar Cells, vol. 90, pp. 2952-2959, 2006.
[8]
F. L. Weichman, "Photoconductivity of Cuprous Oxide in Relation to Its Other
Semiconducting Properties," PhysicalReview, vol. 117, pp. 998-1002, 1960.
[9]
A. Mittiga, et al., "Heteroj unction solar cell with 2% efficiency based on a Cu 20
substrate," Applied Physics Letters, vol. 88, p. 163502, 2006.
[10] J. Mattheis, et al., "Finite mobility effects on the radiative efficiency limit of pnjunction solar cells," Physical Review B, vol. 77, p. 085203, 2008.
[11] S. 0. Ishizuka, et al., "Nitrogen doping into Cu 2 0 thin films deposited by reactive
radio-frequency magnetron sputtering," Japanese Journal of Applied Physics, vol.
40, pp. 2765-2768, 2001.
[12] Z. G. Yin, et al., "Two-dimensional growth of continuous Cu 2 O thin films by
magnetron sputtering," Applied Physics Letters, vol. 86, p. 061901, 2005.
[13] K. Matsuzaki, et al., "Epitaxial growth of high mobility Cu 2 0 thin films and
application to p-channel thin film transistor," Applied Physics Letters, vol. 93, p.
202107, 2008.
[14] Z. Q. Yu, et al., "Epitaxial growth and microstructure of Cu 2 0 nanoparticle/thin
films on SrTiO 3(100)," Nanotechnology, vol. 18, p. 115601, 2007.
111
[15] D. Arana-Chavez, et al., "Modeling the Transport and Reaction Mechanisms of
Copper Oxide CVD," Chemical Vapor Deposition, vol. 16, pp. 336-345, 2010.
[16] R. Liu, et al., "Epitaxial electrodeposition of Cu 2 0 films onto InP(001)," Applied
Physics Letters, vol. 83, pp. 1944-1946, 2003.
[17] M. Imaizumi, et al., "Effect of grain size and dislocation density on the
performance of thin film polycrystalline silicon solar cells," Journal of Applied
Physics, vol. 81, pp. 7635-7640, 1997.
[18] B. A. Movchan and A. V. Demchishin, "Study of Structure and Properties of Thick
Vacuum Condensates of Nickel, Titanium, Tungsten, Aluminium Oxide and
Zirconium Dioxide," The Physics of Metals and Metallography, vol. 28, p. 83,
1969.
[19] J. A. Thornton, "High rate thick film growth," Annual Review of Materials Science,
vol. 7, pp. 239-260, 1977.
[20] A. Mittiga, et al., "Intrinsic defects and metastability effects in Cu 20," Thin Solid
Films, vol. 517, pp. 2469-2472, 2009.
[21] W. H. Brattain, "The Copper Oxide Rectifier," Reviews of Modern Physics, vol. 23,
pp. 203-212, 1951.
[22] J. W. Hodby, et al., "Cyclotron-Resonance of Electrons and of Holes in CuprousOxide, Cu 2 O," Journal of Physics C: Solid State Physics, vol. 9, pp. 1429-1439,
1976.
[23] K. Matsuzaki, et al., "Effects of post-annealing on (110) Cu 20 epitaxial films and
origin of low mobility in Cu 20 thin-film transistor," Physica Status Solidi A, vol.
206, pp. 2192-2197, 2009.
[24] G. P. Pollack and D. Trivich, "Photoelectric Properties of Cuprous-Oxide," Journal
ofApplied Physics, vol. 46, pp. 163-172, 1975.
[25] J. S. Blakemore, Semiconductorstatistics. New York: Dover, 1987.
[26] G. K. Paul, et al., "Defects in Cu 20 studied by deep level transient spectroscopy,"
Applied Physics Letters, vol. 88, pp. 141901-3, 2006.
[27] M. Tapiero, et al., "Photo-Memory Effect in Cu2 0 Single-Crystals Phenomenology and Interpretation," Physica Status Solidi A, vol. 33, pp. 155-166,
1976.
[28] H. Raebiger, et al., "Origins of the p-type nature and cation deficiency in Cu 2 0 and
related materials," Physical Review B, vol. 76, p. 045209, 2007.
[29] D. 0. Scanlon, et al., "Acceptor Levels in p-Type Cu 20: Rationalizing Theory and
112
Experiment," Physical Review Letters, vol. 103, p. 096405, 2009.
[30] E. Fortin and F. L. Weichman, "Hall Effect and Electrical Conductivity of Cu2 0
Monocrystals," CanadianJournalofPhysics, vol. 44, p. 1551, 1966.
[31] H. L. Mckinzie and M. Okeeffe, "High Temperature Hall Effect in Cuprous Oxide,"
Physics Letters A, vol. A 24, p. 137, 1967.
[32] H. Shimada and T. Masumi, "Hall-Mobility of Positive Holes in Cu 2 0," Journalof
the PhysicalSociety ofJapan,vol. 58, pp. 1717-1724, 1989.
[33] B. A. Tazenkov and F. A. Grudzdev, Soviet Physics - Solid State, vol. 16, p. 460,
1974.
[34] M. Zouaghi, et al., "Hall mobility and hole density in photoactivated Cu 20 single
crystals," Solid State Communications, vol. 8, pp. 1823-1825, 1970.
[35] D. Chattopadhyay and H. J. Queisser, "Electron-Scattering by Ionized Impurities in
Semiconductors," Reviews ofModern Physics, vol. 53, pp. 745-768, 1981.
[36] Y. Nishi, et al., "The impact of heterojunction formation temperature on obtainable
conversion efficiency in n-ZnO/p-Cu 20 solar cells," Thin Solid Films, vol. 528, pp.
72-76, 2013.
[37] M. Izaki, et al., "Electrochemically constructed p-Cu 20/n-ZnO heterojunction
diode for photovoltaic device," Journalof Physics D - Applied Physics, vol. 40, pp.
3326-3329, 2007.
[38] S. Jeong, et al.,
"An analysis of temperature
dependent current-voltage
characteristics of Cu 2 0-ZnO heterojunction solar cells," Thin Solid Films, vol. 519,
pp. 6613-6619, 2011.
[39] K. P. Musselman, et al., "A Novel Buffering Technique for Aqueous Processing of
Zinc Oxide Nanostructures and Interfaces, and Corresponding Improvement of
Electrodeposited ZnO-Cu 2 0 Photovoltaics," Advanced Functional Materials, vol.
21, pp. 573-582, 2011.
[40] A. Shah, et al., "Photovoltaic technology: The case for thin-film solar cells,"
Science, vol. 285, pp. 692-698, 1999.
[41] K. L. Chopra, et al., "Thin-film solar cells: an overview,"
Photovoltaics:Research andApplications, vol. 12, pp. 69-92, 2004.
Progress in
[42] M. Taguchi, et al., "Obtaining a higher Voc in HIT cells," Progress in
Photovoltaics:Research andApplications, vol. 13, pp. 481-488, 2005.
[43] A. Kanevce and W. K. Metzger, "The role of amorphous silicon and tunneling in
heterojunction with intrinsic thin layer (HIT) solar cells," Journal of Applied
113
Physics, vol. 105, pp. 094507-7, 2009.
[44] T. Minemoto, et al., "Theoretical analysis of the effect of conduction band offset of
window/CIS layers on performance of CIS solar cells using device simulation,"
Solar Energy MaterialsandSolar Cells, vol. 67, pp. 83-88, 2001.
[45] J. Robertson, "Disorder and instability processes in amorphous conducting oxides,"
Physica Status Solidi B, vol. 245, pp. 1026-1032, 2008.
[46] W. Aron, et al., "Multi-component transparent conducting oxides: Progress in
materials modelling,"Journalof Physics: Condensed Matter, vol. 23, p. 334210,
2011.
[47] T. Minemoto, et al., "Theoretical analysis of the effect of conduction band offset of
window/CIS layers on performance of CIS solar cells using device simulation,"
Solar Energy MaterialsandSolar Cells, vol. 67, pp. 83-88, 2001.
[48] E. Fortunato, et al., "Oxide Semiconductor Thin-Film Transistors: A Review of
Recent Advances," Advanced Materials,vol. 24, pp. 2945-2986, 2012.
[49] K. Nomura, et al., "Room-temperature fabrication of transparent flexible thin-film
transistors using amorphous oxide semiconductors," Nature, vol. 432, pp. 488-492,
2004.
[50] J. Heo, et al., "Atomic layer deposited zinc tin oxide channel for amorphous oxide
thin film transistors," Applied Physics Letters, vol. 101, pp. 113507-5, 2012.
[51] J. Lindahl, et al., "The effect of Zni-xSnxOy buffer layer thickness in 18.0%
efficient Cd-free Cu(In,Ga)Se 2 solar cells," Progress in Photovoltaics: Research
and Applications, p. doi: 10.1002/pip.2239, 2012.
[52] Z. Zhang, et al., "The effect of complexing agents on the oriented growth of
electrodeposited microcrystalline cuprous oxide film," MaterialsResearch Bulletin,
vol. 47, pp. 2561-2565, 2012.
[53] J. Heo, et al., "Low Temperature Atomic Layer Deposition of Tin Oxide,"
Chemistry of Materials,vol. 22, pp. 4964-4973, 2010.
[54] H. Q. Chiang, et al., "High mobility transparent thin-film transistors with
amorphous zinc tin oxide channel layer," Applied Physics Letters, vol. 86, p.
013503, 2005.
[55] J.-Y. Kwon, et al., "Review paper: Transparent amorphous oxide semiconductor
thin film transistor," Electronic MaterialsLetters, vol. 7, pp. 1-11, 2011.
[56] T. Moriga, et al., "Transparent conducting amorphous Zn-Sn-O films deposited by
simultaneous dc sputtering," Journalof Vacuum Science & Technology A, vol. 22,
pp. 1705-1710, 2004.
114
[57] Y.-H. Kim, et al., "Flexible metal-oxide devices made by room-temperature
photochemical activation of sol-gel films," Nature, vol. 489, pp. 128-132, 2012.
[58] S. Narushima, et al., "A p-type amorphous oxide semiconductor and room
temperature fabrication of amorphous oxide p-n heterojunction diodes," Advanced
Materials,vol. 15, pp. 1409-1413, 2003.
[59] K. Satoh, et al., "Influence of Oxygen Flow Ratio on Properties of Zn 2 SnO 4 Thin
Films Deposited by RF Magnetron Sputtering," Japanese Journal of Applied
Physics, vol. 44, pp. L34-L37, 2005.
[60] J. S. Rajachidambaram, et al., "Characterization of amorphous zinc tin oxide
semiconductors," JournalofMaterialsResearch, vol. 27, pp. 2309-2317, 2012.
[61] J. R. Waldrop, et al., "Measurement of semiconductor heterojunction band
discontinuities by x-ray photoemission spectroscopy," Journal of Vacuum Science
& Technology A: Vacuum, Surfaces, and Films, vol. 3, pp. 835-841, 1985.
[62] P. T. Erslev, et al., "Mapping out the distribution of electronic states in the mobility
gap of amorphous zinc tin oxide," Applied Physics Letters, vol. 95, pp. 192115-3,
2009.
[63] M. Batzill, et al., "Gas-phase-dependent properties of Sn0 2 (110), (100), and (101)
single-crystal surfaces: Structure, composition, and electronic properties," Physical
Review B, vol. 72, p. 165414, 2005.
[64] K. G. Godinho, et al., "Energetic and Electronic Structure Analysis of Intrinsic
Defects in Sn0 2," The Journal of Physical Chemistry C, vol. 113, pp. 439-448,
2008.
[65] J. M. Themlin, et al., "Resonant-photoemission study of Sn0 2 : Cationic origin of
the defect band-gap states," PhysicalReview B, vol. 42, pp. 11914-11925, 1990.
[66] Z. Q. Duan, et al., "Effects of Mg composition on open circuit voltage of Cu 2 0MgxZn IxO heterojunction solar cells," Solar Energy Materials and Solar Cells, vol.
96, pp. 292-297, 2012.
[67] L. M. Wong, et al., "Growth of Cu 2 0 on Ga-doped ZnO and their interface energy
alignment for thin film solar cells," JournalofApplied Physics, vol. 108, 2010.
[68] T. Walter, et al., "Determination of defect distributions from admittance
measurements and application to Cu(In,Ga)Se 2 based heterojunctions," Journal of
Applied Physics, vol. 80, pp. 4411-4420, 1996.
[69] T. Eisenbarth, et al., "Interpretation of admittance, capacitance-voltage, and
current-voltage signatures in Cu(In,Ga)Se 2 thin film solar cells," JournalofApplied
Physics, vol. 107, pp. 034509-12, 2010.
115
[70] M. Saad and A. Kassis, "Effect of interface recombination on solar cell
parameters," Solar Energy Materials and Solar Cells, vol. 79, pp. 507-517, 2003.
[71] K. P. Musselman, et al., "Incompatible Length Scales in Nanostructured Cu 2 0
Solar Cells," Advanced FunctionalMaterials,vol. 22, pp. 2202-2208, 2012.
[72] Y. C. Liu, et al., "Minority carrier transport length of electrodeposited Cu 20 in
ZnO/Cu 20 heterojunction solar cells," Applied Physics Letters, vol. 98, p. 162105,
2011.
[73] P. E. de Jongh, et al., "Cu 2 0: Electrodeposition and characterization," Chemistry of
Materials, vol. 11, pp. 3512-3517, 1999.
[74] Y. Roussillon, et al., "Back contact and reach-through diode effects in thin-film
photovoltaics," JournalofApplied Physics, vol. 96, pp. 7283-7288, 2004.
[75] D. K. Schroder and D. L. Meier, "Solar cell contact resistance: A review," IEEE
Transactionson Electron Devices, vol. 31, pp. 637-647, 1984.
[76] D. Abou-Ras, et al., "Formation and characterisation of MoSe 2 for Cu(In,Ga)Se 2
based solar cells," Thin Solid Films, vol. 480-481, pp. 433-438, 2005.
[77] D. L. Batzner, et al., "Development of efficient and stable back contacts on
CdTe/CdS solar cells," Thin Solid Films, vol. 387, pp. 151-154, 2001.
[78] Y.-J. Lee, et al., "Low-Temperature Solution-Processed Molybdenum Oxide
Nanoparticle Hole Transport Layers for Organic Photovoltaic Devices," Advanced
Energy Materials,vol. 2, pp. 1193-1197, 2012.
[79] Y. Sun, et al., "Efficient, Air-Stable Bulk Heterojunction Polymer Solar Cells
Using MoOx as the Anode Interfacial Layer," Advanced Materials, vol. 23, pp.
2226-2230, 2011.
[80] V. Shrotriya, et al., "Transition metal oxides as the buffer layer for polymer
photovoltaic cells," Applied Physics Letters, vol. 88, p. 073508, 2006.
[81] K. Zilberberg, et al., "Solution Processed Vanadium Pentoxide as Charge
Extraction Layer for Organic Solar Cells," Advanced Energy Materials, vol. 1, pp.
377-381, 2011.
[82] J. Meyer, et al., "Transition Metal Oxides for Organic Electronics: Energetics,
Device Physics and Applications," Advanced Materials, vol. 24, pp. 5408-5427,
2012.
[83] Y. Tsur and I. Riess, "Self-compensation in semiconductors," Physical Review B,
vol. 60, pp. 8138-8146, 1999.
[84] Y. Tsur and I. Riess, "Doping of ionic compounds: solubility limit and self116
compensation," Solid State Ionics, vol. 119, pp. 37-42, 1999.
[85] J. Robertson and S. J. Clark, "Limits to doping in oxides," Physical Review B, vol.
83, p. 075205, 2011.
[86] L. C. Olsen, et al., "Experimental and theoretical studies of Cu 20 solar cells," Solar
Cells, vol. 7, pp. 247-279, 1982.
[87] S. Ishizuka and K. Akimoto, "Control of the growth orientation and electrical
properties of polycrystalline Cu 2 0 thin films by group-IV elements doping,"
Applied Physics Letters, vol. 85, pp. 4920-4922, 2004.
[88] S. Ishizuka, et al., "Nitrogen doping into Cu2 0 thin films deposited by reactive
radio-frequency magnetron sputtering," Japanese Journal of Applied Physics Part
1, vol. 40, pp. 2765-2768, 2001.
[89] N. Kikuchi and K. Tonooka, "Electrical and structural properties of Ni-doped Cu 2 0
films prepared by pulsed laser deposition," Thin Solid Films, vol. 486, pp. 33-37,
2005.
[90] C. Malerba, et al., "Nitrogen doped Cu 2 0: A possible material for intermediate
band solar cells?," Solar Energy Materials and Solar Cells, vol. 105, pp. 192-195,
2012.
[91] M. Burgelman, et al., "Modelling polycrystalline semiconductor solar cells," Thin
Solid Films, vol. 361-362, pp. 527-532, 2000.
[92] Y. S. Lee, et al., "Hall mobility of cuprous oxide thin films deposited by reactive
direct-current magnetron sputtering," Applied Physics Letters, vol. 98, p. 192115,
2011.
[93] T. Maruyama and T. Morishita, "Copper Nitride Thin-Films Prepared by
Radiofrequency Reactive Sputtering," Journal of Applied Physics, vol. 78, pp.
4104-4107, 1995.
[94] J. Ghijsen, et al., "Electronic structure of Cu2 0 and CuO," Physical Review B, vol.
38, pp. 11322-11330, 1988.
[95] H. Wu and W. Chen, "Copper Nitride Nanocubes: Size-Controlled Synthesis and
Application as Cathode Catalyst in Alkaline Fuel Cells," Journal of the American
ChemicalSociety, vol. 133, pp. 15236-15239, 2011.
[96] L. C. Olsen, et al., "Explanation for low-efficiency Cu2O Schottky-barrier solar
cells," Applied Physics Letters, vol. 34, pp. 47-49, 1979.
[97] W.-Y. Yang and S.-W. Rhee, "Effect of electrode material on the resistance
switching of Cu 20 film," Applied Physics Letters, vol. 91, p. 232907, 2007.
117
[98] S. C. Siah, et al., "Low contact resistivity of metals on nitrogen-doped cuprous
oxide (Cu 2O) thin-films," Journal ofApplied Physics, vol. 112, p. 084508, 2012.
[99] G. K. Reeves, "Specific contact resistance using a circular transmission line
model," Solid-State Electronics, vol. 23, pp. 487-490, 1980.
[100]
Z. Zhao, et al., "First-principles study on the doping effects of nitrogen on the
electronic structure and optical properties of Cu 2 0," RSC Advances, vol. 3, pp. 8490, 2013.
[101]
N. Gordillo, et al., "Free-carrier contribution to the optical response of N-rich
Cu 3N thin films," JournalofPhysics D-Applied Physics, vol. 42, p. 165101, 2009.
[102] Y. S. Lee, et al., "Ultrathin amorphous zinc tin oxide buffer layer for interface
engineered metal-oxide solar cells," submitted, 2012.
[103]
A. Niemegeers and M. Burgelman, "Effects of the Au/CdTe back contact on IV
and CV characteristics of Au/CdTe/CdS/TCO solar cells," Journal of Applied
Physics, vol. 81, pp. 2881-2886, 1997.
[104] M. Chelvayohan and C. H. B. Mee, "Work Function Measurements on (110),
(100) and (111) Surfaces of Silver," Journal of Physics C-Solid State Physics, vol.
15, pp. 2305-2312, 1982.
[105]
M. H. Huang, et al., "Room-Temperature Ultraviolet Nanowire Nanolasers,"
Science, vol. 292, pp. 1897-1899, 2001.
[106]
P. J. Pauzauskie and P. Yang, "Nanowire photonics," Materials Today, vol. 9, pp.
36-45, 2006.
[107] R. Yan, et al., "Nanowire photonics," Nature Photonics, vol. 3, pp. 569-576,
2009.
[108]
M. Law, et al., "Nanowire dye-sensitized solar cells," Nature Materials, vol. 4,
pp. 455-459, 2005.
[109]
K. P. Musselman, et al., "Strong Efficiency Improvements in Ultra-low-Cost
Inorganic Nanowire Solar Cells," Advanced Materials, vol. 22, pp. E254-E258,
2010.
[110] S. E. Han and G. Chen, "Toward the Lambertian Limit of Light Trapping in Thin
Nanostructured Silicon Solar Cells," Nano Letters, vol. 10, pp. 4692-4696, 2010.
[111]
L. E. Greene, et al., "Low-Temperature Wafer-Scale Production of ZnO
Nanowire Arrays," Angewandte Chemie InternationalEdition, vol. 42, pp. 30313034, 2003.
[112] J. Joo, et al., "Face-selective electrostatic control of hydrothermal zinc oxide
118
nanowire synthesis," Nature Materials, vol. 10, pp. 596-601, 2011.
[113] K. Tomioka, et al., "Control of InAs Nanowire Growth Directions on Si," Nano
Letters, vol. 8, pp. 3475-3480, 2008.
[114] Y. Wei, et al., "Wafer-Scale High-Throughput Ordered Growth of Vertically
Aligned ZnO Nanowire Arrays," Nano Letters, vol. 10, pp. 3414-3419, 2010.
[115] A. R. Madaria, et al., "Toward Optimized Light Utilization in Nanowire Arrays
Using Scalable Nanosphere Lithography and Selected Area Growth," Nano Letters,
vol. 12, pp. 2839-2845, 2012.
[116] C.-M. Hsu, et al., "Wafer-scale silicon nanopillars and nanocones by Langmuir-Blodgett assembly and etching," Applied Physics Letters, vol. 93, pp. 133109-3,
2008.
[117]
[118]
Q. Xie, et al., "Atomic layer deposition of TiO 2 from tetrakis-dimethyl-amido
titanium or Ti isopropoxide precursors and H20," Journal of Applied Physics, vol.
102, pp. 083521-6, 2007.
B.
Kramm,
et al.,
"The band alignment of Cu 2 0/ZnO and Cu 2 0/GaN
heterostructures," Applied Physics Letters, vol. 100, pp. 094102-3, 2012.
119