Sputter deposition and characterization of indium-tin-oxide thin films and

Sputter deposition and characterization of indium-tin-oxide thin films and
Sputter deposition and characterization of
indium-tin-oxide thin films and
indium-tin-oxide/silicon interfaces for silicon solar
cell application
Margrethe Holmer Rein
October 1, 2015
To my family
ii
Preface
As I’m writing this preface, I’m about to accomplish a work carried out in one of the
most exciting and inspiring positions that I can imagine. During the years working
as a PhD student I have had the opportunity to work with an important topic that
has a huge impact globally, as photovoltaics will play an essencial role in how we will
generate electric power in the future.
However, researchers are dependent on their colleagues, as discussions, assistance
and network are essential elements in a researcher’s daily work. I will therefore use the
opportunity to say how much I have appreciated their contributions.
First of all I will proclaim a special thanks to my three supervisors Arve O. Holt,
Jeyanthinath Mayandi and Edouard Monakhov. Thanks to Arve for having faith in me
and giving me this opportunity to work with material research and photovoltaics. Also
thanks for always listening to any kind of problem and suggesting proper solutions,
and giving advise when the work was not straight forward. Thanks to Jeyanthinath
for supervising and inspiring me in the laboratory, in addition to discussion and interpretation of results. Also thanks for briefing me in cultural differences between India
and Norway. And thanks to Edouard who always keeps his head calm and sees everything in the big picture. Thanks for careful advising and reading through publications
and thesis. All three have complemented each other, which has given me supervision
including experiences and challenges that I appreciate very much. Thank you.
Second of all, a huge thanks to Andreas Klein and his Oxide group for granting me
a stay and making me feel welcome at Technische Universität Darmstadt/Technical
University of Darmstadt (TUD). A special and warm thank you to Mareike Hohmann
who spent late evening hours with my experiments (so I could take care of and spend
time with my little son) and for introducing me to DAISY-MAT and teaching and
giving me support on XPS and UPS. You all made my stay at TUD and my PhD
special.
I would also like to thank the young and inspiring group at the Solar energy department at Institutt for Energiteknikk/Institute for Energy Technology (IFE). A special
thanks to Smagul Karazhanov for always having his door open for discussions. Thanks
to Annett Thøgersen who has given me precious support and information regarding
interpretation of XPS results.
iv
Preface
At last, but not least, thanks to my Morten, which whom I have and forever will
share all my problems and pleasures in life.
Abstract
Indium-tin-oxide (ITO) is a material that conducts electricity and transmits light better
than any other transparent conducting oxides (TCO). Due to these excellent properties,
applications of the material are several. Applied as a transparent coating on thin film
silicon solar cells, a TCO layer can serve as a window into the absorbing part of the
solar cell and as an efficient material for light trapping. Thin film silicon solar cells
also utilize TCO for contacting, where the TCO layers help to improve charge carrier
transport, due to high series resistance of the doped silicon layers.
In this thesis, a study of ITO as a front contact for silicon solar cell application is
presented. The material has been deposited onto glass and silicon substrates by the
use of magnetron sputtering. Two different sputtering devices have been applied. The
sputtering device at Institute for Energy Technology (IFE) was of ”industrial size”,
while the sputtering device at Technical University of Darmstadt (TUD) was of ”research size”, but connected to a larger research tool, which allowed sputter deposited
samples to be transferred to an X-ray Photoelectron Spectroscopy (XPS) characterization chamber without breaking vacuum.
The aim of the project was to develop a low-temperature process in order to produce
a low cost heterojunction silicon solar cell. Hence, the focus of the work in this thesis
was to develop a low-temperature sputter deposition process to deposit ITO thin films
with sufficient electrical and optical properties. The large sputter device at IFE enabled
an easy transfer of the process to industrial scale.
The resulting films have been characterized by a variety of characterization methods. First of all, measurements of the transmittance and conductivity have been carried
out, as these properties are two of the most important properties when ITO is used
as a front contact on silicon solar cells. Other measurements such as Secondary Ion
Mass Spectroscopy and X-ray Diffraction have been carried out in order to examine
properties such as chemical composition and crystal structure. The interface between
the ITO film and silicon substrates has been characterized by utilizing Transmission
Electron Microscopy (TEM) and XPS.
The properties of the ITO films have been studied as a function of deposition
parameters. A preliminary study was carried out in order to uncover the best suited
parameters, such as deposition power, pressure and temperature, for low temperature
vi
Abstract
deposition of ITO. Another study was carried out in order to examine the influence of
addition of different deposition gases. The gases examined in this study were oxygen,
hydrogen and nitrogen. The highest transmittance and lowest resistivity was achieved
when ITO was deposited with a 1.6 flow% of oxygen.
An in situ XPS study of the initial growth of ITO on a silicon substrate was carried
out at TUD by utilizing the unique DArmstadt Integrated SYstem for MATerial science
(DAISY-MAT). This study made it possible to uncover the interaction between the
two materials during the initial stages of the establishment of an ITO film on a silicon
substrate. The study revealed the presence of elemental indium and tin and that several
silicon sub-oxides constituted the interfacial silicon oxide layer between the ITO film
and the silicon substrate, when the deposition was performed at room-temperature.
The presence of indium and tin at the interface between silicon and ITO was confirmed by a TEM and XPS study carried out on one of the ITO/Si samples processed at
room-temperature at IFE. During the TEM it was observed that indium nanoclusters
were crystallized and that the thickness of the interface oxide increased, due to high
energy electron irradiation. The TEM characterization also revealed oxygen deficient
regions at the interface and confirmed that the room-temperature deposited ITO films
were crystalline.
List of Figures
1.1
Women educated at the Barefoot College installing solar cell modules.
Reprinted with permission from the Barefoot College. . . . . . . . . . .
2
Al doped ZnO with different carrier concentration and hence, different
optical properties. [1] . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
2.2
The cubic bixbyite structure of In2 O3 . [2] (In2=d-site and In1=b-site.)
13
2.3
The history and forecast of the worlds energy consumption (in British
thermal units) presented by U.S. Energy Information Administration. [3] 15
2.4
Record efficiencies of solar cells at research level. [4] . . . . . . . . . .
16
2.5
An ideal band diagram for ITO/Si SIS solar cells. [5]
. . . . . . . . .
18
2.6
Temperature coefficient of HIT solar cells. [6] . . . . . . . . . . . . . .
19
2.7
The structure of a HIT solar cell. Adapded from Ref. [7] . . . . . . . .
20
3.1
The sputtering technique. Adapted from Ref. [8]
. . . . . . . . . . . .
25
3.2
The sputtering process is a cascade of incidents, which finally ejects a
target atom. [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
3.3
The Crooke’s, Faraday and anode dark spaces. Adapted from Ref. [10].
27
4.1
The sputter equipment at IFE.
4.2
An SiOx free ITO/Si interface we were not able to reproduce.
2.1
. . . . . . . . . . . . . . . . . . . . . .
32
. . . . .
35
4.3
DAISY-MAT. [11] . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
4.4
An energy scheme of a sample in contact with a XPS spectrometer. [12]
40
4.5
The method for the extraction of the EF − EVBM values. . . . . . . . . .
41
5.1
Resistivity of as-deposited (filled symbols) and annealed (15 min at 300◦ C
in air) (open symbols) films deposited with additional supply of oxygen,
hydrogen and nitrogen. . . . . . . . . . . . . . . . . . . . . . . . . . . .
46
5.2
Average transmittance in the wavelength range 400-1000 nm of as-deposited
(filled symbols) and annealed (15 min at 300◦ C in air) (open symbols)
films deposited with additional supply of oxygen, hydrogen and nitrogen. 47
vii
viii
List of Figures
Counts of secondary 1 H ions per second as a function of depth for asdeposited (filled symbols) and annealed (open symbols) samples of ITO
deposited with hydrogen/argon or argon as sputtering gas. The noise
that occurs at about 0.07 to 0.09 μm is due to the interface between the
ITO film and the glass substrate. . . . . . . . . . . . . . . . . . . . . .
5.4 Counts of secondary 1 H ions per second as a function of depth in asdeposited (filled symbols) and annealed (open symbols) samples of ITO
deposited with oxygen/argon, hydrogen/argon or argon as sputtering gas.
The noise that occurs at about 0.08 to 0.09 μm is due to the interface
between the ITO film and the glass substrate. . . . . . . . . . . . . . . .
5.5 Valence band spectra of as-deposited (top) and annealed (bottom) ITO
films versus oxygen flow% during deposition. . . . . . . . . . . . . . . .
5.6 Content of nitrogen shown by 14 N16 O- ions measured by SIMS on asdeposited (Ad) and heat treated (Ht) films. N0 , N2 and N4 corresponds
to 0, 3.2 and 25.0 flow% nitrogen. The noise that occurs at 700 to 1000
Å is due to the interface between the ITO film and the glass substrate. .
5.7 Valence band spectra of as-deposited (top) and annealed (bottom) ITO
films versus nitrogen flow% during deposition. . . . . . . . . . . . . . .
5.8 XRD pattern of as-deposited (upper row) and annealed (lower row) ITO
films deposited with different flow of oxygen, hydrogen and nitrogen. . .
5.9 HRTEM images of the ITO/Si interface, (A) before and (B) after electron beam exposure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.10 In 3d5/2 , Sn 3d5/2 , O 1s and Si 2p spectra of substrate, after 2, 10, 40
and 940 seconds of ITO deposition. Some of the peaks are multiplied
with an integer in order to improve the illustration of the spectra. The
insets are expansions of the silicon oxide established after 2 and 10 seconds.
5.3
49
50
53
55
56
58
61
63
List of Tables
4.1
4.2
4.3
33
33
4.5
Deposition and annealing parameters used for sample preparation. . . .
Pressure values examined in the three experiments. . . . . . . . . . . . .
Power examined in the two sets of experiments with different sputtering
gas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Flow of oxygen and hydrogen during sputter deposition at different temperatures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The standard deviation of the measurements. . . . . . . . . . . . . . . .
5.1
5.2
Flow% of different gases in the experiments described below. . . . . . .
Deposition and annealing parameters used in the ”best of” recipe. . . .
45
66
4.4
ix
34
34
39
x
List of Tables
Contents
Preface
iii
Abstract
v
1 Introduction
1
1.1
Energy and society . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1.2
Why Indium Tin Oxide and solar cells? . . . . . . . . . . . . . . . . . .
3
1.3
Outline of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
2 Transparent conducting oxides
2.1
2.2
2.3
7
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.1.1
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.1.2
Basic physics . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
Indium tin oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
2.2.1
Physical properties . . . . . . . . . . . . . . . . . . . . . . . . .
12
TCO and solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
2.3.1
Our energy requirements . . . . . . . . . . . . . . . . . . . . . .
14
2.3.2
Solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
2.3.3
TCO requirements for solar cell application
. . . . . . . . . . .
15
2.3.4
The ITO/Si solar cell . . . . . . . . . . . . . . . . . . . . . . . .
18
2.3.5
The Heterojunction with Intrinsic Thin layer (HIT) solar cell . .
19
3 Magnetron sputtering
23
3.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
3.2
Sputtering physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
3.2.1
Sputter yield . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
3.2.2
Energy regimes . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
Plasma theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
3.3.1
Direct current (DC) diode plasma . . . . . . . . . . . . . . . . .
26
3.3.2
Radio frequency (RF) diode plasma . . . . . . . . . . . . . . . .
27
3.3.3
Magnetron plasma . . . . . . . . . . . . . . . . . . . . . . . . .
28
Sputter deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
3.3
3.4
xi
xii
Contents
3.4.1
Deposition of thin films and thin film formation . . . . . . . . .
28
3.4.2
Deposition rates . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
3.4.3
Reactive sputtering . . . . . . . . . . . . . . . . . . . . . . . . .
29
4 Thin film fabrication and characterization
31
4.1
Sputter deposition at IFE . . . . . . . . . . . . . . . . . . . . . . . . .
31
4.2
Sample preparation at IFE . . . . . . . . . . . . . . . . . . . . . . . . .
31
4.2.1
32
4.3
Sputter deposition at Technische Universität Darmstadt
. . . . . . . .
36
Sample preparation at TUD . . . . . . . . . . . . . . . . . . . .
37
Characterization methods . . . . . . . . . . . . . . . . . . . . . . . . .
38
4.4.1
Electrical characterization techniques . . . . . . . . . . . . . . .
38
4.4.2
Optical characterization techniques . . . . . . . . . . . . . . . .
38
4.4.3
X-ray Photoelectron Spectroscopy . . . . . . . . . . . . . . . . .
38
4.4.4
X-ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . .
41
4.4.5
Transmission Electron Microscopy . . . . . . . . . . . . . . . . .
42
4.4.6
Secondary Ion Mass Spectroscopy . . . . . . . . . . . . . . . . .
43
4.3.1
4.4
Optimizing film properties - preliminary studies . . . . . . . . .
5 Summary of results and conclusions
5.1
Optimizing film properties - addition of gases . . . . . . . . . . . . . .
45
5.1.1
Paper I - Hydrogenated ITO and its properties
. . . . . . . . .
47
5.1.2
Paper II - Annealing effect on ITO films sputtered with argon,
oxygen and hydrogen . . . . . . . . . . . . . . . . . . . . . . . .
49
Paper III - Annealing of ITO films sputtered with argon and
oxygen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
Paper IV - Optical electrical, chemical and structural properties
of nitrogen doped indium tin oxide thin films . . . . . . . . . . .
54
Electrical and optical properties versus structural properties . .
57
Interfacial properties . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
5.1.3
5.1.4
5.1.5
5.2
45
5.2.1
Paper V - Elemental distribution and oxygen deficiency of magnetron sputtered indium tin oxide films . . . . . . . . . . . . . .
60
Paper VI - An in situ X-ray photoelectron spectroscopy study of
the initial stages of rf magnetron sputter deposition of indium
tin oxide on p-type Si substrate . . . . . . . . . . . . . . . . . .
61
5.3
General conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
65
5.4
Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
66
5.2.2
Bibliography
78
List of publications
81
Contents
xiii
Appended Papers
83
Paper I
85
Paper II
91
Paper III
97
Paper IV
105
Paper V
117
Paper VI
127
xiv
Contents
Chapter 1
Introduction
1.1
Energy and society
”Energy is a critical part of boosting prosperity and eradicating poverty.”
Jim Yong Kim, President, World Bank Group [13]
In Tilonia in India, a college for poor people, named the Barefoot College, is educating the poorest of the poorest people on earth. At this college, less advantaged
women are educated in order to develop livelihood in rural villages. Among a handful
of educations leading to professions such as midwives, hand pump mechanics, artisans,
weavers, crèche teachers, parabolic solar cooker engineers, FM radio operators and fabricators, dentists, masons, and day and night school teachers, they also educate solar
engineers.
Generally, the women who are educated within solar electricity are of the elderly
generation in their home villages and do not know how to read or write. They are
carefully selected from all over the world by delegates from the college and sent to
India for six months of training and education to be able to install and maintain solar
cell modules in their home villages. The women are preferred as they do not ask for
any certificate, which they will not get, and they do not run away from the village in
order to get a job in the city like experience has shown men often do. [14]
Among many benefits, the education of the female solar engineers has resulted
in solar electrification of hundreds of schools and education centers across India. In
addition to 350 villages in India, several villages in sub-Saharan Africa and Afghanistan
and the total campus area of the Barefoot college, of about 80 000 square feet, have
been solar electrified. Due to the replacement of generators and oil lanterns with solar
power, expenditures for over 500,000 liters of kerosene are saved annually. Deforestation
close to the villages has decelerated. The results are reduction of total 13 tons of CO2
per day and saving time on collecting and carrying fuel over long distances. [14]
The advantages of the electrification of the households are several. The light inside
2
Introduction
the houses is brighter and does not emit any soot or particles that can cause health
damages. Children that have to work during the day, have the possibility to go to school
in the evenings. The cooking can be done on electrical hot-plates, reducing the need for
biomass to make fire. At a personal level, the electrification work and responsibility for
the village’s energy generation are changing the women’s position in the village, giving
them an income, more dignity and acknowledgment. It is evident that the ripple
effects of the education that the women receive at the Barefoot College are several
and positive. All this contributes to a positive development and an increasing living
standard in the poor areas of the world. [14]
Figure 1.1: Women educated at the Barefoot College installing solar cell modules.
Reprinted with permission from the Barefoot College.
Most people in developing countries are living close to the equator and areas which
receive up to 2500 kWh/m2 a year from the sun [15]. If electricity from solar panels can
meet the increasing needs of energy in the developing countries, this would be the best
solution to prevent more global warming caused by increasing energy consumption. In
addition to less pollution due to combustion of fossil fuels, vegetation in areas close
to the rural villages will now contribute to the binding of CO2 and construction of
new electrical networks for energy supply to the rural villages will be superfluous. The
latter is a rather expensive investment and the probability for making such investments
in many developing countries is low. Hence, electrification via solar panels is giving
more people access to all the benefits of electrical power.
One of the most comprehensive challenges in the world today is global warming,
due to emission of CO2 and other greenhouse gases, and the consequences thereof. As
the main part of CO2 emission is not taking place in developing countries, it is evident
that the responsibility for a huge reduction in CO2 emission lies on the industrialized
countries [16]. IPCC states in their report Climate Change 2014: Mitigation of Climate Change that ”Integrated modelling studies indicate that decarbonizing electricity
Why Indium Tin Oxide and solar cells?
3
supply will play an important role in achieving low CO2 equivalent (eq) concentration
stabilization levels”. This emphasizes the importance of directing our production of
electrical power to renewable sources [17].
Generating electricity via solar cells is one of the most environmental friendly ways
to produce electrical power, since solar power plants have no emissions. Although
production of solar cells requires some energy, today the energy payback time is less
than 2 years and the carbon footprint of the lifecycle of PV systems are less than
30 g CO2 -eq/kWh [18]. Replacing electrical power based on gas and coal with solar
power, CO2 emissions will be reduced by 10-20 times, depending on the fossil fuel and
technology used. [17] (Chemical waste, emission of criteria pollutants and heavy metals
are other issues associated with production of PV systems [18]. This requires cleaning
technology at every production step, in order to achieve low emissions and minimal
pollution.)
1.2
Why Indium Tin Oxide and solar cells?
As the importance of photovoltaics hereby is stated, it is evident that research on PV
is crucial. This is the only way we can develop solar cell technology and make the
production process of the solar cells even more environmental friendly. This can be
done by optimizing solar cell production methods and/or improve solar cell efficiencies.
Cost reduction of solar power is an implicit advantage.
One example of how research for almost forty years has resulted in one of the highest
efficiencies for solar cells produced for practical application is the Heterojunction with
Intrinsic Thin layer (HIT) solar cell by Panasonic/Sanyo. Even though this solar cell
structure has been continuously developed throughout the years, it still has a great
potential for further efficiency improvement. The latest record is 25.6% and the solar
modules are produced with one of the highest efficiencies on the market.
One of the materials used in HIT solar cells is the transparent conducting oxide
(TCO) ”indium tin oxide” (ITO). This is a material which has several applications,
and its usefulness is mostly due to its excellent electrical and optical properties. For
solar cell purposes, no other TCO has shown better electrical and optical properties.
This thesis is a study of sputter deposited ITO thin films. The work was a part of
a collaboration project on highly efficient silicon based heterojunction (HJ) solar cells.
The members of the project group were Renewable Energy Corporation (REC), the
University of Oslo (UiO) and the Institute for Energy Technology (IFE). The goal of the
project was to develop cost-effective HJ solar cells, by reducing the energy consumption
in solar cell production. Hence, low-temperature processes at every processing step of
the solar cell production were required.
The object of the PhD work was to develop a low-temperature deposition process
4
Introduction
of an ITO layer used as an electrode on top of the HJ solar cell structure. It was
highlighted that the process would be done by the use of a magnetron sputter of
industrial size, in order to easily transfer the processes to industrial scale. Several
properties of the ITO layers have been studied, including the interface between ITO
and silicon.
1.3
Outline of thesis
The structure of the thesis is as follows. It starts with giving the reader essential
background information (chapter 2 and 3). The first part is a review chapter (chapter
2) of TCOs, giving the reader a touch of the basic properties and the importance of
the material studied in this work. The second part (chapter 3) is an introduction to
the physics of sputtering, which is the deposition method utilized for film deposition
in the present work.
The thesis continues with a report of the experimental work carried out during
the PhD period. In chapter 4, the experimental methods used for deposition and
characterization of the films are thoroughly explained. In chapter 5, a summary of
the main results and conclusions from the PhD work is presented. The results from
ITO/Si studies and different properties of ITO films deposited with addition of oxygen,
hydrogen and nitrogen are illustrated. The fully and detailed description of the results
are found in the appended papers. Chapter 5 also includes a section about future work.
The papers have been carried out with assistance from supervisors and colleagues
with special education in operating advanced characterization equipment. All TEM
characterizations have been carried out by Annett Thøgersen. The XPS performed at
the Technical University of Darmstadt was carried out by Mareike Hohmann and me,
while all other XPS was carried out by Spyros Diplas at the University of Oslo/SINTEF.
Lasse Vines and Claus Magnus Johansen performed the secondary ion mass spectroscopy at the University of Oslo. Jeyanthinath Mayandi produced some of the indium tin oxide samples doped with hydrogen and did the electrical characterization of
these. Except for this, sample preparation, electrical and optical characterization and
XRD were carried out by me.
I’m the first author of Paper I, II, III, IV and VI. In paper VI, Mareike Hohmann
contributed with the technical information of DAISY-MAT. The first author of Paper
V is Annett Thøgersen. This is a paper on TEM and XPS characterization of the
interface of a ITO/Si sample that I prepared, i.e the sputter deposition and annealing of
the sample. In the manuscript, I wrote the text about the experimental part concerning
the sample preparation. In addition, I took part in the discussion of the results and
proof reading of the manuscript.
Outline of thesis
5
6
Introduction
Chapter 2
Transparent conducting oxides
2.1
Introduction
Transparent conducting oxides (TCOs) are materials having the property of simultaneously conducting electricity and exhibiting transparency in the visible region of the
electromagnetic spectrum. The materials are often a compound of oxygen and one
or two metallic elements and the material properties are highly dependent on the elements used. The properties are also affected by the quality of the material, which
often is dependent on the deposition method. High quality TCO films (thickness 30420 nm) deposited by pulsed laser deposition and magnetron sputtering have shown
transparency as high as ∼90% and a resistivity in the 10−5 Ω cm range [19, 20].
TCOs were first reported by Badeker in 1907. He discovered that a thin film of
cadmium could be made transparent by oxidizing it. Fortunately, the film was still
conducting electricity. Since then, several other materials have been shown to exhibit
this unique combination of properties. Some of the most frequently applied TCOs
today are antimony doped tin oxide (ATO), tin doped indium oxide (ITO), aluminium
doped zinc oxide (AZO), fluorine doped tin oxide (FTO) and niobium doped titanium
oxide [21].
Over the last decade, new transparent and conducting materials for flexible substrate application have been studied intensively. Fortunately, research on materials like
nanowires of pure metals, plasmonic graphene and highly doped conjugated organic
polymers has shown that they exhibit sufficient optical transparency and electrical
conductivity as transparent conducting materials (TCMs) [22, 23, 24].
2.1.1
Applications
TCOs are utilized for a large quantity of applications. Though, the applications are
dominated by only a handful of TCOs: FTO, ITO and a variety of different doped
ZnO.
8
Transparent conducting oxides
The most common TCO applied in the industry today is FTO. This material is
used in the production of windows having a low radiative heat loss, called ”energy
efficient windows” or ”low-e windows” (low emissivity windows). Such windows are
able to maintain the heat indoor during the winter months and reflect or absorb the
infrared light from outside during the summer months. This way, both heating and
cooling costs will decrease [21, 25].
The second largest application area for TCOs is flat panel displays. This is a
continuously growing market and ITO is the main TCO applied. Though, recently
amorphous indium doped zinc oxide has partly replaced ITO due to better patterning
properties and thermal stability [21].
The third largest industry application for TCOs is front electrodes in photovoltaics.
ITO is the most utilized TCO for this purpose. However, indium is an expensive
substance and the abundance of indium is low. This has raised zinc oxide as an
alternative TCO in photovoltaics [21].
Other important application areas of TCOs are for instance gas sensors, oxide
based thin film transistors and electrochromic devices, e.g. windows with adjustable
transparency, and automatically dimming of rear-view mirrors for automobiles [26, 27,
28]. The applications of TCOs show that such materials play an important role in
the development of ”green” technologies. Energy efficient windows and light-emitting
diodes (LED) for indoor lightning [29] have a significant impact on lowering energy
consumption, while photovoltaics are contributing to a more sustainable and clean
power generation [21, 30].
2.1.2
Basic physics
As previously stated, the unique combination of transmitting and conducting properties of TCOs makes the materials suitable for several applications. However, the two
properties are not independent. A change of one of the properties will affect the other.
The nature of the transmitting property is associated with the wide band gap exhibited by the TCO. In order to be transparent in the visible range, the TCO material
generally has a band gap larger than 3 eV, which implies an absorption onset at approximately 400 nm. At the long wavelength side of the visible spectrum (near infrared), the
transmission is limited by the electron plasma oscillation [31]. Both limits are affected
by changes in the electron concentration. One effect of increased electron concentration (N) is a shift of the Fermi level to higher energies, called the Burstein-Moss shift
[32, 33]. When the Fermi level is increased, the lowest states in the conduction band
are blocked. This requires incident photons of higher energy in order to be absorbed,
and hence the onset of the absorption is shifted toward higher energies [21, 34, 35, 36].
The second effect of increasing the electron concentration is a shift of the plasma
9
Introduction
frequency towards higher frequencies, as seen in Eq. 2.1:
ωp ∝
√
N
(2.1)
where ωp is the plasma frequency and N is the number of electrons [37]. This introduces
absorption and hence, a reduction in transparency at longer wavelengths. Figure 2.1
illustrates an example of Al doped ZnO with different absorption and transmission
spectra due to a change in carrier concentration. A higher level of dopants and hence,
a higher carrier concentration induces a shift of the transmission and absorption toward
shorter wavelengths. The figure also illustrates the proportion of the two effects. On
one hand, the Burstein-Moss effect is hardly seen at the short wavelength side. This
is explained by the compensating many-body effect in highly doped TCO films, which
induces a band gap shrinkage [38, 39]. On the other hand, one can observe significant
reduction in transmission at the longer wavelengths due to the plasma frequency shift.
Figure 2.1: Al doped ZnO with different carrier concentration and hence, different
optical properties. [1]
Understanding the nature of the conducting property of TCO materials has been
attemted for years. Bellingham et al. [40] calculated the resistivity limit of the In2 O3 ,
SnO2 and ZnO systems. Based on the assumption that ionized impurity scattering
was the limiting factor of the resistivity that could be achieved, the limit was found
to be 4 × 10−5 Ω cm. The types of defects and impurities which may be the origin
of this scattering are discussed by King et al. [34]. Usually, the main contribution
to the conductivity of TCOs is assigned to native defects such as oxygen deficiencies.
However, King et al. [34] discussed whether other defects like cation interstitials,
cation vacancies and defect complexes could explain the conductivity’s well-known
oxygen dependency. Nevertheless, they conclude that the origin to conductivity in
TCO materials is rather complex, suggesting that substitutional hydrogen, in addition
to native defects and interstitial hydrogen trapped at native defect centers are most
10
Transparent conducting oxides
likely the main contributors to the total conductivity of TCO materials.
Another aspect which should not be ignored when discussing the conductivity of
a TCO material is the surface conductivity. As for semiconductors such as silicon,
the surface conductivity differs from the conductivity in bulk. Unlike semiconductors,
which can exhibit a carrier depletion at the surface, TCOs may exhibit an accumulation
of carriers at the surface. This certainly has an effect on the electrical properties at
surfaces and interfaces of TCO materials. Gassenbauer et al. [41, 42] and others
(see references in [41]), have shown that the metallic nature of the surface of highly
conductive ITO is due to the presence of surface states and not due to the electrons in
the conduction band. They also showed that the Fermi level at an ITO surface could
be lowered by a higher oxygen content in the deposition gas, causing an increase in
the surface work function. This was attributed to a passivation effect of the surface
states at the ITO surface. (Some TCOs like SnO2 and ZnO may exhibit depletion or
accumulation properties, depending on the bulk Fermi level and the stoichiometry of
the surface. This property makes them sufficient for application in gas sensing devices
[34]).
As the electrical and optical properties are highly dependent on each other, one
interesting issue would be: is it possible to improve the optical and electrical properties simultaneously? T. J. Coutts et al. [37] made a theoretical approximation to this
problem. They performed a modelling of optical properties as a function of parameters
like relaxation time τ and carrier concentration n. Their starting point was the Drude
theory, where it is presumed that the electrons are put in motion due to an electromagnetic field. The time and frequency dependent position, velocity and acceleration
of an average electron are modelled by the Lorentz oscillator equation of motion [37]:
→
−
d2 x m∗ dx
+
+ Kx = e E (t)
(2.2)
2
dt
τ dt
where m∗ is the electron effective mass, τ is the relaxation time of the electron, e is the
electronic charge, K is the restoring force per unit displacement between the electrons
→
−
and their host ion cores and E is the electric field strength [37]. The relaxation time
τ is interpreted as the time interval between the electron collisions. It is further shown
that the imaginary part of the frequency-dependent dielectric permittivity is inversely
proportional to the relaxation time. Equation 2.3 and 2.4 show the real and imaginary
part of the permittivity, respectively:
m∗
1 = ∞ −
2 =
ne2
0 m ∗ ω 2
ne2
0 m∗ ω 3 τ
(2.3)
(2.4)
11
Introduction
where n is the carrier concentration, ω is the plasma frequency, ∞ is the high-frequency
permittivity [37] and 0 is the permittivity of vacuum. The definition of the absorption
coefficient shows the correlation between the relaxation time and the absorption in a
material:
4πk
(2.5)
α=
λ
where λ is the wavelength and k is the extinction coefficient defined as:
k=
1
1 2
1
(1 + 22 ) 2 −
2
2
(2.6)
The modelling showed that increasing the carrier concentration caused a shift of the
absorption peak toward higher frequencies, reducing the transmission, which coincides
with the theory above. However, increasing the relaxation time decreased both the
maximum and the half-width of the absorption peak. Thereof, they concluded from
the modelling that the focus on improving the electrical and optical properties of TCOs
should be on developing materials with long relaxation times and hence, high carrier
mobilities.
T. J. Coutts et al. [37] also showed that there might be another way to improve
the TCO properties through the definition of a figure of merit ψ:
σ
α
(2.7)
ne2 τ
m∗
(2.8)
ψ=
where σ is the d.c. conductivity defined as:
σ=
By some calculations (see reference [37] for details) the formula for ψ takes the form:
ψ=
√
1 c0 τ 2 ω 2
(2.9)
where c is the speed of light. The equation highlights the importance of the relaxation time parameter and indicates one more parameter of importance, the 1 , which
is directly proportional to the high-frequency permittivity (∞ ). An attempt to produce films with higher ∞ showed that the near infrared transmittance was increased.
They also observed that this change in the optical property was a function of carrier
concentration. The lower the n, the greater was the effect of increased ∞ .
12
2.2
Transparent conducting oxides
Indium tin oxide
ITO transmits light in the visible range and conducts electricity better that any other
TCO. Nevertheless, there is a vast amount of research on other TCOs and materials
that can substitute ITO. This is mostly due to the fact that indium is a rare and
expensive element. The increasing production of flat panel displays, commonly liquid
crystal displays, has accelerated the price to higher levels. Last decade, the price of
indium increased from below US$100/kg in 2002 up to US$715/kg in 2013 [43]. The
high prices and low abundance of indium (approximately 0.05 part per million in the
continental crust [44]) have made recycling of indium and especially ITO extremely
important. In 2007, recycling constituted about 650 tons indium, while the production
was 476 tons [45]. (World total refinery production for 2013 was 770 tons.) The
consumption of indium is expected to increase in the years to come due to increased
production of flat panel displays. Nevertheless, as the price increases the reclaim of
indium becomes more profitable. In the future, it is assumed that consumption can
be balanced by production and recycling, and the price of indium will remain high.
However, this will depend on the development of ITO substitutes. [43, 45]
2.2.1
Physical properties
While the thin films of ITO are transparent, a sintered sputtering ITO target can
appear as yellowish-grey. The crystal structure of ITO is body centred cubic bixbyite,
as for In2 O3 , when not heavily tin doped (Figure 2.2). A corundum structure or
perovskite-like structure is exhibited when the material has been processed under high
temperature and pressure [46]. The cubic bixbyite structure has 80 unit cell atoms
[47]. The lattice constant is tuneable via changes in oxygen partial pressure during
deposition. This was shown by Fan et al. [48] who reported values between 1.015 and
1.023 nm, dependent on the oxygen partial pressure applied. The space group is Ia3
[49, 50] and the indium atoms are either positioned in b-sites or d-sites. The b-sites
have a body diagonal of oxygen vacancies, while the d-sites have vacancies at a face
diagonal. There are eight and 24 b- and d-sites per unit cell, respectively. Figure 2.2
illustrates the bixbyite structure of In2 O3 with its two different In sites. In ITO, tin
is positioned at substitutional indium sites in the Sn+4 form [51]. The Sn atom is
contributing with one electron to the carrier concentration. The oxygen vacancies are
assumed to be the contributor of two electrons [52, 53, 46].
As ITO is a heavily doped (degenerated) n-type semiconductor, the Fermi-level is
aligned with or lies above the conduction band. The band gap of ITO has been shown
to be direct [54] and in the range of 3.5-4.3 eV [55] based on observed band edge absorption. However, Klein et al. [56] showed recently by photoelectron spectroscopy that
the fundamental band gap of ITO is 2.8 eV. One approach to explain this divergency
Indium tin oxide
13
Figure 2.2: The cubic bixbyite structure of In2 O3 . [2] (In2=d-site and In1=b-site.)
is the presence of an indirect band gap and hence, an energy difference of ∼0.85-0.90
eV between weak and strong optical absorption [57]. It has not been clarified whether
the weak absorption is inhibited by low dipole intensity or that the transition from the
highest valence band to the conduction band is symmetry-forbidden [56]. Thus, the
presence of the indirect band gap in In2 O3 and ITO is not yet explained [21, 56, 58, 57].
The ionization potential of ITO can vary in the range 7.0-8.1 eV, though a common
value is 7.6 eV [56]. It has been shown that the work function can be manipulated
either through doping, which changes the distance between the Fermi-level and the
vacuum level, or through the surface dipole which changes the ionization potential.
For conducting purpose, it is obvious that the work function should be as small as
possible, which implies that the Fermi-level is shifted upwards in the conduction band
and the material is highly degenerated. The work function of ITO is predicted to be
4.7 eV [59], but can be tuned between 3.6 eV and 5.3 eV by post-deposition oxidation
treatments [60] or by adjusting the oxygen present during deposition [61].
Due to the excellent conductivity, ITO is assumed to have a metal-like conducting
behaviour and hence, the resistivity is defined as proportional to the inverse of the
product of the carrier mobility and carrier density. (Typical mobility and electron
density for a ∼ 80 nm thick sputter deposited ITO film is μ = 20 − 40 cm2 /V and
n = 1019 − 1021 , respectively [62].) Based on this resemblance, the Drude theory
is often applied as a model for the conductivity in ITO. Nevertheless, according to
Hamberg and Granqvist [51], the Drude theory is not sufficient as a free electron model
for ITO. The Drude model describes the electrons as a gas of free electrons that are
occasionally scattered by the metal ions. In ITO, other scattering mechanisms such as
electron-defect scattering, electron-lattice scattering, and electron-electron scattering
must be considered as well. For films with good crystallinity Hamberg and Granqvist
claims that ionized dopant impurity scattering is the most important and cannot be
ignored. They showed that this type of scattering and electron-electron scattering have
a band gap narrowing effect and hence, a compensating effect on the Burstein-Moss
14
Transparent conducting oxides
shift [51, 36].
2.3
2.3.1
TCO and solar cells
Our energy requirements
The energy consumption in the world today is higher than ever and it is likely to
increase in the years to come. Predications of such an increment have been attempted
by the U.S. Energy Information Administration. They have proposed a growth of
the world’s total energy demand from 2010 until 2040 to be as high as 56% [3]. An
illustration of this is shown in Figure 2.3. In the same timeframe, Exxon Mobile has
proposed that the electricity demand will increase by 90% [63]. They mainly ascribe
this increment to improved living standards in developing countries, although the power
consumption per capita will still be dominated by the western countries.
In order to meet such energy demands, an increased power generation is needed.
Exxon Mobile indicates in their energy outlook that the use of coal will still increase
in developing countries, although the use of natural gas and renewable energy sources
as wind and solar will expand [63].
The focus on high CO2 emission, global warming and pollution force the energy
consumption and power generation toward a sustainable and clean direction. This
makes renewable energy the most promising energy supply. The biggest challenge with
renewable energy today is the large variations in power generated due to changes in
weather, as is the case for solar and wind power generation. Solutions to this problem
can be distributed storage, such as batteries or balancing power generation from for
instance a gas power plant. Still, renewable power generation has gained a footing
in countries like Germany, where it has been shown that power generated by wind is
compensating the loss in power generated by the sun and vice versa [64].
As the competitiveness of renewable energy increases, the number of installed wind
turbines and solar systems will continue to rise. The U.S. Energy Information Administration states in ”The International Energy Outlook 2013” that renewable energy is one
of the fastest growing power suppliers and predicts an annual growth of approximately
2.5% [3].
2.3.2
Solar cells
In order to increase the competitiveness of solar cells research aims at lowering the
production costs and/or increase the solar cell efficiency. The research includes several
kinds of solar cells. This is illustrated by the figure from National Renewable Energy
Laboratory (NREL) (Figure 2.4) giving an overview of record efficiencies on research
solar cells.
TCO and solar cells
15
Figure 2.3: The history and forecast of the worlds energy consumption (in British
thermal units) presented by U.S. Energy Information Administration. [3]
Solar cells can be based on compounds of semiconductors and can be divided into
three groups. These are the III-V type like GaAs, InP, (Al,Ga)As and (In,Ga)P,
the II-VI type like CdTe and Cd(S,Te), and the I-III-IV2 type such as the CuInSe2 ,
Cu(In,Ga)Se2 and Cu(In,Ga)(S,Se)2 solar cells. Lately, new types of solar cells like
dye-sensitized solar cells, organic solar cells, quantum dot solar cells, plasmonic solar
cells and very recently the promising perovskite solar cells have also been added to the
multiplicity of photovoltaics. Still, in spite of the great variety of photovoltaics, the
silicon based solar cells are preferable. This because silicon is the second most abundant
element in the Earth’s crust (about 28%) [65], which makes silicon highly suitable as a
photovoltaic material in a sustainable and access point of view. This is reflected by the
solar cell production, where crystalline silicon solar cells are constituting more than
90% of the PV market [21].
Solar cells are energy converters based on carrier generation. A carrier is generated
when incident light reaches the surface of a solar cell and a photon with sufficient
energy is absorbed. The absorption implies a transfer of the photon energy to the
carrier which is exited from a ground state to a higher energy level. The next step of
the power generation is the extraction of the carrier from the solar cell. This can be
done via metal fingers or a thin film of TCO or a combination of these.
2.3.3
TCO requirements for solar cell application
The requirements for TCOs used as current collectors in photovoltaics are dependent
on the type of solar cell and hence, various TCOs are applied. The main TCOs utilized
for solar cells with amorphous or microcrystalline silicon solar cells are doped In2 O3 ,
Transparent conducting oxides
16
Figure 2.4: Record efficiencies of solar cells at research level. [4]
17
TCO and solar cells
SnO2 and ZnO. These are utilized in solar cells like the Heterojunction with Intrinsic
Thin layer (HIT) solar cells, the Copper Indium Gallium Selenide (CIGS) and the
dye-sensitized Graetzel solar cell. Other TCOs applied in photovoltaics are TiO2 and
Zn2 SnO4 /Cd2 SnO4 [35, 56]. Research has shown that new transparent conducting
materials like carbon nanotubes, metal nanowire networks, graphene films and ultrathin metal films may replace TCOs in solar cells [66, 67].
In addition to the tasks as a front window and a conducting material, the TCO
can be selected due to other needs as well. For instance, the TCO’s ability of forming
a desired interface with organic or inorganic materials could be one requirement. It
could also be desirable that the TCO can be employed as a diffusion barrier (CIGS) or
as a light trapping material (HIT). For a nano-hybrid polymer cell it is important that
the TCO must be chosen as a material for controlling the contact work function [35].
Beyer et al. [1] have highlighted requirements for TCOs used in thin film silicon
photovoltaics mainly utilizing amorphous silicon as absorber. First of all, they point
out that the transmitting and conducting properties should be as high as possible.
Second, the refractive index should optimize the probability for light to reach the
absorbing part of the solar cell. Other aspects are high carrier mobility and ability of
surface patterning. In addition to this, desired requirements are low price, a non-toxic
material and that the material is sustainable and chemically stable both at interfaces
and in bulk.
According to Beyer et al. [1] the conductivity of the TCO layer as a front contact
should be as high as possible. As the purpose of the TCO layer is to conduct charge
laterally to the contact points (such as metal fingers), high conductivity implies that
the sheet resistance in a TCO layer should be minimized [68].
In order to reduce the sheet resistance of a TCO layer, the three parameters that
can be adjusted, according to equation 2.10, are thickness, mobility and carrier density.
Rsh =
1
eμnt
(2.10)
In this equation, e is the electronic charge, μ is the mobility, n is the carrier density
and t is the thickness of the TCO layer.
A thickness of 65-80 nm is an optimized thickness for TCOs having refractive indices
of 1.7-2.1 as this induces maximum transmittance (minimum reflectance) at 600 nm
[68, 69]. Increasing the carrier density involves lower transparency due to free carrier
absorption. Mobility is therefore the only parameter that can be adjusted in order
to achieve a sufficiently low sheet resistance for minimizing the electrical power loss.
Koida et al. have demonstrated a TCO (hydrogen doped indium oxide) with a carrier
mobility as high as >100 cm2 /Vs [70]. The necessity of high carrier mobility was also
pointed out by Coutts et al. [37], as discussed in the last paragraph in subsection 2.1.2.
18
Transparent conducting oxides
The possibility of patterning the TCO layer is desired for some solar cell structures,
as the TCO can be applied as a scattering material in order to extend the path way of
the incident light. One material studied for this purpose is AZO. By applying appropriate deposition parameters and wet etching of AZO, sufficient scattering properties
can be achieved [66].
2.3.4
The ITO/Si solar cell
The requirements pointed out by Beyer et al. ([1]) can partly be fulfilled by TCOs
like ZnO, SnO2 and ITO. Still, the preferred TCO for a multitude of solar cells due
to the excellent conducting and transmitting properties and chemical stability is ITO.
One of the first solar cell applications of ITO was initiated already in the late 1970s,
when the photovoltaic effect of a ITO/Si junction was studied. This was a variety
of the Schottky barrier and metal insulator semiconductor (MIS) solar cell. It was
stated that the ITO/Si junction had an excellent potential for photovoltaic applications [71] and conversion efficiencies up to 13 % were reported [72]. The junction
of the highly degenerated n-doped ITO material and the c-Si was labelled degenerate semiconductor-insulator-semiconductor (SIS) cells. An ideal band diagram of this
structure is illustrated in Figure 2.5.
Figure 2.5: An ideal band diagram for ITO/Si SIS solar cells. [5]
A real ITO/Si band structure includes a thin (1-3 nm) silicon oxide layer between
the ITO layer and the Si substrate. This has obviously an impact on the electrical
property of the ITO/Si junction. Goodnick et al. [73] showed a thermal degradation
of ITO/p-Si solar cells due to growth of additional interfacial SiOx , while Kobayashi et
al. [74] states that the presence of the thin insulating layer increases the photovoltage
of such a solar cell. The latter could be due to effects like i) number of interface states
in the semiconductor band gap decreases due to the silicon oxide layer or ii) dark
current density decreases due to the suppression of the transfer probability of majority
TCO and solar cells
19
carriers. Kobayashi et al. [74] studied the mechanisms of carrier transport through a
silicon-oxide layer for ITO/SiOx /Si solar cells. When they deposited an ITO layer at
elevated temperature on a flat Si surface, they found that the thickness of the SiOx
layer affected the dark current density under a depletion condition. They concluded
the study by stating that quantum mechanical tunnelling in this case is the dominant
mechanism for the charge carrier transport through the silicon-oxide layer.
One SIS structure that has shown promising efficiencies is the Triex hybrid Si solar
cell from Silevo. This structure includes a thin tunnelling oxide layer and a thin-film
passivation layer. Silevo has reported up to 22% efficiency on this solar cell structure
[75], which has a similarity to the HIT solar cell [5].
2.3.5
The Heterojunction with Intrinsic Thin layer (HIT) solar cell
The HIT solar cell was developed by Sanyo Ltd. in 1975. In 1997 they started mass
production. HIT solar cells exhibit one of the highest solar cell module efficiencies found
on the market [76, 77] and offers several benefits. Today, modules are fabricated with
an efficiency of 19.4% [78] and the latest reported efficiency on a 100 cm2 research solar
cell was 24.7% (Feb. 2013) for a standard HIT solar cell. In April 2014, Panasonic
achieved 25.6% efficiency on a back contacted HIT solar cell [79]. Another benefit
of the HIT solar cell is the low degradation of the conversion efficiency during high
temperature operation (<0.3 %/◦ C) compared to conventional crystalline Si solar cells
[6] (Figure 2.6). In addition, the fabrication process is relatively simple and low-priced
and is performed at low temperature (<200 ◦ C) [80].
Figure 2.6: Temperature coefficient of HIT solar cells. [6]
As the name indicates, the HIT solar cell includes a heterojunction. In this case,
the crystalline silicon substrate and the n-type and p-type doped amorphous silicon
20
Transparent conducting oxides
(a-Si:H) layers constitute this junction.
Figure 2.7 depicts a HIT solar cell structure. Both sides of the c-Si substrate are
covered with a thin intrinsic a-Si:H layer, in order to passivate the crystalline substrate
surface. The intrinsic layers are coated by the p-type and n-type doped a-Si:H layers.
On top of the doped a-Si:H layers the TCO film provides the grid electrodes with the
current generated by the junction. The symmetrical structure makes it suitable for
bifacial solar cells (called HIT Double).
Figure 2.7: The structure of a HIT solar cell. Adapded from Ref. [7]
As the structure does not include any high-hardness alloy metals it is a stress-free
structure, which is appropriate for thin c-Si substrates. The latest efficiency record
on a research solar cell was achieved with a 98 μm thick c-Si substrate. Simulations
made by Dwivedi et al. [77] have shown that high efficiencies can be achieved with a
substrate thickness as low as 58 μm. The thickness of the a-Si:H layers were as low
as 6 nm and 3 nm for the doped and intrinsic layers, respectively and the maximum
theoretical efficiency obtained was 27%.
Improved efficiencies and reduction in fabrication costs are the benefits of research
performed on this structure. The role of the different layers and how materials interact
as they are brought in contact and during solar cell operation are the main issues in
order to understand the electrical properties of the HIT solar cell.
To summarize, we have seen that ITO is a TCO material utilized for several applications. It is a sufficient material as a front contact on solar cells, due to its excellent conducting and transmitting properties. However, ITO is a complex material.
The properties depend on the deposition method, the process and the post-deposition
treatment. In order to know how to reach the potential of the ITO properties, a good
understanding of the material is required. What we know about ITO is also essential
for research on other TCO materials, for instance in order find an element that can
substitute indium.
TCO and solar cells
21
In this thesis, we have worked toward a better understanding of the properties of
ITO and the ITO/Si interface. We studied how the electrical and optical properties
of the ITO films were affected by adjusting deposition parameters and post-deposition
treatment. We have also studied the chemical composition of ultrathin ITO layers
deposited onto a silicon substrate and the resulting interface, in order to try to accomplish a better understanding of the charge carrier transport between the two materials,
which is essential in a solar cell.
22
Transparent conducting oxides
Chapter 3
Magnetron sputtering
3.1
Introduction
Thin films of ITO can be deposited by several techniques. For instance, methods
such as wet chemical deposition like the sol-gel process or physical depositions like
electron beam evaporation, physical vapour deposition as well as RF and DC magnetron
sputtering are frequently applied. The resulting film properties are dependent on the
deposition method and the parameters applied during processing.
Due to the ability of large area deposition at high through-puts and low temperature, the preferred deposition method for several applications is magnetron sputtering.
By sputtering, parameters like substrate temperature, gas pressure during processing,
the power applied or introduction of gases can modify the transparency, conductivity,
structural properties or electronic properties like the work function or ionization potential of ITO. For instance, the Fermi-level is shown to be strongly dependent on the
partial pressure of oxygen present during deposition [56]. In addition, the percentage of
Sn in the film and the film thickness are of importance for determining the properties
[51].
Sputtering is a physical deposition technique. The process of sputtering can be
described as an erosion of a target surface by bombardment of energetic particles. The
dislodged target atoms will deposit on the surroundings. Compared to the sputtering
alternative - evaporation - sputtering produces layers of higher quality when the target
is a compound of materials or an alloy and has, in addition, a much better step coverage.
Sputtering is widely utilized for etching as well. Application areas for etching are,
among other things, cleaning, depth profiling and patterning. Cleaning is a procedure
often used in order to remove contamination from a surface, such as pre-sputtering of
a target surface before sputter deposition or sample surface before deposition of a layer
or characterization of the surface. Drawbacks of sputter cleaning of samples can be
damage to the top surface layer and re-sputtering of contamination. Depth profiling
24
Magnetron sputtering
is frequently performed prior to characterization techniques like Secondary Ion Mass
Spectroscopy (SIMS) or X-ray Photoelectron Spectroscopy (XPS), in order to extract
information from bulk materials and interfaces. Sputter etching can also be utilized
for improvement of step-coverage or redistribution of deposited films.
3.2
Sputtering physics
In order to avoid interaction between the sputtered material and the sputtering gas, a
noble gas such as argon, xenon, neon and krypton is applied. The most common gas
used is argon. The typical operation range of gas pressure is mTorr - 1 Torr, and the gas
is naturally partly ionized. Paschen’s law (Eq.3.1) describes the relationship between
the voltage applied in order to ignite the plasma, called the breakdown voltage (Vbd ),
the gas pressure (P) and the distance between the electrodes (L). b is an empirically
determined constant. [10]
Vbd ∝
P ×L
log(P × L) + b
(3.1)
Once the plasma is ignited, the electric field between the electrodes accelerate the
ions toward the cathode. As the ions impinge the target surface, both target atoms
and secondary electrons are dislodged. Generally, only a small part of the gas consists
of molecules (∼5%). The secondary electrons are accelerated toward the anode, which
is either the sample or the surrounding chamber walls. On their way to the anode the
electrons eventually gain enough energy to either ionize the sputter gas atoms or excite
them to an energetic state. The latter will subsequently cause relaxation of the sputter
atom through an optical transition, seen as the plasma glow. The sputtering process
is represented in Figure 3.1.
3.2.1
Sputter yield
The sputter yield S is defined as [81, 10]
S=
number of emitted particles
number of incident particles
(3.2)
and is dependent on the energy of the incident particles, the angle of incidence of the
particles, the target material and the crystal structure of the target. The maximum
sputter yield normally occurs approximately at 1 keV [10].
3.2.2
Energy regimes
Sputtering is not just an ejection of particles from the topmost surface of the target,
but rather a cascade of incidents in a surface region, as illustrated in Figure 3.2.
Sputtering physics
25
Figure 3.1: The sputtering technique. Adapted from Ref. [8]
Figure 3.2: The sputtering process is a cascade of incidents, which finally ejects a target
atom. [9]
26
Magnetron sputtering
When the plasma is initiated, ions with different kinetic energies are accelerated
toward the cathode (target). Dependent on the kinetic energy of the incident ion,
different events will occur. At very low energies up to 40 eV, the ions bounce of the
surface and little sputtering occur [81]. The sputtering yield at 30-40 eV is ∼ 10−2
[81].
The energy range 40-1000 eV is called the ”Knock-on” energy [81]. Even though
the ions have enough energy to dislodge ten to hundreds of atoms per incident ion in
this regime, the number of dislodged atoms from the surface is only in the range of
0-10 atoms. This is due to the fact that most of the energy is transferred to the target
atoms by collisions. Several features characterize this regime. Firstly, the sputter yield
is linearly dependent on both the ion energy and the flux of ions [81]. Secondly, the
yield maximum occurs when the mass of sputtering ions is similar to the target atoms.
At last, there is a clear dependence between the angle of incidence and the sputter
yield. The Knock-on energy regime is used for most sputtering purposes.
When the particles have energies at 1-50 keV the sputtering regime can be called
collision-cascade sputtering. At impact, the particles have enough energy to dislodge all
target atoms in a spherical region in close vicinity to the impact site. Nevertheless, this
type of sputtering demands supply of high energy, which is impractical for industrial
usage. Above 50 keV the particles will cause high energy implantation. Since all the
energy is transferred to the bulk atoms little or no sputtering occurs. [81]
3.3
Plasma theory
A plasma is defined as a gas which partially consist of ions, atoms and electrons [10].
The characteristic glow is given by the relaxation of the atoms excited by the electrons
accelerated toward the anode. There are different ways to create plasma.
3.3.1
Direct current (DC) diode plasma
Plasma can simply be created in a vacuum between two electrodes where there is a
voltage difference equal to the breakdown field of the gas. When the plasma is ignited,
secondary electrons are created as the ions strikes the cathode. These electrons are
accelerated. When the kinetic energy is sufficiently high, they create more ions on the
way toward the anode. Close to the cathode, the electrons have not yet gained enough
energy to excite atoms and the density of ions is considerable. This creates a positive
net charge, which contributes to a shielding of the plasma behind the net charge by
reducing the field between this area and the anode. Other distinctive features of DC
plasma are the three dark spaces. Close to the cathode the plasma is not visible due
to the fact that the secondary electrons have not yet gained enough energy to excite
Plasma theory
27
atoms. This region is called the Crooke’s dark space. A similar dark space called the
anode dark space is created close to the anode as the electron’s kinetic energy is high
and hence, the density of electrons is low. The last dark space in DC diode plasma is
the Faraday’s dark space. This is the region where the electrons have achieved sufficient
energy to ionize the gas atoms and the density of ions are significantly higher than the
density of excited atoms.
Figure 3.3: The Crooke’s, Faraday and anode dark spaces. Adapted from Ref. [10].
There are some drawbacks of DC plasma. The probability of electrons striking the
anode without ionizing the gas atoms is high and increases with the voltage applied.
This implies that scaling up the dimensions or increasing the power will not provide
increased sputtering rates. The other drawback is that it cannot be utilized for insulators. When the ions bombard an insulating cathode a charge is built up on the cathode
surface and reduces the field between the electrodes. As a result, the plasma ceases to
exist.
3.3.2
Radio frequency (RF) diode plasma
If the cathode is an insulating target, sputtering with radio frequency (RF) diode
plasma is preferred. Such plasma is created with an alternating field between the
cathode and anode, in an otherwise similar set-up as the DC set-up. This method
prevents the charge build-up on the target surface by frequently neutralizing the surface
with electrons or ions. Normally, the frequency used is at 13.56 MHz [81].
The alternating field has a second benefit as well. When the field is changed, the
electrons which have not yet reached the anode will be accelerated toward the ”new”
anode and hence, increase the probability of a collision with an atom. This implies a
higher density of ions and higher sputtering rate.
28
3.3.3
Magnetron sputtering
Magnetron plasma
If a magnetic field is applied parallel to the cathode in a DC set up, a Lorenz force will
act on the charged particles in the plasma. This way, the electrons are prevented from
accelerating toward the anode as they are forced in a helical path back to the cathode
and therefore trapped in a small region close to the cathode. The ions are only weakly
affected by the magnetic field, due to their large masses. As the density of electrons
is increased, the probability of ionization collisions and hence, the density of ions
is increased. Eventually, the ions will impinge the target surface and more secondary
electrons are generated. This makes it possible to obtain high deposition and sputtering
rates at low pressure. Other benefits of magnetron sputtering are low contamination
of the deposited films and neglectable bombardment of electrons onto the substrate.
The use of magnetron covers 95% of all sputtering deposition applications [81].
3.4
3.4.1
Sputter deposition
Deposition of thin films and thin film formation
Deposition of a thin film is a process of several steps. As soon as the atoms, which are
to be deposited reach the substrate surface their velocity component perpendicular to
the substrate is lost. However, the component parallel to the substrate is still present
and the adsorbed atom is free to move on the surface.
Depending on the adhesion of the deposit atoms and substrate, the film formation
will take place in one of the following three modes.
In the Frank-van der Merwe growth mode, the films are grown layer by layer as the
deposited atoms are more attached to the substrate than to themselves. This is also
called 2D-growth.[82]
A 3D-growth takes place in the Volmer-Weber growth mode, where the starting
point of the film formation is island growth. During this process the atom will interact
with other atoms and start to form clusters. As the number of clusters increases, the
clusters start to grow into nuclei, and at some point the nucleation density is saturated.
The nucleation density is dependent on several parameters, which are listed in Ref. [83].
The nucleus increases in size either parallel or perpendicular to the substrate and finally
the process of agglomeration starts. The nucleuses are formed into a continuous film
and the film growth proceeds in cylindrical columns normal to the substrate. Unless a
recrystallization takes place, the initial nucleuses determine the crystal size. For films
≥ 1μm the film thickness determines the size normal to the substrate. [83]
The third film formation mode is the Stranski-Krastanov growth mode. This is
a mix of the first two, and starts with a layer formation, but converts into island
formation [82].
Sputter deposition
29
Gas pressure, power, temperature and the distance between the target and sample
are of importance for the properties of the film, as they affect the kinetic energy of
the impinging particles. A high gas pressure will increase the probability of collisions
between the particles and hence, reduce the kinetic energy. Likewise, the temperature,
power and travelling distance for the particles decides the amount of kinetic energy
when the particles reach the substrate. The lower kinetic energy, the lower is the
surface mobility. The consequence can be a large number of nucleuses at smaller size.
The grain size and orientation of the final film is affecting the properties of the film.
Terzini et al. studied the effect of RF power on crystal size and orientation and how
this affected the film properties of ITO thin films [84]. They showed that the lower the
RF power, the lower the crystal size.
3.4.2
Deposition rates
For industrial purposes, high deposition rates are desired. Hence, prediction of how
the deposition rate is affected due to modifications of sputtering parameters is important. The deposition rate is dependent on sputtering yield and the probability of the
dislodged atom to reach the sample.
In case of magnetron sputtering, the plasma is dense in a E × B drift loop close to
the cathode. This and other factors like system geometry and emission profile makes
exact prediction of sputtering deposition rate complex. The most practical method to
predict the deposition rate is to calculate the ”rate per watt”. This is due to the fact
that in the operation region of most magnetrons the deposition rate increases linearly
with discharge power as the sputter yield increases linearly with discharge voltage,
which changes slowly. [81]
3.4.3
Reactive sputtering
Reactive sputtering is usually applied for deposition of oxide and nitride films. The
sputtering gas is usually a mix of argon and oxygen or nitrogen gas. The method can
be used for metallic targets or for compound targets of oxides and nitrides. Normally,
the film deposited consists of the target atoms and the reactive species are absorbed
at the film surface during deposition.
If the flow of reactive gas exceeds a critical flow point where the absorption of the
reactive gas by the film has extinct, the reactive gas starts to form an insulating film
on the cathode. The more gas the less target atoms are deposited as a film and the
fewer atoms can react with the reactive gas, which eventually increases the thickness
of the insulating layer on the cathode. This changes the discharge voltage and the
sputter yield is reduced up to an order of magnitude or more. In order to reach the
normal discharge voltage and sputter yield, the flow of reactive gas must be reduced
30
Magnetron sputtering
to a level further below the critical point where the formation of the insulating layer
started. There are different methods to circumvent the formation of an insulating layer
during reactive sputtering. See for instance Ref. [81].
The work in this thesis has mainly utilized DC magnetron sputtering as the film
deposition process, in order to produce high quality ITO thin films at high deposition
rates. In addition was the effect of sputtering with oxygen, nitrogen and hydrogen on
the properties of the ITO films studied with the aim of achieving highly transparent and
conducting films. RF sputtering performed at the Technical University of Darmstadt
was used during the study of ultrathin ITO films.
Chapter 4
Thin film fabrication and
characterization
4.1
Sputter deposition at IFE
The sputtering equipment at IFE was purchased in 2008 from Leybold Optics and is of
industrial size (target size 755 cm2 ). See Figure 4.1. It has three DC magnetron and one
RF magnetron target positions. The latter can be applied for co-sputtering together
with one of the DC targets. The films are deposited on substrates mounted to a sample
holder, which is tilted 7◦ . The tilting makes it possible to deposit on substrates mounted
on rails. Deposition is performed as the sample holder is transferred in front of a target
at a given velocity. In order to achieve the desired film thickness, either the velocity of
the sample holder can be adjusted or the sample holder can be oscillated several times
in front of the target. An area of 40 cm x 40 cm centered at the sample holder defines
the region where the produced films will have homogeneous film properties. The weight
ratio of In2 O3 /SnO2 target used in this work was 90/10% and the target was produced
by GfE Metalle und Materialien GmbH.
4.2
Sample preparation at IFE
All experiments carried out at IFE followed the same recipe for sample preparation.
In order to have a proper set of data, up to nine glass substrates of size 72 mm x 26
mm x 1 mm (MENZEL microscope slides) were coated with ITO film using DC magnetron sputtering. In addition, two or three polished p-type Cz silicon monocrystalline
silicon substrates (orientation: < 100 > and resistivity: 1-3 Ω cm) were included in
the experiments. One silicon substrate was covered with a mask before deposition in
order to make isolated areas with ITO (an attempt to produce diodes). In order to
remove contamination (dust, particles, etc.) from the substrate surfaces, all substrates
32
Thin film fabrication and characterization
Figure 4.1: The sputter equipment at IFE.
were cleaned 30 min in ultrasonic bath with DI-water. Subsequently, all samples were
dried by the use of compressed air. In addition, the silicon substrates were etched one
minute in HF and rinsed two minutes in DI-water subsequent to the ultrasonic bath.
This was done in order to remove any native oxide on the silicon surface.
The samples were mounted with adhesive tape, dispersed on the 40 cm x 40 cm
area on the sample holder. As the sample holder was positioned in the load lock (L1),
the heaters in L1 were turned on for 30 min at 100◦ C in order to dehumidify the
samples and L1. A temperature measurement with temperature stickers (”Thermax”)
on the back of some glass samples showed that this heat treatment could raise the
temperature of the glass substrates up to ∼45◦ C. Film deposition was usually performed without additional heating in the processing chamber and hence, we called this
room temperature (RT) deposition. The argon flow used during sputtering was kept
at 180 sccm. The power was usually 4 kW, giving a power density of 5.3 W/cm2 . The
pressure in the processing chamber was held at 3.1 x 10−3 mbar. The time for ramping up the sputtering power and presputtering of the target was 15 min. Deposition
was always performed with only one pass of the sample holder in front of the target.
The velocity of the sample holder was 0.7 m/min and the target to substrate distance
was held constant at 80 mm. These procedures and parameters constitute the main
sample preparation recipe, from now called standard recipe. Table 4.1 summarizes the
parameters in the standard recipe.
4.2.1
Optimizing film properties - preliminary studies
As the sputtering machine was acquired as our project was initiated, several preliminary
experiments were required in order to uncover the best suited sputtering parameters
and hence, develop the standard recipe. The determination of the parameters was
based on the characteristics of the optical and electrical properties of the ITO films.
33
Sample preparation at IFE
Gas and flow
Power
Pressure
Temperature
Target to sample distance
Velocity of sample holder
Ramping up time
Dehumidifying in L1
Annealing
180 sccm Ar
4 kW
3.1x10−3 mbar
RT
80 mm
0.7 m/min
15 min
30 min at 100◦ C
15 min at 300◦ C in air
Table 4.1: Deposition and annealing parameters used for sample preparation.
The experiments are described in detail below.
Effects of pressure
One set of experiments was performed in order to uncover the effect of the working
pressure. The pressures examined were 2.6 x 10−3 , 3.1 x 10−3 and 3.6 x 10−3 mbar.
Other processing parameters were according to the standard recipe in Table 4.1. Table
4.2 summarizes the parameters in the pressure experiments.
Experiment
1
Pressure in mbar 2.6x10−3
2
3.1x10−3
3
3.6x10−3
Table 4.2: Pressure values examined in the three experiments.
The results showed that changing the working pressure in the given interval did not
affect the transmittance, resistivity or thickness of the films significantly (deviations of
maximum 2%).
Effects of working power
Two sets of experiments were performed in order to uncover the effect of the working
power on the electrical and optical properties. The growth rate of the films was also
monitored. There were two differences between the two sets of experiments. One was
the addition of oxygen and the other was the power interval examined. In the first set
of experiments, the sputtering gas was argon only and the sputtering power was varied
between two and six kW. In the second set of experiments, the flow% of oxygen was
3.2 (flow% is defined as the percentage flow of additional gas (oxygen, hydrogen and
nitrogen) of the total flow into the chamber, i.e. 3.2 flow% of oxygen corresponds to 6
sccm oxygen and 180 sccm Ar) and the power was varied from three to five kW. Table
4.3 summarizes the two sets of experiments.
Characterization of the film properties showed a correlation between the spectral
transmittance (not the average) and the change in power. The resistivity of the films
34
Thin film fabrication and characterization
Gas and flow%
0 oxygen
3.2 oxygen
2 kW
x
-
3 kW
x
x
3.5 kW 4 kW
x
x
x
4.5 kW 5 kW
x
x
x
6 kW
x
-
Table 4.3: Power examined in the two sets of experiments with different sputtering gas.
deposited at 2 kW was significant higher (3 Ω cm compared to 1.3 Ω cm at 3 kW). In
the power interval 3-6 kW, the resistivity varied 5-10%. However, the thickness shows
an approximately linear correlation with the sputtering power, which likely explains
both the variation in spectral transmittance and high resistivity at low power.
Effects of temperature during sputtering
The effect of elevated temperature was examined in four different sets of experiments.
Our intention was to develop a solar cell fabrication process at low temperatures (see
section 1.2) and the temperatures examined were 100◦ C, 200◦ C and 300◦ C. Other
processing parameters were according to the standard recipe described above.
The samples were transferred in front of the heaters in the processing chamber
when the heaters had reached the set temperature. The heating time was one hour
prior to deposition. The actual reached substrate temperature was not measured, but
is assumed to be somewhat below the heater temperature. The resulting films were
compared with similar films deposited at room temperature.
The difference between the four sets of experiments was the addition of gases. One
set of experiments was carried out with argon only (reference samples), one with a 3.2
flow% of oxygen and the last two were performed with a 3.2 and 9.7 flow% of hydrogen
(180 sccm Ar and 6 sccm H, and 168 sccm Ar and 18 sccm H, respectively). The
temperature and flow of gases examined in the experiments are listed in Table 4.4.
Gas and flow%
0 oxygen/hydrogen
3.2 oxygen
3.2 hydrogen
9.7 hydrogen
RT
x
x
x
x
100◦ C
x
-
200◦ C
x
-
300◦ C
x
x
x
x
Table 4.4: Flow of oxygen and hydrogen during sputter deposition at different temperatures.
The results showed that in the case of the reference samples and the samples deposited with oxygen we did not observe any significant change in resistivity nor in
transmittance, as the temperature was changed between RT and 300◦ C. In case of hydrogen we observed a slight effect in resistivity, due to changes in temperature. This
effect increased as the flow% of hydrogen was increased. The transmittance was more
Sample preparation at IFE
35
dependent on the hydrogen flow% than on the temperature.
Effects of annealing
Two sets of experiments were performed in order to examine the effect of annealing.
In the first set, ITO was deposited at room temperature and subsequently annealed in
the processing chamber. Hence, the samples were not exposed to air before annealing.
The oxygen flow% applied during deposition in these experiments was 3.2. The heaters
positioned at the other end of the processing chamber (compared to the ITO target)
were turned on after deposition. The ITO samples were transferred in front of the
heaters when the heater temperature had reached the set point temperature of 300◦ C.
This annealing method was performed five times in different ambient atmospheres.
One annealing was performed in vacuum at 10−7 mbar and four in atmospheres of pure
argon, oxygen, nitrogen and hydrogen at 10−3 mbar. The duration of the annealing
was one hour.
The results showed that the transmittance and resistivity of these films were not
affected significantly by the different annealing methods. Compared with the properties of as-deposited samples, effects of the annealing are scarcely seen (less than 5%
deviation from average). However, TEM of the ITO/Si samples showed one interesting
feature of the sample annealed in vacuum. It appeared as there was no SiOx . Hence,
this experiment was repeated in order to recreate the same interface. Unfortunately, it
was no success. It was not concluded why we were not able to reproduce the interface,
but one hypothesis was that such interfaces occasionally occur. The TEM picture of
the SiOx free ITO/Si interface is shown in Figure 4.2.
Figure 4.2: An SiOx free ITO/Si interface we were not able to reproduce.
36
Thin film fabrication and characterization
In the other set of experiments, annealing was performed on a hot-plate in ambient
air for 15 min. The sample was covered by a beaker in order to protect the surface
from dust during annealing and the temperatures examined were 250◦ C, 300◦ C and
350◦ C. This set of experiments included samples from depositions performed with a 0,
0.3, 1.6, 3.2, 6.3, 9.1, 14.3 and 25.0 flow% of oxygen.
The results of the hot-plate annealing experiments are published in Paper III and
show that the transmittance is improved more at 300◦ C than at the other annealing
temperatures examined. The annealing temperature is less significant for the resistivity
of the ITO films.
By comparing the annealing methods, the results show that hot-plate annealing was
best suited to improve the film properties. The transmittance was about 5% higher
and the resistivity was about 5% lower when the samples were annealed on a hot-plate.
Summary of the preliminary studies
The results from the different preliminary experiments lead to the conclusions:
ˆ changing the working pressure from 2.6 x 10−3 to 3.6 x 10−3 mbar did not improve
the film properties
ˆ increasing the working power increased the film thickness, which affected the
spectral transmittance and resistivity at low power
ˆ sample heating (RT to 300◦ C) had only minor impacts on the film properties
ˆ annealing on hot plate had the best effect on the film properties and the trans-
mittance was the highest when the samples were annealed at 300◦ C
Thus, the results from these preliminary studies hereby justify the choice of the deposition parameters used in the standard recipe. Table 4.1 summarizes the parameters
used during sample preparation.
4.3
Sputter deposition at Technische Universität
Darmstadt
The Surface Science division of Prof. Dr. Jaegermann at the Technical University in
Darmstadt (TUD) is fortunate to possess a set of powerful integrated systems. One
of these systems is the DArmstadt Integrated SYstem for MATerial science (DAISYMAT). This is an ultrahigh vacuum system (10−9 mbar as base pressure), which combines several deposition techniques and a photoelectron spectroscopy characterization
chamber, including both X-ray and UV photoelectron spectroscopy. The deposition
Sputter deposition at Technische Universität Darmstadt
37
techniques cover Plasma Enhanced Chemical Vapor Deposition (PECVD), Molecular
Beam Epitaxy (MBE), Metalorganic Chemical Vapor Deposition (MOCVD) and sputter deposition. The unique thing about this system is the option of sample transfer
between the deposition chamber and the characterization chamber, via a cylindrical
distribution chamber without breaking the vacuum. Hence, it is possible to carry out
detailed studies of thin film formation processes on surfaces, interaction between different materials and chemical properties of surfaces and interfaces. A sketch of the
system is illustrated in Figure 4.3.
Figure 4.3: DAISY-MAT. [11]
4.3.1
Sample preparation at TUD
The ITO/Si interface is of great importance as this is the junction where electrons are
extracted from the current generating part of the solar cell. Hence, as little current loss
as possible is desired. The mechanisms of current loss can be due to different factors.
One factor is the presence of a thin insulating silicon oxide layer. This was discussed
in chapter 2, subsection 2.3.4. At TUD, we were fortunate to have the opportunity
to carry out a thorough XPS study of this interfacial silicon oxide layer by the use of
DAISY-MAT.
In our study, we used a Si sample with the orientation < 100 > and resistivity 1-3
Ω cm. This was etched for two min and rinsed one min in DI-water. The cleaning
procedure was repeated twice in order to completely remove any native oxide. The
substrate was blown dry with nitrogen. Subsequent to this treatment, the substrate
38
Thin film fabrication and characterization
was mounted to the sample holder and transferred quickly (in nitrogen atmosphere) to
the load lock of DAISY-MAT. XPS was carried out on the silicon substrate prior to
the depositions and subsequent to each deposition. The ITO layers were deposited at
room temperature in four steps of 2, 8, 30 and 900 s, giving a total deposition time of
940 s. The substrate to target distance was 10 cm, the working power 25 W (power
density of ∼1.23 W/cm2 ) and the working pressure was 5x10−3 mbar. The flow of
argon during sputtering was 6 sccm and the target composition was 90wt% In2 O3 and
10wt% SnO2 . The results are published in Paper VI.
4.4
4.4.1
Characterization methods
Electrical characterization techniques
The resistivity of the ITO films was measured by the use of a four-point-probe set-up.
This consists of four metal tips mounted on springs to reduce damage to the sample
upon measurements. The two outer tips are supplied with current and the two inner
tips are connected to a voltmeter in order to measure the resulting voltage and hence,
calculate the resistivity by multiplying the thickness and the sheet resistance, according
to equation 4.1. The thickness was measured by an alpha-step profilometer. The
graphical results presented in the papers are based on the average of the measurements.
The calculated standard deviations of the measurements are listed in the table 4.5.
ρ = 4.53 × Rsh t
4.4.2
(4.1)
Optical characterization techniques
For solar cell application the transmittance of the ITO film is an important factor,
while reflectance should be at a minimum. Spectral transmittance and reflectance
measurements were performed with a spectral response in the wavelength range 4001000 nm. The reference of the measurements was air, but the loss in transmittance
due to the glass substrate was added to the transmittance measurements in order to
calculate the transmittance of the films only. However, any optical interface effects at
the film/substrate interface were not included. The results are based on the average of
the measurements. The calculated standard deviations of the measurements are listed
in the table 4.5.
4.4.3
X-ray Photoelectron Spectroscopy
X-ray Photoelectron Spectroscopy (XPS), also known as Electron Spectroscopy for
Chemical Analysis (ESCA), is a characterization technique widely used to examine the
39
Characterization methods
Property
Transmittance
Sheet resistance
Thickness
Standard deviation
<2%
7%
5%
Table 4.5: The standard deviation of the measurements.
chemical states of materials. The technique is based on the principle of the photovoltaic
effect, where electrons are ejected from a surface when irradiated by energetic photons.
Practically, when applying XPS the energetic photons consist of a beam of monochromatic characteristic X-rays usually having the energy hν = 1486.6 eV (Al Kα) or
hν = 1253.6 eV (Mg Kα). When the material is bombarded by photons, secondary
electrons are ejected and captured in an analyzer, usually a concentric hemispherical
analyzer, where their kinetic energy is measured. As the kinetic energy is equal to the
difference between the incident photon energy and the energy required for ejecting the
electron, it is possible to calculate the binding energy of the electron. By comparing the
binding energy with reference literature the chemical composition of the material can
be determined. The energy required for ejecting the electron from the material is equal
to the work function of the material and the binding energy. During measurement, the
sample is usually in electrical contact with the spectrometer and the Fermi levels of
the sample and spectrometer are aligned. Though, there is a difference between the
work function of the spectrometer and the work function of the sample. This difference must be subtracted from the measured kinetic energy in order to find the actual
kinetic energy of the electrons ejected from the sample, as described in equation 4.2
and illustrated in Figure 4.4 [85, 86].
Ek = hν − Eb − φs − (φspec − φs ) = hν − Eb − φspec
(4.2)
Ek is the kinetic energy, Eb is the binding energy, φspec is the work function of the
spectrometer, φs is the work function of the sample and hν is the photon energy. The
work function of the spectrometer is a more appropriate parameter to work with as
this is usually a known value.
Auger electron emission usually follows the photoelectric process and occurs ∼
seconds after the photoelectric event. The Auger electrons are ejected due to a
10
relaxation event where an electron from an outer orbital fills up an inner orbital vacancy
and gives its energy to a second electron, simultaneously. The intensity of these ejected
electrons are seen as peaks in the XPS spectra as well, but can be distinguished from
the photoelectron peaks by applying different photon energy. The photoelectron peaks
will be shifted while the peaks from the Auger electrons will remain at the same binding
energy.
−14
XPS is a surface characterization technique. Only the first 5-10 nm of the sample
40
Thin film fabrication and characterization
Figure 4.4: An energy scheme of a sample in contact with a XPS spectrometer. [12]
is characterized as the secondary electrons lose their energy due to inelastic collisions
in the material. The distance they can travel is given by the mean free path. This
parameter is dependent on the material and the kinetic energy of the electrons. Thus,
in order to extract information from bulk and interfaces, XPS depth profiling is a widely
used technique. The profiling is usually performed with Ar+ sputtering in sequences,
intercepted by XPS measurements, resulting in XPS spectra at given depths. As this
is a destructive method, the sample will not be measured in its original state and the
measurements may be affected by this.
Usually, a survey scan is performed before the high resolution scan. This is done
in order to get information about which elements are present in the sample. The high
resolution scan gives a detailed spectrum of the different elements. Information about
the chemical states is found by peak fitting. Shifts of the binding energy and peaks
consisting of several components will inform the user about the chemical state and
binding conditions of the given elements. The Fermi level is usually defined at zero
binding energy.
The XPS in paper III and V was performed by Spyros Diplas at MiNALab in Oslo,
while in paper IV the measurements were performed by Mareike Hohmann and me
at the University of Darmstadt. The instrument used in Oslo was a KRATOS AXIS
ULTRADLD and a monochromatic Al Kα radiation with hν =1486.6eV was used at 15
kV and 10 mA. To achieve contact between the film and the spectrometer and hence,
aligned Fermi levels, all samples were mounted to the sample holder with a metallic
clips. Charge compensation by an electron flood gun was used to prevent the samples
from being charged during the characterization.
In paper IV, Figure 4 C shows the difference between the Fermi level position and
the valence band maximum (VBM). VBM is determined by linear extrapolation of the
Characterization methods
41
slope of the valence band spectra data. See Figure 4.5. As the Fermi level is defined
at zero binding energy, the difference between the Fermi level and the valence band
maximum (EF − EVBM ) is easily measured from the figures.
Figure 4.5: The method for the extraction of the EF − EVBM values.
At the University of Darmstadt, XPS was carried out using a Physical Electronics
PHI 5700. A monochromatic Al Kα radiation (hν = 1486.6 eV) was employed for
excitation here as well. X-ray photoelectron (XP) spectra of In 3d, Sn 3d, O 1s and Si
2p were recorded with a take-off angle of 45◦ . The XPS results were peak fitted using
CasaXPS [87]. The background type was Shirley and the fitting components were
Gaussian/Lorentzian (30%/70%) product formulas. Some of the components were also
modified by the exponential blend, due to the metallic state of elements.
4.4.4
X-ray Diffraction
X-ray diffraction (XRD) is a nondestructive technique and is mainly utilized for examining the orientation of crystal planes in crystalline solids. In addition, it can be
applied in order to extract information about material structure, chemical composition of an unknown material, crystallite size, lattice strain, and surface and interface
roughness.
The technique was developed by Bragg (father and son) in 1913, when they discovered constructive interference from X-rays reflected by crystal planes at certain
42
Thin film fabrication and characterization
wavelengths and incident angles. Bragg’s law describes a model for the diffraction
2d × sinθB = nλ
(4.3)
where d is the distance between the crystal planes (d-spacings), θB is Bragg’s angle
and λ is the wavelength of the incident x-ray beam. The n is an integer. By analyzing
the d-spacings and their intensity, the phase of the state of the material can be determined. Data from International Center for Diffraction Data (ICDD) or cards from
Joint Committee on Powder Diffraction Standards (JCPD) is used for comparing and
determine crystall orientation and composition of materials. [88, 89]
In the present work, two different XRD instruments were used. One was located
at the Department of Chemistry at the University of Oslo and the other at Institute
for Energy Technology, both a Bruker-Siemens D5000 equipment with parallel beam
provided by a Göbbel mirror. The equipment at the University of Oslo was used on the
samples deposited with addition of hydrogen and nitrogen and the measurements were
performed with Grazing incidence. The equipment at IFE was used on the samples
deposited with oxygen with a theta-theta geometry.
4.4.5
Transmission Electron Microscopy
Transmission Electron Microscopy (TEM) is a powerful and direct characterization
method for investigating the structure, morphology and local chemistry of interfaces, in
the way of producing pictures of interface cross-sections at high resolution (approaching
0.08 nm) [90, 91]. This is due to the fact that the system utilizes electrons rather than
light in order to interfere with the materials in a sample.
Generally, a TEM instrument consists of a vacuum system, an electron gun, condensers and imaging devices. The electron emission from the electron gun originates
from either a LaB6 thermionic emission source or a field emission source [92]. The
lenses focus the electrons onto the sample, which is sufficiently thin for electrons to
be transmitted. A magnified image and a diffraction pattern are formed from the
transmitted and the scattered electrons. The images formed from only the transmitted
electrons are called bright-field image. Dark-field images are formed with a specific
diffracted beam. [91]
The TEM carried out in my work was performed by Annett Thøgersen at the
University of Oslo. The equipment used was a 200 keV JEOL 2010F microscope with
a Gatan imaging filter and detector, and a NORAN Vantage DI+ Electron Dispersive
Spectroscopy (EDS) system.
The TEM was performed with low electron doses in order to confirm crystallinity of
the film, without inducing crystallization. Electron diffraction confirmed crystallinity
at very short electron exposures, and no changes in the electron diffraction patterns
Characterization methods
43
were observed.
4.4.6
Secondary Ion Mass Spectroscopy
Secondary Ion Mass Spectroscopy (SIMS) is a destructive characterization technique,
which is utilized for examining the composition of materials. By the use of an ion beam
for sputtering, atoms are ejected from the sample material. A fraction of the ejected
atoms are ions. Generally, the mass analyzer determines the mass of the ion based on
the charge of the ion and the number of ions is counted. By measuring the crater after
sputtering, it is possible to present a depth profile of the material composition.
The Secondary Ion Mass Spectroscopy (SIMS) measurements presented in the papers were carried out at the Center for Materials Science and Nanotechnology (University of Oslo) by Lasse Vines and Klaus Magnus Johansen. The equipment used
was a CAMECA ims 7f instrument and it revealed the intensity (counts per second) of
the elements In, Sn, O, N and H. All elements were measured in negative mode with
a 15 keV primary ion beam of Cs+. The sputtering time was kept constant at 3768
s, and the depth of the SIMS craters was measured with a Dektak 8 surface stylus
profilometer.
44
Thin film fabrication and characterization
Chapter 5
Summary of results and conclusions
This chapter presents a summary of the appended papers (section 5.1 and 5.2) , general
conclusions (section 5.3) and future work (section 5.4).
The essential elements for ITO applied as a transparent conducting electrode on
top of a solar cell structure are the bulk properties transmittance and resistivity, in
addition to the properties of the interface between the ITO and the substrate. The
work in this thesis has therefore focused on these two elements. Paper I-IV cover
the study of the bulk properties and paper V and VI cover the interfacial properties
between ITO and silicon.
5.1
Optimizing film properties - addition of gases
This section presents the four papers (I-IV) where we examined the effect of adding
different gases. In the papers, results from SIMS, XPS and optical, electrical and
structural characterization are illustrated for both as-deposited and annealed samples.
The sample preparation followed the standard recipe and the annealing was performed
at 300◦ C in air for 15 min. All ITO films studied in this section were deposited onto
glass substrates.
By adding gases during sputter deposition, the properties of ITO films can be altered. We examined how the addition of three different gases, oxygen, hydrogen and
nitrogen, influenced the electrical and optical film properties. Table 5.1 summarizes
the flow% used in the different experiments.
Gas
Oxygen
Hydrogen
Nitrogen
Flow%
0, 0.3, 1.6, 3.2, 6.3, 9.1, 14.3 and 25.0
0, 3.2, 6.3, 11.8, 14.3, 21.1 and 25.0
0, 1.6, 3.2, 14.3 and 25.0
Table 5.1: Flow% of different gases in the experiments described below.
45
46
Summary of results and conclusions
Figure 5.1 summarize the electrical properties of the as-deposited and annealed
films presented in Paper I-IV.
Figure 5.1: Resistivity of as-deposited (filled symbols) and annealed (15 min at 300◦ C
in air) (open symbols) films deposited with additional supply of oxygen, hydrogen and
nitrogen.
Figure 5.2 summarize the optical properties of the as-deposited and annealed films
presented in Paper I-IV.
For solar cell application the transmittance of the ITO film is an important factor,
while reflectance should be at a minimum. Spectral transmittance measurements were
performed with a spectral response in the wavelength range 400-1000 nm. The reference
of the measurements was air, but the loss in transmittance due to the glass substrate
was added to the transmittance measurements in order to calculate the transmittance
of the films only. However, any optical interface effects at the film/substrate interface
were not included. The results are based on the average of the samples.
The transmission measurements were carried out in order to make a qualitative
characterization of the optical properties of the films as a function of different parameters. It is difficult to make a quantitative conclusion based on these measurements,
as the reflectance and absorptance of the films are not included.
Optimizing film properties - addition of gases
47
Figure 5.2: Average transmittance in the wavelength range 400-1000 nm of as-deposited
(filled symbols) and annealed (15 min at 300◦ C in air) (open symbols) films deposited
with additional supply of oxygen, hydrogen and nitrogen.
5.1.1
Paper I - Hydrogenated ITO and its properties
Some researchers claim that addition of water vapour may have the same effect as
addition of oxygen, i.e. increment in film conductivity and transmittance, and relate
the effect to the hydrogen introduced by the water. Other studies have shown that
hydrogen doping can improve the mobility in TCOs. This has been demonstrated
by Barraud et al. and Koida et al. [68, 70, 93], which have produced films with
mobilities > 100 cm2 /Vs in combination with sufficient carrier density (1020 cm−3 ) by
adding water vapor as a hydrogen source during sputter deposition of In2 O3 at low
temperature.
We performed a study where only hydrogen was added during ITO deposition, in
order to examine the effect of hydrogen on our samples. The flow% of hydrogen used is
listed in Table 5.1. Other parameters were according to the standard recipe described
previously. SIMS, XRD, electrical and optical characterization were carried out on
these samples.
The results published in Paper I show that addition of hydrogen had a darkening
effect on the as-deposited samples. As the hydrogen flow increase, the lower is the
transmittance. This is also the case for conductivity measurements. The higher the
hydrogen flow, the lower the conductivity. When the samples are annealed on a hotplate, the darkening effect disappeares. The conductivity increases to the same order of
48
Summary of results and conclusions
magnitude as the reference sample. See Figure 5.1 and Figure 5.2. The transmittance
deviate at a maximum of 2% compared to the reference sample.
Figure 5.3 shows the relative hydrogen concentration of the hydrogenated films.
When studying these SIMS results, one may attribute the improvement of the electrical
and optical properties of the hydrogenated samples by annealing to the reduction in
the relative hydrogen concentration.
Sputtering with a hydrogen and argon mixture is previously reported by K. Zhang
et al. [94], but at a substrate temperature of 300◦ C. By SIMS measurements, they
also reveale a lower relative oxygen concentration when sputtering with a hydrogenargon mixture than with pure argon. They claim that hydrogen with high energy can
remove weakly bound oxygen in the growing film and that the consequence is a higher
concentration of oxygen vacancies. As oxygen vacancies are said to be a source for
charged carriers, the increase in oxygen vacancies will lead to a lower resistivity, which
is found for a certain hydrogen partial pressure in Ref [94].
An increased transmittance and diminished resistivity at hydrogen flow rates below 3.3% is reported by D. G. Kim et al. [95]. The improvements are addressed to
a hydroxyl formation when adding hydrogen during sputtering. This formation is attributed to a removal of excess oxygen and increment in the concentration of oxygen
vacancies.
These results from the literature are different from the observations in Paper I, and
may be attributed to the high flow of hydrogen during sputtering and/or the structural
phase of the films. K. Zhang et al. [94] found a crystalline phase for all hydrogen partial
pressures up to 1.6 10−5 Torr at a substrate temperatures of 300◦ C, and D.G. Kim et
al. [95] found a crystalline phase for the films deposited with a hydrogen flow rate
below 5% at a substrate temperature below 70◦ C.
The electrical and optical properties of the hydrogenated films in our study may
indicate that the incorporation of hydrogen into ITO at low temperature may result
in an amorphous phase, and that hydrogen is annealed out of the film and the film
crystallizes during the annealing.
Conclusion - Paper I
Paper I examines the influence of an increased flow% of hydrogen present during room
temperature (RT) deposition on the electrical and optical properties of ITO films. The
main conclusion of this paper is that the supply of hydrogen during sputtering increased
the resistivity and decreased the transmittance of the as-deposited films. When the
films were annealed the addition of hydrogen had only a minor effect on the same
properties. In addition, a lower concentration of 1H ions was found by the use of SIMS
after annealing.
Optimizing film properties - addition of gases
49
Figure 5.3: Counts of secondary 1 H ions per second as a function of depth for asdeposited (filled symbols) and annealed (open symbols) samples of ITO deposited with
hydrogen/argon or argon as sputtering gas. The noise that occurs at about 0.07 to 0.09
μm is due to the interface between the ITO film and the glass substrate.
5.1.2
Paper II - Annealing effect on ITO films sputtered with
argon, oxygen and hydrogen
Paper II makes a comparison of the effect that a similar flow of oxygen and hydrogen
have on the ITO films and shows that addition of oxygen has the best effect on the
electrical and optical properties on as-deposited samples. When the samples were
annealed, the reference sample and the hydrogenated sample show the best properties.
The results from SIMS, and electrical and optical characterization are presented in the
paper.
Figure 5.4 shows the relative hydrogen concentration in the three different samples. The figure shows that, unlike the hydrogenated sample, the sample deposited
with addition of oxygen experiences an increase in hydrogen content due to annealing.
However, the measurements of the optical and electrical properties of the oxygenated
sample show a minor improvement due to heat treatment, compared to the hydrogenated sample. See Figure 5.2. Hence, there is no clear correlation between the SIMS
measurements of hydrogen concentration in the samples and the optical and electrical properties of the films. A correlation between the annealing and electrical and
optical properties may be addressed to other film properties, such as structure and
crystallinity.
50
Summary of results and conclusions
Figure 5.4: Counts of secondary 1 H ions per second as a function of depth in asdeposited (filled symbols) and annealed (open symbols) samples of ITO deposited with
oxygen/argon, hydrogen/argon or argon as sputtering gas. The noise that occurs at
about 0.08 to 0.09 μm is due to the interface between the ITO film and the glass
substrate.
Optimizing film properties - addition of gases
51
Conclusion - Paper II
Paper II compares the influence of the presence of one exact flow% of oxygen and hydrogen during the deposition and post-deposition annealing of the resulting films. The
presence of oxygen increased the transmittance of the film, while hydrogen had a darkening effect. The transmittance of the reference film was lower than the transmittance
of the sample deposited with oxygen and higher than the transmittance of the sample
deposited with hydrogen. The same relation is observed for the resistivity of the films,
the oxygen sample having the lowest resistivity. Annealing the samples improved the
resistivity and transmittance of all the samples.
5.1.3
Paper III - Annealing of ITO films sputtered with argon
and oxygen
It is well known that addition of oxygen enhances both the transmittance and the
resistivity in ITO films [96]. Due to higher volatility of oxygen than In and Sn it
is necessary to compensate some of the oxygen supplied by the In2 O3 /SnO2 target,
which is removed during pumping in order to obtain a low pressure in a deposition
chamber [97]. Hence, experiments were performed in order to uncover the best suited
flow of oxygen during sputtering and examine the effect of additional oxygen supply
on electrical and optical properties. The flow% of oxygen applied in the experiments is
listed in Table 5.1. Other parameters were according to the standard recipe described
above. In addition to electrical and optical characterization, XRD and XPS were
performed on these samples.
The results published in Paper III show that increased flow% induce band gap
states, a reduction in film thickness and an increase in resistivity. Addition of oxygen
also make the films more crystalline compared to a reference sample. A 1.6 flow% of
oxygen is probably the most optimal flow, giving the highest transmittance for both
as-deposited and annealed samples and the resistivity is improved compared with the
reference sample in the as-deposited case. As the samples are annealed, there are only
minor differences between the reference sample and the sample deposited with a 1.6
flow% of oxygen.
From the XPS measurements of the films we extracted the emissions from the
valence band region. As zero binding energy corresponds to the Fermi level position,
we determined the Fermi level position (EF ) relative to the valence band maximum
(EVBM ). The method we used for finding the difference between EF and EVBM is
explaied in subsection 4.4.3.
Figure 5.5 shows the valence band spectra of as-deposited and annealed (15 min
at 300◦ C on a hot-plate) samples. It is evident from the figure that the oxygen flow%
during deposition affects the difference between the valence band maxima and Fermi
52
Summary of results and conclusions
level. Higher oxygen flow induces lower binding energy of the valence bands, i.e. the
Fermi energy level lies closer to the valence band edge. A lower Fermi level is assumed
to indicate lower doping. This coincides with the resistivity measurements, which show
that increased oxygen flow leads to increased resistivity.
On one hand, the shift indicates formation of oxygen induced acceptor states close to
the valence band (such as oxygen interstitials and indium vacancies), and the number of
states increases as the flow of oxygen increases. On the other hand, poor crystal quality
with abundant structural defects can also result in effective doping compensation.
Frank and Köstlin [98] presents a theory of interstitial defects consisting of one
oxygen and two tin ions. The dissociation of these defects will give two free electrons
according to equation 5.1.
”
(Sn2̇ 2Oi )
(g)
2Sn˙+ 2e + 1/2O2
(5.1)
The Sn˙2 is tin ions, which has delivered their conduction electron, i.e. is positively
charged, O”i is double negatively charged oxygen interstitials and e’ is electrons, i.e.
negatively charged particle.
As our films are annealed, the shift in binding energy disappears and the resistivity
drops. This can be attributed to the dissociation of the defects presented by Frank
and Köstlin [98]. Hence, the shift in VBM can be associated with the resistivity of the
films, as a drop in resistivity coincides with the disappearance of the shift in VBM.
Gassenbauer and Klein [41] do a similar observation of shift in VBM and increase
in resistivity as a function of oxygen content in the sputtering gas. They state that a
lower Fermi energy position and a lower doping level is expected due to formation of
neutral defect complexes, according to the theory of Frank and Köstlin [98].
Conclusion - Paper III
Paper III examines the influence of oxygen and an increasing supply of this gas. Increased flow% induced states close to the valence band, a reduction in film thickness
and an increase in resistivity. Addition of oxygen also made the films more crystalline
compared to a reference sample. Annealing of the films improved the transmittance
and resistivity and the number of states close to the valence band decreased. The
increment in resistivity as a function of oxygen flow in the as-deposited samples was
attributed to formation of oxygen-related interstitial defects. The paper confirms that
addition of oxygen is essential for sputter deposition of ITO. The best suited flow%
of oxygen was 1.6 in order to achieve optimal electrical and optical properties for RT
deposited films.
Optimizing film properties - addition of gases
53
Figure 5.5: Valence band spectra of as-deposited (top) and annealed (bottom) ITO films
versus oxygen flow% during deposition.
54
5.1.4
Summary of results and conclusions
Paper IV - Optical electrical, chemical and structural
properties of nitrogen doped indium tin oxide thin films
P-type TCO is a highly relevant topic among researchers. Some researchers have
demonstrated that nitrogen can be added in order to produce p-type TCOs [99, 100].
We performed one study, in order to examine the influence of addition of nitrogen
during sputtering of ITO. The flow% of nitrogen used in the experiments is listed in
Table 5.1. Other parameters were according to the standard recipe described above.
In addition to electrical and optical characterization, SIMS, Hall, XRD and XPS were
carried out on these samples.
The results published in Paper IV show that a 1.6 and 3.2 flow% of nitrogen increase
both the transmittance and conductivity of the as-deposited films, compared to the
reference sample. When the samples are annealed, the reference sample show the best
electrical and optical properties. Hall measurements (not included in the paper) do
not indicate a p-type, but n-type ITO.
Figure 5.6 shows the relative concentration of nitrogen in the as-deposited and annealed samples. N0 , N2 and N4 correspond to 0, 3.2 and 25.0 flow% nitrogen. The
SIMS data illustrate a non-linear dependence of incorporated nitrogen versus flow% of
nitrogen into the sputtering chamber. In addition, Figure 5.6 shows that the incorporated nitrogen is not significantly affected by the annealing. XPS measurements of the
N 1s peak (not presented in the paper) confirm that the content of nitrogen does not
increase as the supply of nitrogen increases from 14.3 to 25.0 flow%. These findings
indicate that there is a solubility limit of the quantity of nitrogen which is possible to
incorporate into ITO.
A similar observation is made in Figure 5.7, which shows a shift in valence band
maximum of the as-deposited and annealed ITO samples due to introduction of nitrogen. This shift is interpreted as formation of nitrogen induced acceptor states close
to the valence band, and the number of states increases as the supply of nitrogen increases, up to 14.3 flow% (illustrated by the arrow). For a high nitrogen supply of
25.0 flow%, the amount of the induced states does not seem to increase. In addition,
the heat treatment does not eliminate the states. Hence, these results correlate to the
non-linear dependence and the slight effect of annealing found by SIMS in Figure 5.6.
Conclusion- Paper IV
Paper IV examines the effect of nitrogen being present in the deposition gas. The results
show that adding nitrogen had an impact on both the resistivity and transmittance
of the as-deposited films. Small flows of nitrogen (1.6 and 3.2 flow%) improved both
properties compared to the reference sample, while larger nitrogen flows resulted in
lower transmittance and higher resistivity. Annealing had a minor impact on the
Optimizing film properties - addition of gases
55
Figure 5.6: Content of nitrogen shown by 14 N16 O- ions measured by SIMS on asdeposited (Ad) and heat treated (Ht) films. N0 , N2 and N4 corresponds to 0, 3.2 and
25.0 flow% nitrogen. The noise that occurs at 700 to 1000 Å is due to the interface
between the ITO film and the glass substrate.
nitrogen doped ITO samples, compared to the reference sample.
One of the main conclusions is that there is a non-linear dependence between the
nitrogen content in the films and the nitrogen concentration in the sputtering plasma.
This conclusion was based on Figure 5.6 and 5.7, which indicate that there is a solubility
limit of nitrogen in ITO. However, our main conclusion was the observation of acceptor
states induced by the nitrogen.
56
Summary of results and conclusions
Figure 5.7: Valence band spectra of as-deposited (top) and annealed (bottom) ITO films
versus nitrogen flow% during deposition.
Optimizing film properties - addition of gases
5.1.5
57
Electrical and optical properties versus structural properties
Figure 5.8 illustrates the XRD patterns of the as-deposited and annealed samples deposited with oxygen, hydrogen and nitrogen on glass substrates. The reference sample
shown in the figure with the samples deposited with oxygen exhibit a prominence between 10 and 40 2θ, which is not observed for the reference sample shown together
with the hydrogen and nitrogen samples. This is due to the fact that two different
XRD set-ups were used. However, the same reference sample was measured with both
set-ups and the respective pattern of the reference sample is shown.
Crystalline films are observed for as-deposited films when a 3.2 and 25.0 flow%
of oxygen is added during deposition. Likewise is a crystalline film obtained when a
3.2 flow% of nitrogen is added. The remaining samples exhibit XRD patterns, which
indicate an amorphous like structure.
We observe crystallinity of all annealed samples, except for the sample deposited
with a 25.0 flow% of nitrogen. The samples deposited with hydrogen and nitrogen
show more crystal orientations than the reference sample and samples deposited with
oxygen. This could be linked to the formation of grains from an amorphous structure
due to the annealing.
It is evident from the figure that addition of small amounts of oxygen and nitrogen
enhance the crystallinity of as-deposited samples deposited at RT. This can be seen in
correlation with the only slight improvements in both transmittance and resistivity for
these films due to annealing. As these films are annealed, the crystallinity is already
present and there is little potential for improvement left.
Figure 5.8: XRD pattern of as-deposited (upper row) and annealed (lower row) ITO films deposited with different flow of oxygen, hydrogen
and nitrogen.
58
Summary of results and conclusions
Optimizing film properties - addition of gases
59
Figure 5.2 shows the transmittance of the as-deposited and annealed (15 min at
300◦ C in air) films produced with addition of oxygen, hydrogen and nitrogen. It is evident from the figure that the highest transmittance is exhibited by the samples which
are deposited with a 1.6 flow% of oxygen. This is true for both the as-deposited and
annealed samples and we see that there is almost no change of this value due to annealing. Other samples exhibit an increase in transmittance after annealing. However,
some are more affected and others are very little affected by the heat treatment.
If we compare Figure 5.2 and 5.8, it is plausible to partly relate the variation
in the effect of the annealing to the improvement in crystallinity due to the heat
treatment. However, this explanation is probably not valid for the improvement in the
transmittance of the films deposited with 25.0% nitrogen flow, as Figure 5.8 not shows
any improvement in crystallinity of this film. Likewise is the transparency of the films
deposited with 3.2% nitrogen flow not affected by the heat treatment, although there
is an improvement in the crystallinity.
The reference sample and the hydrogenated samples show a stronger correlation
between the transparency and the crystallinity of the films. This change in transparency and structure may be correlated to changes in the band gap. However, this is
an assumption as we do not have any data that can confirm this postulation.
Figure 5.1 shows the resistivity of the samples. The as-deposited samples which
show the best electrical properties are those deposited with a 1.6 flow% of oxygen. As
this value coincides with the flow which leads to the highest transmittance, we can
conclude from this study that a 1.6 flow% of oxygen is the most suitable gas supply for
RT ITO deposition in our sputtering system, in order to achieve optimal electrical and
optical properties. (A small supply of oxygen is required in order to replace oxygen
removed by pumping and achieve a stoichiometric film, as discussed in subsection 5.1.3.)
When the samples are annealed, there are only small differences between the resistivity of the samples deposited with a 1.6 flow% of oxygen and a 1.6 flow% of nitrogen.
From the almost linear dependence between the flow% of hydrogen and the resistivity,
it is most likely that the resistivity of the films deposited at a 1.6 flow% of hydrogen
would not deviate from the resistivity at a 1.6 flow% of oxygen or nitrogen. The reference sample exhibits the lowest resistivity compared to all other samples. The figure
shows that annealing has overall a positive effect on the resistivity. (The nitrogen
samples less affected.)
To conclude, in order to achieve the best transmittance and conductivity for ITO
films deposited at RT, the standard recipe with a 1.6 flow% of oxygen is recommended.
If the films are subsequently heated in air, the standard recipe with a 1.6 flow% oxygen
is recommended in order to achieve the best transmittance and only Ar (no additional
gas) is recommended in order to achieve the best conductivity.
60
5.2
Summary of results and conclusions
Interfacial properties
This section presents the two papers where we examined the ITO/Si interface. Both
papers include results from TEM and XPS, paper V on a sample prepared at IFE and
paper VI on a sample prepared at TUD.
The ITO/Si interface is of great importance as this is the junction where electrons
are extracted from the current generating part of the solar cell. Hence, as little current
loss as possible is desired. The mechanisms of current loss can be due to different
factors. One factor is the presence of a thin insulating silicon oxide layer. This was
discussed in chapter 2, subsection 2.3.4. Paper V and VI studies the ITO/Si interface
and the thin silicon oxide layer.
Paper VI describes, in our opinion, a unique experiment with solid and original
results, and is reviewed in the thesis in more detail.
5.2.1
Paper V - Elemental distribution and oxygen deficiency
of magnetron sputtered indium tin oxide films
A TEM study of the interface between the ITO layer and a Si substrate was carried out
on one of the ITO/Si samples, as-deposited and annealed, made at IFE. The ITO was
deposited by using the standard recipe with a 3.2 flow% of oxygen and the annealing
was made in air on a hot-plate for 15 min at 300 ◦ C.
The results published in Paper V show that both the RT deposited and annealed
films are crystalline. XPS depth profiling and TEM of the ITO/Si interface reveal
increasing amounts of elemental In and Sn toward the interface, as well as the presence
of a SiOx layer. There is no difference in the interface oxide between the as-deposited
and annealed sample.
The TEM shows that the ITO/Si interface contain oxygen deficient areas and small
clusters of In and Sn. During the TEM examination the small clusters of In are
crystallized and both the In clusters and the silicon oxide layer are enlarged. See
Figure 5.9. This is related to the electron beam exposure and induced splitting and
reconfiguration of atomic bonds.
Conslusion - Paper V
In Paper V we examined one of the ITO/Si samples by TEM. The TEM results showed
that the films were crystalline. At the interface, we observed small clusters of In and Sn,
a thin silicon oxide layer and oxygen deficient areas. During the TEM characterization
we discovered that the small clusters of indium and the silicon oxide layer were enlarged.
This was expected to be due to the heat induced by the electron beam exposure.
Interfacial properties
61
Figure 5.9: HRTEM images of the ITO/Si interface, (A) before and (B) after electron
beam exposure.
5.2.2
Paper VI - An in situ X-ray photoelectron spectroscopy
study of the initial stages of rf magnetron sputter deposition of indium tin oxide on p-type Si substrate
XPS is a characterization method which is sensitive to surface contamination. For
instance, carbon impurities is previously proven to affect the work function value of
ITO measured by XPS [101]. Thus, in order to avoid influence by any contamination
a cleaning procedure prior the spectroscopy is required. Cleaning the sample by ion
bombardment will remove most of the impurity elements, though, some atoms will
still remain at the surface. A drawback of ion bombardment is that it may induce
side-effects such as chemical changes in the material, as described in Ref. [97] and
references therein. One type of chemical change can be depletion of certain elements
when the material is a compound. This can either be due to different sputtering yield
or due to preferential sputtering caused by higher volatility of one element. In the
case of a metal oxide, such depletion might induce a reduction in the oxidation state
of the metal. Other chemical changes would be the effect of ’knock on’ (Ref. [97]),
which is an incorporation of elements further into the material during sputtering. Another side-effect of ion bombardment is the topological changes at the sample surface,
also described in Ref. [97]. For both sample cleaning and depth profiling XPS, the
possibility of chemical changes must be taken into consideration.
A method which can overcome the influence of contamination and cleaning procedures is in situ XPS. Without breaking vacuum after material deposition, it is possible
to carry out XPS studies of ultrathin layers step by step during growth of a thicker
film, not influenced by contamination.
At the Technical University of Darmstadt, we were fortunate to have the opportunity to carry out a thorough in situ XPS study of the ITO/Si interface and the
interfacial silicon oxide layer by the use of DAISY-MAT. See section 4.3 for detailes
62
Summary of results and conclusions
about DAISY-MAT.
In our study, we used a Si sample with the orientation < 100 > and resistivity 1-3
Ω cm. This was etched for two min and rinsed one min in DI-water. The cleaning
procedure was repeated twice in order to completely remove any native oxide. The
substrate was blown dry with nitrogen. Subsequent to this treatment, the substrate
was mounted to the sample holder and transferred quickly (in nitrogen atmosphere) to
the load lock of DAISY-MAT. XPS was carried out on the silicon substrate prior to the
depositions and subsequent to each deposition. The ITO layers were deposited at RT
in four steps of 2, 8, 30 and 900 s, giving a total deposition time of 940 s. The substrate
to target distance was 10 cm, the working power 25 W (power density of ∼1.23 W/cm2 )
and the working pressure was 5x10−3 mbar. Only argon was used as sputtering gas and
the flow of argon during sputtering was 6 sccm. The target composition was 90wt%
In2 O3 and 10wt% SnO2 .
Figure 5.10 shows the XPS spectra of the ITO/Si interface study. It shows the In
3d5/2 , Sn 3d5/2 , O 1s and Si 2p spectra of a clean Si substrate and after 2, 10, 40 and
940 seconds of ITO deposition.
The In 3d5/2 , Sn 3d5/2 and O 1s spectra are fitted with two or three components,
while the Si 2p spectra from substrate and after 2, 10 and 40 s deposition are fitted
with two, four, five and one Si3/2 and Si1/2 peaks, respectively. No In or Sn peaks are
observed prior deposition and no Si peaks after the last deposition. The thicknesses
are estimated from TEM images and calculations based on the integrated area of the
Si peaks in the XPS spectra [102].
The predominant peaks in the In 3d5/2 and Sn 3d5/2 spectra after the first deposition
with a duration of 2 s are attributed to elemental In and Sn. This was done by following
a similar analysis procedure reported by Thøgersen et al. [103]. Elemental In is also
found after 10 s, while elemental Sn is present both after 10 and 940 s. A segregation
of tin to the film surface is previously reported by others [104]. The mentioned analysis
is performed for the predominant InII and SnII peaks after 10, 40 and 940 s as well,
and these peaks are assigned to In and Sn in crystalline ITO.
In paper V we argue that In components at 444.9 eV and 446.0 eV (energy separation of 1.1 eV) could be attributed to crystalline and amorphous ITO, respectively
[103]. By further examination, applying the electroneutrality principle, the presence
of amorphous ITO is excluded [104]. We confirm this conclusion in the present study
by analysis of TEM images (Fig. 2 in paper VI) which show no diffuse circles in the
diffraction pattern.
The origin of the minor peaks InIII (at 445.9, 445.7 and 446.0 eV after 2, 40 and
940 s, respectively) and SnIII (487.7 and 487.5 eV after 40 and 940 s, respectively)
is assigned to hydrogen. Hydrogen is detected by SIMS in both bulk and interfacial
ITO (see Figure 3 in paper VI). The position of the InIII component coincides with
Interfacial properties
63
Figure 5.10: In 3d5/2 , Sn 3d5/2 , O 1s and Si 2p spectra of substrate, after 2, 10, 40
and 940 seconds of ITO deposition. Some of the peaks are multiplied with an integer in
order to improve the illustration of the spectra. The insets are expansions of the silicon
oxide established after 2 and 10 seconds.
64
Summary of results and conclusions
the In(OH)3 state (445.8 eV) reported by C. D. Wagner [105]. Similarly, the high
binding energy peak SnIII can be assigned to bonds between hydrogen and Sn. Another
explanation can be plasmon excitation. This is previously discussed by Christou et al.
and Gassenbauer et al. [106, 41].
The two O 1s components found at the Si wafer surface (531.9 and 532.6 eV) are
most likely due to water molecules and contamination adsorbed at the substrate surface, originated from the DI water rinsing or exposure to air. The values coincide with
the binding energies for hydroxides, contamination and oxygen due to air exposure
(531.7±0.2 and 532.7±0.2 eV) as reported by Plá et al. [107]. Hydroxides are plausible as the substrate was cleaned in HF and rinsed in DI water. OH groups at Si (100)
wafers after HF etch and DI water rinsing were previously detected using High Resolution Electron Energy Loss Spectroscopy [108]. Contamination due to air exposure
is also plausible as traces of carbon are detected at the substrate surface and after 2 s
deposition (not illustrated).
After 2 s of deposition, the predominant O 1s peak is at BE=532.6 eV (OIII ). The
intensity of this peak increases slightly after the next deposition (not illustrated in the
figure) and disappears after further deposition. Hence, this binding energy is most
likely due to other chemical states than found at the substrate surface. An obvious
explanation of the origin can be compounds of oxygen and silicon as SiOx is observed
as a shoulder after 2 and 10 s in the Si 2p spectra (discussed below). A binding energy
of 532.6 eV is reported by Wagner et al. of thermally oxidized silicon wafers [109].
The interfacial SiOx layer is also observed by TEM and SIMS in Figure 2 and 3 in the
paper (VI).
The analysis of the OI and OII components found after deposition is similar to the
work reported by Thøgersen et al. [103]. Both peaks are attributed to crystalline ITO,
the OII component related to O2− ions at oxygen deficient sites.
The binding energy, FWHM and chemical shift of the Si 2p and SiOx peaks were
fitted as shown by Thøgersen et al. [110]. In addition to the Si 2p peaks corresponding
to the 2p1/2 and 2p3/2 spin states, two small components with binding energy in the
range 99.8 - 100.6 eV are found. These small components are attributed to silicon
bonded to hydrogen, according to Thøgersen et al. [110]. In that work, a peak with
binding energy at 99.7 is attributed to Si3 SiH and is close to our value (99.8 eV). The
two small components at 100.1 and 100.6 eV after 10 s can be attributed to Si2 SiH2 ,
which has a binding energy 0.57 eV higher than elemental Si [111].
The SiOx shoulder with a peak position at 102.6 eV is fitted by six and eight components after 2 and 10 s deposition, respectively. The components in the decomposed
SiOx peak correspond to the Si2 O, SiO, Si2 O3 (suboxides) and SiO2 states. Two components are used for each oxide state, due to the spin-orbit coupling. No SiO2 is found
after 2 s deposition and the predominant peaks correspond to the Si2 O and Si2 O3
General conclusions
65
suboxide states. After 10 s deposition, the Si2 O3 peaks are predominant and small
peaks corresponding to the SiO2 state are present. These results show that SiO2 is not
the dominating oxide in the SiOx interfacial layer between ITO and Si. An interface
consisting of purely SiO2 is previously assumed in the literature [112].
The presence of suboxides may introduce another band gap than SiO2 in the SiOx
layer. A narrower band gap for suboxides in a SiO2 /n-type Si (111) interface region is
calculated by Yamashita et al. [113]. They show that the main oxides in a thin interface
region between a Si substrate and a SiO2 film are Si2 O and Si2 O3 and that the band
gap increases as the oxidation state increases (Eg (Si2 O)<Eg (Si2 O3 )<Eg (SiO2 )).
A different band gap alignment of the ITO/Si interface than in the case of SiO2 as
the main silicon oxide and the presence of metallic In and Sn at the interface may play
a significant role in the carrier transport mechanism in the ITO/Si junction.
Conclusion - Paper VI
In Paper VI we studied the initial growth of ITO on Si by in situ XPS examination.
The results showed how the oxygen and silicon readily reacts, whereas the tin and
indium clusters in the early start of the formation of an ITO film at a Si substrate.
The analysis of the XPS data revealed that the silicon oxide layer formed between the
Si and ITO film was not a silicon dioxide layer, but a mixture of different suboxides.
According to the literature, such suboxides exhibit a different band gap than silicon
dioxide. Such information is important to establish a realistic model of the band gap
alignment of the ITO/Si structure, which can be used for a reliable predication of the
electronic transfer between the two materials.
5.3
General conclusions
Regarding the results and conclusions from the first four papers it is possible to extract
a ”best of” recipe for RT sputter deposition of ITO for solar cell application, considering
the electrical and optical properties as the key properties. Hence, for application of
these results, I would recommend the film deposition process including a 1.6 flow%
of oxygen. The resulting films exhibit high transparency and low resistivity as both
as-deposited and annealed. Table 5.2 summarize a ”best of” process.
The study of electrical and optical properties showed that RT deposition with pure
argon and post-deposition annealing of the ITO film in air resulted in the lowest resistivity. Hence, if the resisitivity is more important than the transmittance of ITO for a
special application, addition of oxygen is superfluous when the samples are annealed.
Another important aspect of ITO for solar cell application is the interfaces. As
previously discussed, the interface between ITO and Si can be of significance for the
66
Summary of results and conclusions
Gas and flow
Power
Pressure
Temperature
Target to sample distance
Velocity of sample holder
Ramping up time
Dehumidifying in L1
Annealing
3 sccm oxygen and 180 sccm Ar
4 kW
3.1x10−3 mbar
RT
80 mm
0.7 m/min
15 min
30 min at 100◦ C
15 min at 300◦ C in air
Table 5.2: Deposition and annealing parameters used in the ”best of” recipe.
current transport properties in the solar cell structure.
Paper V and VI indicate that oxygen in the ITO easily react with the surface of
the silicon substrate during sputter deposition, forming the silicon oxide and leaving
Sn and In unbound. This is independet on the supply of oxygen in the two studies (0
flow% and 3.2 flow%).
5.4
Future Work
Of course there are more issues to explore in the field of TCOs. After the years I’ve
spent on working with ITO there are several things I would have done in addition, if
time would let me. This concerns mostly ITO/Si interface studies.
While working with Paper VI I realized that it would have been quite informative
if we had included an ultraviolet photoelectron spectroscopy (UPS) scan in addition
to XPS. This could have given us an indication of the actual band alignment of the
ITO/Si interface. A similar study with another substrate morphology like for instance
a-Si, m-Si and textured wafers would have revealed a more realistic interface used in
solar cells, e.g. the HIT solar cell.
Another interesting issue could be a XPS/UPS ITO/Si interface study of different
deposition parameters like for instance working power, pressure, temperature and addition of gases. One XPS/UPS study could uncover differences at the interface between
pre-annealed substrate with ITO deposited on it and a post-annealed sample. Such
studies would probably reveal more essential information about the physics behind the
film formation of ITO on Si.
As mentioned previously, indium is an expensive element and alternative materials
to replace ITO is needed. Much research has already been done in the field of ZnO, but
there are still issues correlated to the replacement of ITO. Central aspects are chemical
stability of ZnO and the resistivity, which can still not surpass ITO. New TCMs, like
for instance metal nanowires, are other alternatives that could replace ITO. However,
more research is required in order to compete with the excellent optical and electrical
Future Work
67
properties of ITO and other TCOs.
Another field of TCOs is production of p-type TCO. One of the challenges in
producing p-type TCOs lies in the introduction of acceptor states. When the acceptor
states are introduced the conductivity is often temperature dependent over a broad
temperature interval. Mobility of carriers is usually one order of magnitude lower
compared to the mobility in n-type TCO. Other challenges are low reproducibility, film
instability and low hole concentrations. Overcoming the aspects with p-type TCOs will
expand the field of new applications within transparent electronics.
68
Summary of results and conclusions
Bibliography
[1] W. Beyer, J. Hüpkes, and H. Stiebig. Transparent conducting oxide films for thin
film silicon photovoltaics. Thin Solid Films, 516(2-4):147–154, 2007.
[2] M. Quaas, C. Eggs, and H. Wulff. Structural studies of ITO thin films with the
Rietveld method. Thin Solid Films, 332(12):277 – 281, 1998.
[3] [online], U.S Energy Information Administration. International Energy Outlook
2013. http://www.eia.gov/forecasts/archive/ieo13/ (Dec. 2013).
[4] [online], National Renewable Energy Laboratory. Best Research-Cell Efficiencies.
http://www.nrel.gov/ncpv/images/efficiency chart.jpg (Jan. 2014).
[5] R. Har-Lavan and D. Cahen. 40 Years of Inversion Layer Solar Cells: From
MOS to Conducting Polymer/Inorganic Hybrids. IEEE Journal of Photovoltaics,
3(4):1443–1459, 2013.
[6] T. Mishima, M. Taguchi, H. Sakata, and E. Maruyama. Development status of
high-efficiency HIT solar cells. Solar Energy Materials and Solar Cells, 95(1):18–
21, 2011.
[7] [online], Panasonic.
HIT Power 205N(NKHA5) Datasheet - Panasonic.
http://www.panasonic.com/business/pesna/includes/pdf/eco-constructionsolution/HIT Power 205N%28NKHA5%29 Datasheet.pdf (Feb. 2014).
[8] [online], Weizmann Institute of Science.
Materials and
http://www.weizmann.ac.il/materials/igorl/Sputter (Nov. 2014).
interfaces.
[9] [online], Oxford Vacuum Science. Sputter deposition. http://www.oxfordvacuum.com/background/thin film/sputtering.htm (Jan. 2014).
[10] S. A. Campbell. The science and engineering of microelectronic fabrication. Oxford University Press, New York, 2001.
[11] [online],
Technische Universität Darmstadt.
Darmstadt’s integated
system
for
material
science.
http://www.mawi.tudarmstadt.de/of/of/methoden/daisy mat/index.en.jsp (Feb. 2014).
69
70
Bibliography
[12] A. Thißen. Grundlagen der Röntgenphotoelektronenspektroskopie(X-Ray Photoelectron Spectroscopy ”XPS”), A compendium used in the teaching of students
at Technische Universität Darmstadt. 2010.
[13] [online], Exxon Mobile. The Outlook for Energy: A View to 2040, 2014.
http://cdn.exxonmobil.com/∼/media/Reports/Outlook%20For%20Energy/2014/2014Outlook-for-Energy-low-resolution.pdf (Jan. 2014).
[14] [online], barefoot college. Barefoot college. http://www.barefootcollege.org/
(June 2014).
[15] [online], GoeModel SOLAR - solargis. World map of direct normal radiation. http://solargis.info/doc/ pics/freemaps/1000px/dni/SolarGIS-Solar-mapDNI-World-map-en.png (June 2014).
[16] [online], The International Energy Agency. Key world energy statistics.
http://www.iea.org/publications/freepublications/publication/KeyWorld2013.pdf
(June 2014).
[17] [online],
Intergovernmental Panel on Climate Change.
Climate change 2014:
Mitigation of climate change - chapter
7.
http://report.mitigation2014.org/drafts/final-draftpostplenary/ipcc wg3 ar5 final-draft postplenary chapter7.pdf (June 2014).
[18] [online], International Energy Agency.
Life cycle inventories and
life cycle assessments of photovoltaic systems.
http://www.ieapvps.org/index.php?id=3&eID=dam frontend push&docID=881 (June 2014).
[19] A. Suzuki, T. Matsushita, T. Aoki, Y. Yoneyama, and M. Okuda. Pulsed laser
deposition of transparent conducting indium tin oxide films in magnetic field
perpendicular to plume. Japanese Journal of Applied Physics Part 2-Letters,
40(4B):L401–L403, 2001.
[20] Swati Ray, Ratnabali Banerjee, N Basu, A K Batabyal, and A K Barua. Properties of tin doped indium oxide thin films prepared by magnetron sputtering.
Journal of Applied Physics, 54(6):3497–3501, 1983.
[21] D. S. Ginley and J. D. Perkins. Transparent conductors. In David S Ginley,
editor, Handbook of Transparent Conductors. Springer Science+Business Media,
LLC, Boston, MA, 2010.
[22] C. Liu and X. Yu. Silver nanowire-based transparent, flexible, and conductive
thin film. Nanoscale research letters, 6(1):75, 2011.
Bibliography
71
[23] G. Xu, J. Liu, Q. Wang, R. Hui, Z. Chen, V. a Maroni, and J. Wu. Plasmonic
graphene transparent conductors. Advanced Materials, 24(10):OP71–6, March
2012.
[24] A. Pron and P. Rannou. Processible conjugated polymers: from organic semiconductors to organic metals and superconductors. Progress in Polymer Science,
27:135–190, 2002.
[25] [online], PPG Glass Education Center. Glass topics - what is low-e glass?
http://educationcenter.ppg.com/glasstopics/how lowe works.aspx (Aug. 2014).
[26] J. Livage and D. Ganguli. Solgel electrochromic coatings and devices: A review.
Solar Energy Materials and Solar Cells, 68(3-4):365–381, 2001.
[27] T. Miyata, T. Hikosaka, and T. Minami. High sensitivity chlorine gas sensors
using multicomponent transparent conducting oxide thin films. Sensors and Actuators B: Chemical, 69:16–21, 2000.
[28] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, and H. Hosono. Roomtemperature fabrication of transparent flexible thin-film transistors using amorphous oxide semiconductors. Nature, 432(7016):488–92, 2004.
[29] Y. Tak, K. Kim, H. Park, K. Lee, and J. Lee. Criteria for ITO (indiumtin-oxide)
thin film as the bottom electrode of an organic light emitting diode. Thin Solid
Films, 411(1):12–16, 2002.
[30] R.G. Gordon. Criteria for choosing transparent conductors. MRS Bulletin,
25(8):52–57, 2000.
[31] J.D. Perkins, J.A. del Cueto, J.L. Alleman, C. Warmsingh, B.M. Keyes, L.M.
Gedvilas, P.A. Parilla, B. To, D.W. Readey, and D.S. Ginley. Combinatorial
studies of zn-al-o and zn-sn-o transparent conducting oxide thin films. Thin
Solid Films, 411(1):152 – 160, 2002.
[32] E. Burstein. Anomalous optical absorption limit in InSb. Physical Review,
93(3):632–633, 1954.
[33] T. S. Moss. The interpretation of the properties of indium antimonide. Proceedings of the Physical Society of London Section B, 67(418):775–782, 1954.
[34] P. D. C. King and T. D. Veal. Conductivity in transparent oxide semiconductors. Journal of Physics. Condensed matter : an Institute of Physics journal,
23(33):334214, 2011.
72
Bibliography
[35] E. Fortunato, D. Ginley, H. Hosono, and D. C. Paine. Transparent Conducting
Oxides for Photovoltaics. MRS Bulletin, 32(March):242–247, 2007.
[36] I. Hamberg, C. G. Granqvist, K. F. Berggren, B. E. Sernelius, and L. Engstrom.
Band-gap widening in heavy Sn-doped In2 O3 . Physical Review B, 30(6):3240–
3249, 1984.
[37] T. J. Coutts, D. L. Young, and T. A. Gessert. Modeling, characterization, and
properties of transparent conducting oxides. In D. S. Ginley, editor, Handbook of
Transparent Conductors. Springer Science+Business Media, LLC, Boston, MA,
2010.
[38] K. Ellmer. Transparent conductive zinc oxide and its derivatives. In D. S. Ginley,
editor, Handbook of Transparent Conductors. Springer Science+Business Media,
LLC, Boston, MA, 2010.
[39] B.E. Sernelius, K. F. Berggren, Z. C. Jin, I. Hamberg, and C.G. Granquist.
Band-gap tailoring of ZnO by means of heavy Al doping. Physical Review B,
37(17):10244–10248, 1988.
[40] J.R. Bellingham, W.A. Phillips, and C.J. Adkins. Intrinsic performance limits in
transparent conducting oxides. Journal of Materials Science Letters, 11:263–265,
1992.
[41] Y. Gassenbauer and A. Klein. Electronic and chemical properties of tin-doped
indium oxide (ITO) surfaces and ITO/ZnPc interfaces studied in-situ by photoelectron spectroscopy. The journal of physical chemistry. B, 110(10):4793–801,
2006.
[42] Y. Gassenbauer, R. Schafranek, A. Klein, S. Zafeiratos, M. Havecker, A. KnopGericke, and R. Schlogl. Surface states, surface potentials, and segregation at
surfaces of tin-doped In2 O3 . Physical Review B, 73(24), 2006.
[43] [online], U.S. Geological Survey. Minerals information - indium, statistics and
information. http://minerals.usgs.gov/minerals/pubs/commodity/indium/ (July
2012).
[44] [online], Royal Society Of Chemistry.
Periodic Table http://www.rsc.org/periodic-table/element/49/indium (Aug. 2014).
Indium.
[45] G. Phipps, C. Mikolajczak, and T. Guckes. The sustainability of indium and
gallium supply. In 22nd European Photovoltaic Solar Energy Conference, pages
2389–2392, 2007.
Bibliography
73
[46] J. H. W. Dewit. Structural aspects and defect chemistry in In2 O3 . Journal of
Solid State Chemistry, 20(2):143–148, 1977.
[47] M. Marezio. Refinement of crystal structure of In2 O3 at two wavelengths. Acta
Crystallographica, 20:723–728, 1966.
[48] J. C. C. Fan, F. J. Bachner, and G. H. Foley. Effect of O2 pressure during
deposition on properties of rf-sputtered Sn-doped In2 O3 films. Applied Physics
Letters, 31(11):773–775, 1977.
[49] T. O. Mason, G. B. Gonzlez, J.-H. Hwang, and D. R. Kammler. Point defects
and related properties of highly co-doped bixbyite In2 O3 . Physical Chemistry
Chemical Physics, 5(11):2183–2189, 2003.
[50] C. Janowitz, V. Scherer, M. Mohamed, A. Krapf, H. Dwelk, R. Manzke,
Z. Galazka, R. Uecker, K. Irmscher, R. Fornari, M. Michling, D. Schmeißer,
J. R. Weber, J. B. Varley, and C. G. Van De Walle. Experimental electronic
structure of In2 O3 and Ga2 O3 . New Journal of Physics, 13(8):085014, 2011.
[51] I. Hamberg and C. G. Granqvist. Evaporated Sn-doped In2 O3 films: Basic optical properties and applications to energy-efficient windows. Journal of Applied
Physics, 60(11):R123–R159, 1986.
[52] K. Okada, S. Kohiki, S. Luo, D. Sekiba, S. Ishii, M. Mitome, A. Kohno, T. Tajiri,
and F. Shoji. Correlation between resistivity and oxygen vacancy of hydrogendoped indium tin oxide thin films. Thin Solid Films, 519(11):3557–3561, 2011.
[53] J. H. W. Dewit. Electrical properties of In2 O3 . Journal of Solid State Chemistry,
8(2):142–149, 1973.
[54] L. F. J. Piper, A. DeMasi, S. W. Cho, K. E. Smith, F. Fuchs, F. Bechstedt,
C. Korber, A. Klein, D. J. Payne, and R. G. Egdell. Electronic structure of In2 O3
from resonant x-ray emission spectroscopy. Applied Physics Letters, 94(2):022105,
2009.
[55] H. Kim, C. M. Gilmore, A. Pique, J. S. Horwitz, H. Mattoussi, H. Murata,
Z. H. Kafafi, and D. B. Chrisey. Electrical, optical, and structural properties of
indiumtinoxide thin films for organic light-emitting devices. Journal of Applied
Physics, 86(11):6451, 1999.
[56] A. Klein, C. Körber, A. Wachau, F. Säuberlich, Y. Gassenbauer, S. P. Harvey,
D. E. Proffit, and T. O. Mason. Transparent Conducting Oxides for Photovoltaics: Manipulation of Fermi Level, Work Function and Energy Band Alignment. Materials, 3(11):4892–4914, 2010.
74
Bibliography
[57] A. Walsh, J. L. F. Da Silva, S. H. Wei, C. Korber, A. Klein, L. F. J. Piper,
A. DeMasi, K. E. Smith, G. Panaccione, P. Torelli, D. J. Payne, A. Bourlange,
and R. G. Egdell. Nature of the band gap of In2O3 revealed by first-principles
calculations and x-ray spectroscopy. Physical Review Letters, 100(16):167402,
2008.
[58] P. Erhart, A. Klein, R. G. Egdell, and K. Albe. Band structure of indium oxide:
Indirect versus direct band gap. Phys. Rev. B, 75:153205, 2007.
[59] K. Sugiyama, H. Ishii, Y. Ouchi, and K. Seki. Dependence of indium-tin-oxide
work function on surface cleaning method as studied by ultraviolet and x-ray
photoemission spectroscopies. Journal of Applied Physics, 87(1):295–298, 2000.
[60] A. Klein, C. Körber, A. Wachau, F. Säuberlich, Y. Gassenbauer, S. P. Harvey,
D. E. Proffit, and T. O. Mason. Transparent Conducting Oxides for Photovoltaics: Manipulation of Fermi Level, Work Function and Energy Band Alignment. Materials, 3(11):4892–4914, 2010.
[61] S. Q. Hussain, W. Oh, S. Ahn, A. H. T. Le, S. Kim, Y. Lee, and J. Yi. Rf
magnetron sputtered indium tin oxide films with high transmittance and work
function for a-si:h/c-si heterojunction solar cells. Vacuum, 101(0):18 – 21, 2014.
[62] M. Buchanan, J. B. Webb, and D. F. Williams. Preparation of conducting and
transparent thin films of tin-doped indium oxide by magnetron sputtering. Applied Physics Letters, 37(2):213, 1980.
[63] [online], Exxon Mobile.
The Outlook for Energy: A View to 2040.
http://corporate.exxonmobil.com/en/energy/energy-outlook (Dec. 2013).
[64] [online],
Fraunhofer-Institut
für
Solare
Energisysteme
ISE.
Stromerzeugung
aus
Solarund
Windenergie
im
Jahr
2013.
http://www.ise.fraunhofer.de/de/downloads/pdffiles/aktuelles/stromproduktion-aus-solar-und-windenergie-2013.pdf
Jan.
2014.
[65] [online], Georgia state University; Department of Physics & Astronomy.
Abundances of the elements in the earth’s crust. http://hyperphysics.phyastr.gsu.edu/hbase/tables/elabund.html (July 2012).
[66] C. G. Granqvist. Transparent conductors as solar energy materials: A panoramic
review. Solar Energy Materials and Solar Cells, 91(17):1529–1598, 2007.
[67] T. M. Barnes, M. O. Reese, J. D. Bergeson, B. A. Larsen, J. L. Blackburn, M. C.
Beard, J. Bult, and J. van de Lagemaat. Comparing the Fundamental Physics
75
Bibliography
and Device Performance of Transparent, Conductive Nanostructured Networks
with Conventional Transparent Conducting Oxides. Advanced Energy Materials,
2(3):353–360, 2012.
[68] L. Barraud, Z. C. Holman, N. Badel, P. Reiss, A. Descoeudres, C. Battaglia,
S. De Wolf, and C. Ballif. Hydrogen-doped indium oxide/indium tin oxide bilayers for high-efficiency silicon heterojunction solar cells. Solar Energy Materials
and Solar Cells, 115:151–156, 2013.
[69] G. Cheek, A. Genis, J.B. DuBow, and V. R Pai Verneker. Antireflection properties of indium tin oxide (ITO) on silicon for photovoltaic applications. Applied
Physics Letters, 35(7):495–497, 1979.
[70] T. Koida, H. Fujiwara, and M. Kondo. Hydrogen-doped In2 O3 as Highmobility Transparent Conductive Oxide. Japanese Journal of Applied Physics,
46(28):L685–L687, 2007.
[71] N.S. Chang and J.R. Sites. Electronic characterization of indium tin oxide-silicon
photo-diodes. Journal of Applied Physics, 49(9):4833–4837, 1978.
[72] H. Kobayashi, T. Ishida, Y. Nakato, and H. Tsubomura. Mechanism of carrier
transport in highly efficient solar cells having indium tin oxide/ Si junctions.
Journal of Applied Physics, 69(3):1736–1743, 1991.
[73] S.M. Goodnick, J.F. Wager, and C.W. Wilmsen. Thermal degradation of indiumtin-oxide/p-silicon solar cells. Journal of Applied Physics, 51(1):527, 1980.
[74] H. Kobayashi, T. Ishida, Y. Nakato, and H. Mori. Mechanism of carrier transport
through a silicon-oxide layer for indium-tin-oxide/silicon-oxide/silicon solar
cells. 78(6):3931–3939, 1995.
[75] [online], Silevo; News and Events. Silevo achieves 22%; Ranks first in PTC
rating for technology with c-Si layers. http://silevosolar.com/2012/12/05/silevoachieves-22-ranks-first-in-ptc-rating-for-technology-with-c-si-layers/ (Nov. 2013).
[76] [online], Panasonic. Experience in solar for almost 40 years.
solar.panasonic.net/en/about-panasonic-solar/ (Nov. 2014).
http://eu-
[77] N. Dwivedi, S.l Kumar, A. Bisht, K. Patel, and S. Sudhakar. Simulation approach
for optimization of device structure and thickness of HIT solar cells to achieve
27% efficiency. Solar Energy, 88:31–41, 2013.
®
[78] [online], Panasonic. The HIT
photovoltaic modules: VBHN245SJ25 /
VBHN240SJ25. http://eu-solar.panasonic.net/en/products/n-240-n-245/ (Jan.
2014).
76
Bibliography
®
[79] [online], Panasonic.
Panasonic HIT
Solar Cell Achieves World’s
Highest Energy Conversion Efficiency*1 of 25.6% at Research Level.
http://panasonic.co.jp/corp/news/official.data/data.dir/2014/04/en1404104/en140410-4.html (June 2014).
[80] V. A. Dao, J. Heo, H. Choi, Y. Kim, S. Park, S. Jung, N. Lakshminarayan,
and J. Yi. Simulation and study of the influence of the buffer intrinsic layer,
back-surface field, densities of interface defects, resistivity of p-type silicon substrate and transparent conductive oxide on heterojunction with intrinsic thinlayer (HIT) solar cell. Solar Energy, 84(5):777 – 783, 2010.
[81] K. Seshan. Handbook of Thin-Film Deposition Processes and Techniques - Principles, Methods, Equipment and Applications (2nd Edition). William Andrew
Publishing/Noyes, 2002.
[82] [online], Leiden University. A comparative study of platinum nanodeposits
on HOPG (0001), MnO(100) and MnOx/MnO(100) surfaces by STM and
AFM after heat treatment in UHV, O2 , CO and H2 - CHAPTER 3,
Growth modes, sintering mechanisms and equilibrium shape of small particles.
https://openaccess.leidenuniv.nl/bitstream/handle/1887/15973/Chapter+3.pdf?
sequence=9 (June 2014).
[83] K. Wasa and S. Hayakawa. Handbook of sputter deposition technology: principles,
technology and applications. Noyes Publications, Park Ridge, N.J., 1992.
[84] E. Terzini, P. Thilakan, and C. Minarini. Properties of ITO thin films deposited
by RF magnetron sputtering at elevated substrate temperature. Materials Science and Engineering B, 77:110 – 114, 2000.
[85] D. G. Brandon and W. D. Kaplan. Microstructural characterization of materials.
Wiley, Chichester, 2008.
[86] J.F. Moulder and J. Chastain. Handbook of X-ray photoelectron spectroscopy: a
reference book of standard spectra for identification and interpretation of XPS
data. Physical Electronics, 1995.
[87] [online], CasaXPS. CasaXPS. http://www.casaxps.com/ (June 2014).
[88] [online], The Harvey M. Krueger Familiy Center for Nanoscience and Nanotechnology. The Unit for Nanocharacterization - X-Ray Diffractometer.
http://nanoscience.huji.ac.il/unc/xrd main.htm (May 2014).
[89] [online], Portland State University.
X-Ray Diffraction (XRD).
http://web.pdx.edu/%7epmoeck/phy381/Topic5a-XRD.pdf (May 2014).
Bibliography
77
[90] J. C. Bravman and R. Sinclair. The preparation of cross-section specimens for
transmission electron microscopy. Journal of Electron Microscopy Technique,
1(1):53–61, 1984.
[91] D. K. Schroder. Semiconductor material and device characterization. John Wiley
& Sons, Inc., 3rd ed. edition, 2006.
[92] Z. L. Wang. Transmission Electron Microscopy of Shape-Controlled Nanocrystals
and Their Assemblies. The Journal of Physical Chemistry B, 104(6):1153–1175,
2000.
[93] T. Koida, H. Fujiwara, and M. Kondo. High-mobility hydrogen-doped In2 O3
transparent conductive oxide for a-Si:H/c-Si heterojunction solar cells. Solar
Energy Materials and Solar Cells, 93(6-7):851–854, 2009.
[94] Keran Zhang, Furong Zhu, C. H. a. Huan, and a. T. S. Wee. Effect of hydrogen partial pressure on optoelectronic properties of indium tin oxide thin films
deposited by radio frequency magnetron sputtering method. Journal of Applied
Physics, 86(2):974, 1999.
[95] D. Kim, S. Lee, G. Lee, and S. Kwon. Electrical, optical, and structural properties
of indiumtinoxide thin films for organic light-emitting devices. Thin Solid Films,
515(17):6949, 2007.
[96] H. N. Cui, V. Teixeira, L. J. Meng, R. Martins, and E. Fortunato. Influence of
oxygen/argon pressure ratio on the morphology, optical and electrical properties
of ITO thin films deposited at room temperature. Vacuum, 82:1507–1511, 2008.
[97] J. C. Riviere. Practical Surface Analysis: by Auger and X-ray Photoelectron
Spectroscopy. Wiley, Chichester, 1983.
[98] G. Frank and H. Köstlin. Electrical Properties and Defect Model of Tin-Doped
Indium Oxide Layers. Applied Physics A, 27:197–206, 1982.
[99] L. L. Kerr, X. Li, M. Canepa, and A. J. Sommer. Raman analysis of nitrogen
doped ZnO. Thin Solid Films, 515(13):5282–5286, 2007.
[100] S. S. Pan, G. H. Li, L. B. Wang, Y. D. Shen, Y. Wang, T. Mei, and X. Hu.
Atomic nitrogen doping and p-type conduction in SnO2 . Applied Physics Letters,
95(22):222112, 2009.
[101] K Sugiyama, H Ishii, Y Ouchi, and K Seki. J. Appl. Phys., 87(1):295, 2000.
[102] M. P. Seah. Practical Surface Analysis: by Auger and X-ray Photoelectron Spectroscopy. Wiley, Chichester, 1983.
78
Bibliography
[103] A. Thøgersen, M. Rein, E. Monakhov, J. Mayandi, and S. Diplas. Journal of
Applied Physics, 109(11):113532, 2011.
[104] J.S. Kim, P.K.H. Ho, D.S. Thomas, R.H. Friend, F. Cacialli, G.W. Bao, and
S.F.Y. Li. Chemical Physics Letters, 315(5-6):307, 1999.
[105] C. D. Wagner. Practical Surface Analysis: by Auger and X-ray Photoelectron
Spectroscopy. Wiley, Chichester, 1983.
[106] V. Christou, M. Etchells, O. Renault, P. J. Dobson, O. V. Salata, G. Beamson,
and R. G. Egdell. Journal of Applied Physics, 88(9):5180, 2000.
[107] J. Plá, M. Tamasi, R. Rizzoli, M. Losurdo, E. Centurioni, C. Summonte, and
F. Rubinelli. Thin Solid Films, 425(1-2):185, 2003.
[108] D. Graf, M. Grundner, R. Schulz, and L. Muhlhoff. Journal of Applied Physics,
68(10):5155, 1990.
[109] C.D. Wagner, D.E. Passoja, H.F. Hillery, T.G. Kinisky, H.A. Six, W.T. Jansen,
and J.A. Taylor. Journal of Vacuum Science Technology, 21(4):933, 1982.
[110] A. Thøgersen, J. H. Selj, and E. S. Marstein. Journal of The Electrochemical
Society, 159(5):D276, 2012.
[111] G.F. Cerofolini, C. Galati, and L. Renna. Surface Interface Analysis, 35(12):968,
2003.
[112] D. Sueva, S.S. Georgiev, N. Nedev, A. Toneva, and N. Chikov. A neutron detector
based on an ITO/p-Si structure. Vacuum, 58(2-3):308–314, 2000.
[113] Y Yamashita, S Yamamoto, K Mukai, J Yoshinobu, Y Harada, T Tokushima,
T Takeuchi, Y Takata, S Shin, K Akagi, and S Tsuneyuki. Phys. Rev. B,
73:045336, 2006.
Bibliography
79
80
Bibliography
List of publications
Publications in conference proceedings
PAPER I
(p. 85)
M. H. Rein, J. Mayandi, B. R. Olaisen, A. Holt, E. Monakhov. Hydrogenated ITO
and its properties. In Proceedings of the 24th European Photovoltaic Solar Energy
Conference, pages 2746 - 2749, 2009.
PAPER II
(p. 91)
M. H. Rein, J. Mayandi, B. R. Olaisen, A. Holt, E. Monakhov. Annealing effect on
ITO films sputtered with argon, oxygen and hydrogen. In Proceedings of the 24th
European Photovoltaic Solar Energy Conference, pages 2817 - 2820, 2009.
PAPER III
(p. 97)
M. H. Rein, J. Mayandi, E. Monakhov and A. Holt. Annealing of ITO films sputtered
with argon and oxygen. In Proceedings of the 26th European Photovoltaic Solar
Energy Conference, pages 2608 - 2612, 2011.
Publications in refereed journals
PAPER V
(p. 117)
Annett Thøgersen, Margrethe Rein, Edouard Monakhov, Jeyanthinath Mayandi and
Spyros Diplas. Elemental distribution and oxygen deficiency of magnetron sputtered
indium tin oxide films. JOURNAL OF APPLIED PHYSICS 109, 113532 (2011)
PAPER VI
(p. 127)
M. H. Rein, M. V. Hohmann, A. Thøgersen, J. Mayandi, A. O. Holt, A. Klein, and E.
V. Monakhov. An in situ X-ray photoelectron spectroscopy study of the initial stages
of rf magnetron sputter deposition of indium tin oxide on p-type Si substrate.
APPLIED PHYSICS LETTERS 102, 021606 (2013)
Unpublished paper
PAPER IV
(p. 105)
M. H. Rein, J. Mayandi, A. O. Holt and E. V. Monakhov. Optical, electrical,
chemical and structural properties of nitrogen doped indium tin oxide thin films.
81
82
List of publications
Appended Papers
83
84
Appended Papers
Paper I
M. H. Rein, J. Mayandi, B. R. Olaisen, A. Holt, E. Monakhov. Hydrogenated ITO
and its properties. In Proceedings of the 24th European Photovoltaic Solar Energy
Conference, pages 2746 - 2749, 2009.
85
HYDROGENATED ITO AND ITS PROPERTIES
M. H. Rein1*, J. Mayandi1, B. R. Olaisen1, A. Holt1, E. Monakhov2,
Department of Solar Energy, Institute for Energy Technology, Pb. 40, 2027 Kjeller,
2
Physical Electronics group, University of Oslo
*
Corresponding e-mail and phone: [email protected], +47 63 80 63 46
1
ABSTRACT: ITO sputtered by an industrial designed dc magnetron sputtering system has been studied as a TCO for solar
cell application. The ITO was deposited onto substrates of corning glass in argon-hydrogen gas mixtures at different
hydrogen flow rates. The sputtering was performed with a substrate temperature below 75 °C. The influence of the hydrogen
content in the sputtering plasma and the effect of annealing at 300 °C for 15 min in air after sputtering have been studied.
The highest transmittance (81.1%) and lowest resistivity (9.7 x 10-4 Ω cm) of the as-deposited films was achieved when
hydrogen was not introduced into the sputtering chamber. This sample also showed the highest transmittance (above 90%)
and lowest resistivity (2.3 x 10-4 Ω cm) after annealing.
Keywords: ITO, hydrogen, annealing.
1
INTRODUCTION
Transparent conductive oxides (TCO) have several
application areas such as flat panel displays, organic light
emitting diodes (OLED), gas sensors and electrodes for
solar cells. Indium tin oxide (ITO) is a TCO with good
conducting and transmitting properties. A resistivity of
1.9×10-4 Ω cm and 87% transparency was achieved by
M. Ando et al. [1].
Several papers, for instance [2-4], have reported
work on sputtering of ITO with a hydrogen-argon
mixture. Some have reported improved electrical
properties due to addition of a specific amount of
hydrogen in the sputtering gas. D. G. Kim et al. [3]
observed a decrease in resistivity and increase in
transmittance at a 0.8-1.7% hydrogen flow ratio,
compared with ITO sputtered without hydrogen.
An often used technique to deposit TCO is by
sputtering. In this study, ITO is deposited by dc
magnetron sputtering performed with an industrial
designed sputtering unit. The sputtering is performed in
argon-hydrogen gas mixtures at different hydrogen flow
rates, and the effects of the hydrogen in the sputtering
plasma and a subsequent 15 min annealing treatment at
300 °C in air are studied. In order to relate the electrical
and optical properties to the elemental composition of the
films, a SIMS analysis of the relative concentration of
hydrogen, indium, tin and oxygen is included.
2
EXPERIMENTAL
2.1 Preparation and deposition
An industrial designed magnetron sputtering
equipment (made by Leybold Optics) was applied for
deposition of ITO. The plasma was ignited in an argonhydrogen gas mixture with hydrogen flow rates of 0%,
3.3%, 14.7% and 25.0%. The processing chamber was
evacuated by a turbo molecular pumping system. The
base pressure of the processing chamber was in the order
10-7 mbar and the working pressure was held constant at
3.1×10-3 mbar. The power density of the target was kept
at 5.3 W/cm2, and the ramp up time of the power (presputtering) was 15 min.
Prior to the deposition,
substrates of corning glass (MENZEL microscope slides)
of size 72 mm x 26 mm x 1 mm were cleaned in DIwater in ultra sound bath for 30 min, and subsequently
dried by compressed air. The samples were mounted on
the carrier and loaded into the load lock which was
pumped down to 10-5 mbar. The carrier was transferred to
the processing chamber after 30 min of heating at 100 °C
in the load lock. The pressure in the load lock at this time
was in the order of 10-6 mbar. The indium-tin-oxide
(In2O3/SnO2) target had a weight ratio of 90/10%, and the
film was deposited by one oscillation of the carrier in
front of the target at a velocity of 0.7 m/min. The
deposition was performed without substrate heating in
the processing chamber, and the maximum substrate
temperature was measured to be below 75 °C.
After deposition, the optical and electrical properties
of the films were characterized. Half of one sample from
each of the different sputtered films was annealed in air
at 300 °C for 15 min, and the characterization was
repeated. The SIMS characterization was then done on
both the as-deposited and annealed (half) samples under
equal conditions within the same day.
2.2 Characterization
The transmittance and reflectance properties were
mapped in the wavelength range 400-1000 nm by the use
of a spectral response. Both reflectance and transmittance
were measured after calibration to air. The transmittance
of the ITO layer is calculated by adding the loss due to
the glass to the total measured transmittance of both glass
and ITO film. The resistivity was calculated from the
thickness, measured by an alpha-step 200 profilometer
(TENCOR Instruments), and the sheet resistance, which
was measured with a four-point-probe (ai alessi).
A CAMECA ims 7f instrument was used for
Secondary
Ion
Mass
Spectroscopy
(SIMS)
measurements. This characterization revealed the relative
concentration (measured as counts per second) of the
elements In, Sn, O and H in the as-deposited and
annealed ITO layers. All elements were ionized in
negative mode with a 15 keV primary ion beam of Cs+,
and the sputtering time was kept at 3768 s. The depth of
the sputtered craters was measured with a Dektak 8
surface stylus profilometer.
3
RESULTS
The optical and electrical properties of the samples are
tabulated in table I and II and also illustrated in figure 1,
2 and 3.
3.1 Optical properties
Figure 1 displays the measured transmittance of the
ITO films and shows a trend of decreasing transmittance
as the flow rate of hydrogen to argon increases. The same
trend is observed for the annealed samples, but at a much
higher transmittance.
Figure 2: Reflectance of the as-deposited and annealed
ITO layers as a function of wavelength and percentage of
hydrogen flow.
3.2 Electrical properties
The illustration of the resistivity of the samples show
that the resistivity of the ITO increases as the percentage
of hydrogen flow during the deposition increases. The
filled symbols illustrate the as-deposited films and the
open symbols illustrate the annealed films. The figure
and the tables reveal that resistivity of the films decrease,
more significant for the highly hydrogenated film, after
annealing.
Figure 1: Transmittance of the as-deposited and
annealed ITO layers as a function of wavelength and
percentage of hydrogen flow.
Table I: Optical and electrical properties of the asdeposited ITO sputtered with different H2 flow rate.
Sample
T [%]
0%
81.1
3.3%
76.4
14.7%
69.0
25.0%
61.0
R [%]
Res [×10-4 Ω cm]
t [nm]
16.3
9.7
91.0
18.5
28.5
91.5
20.5
821.9
96.3
21.5
18000
94.8
Table II: Optical and electrical properties of the
annealed ITO sputtered with different H2 flow rate.
Sample
T [%]
R [%]
Res [×10-4 Ω cm]
t [nm]
0%
91.3
9.0
2.3
96.1
3.3%
90.7
9.2
2.7
95.1
14.7%
88.4
9.4
3.7
101.4
25.0%
87.0
10.5
5.0
92.6
The reflectance of the samples (both film and
substrate) is shown in figure 2. An increasing flow of
hydrogen raises the reflectance, both for the as-deposited
and annealed case, though, the reflectance of the
annealed samples is lower and does not differ
significantly.
Figure 3: Resistivity of the ITO layers before and after
15 min annealing at 300 °C in ambient air.
The thickness of the samples before and after
annealing is tabulated in table I and II. The deposition
time was kept constant, and as the tables show, there is
no significant change (above 5%) in the thickness neither
due to the addition of hydrogen nor due to annealing.
3.3 SIMS characterization
The SIMS results reveal the relative concentration of
the different elements in the samples; hydrogen, indium,
tin and oxygen. The secondary ions are measured as
counts per second and can be converted to the
concentration if a reference sample is present. In our
case, there were no such a reference sample and the
measurements are referred to as the relative
concentration.
The sputtering rate and ionization efficiency of the
elements may vary slightly from film to film due to
different material properties, which can cause a
perceptible uncertainty in the measured relative
concentration of the matrix elements In, Sn and O.
Only the two hydrogenated samples deposited with
the highest and lowest flow of hydrogen are included in
the SIMS characteristics, in addition to the reference
sample.
Figure 4 shows the relative hydrogen concentration
in the hydrogenated films and the film sputtered with 0%
of hydrogen (from this point referred to as the reference
sample) before and after annealing. The measurements
clearly show that the hydrogen content in the samples
increases as the flow of hydrogen into the sputtering
chamber increases. It is also observed that the relative
concentration is lower in the annealed samples. The
hydrogen content in the sample sputtered with the highest
flow of hydrogen (25% of the total flow) reduced by a
factor of 4, and sample sputtered with 3.3 % hydrogen
flow (of the total flow) show a reduction by a factor of 2
compared with the content in the as-deposited sample.
The reference sample shows only small variations in
relative concentration of hydrogen due to annealing.
influence
on
the
relative
concentration
of
tin.
Figure 6: Counts of secondary 120Sn ions per second as a
function of depth.
Figure 4: Counts of secondary 1H ions per second as a
function of depth.
The relative concentration of indium in the samples
is shown in figure 5. The measurements indicate a lower
relative concentration of indium as the flow of hydrogen
during sputtering increases. Another indication is that the
relative concentration increases in the hydrogenated
films, but decreases in the reference sample after
annealing.
Figure 7 shows the measured relative concentration
of oxygen in the films. The hydrogenated samples may
indicate a lower relative concentration as the hydrogen
flow increases (the graphs of the as-deposited and
annealed samples coincide). In case of the reference
sample, it is measured a lower relative concentration of
oxygen after annealing.
Figure 7: Counts of secondary
function of depth.
4
Figure 5: Counts of secondary
function of depth.
115
In ions per second as a
The relative concentration of tin in the samples is
illustrated in figure 6. Neither the addition of hydrogen
nor the annealing treatment seems to have significant
18
O ions per second as a
DISCUSSION
As observed in the tables I and II and figure 1-3,
there is a strong correlation between the resistivity and
the optical properties in this study. The lowest measured
resistivity is found for the sample with highest
transmittance and lowest reflectance, and vise versa. This
is observed for both as-deposited and annealed samples.
The annealing has significant effect on the ITO as it
improves both the optical properties and the resistivity.
There is also a strong correlation between the amount
of hydrogen in the films and the electrical and optical
properties. The higher the flow of hydrogen the higher is
the resistivity and reflectance and lower is the
transmittance.
From the SIMS results, the improvement of the
properties of the hydrogenated samples by annealing may
be attributed by the reduction in the relative hydrogen
concentration.
Sputtering with a hydrogen and argon mixture is
previously reported by K. Zhang et al. [2], but at a
substrate temperature of 300°C. By SIMS measurements,
they also revealed a lower relative oxygen concentration
when sputtering with a hydrogen-argon mixture than
with pure argon. It is claimed that hydrogen with high
energy can remove weakly bound oxygen in the growing
film and that the consequence is a higher concentration
of oxygen vacancies. As oxygen vacancies are said to be
a source for charged carriers, the increase in oxygen
vacancies will lead to a lower resistivity, which is found
for a certain hydrogen partial pressure in [2].
An increased transmittance and diminished resistivity
at hydrogen flow rates below 3.3% was reported by D. G.
Kim et al. [3]. The improvements were addressed to a
hydroxyl formation when adding hydrogen during
sputtering. This formation was attributed to a removal of
excess oxygen and increment in the concentration of
oxygen vacancies.
Our SIMS measurements may indicate a lower
relative concentration of oxygen for the films sputtered
with 3.3% and 25.0 % hydrogen flow, than for the
reference sample. The higher the flow rate of hydrogen is
the lower is the relative concentration of oxygen.
Nevertheless, these films have a higher resistivity and
their lower oxygen content seems to not contribute to the
conducting property.
The different results in our study may be attributed to
the high flow of hydrogen during sputtering and/or the
structural phase of the films. K. Zhang et al. [2] found a
crystalline phase for all hydrogen partial pressures up to
1.6 ×10-5 Torr at a substrate temperatures of 300 °C, and
D.G. Kim et al. [3] found a crystalline phase for the films
deposited with a hydrogen flow rate below 5% at a
substrate temperature below 70 °C.
The electrical and optical properties of the
hydrogenated films in our study may indicate that the
incorporation of hydrogen into ITO at low temperature
may result in an amorphous phase, and that hydrogen is
annealed out of the film and the film crystallizes during
the annealing, due to the high temperature.
5
CONCLUSION
There is a strong correlation between the electrical
and optical properties, which are highly affected by the
content of hydrogen in the films. The resistivity and
reflectance increases and the transmittance decreases as
the flow of hydrogen during sputtering increases.
Annealing at 300 °C for 15 min in air improved the
resistivity and optical properties for both hydrogenated
and not hydrogenated films. It also lowered the relative
concentration of hydrogen in the hydrogenated samples.
The SIMS analysis shows no notable change in relative
concentration of indium and tin, but may indicate a lower
relative concentration of oxygen in the annealed
reference sample than in the as-deposited sample.
6
ACKNOWLEDGEMENT
Thanks to Lasse Vines at the Center for Materials
Science and Nanotechnology (University of Oslo) who
performed the SIMS measurements and assisted us with
helpful information.
7
REFERENCES
[1] M. Ando et al, Journal of Applied Physics 93
(2003), 1023.
[2] K. Zhang et al., Journal of Applied Physics, 86 -2
(1999) 974-980
[3] D.G. Kim et al., Thin Solid Films, 515 (2007) 69496952
[4] R. Das et al., Applied Surface Science, 253 (2007)
6069-6073
Paper II
M. H. Rein, J. Mayandi, B. R. Olaisen, A. Holt, E. Monakhov. Annealing effect on
ITO films sputtered with argon, oxygen and hydrogen. In Proceedings of the 24th
European Photovoltaic Solar Energy Conference, pages 2817 - 2820, 2009.
91
ANNEALING EFFECT ON ITO FILMS SPUTTERED WITH ARGON, OXYGEN AND HYDROGEN
M. H. Rein1*, J. Mayandi1, B. R. Olaisen1, A. Holt1, E. Monakhov2,
Department of Solar Energy, Institute for Energy Technology, Postboks 40, 2027 Kjeller,
2
Physical Electronics group, University of Oslo
*
Corresponding e-mail and phone: [email protected], +47 63 80 63 46
1
ABSTRACT: Sputtered indium tin oxide (ITO) has been studied as a TCO for solar cell application. The deposition was
performed with an industrial designed sputtering unit, and corning glass was used as substrate. The effect of annealing on the
electrical and optical properties of ITO films was studied. The films were sputtered in three different ambient; Ar, a mixture
of Ar and O2 and a mixture of Ar and H2. The annealing was performed in air at 300 °C for 15 min. The highest
transmittance (88%) and lowest resistivity of the as-deposited films was achieved when the ITO was sputtered with a 3.3%
oxygen flow (of total flow) added to the working gas (Ar). After annealing, the highest transmittance (91.3%) and lowest
resistivity (2.3 x 10-4 Ω cm) was achieved when the ITO was sputtered without addition of oxygen or hydrogen gas to the
working gas. SIMS measurements were performed on both as-deposited and annealed samples in order to disclose the
elemental composition.
Keywords: ITO, annealing, hydrogen, oxygen
1
INTRODUCTION
The increased area of application of thin film
materials that transmit light and conduct properly has
increased the interest for and research on TCO materials.
Common applications are transparent electrodes in liquid
crystal displays, in thin film gas sensors, as anodes in
organic light emitting diodes (OLED) and as transparent
electrodes in photovoltaics.
Indium-tin-oxide (ITO) is a commonly used TCO
material due to its low resistivity (1.9×10-4 Ω cm [1]) and
high transparency (87% [1]) in the visible range. A lot of
research work is done to improve the optical and
electrical properties by optimizing the TCO deposition
method. A commonly applied method is dc or rf
magnetron sputtering. Changing the parameter sets used
during sputtering can affect the properties of the films
[2].
One parameter that can contribute to an enhanced
transmittance and lower resistivity is the addition of
reactive gases to the plasma. Such an improvement is
reported by M Rottmann et al. [3], who achieved an
increase in the donor level of Sn and O vacancies when
applying reactive sputtering with H2O vapor as the
reactive species.
Annealing temperature is another parameter that can
influence the properties of the ITO film. One study of the
annealing effect is reported by Y. Hu et al. [4], who
improved both crystallinity and electrical and optical
properties by increasing the annealing temperature. The
annealing was done in both air and vacuum, but the
results showed that the improvements were independent
of the annealing ambient.
A study of the effects of annealing on electrical and
optical properties of ITO films sputtered with addition of
oxygen or hydrogen is presented in this paper. The
deposition is performed with an industrial designed
sputtering unit and the ITO films are studied with the aim
of solar cell application. A SIMS characterization of the
films is performed in order to uncover a relation between
the optical and electrical properties and the elemental
composition.
2
EXPERIMENTAL
2.1 Preparation and deposition
ITO films were deposited on corning glass substrates
(MENZEL microscope slides) by an industrial designed
dc magnetron sputtering equipment made by Leybold
Optics. Before deposition, all glass samples (of size 72
mm x 26 mm x 1 mm) were cleaned in an ultra sound
bath with DI water for 30 min, and dried with
compressed air. They were subsequently placed on the
sputtering carrier and loaded into the load lock and
heated at 100 °C for 30 min. The pressure in the load
lock was 10-6 mbar when the carrier was transferred from
the load lock to the process chamber. The base pressure
in the process chamber was 10-7 mbar, and the working
pressure was always kept at 3.1 x 10-3 mbar. The flow of
Ar was kept at 180 sccm, and the flow of oxygen or
hydrogen was 6 sccm (3.3% of the total flow), when
used. The ramp up time for the target was 15 min and the
power density used was 5.3 W/cm2. The weight ratio of
the indium-tin-oxide (In2O3/SnO2) target was 90/10%.
The carrier velocity was 0.7 m/min and it was performed
only one oscillation of the carrier in front of the target.
All sputtering experiments were performed without
substrate heating in the process chamber, and the ITO
layers were sputtered onto the substrates in three
different ambient gases. One was done in a mixture of O2
and Ar (oxygenated ITO), one in a mixture of H2 and Ar
(hydrogenated ITO) and one in Ar (reference sample).
One of each ITO sample was cut in two pieces after
optical and electrical characterization of the as-deposited
layers. One half of each sample was then annealed on a
hot plate at 300 °C for 15 min in ambient air, and the
electrical and optical characterization was performed
once more. The characterization with the SIMS was then
performed on both the as-deposited and annealed part of
the sample under equal conditions and within one day.
2.2 Characterization
The transmittance and reflectance of the samples
were measured before and after annealing by the use of a
spectral response in the wavelength range 400-1000 nm.
The transmittance of the film was calculated by adding
the loss in transmittance due to the glass substrate to the
total transmittance. The sheet resistance was measured by
a four-point-probe (ai alessi) and the resistivity was
calculated when the thickness had been measured. The
thickness was measured by the use of an alpha-step 200
profilometer (TENCOR Instruments).
The Secondary Ion Mass Spectroscopy (SIMS)
measurements were performed by a CAMECA ims 7f
instrument and revealed the intensity (counts per second)
of the elements In, Sn, O and H in the as-deposited and
annealed ITO layers. All elements were measured in
negative mode with a 15 keV primary ion beam of Cs+.
The sputtering time was kept constant at 3768 s, and the
depth of the SIMS craters was measured with a Dektak 8
surface stylus profilometer.
3
of the reference sample and the hydrogenated sample,
where the reflectance has decreased by approximately
50% and 45%, respectively. In the case of the
oxygenated sample, about 15% reduction in reflectance is
observed. This is also the sample having the highest
reflectance after annealing. The lowest reflectance is
found in the case of the annealed ITO sputtered with no
additional gases added to the working gas.
RESULTS
3.1 Optical properties
The transmittance of the ITO layers before and after
annealing is shown in figure 1 and the average
transmittance is displayed in table I and II. The figure
shows that the highest transmittance of the as-deposited
films is achieved when the sputtering was performed in
an oxygen-argon gas mixture. The lowest transmittance
is seen for the case of deposition in a hydrogen-argon gas
mixture. After annealing the transmittance is shifted
toward higher values and the average transmittance is
now above 90% for all samples. In case of oxygen
incorporated into the ITO, only a small increment (about
3 %) in transmittance due to the annealing is seen, when
compared with the increment in transmittance of about
19% and 13 % for the hydrogenated and the reference
sample, respectively.
Figure 2: Reflectance as a function of wavelength for assputtered and annealed films.
3.2 Electrical properties
The resistivity of the samples is illustrated in figure 3
and tabulated in table I and II. In the case of as-deposited
ITO (the filled symbols), the highest resistivity is seen
for the hydrogenated sample, whereas the ITO sputtered
with O2 exhibit the lowest resistivity. The annealing
improves the resistivity of all samples. The oxygenated
sample has the highest resistivity and the reference
sample has the lowest resistivity after annealing.
Figure 1: Transmittance as a function of wavelength for
as-sputtered and annealed films.
Table I: Average transmittance, reflectance, resistivity
and thickness of as-deposited samples
Average transmittance [%]
Average reflectance [%]
Resistivity [×10-4 Ω cm]
Thickness [nm]
3.3% H2
76.4
18.5
28.5
92
3.3% O2 Reference
88.1
81.1
17.8
16.3
7.9
9.7
95
91
Table II: Average transmittance, reflectance, resistivity
and thickness of annealed samples
Average transmittance [%]
Average reflectance [%]
Resistivity [×10-4 Ω cm]
Thickness [nm]
3.3% H2
90.7
9.2
2.7
95
3.3% O2 Reference
90.9
91.3
15.1
9.0
5.8
2.3
91
96
In figure 2 is the reflectance of both glass and ITO
films of the as-deposited and annealed samples shown.
The average reflectance is tabulated in table I and II. The
most significant effect of the annealing is seen in the case
Figure 3: Resistivity of as-sputtered and annealed films.
The thickness of the as-deposited and annealed
samples is displayed in the tables I and II. The deposition
time was kept constant, and the results indicate that the
thickness variation is small (less than 5 %) between the
different sputtered films. The tables also reveal that the
annealing has no significant effect on the thickness.
3.3 SIMS measurements
The SIMS measurements disclosed the relative
concentration, counting secondary ions per second of the
different elements hydrogen, indium, tin and oxygen in
the samples. The counts per second can be transformed
into the concentration of elements in the samples, if a
calibration sample is present. In our case, there were no
calibration sample and the calculation of the
concentration is not performed. The measurements will
be referred to as the relative concentration.
The sputtering rate and the ionization efficiency may
vary slightly from sample to sample due to different
material properties. This can cause a perceptible
uncertainty in the measurements of the counts of
secondary ions per second of the matrix elements In, Sn
and O.
The measured relative concentration of indium in the
samples is shown in figure 4. The relative concentration
of indium does neither alter significantly due to the
different sputtering conditions nor the annealing.
Figure 6: Counts of secondary
function of depth.
18
O ions per second as a
Figure 7 shows the measured relative concentration
of hydrogen in the samples. The highest relative
concentration is measured in the sample sputtered with
addition of H2 and the lowest relative concentration in
the oxygenated sample. The heat treatment at 300 °C for
15 min at a hotplate in ambient air lowers the relative
concentration of hydrogen in the hydrogenated sample,
but raises the relative concentration of hydrogen in the
oxygenated sample. The relative concentration in the
reference sample is not notably changed due to
annealing.
Figure 4: Counts of secondary
function of depth.
115
In ions per second as a
Figure 5 illustrates the measured relative
concentration of tin in the samples. The films show an
approximately equal relative concentration, which is not
heavily influenced by the annealing treatment.
Figure 7: Counts of secondary 1H ions per second as a
function of depth.
4
Figure 5: Counts of secondary 120Sn ions per second as a
function of depth.
The measured relative concentration of oxygen in
the samples seems to not change significant due to the
different sputtering ambient, as can be seen in figure 6. In
case of the films sputtered with oxygen or hydrogen
added to the working gas the annealing treatment has not
affected the relative concentration. For the sample
sputtered without addition of O2 or H2, a slight reduction
in the relative concentration of oxygen after annealing is
observed.
DISCUSSION
Figure 1, 2 and 3 and table I and II reveal that there is
a strong correlation between the electrical and optical
properties of the films. A high transmittance and low
reflectance imply a low resistivity of the film. This is
observed for the as-deposited as well as for the annealed
samples.
The main source for charge carriers in ITO is
claimed to be both tin dopants and charged/ionized
oxygen vacancies. Our results show little variation in the
relative concentration of tin both in as-deposited and
annealed case. This may indicate that this parameter do
not have a notable impact on the resistivity nor the
optical properties for these samples.
The slightly drop in the relative concentration of
oxygen in the reference sample in figure 7 may indicate a
creation of oxygen vacancies during annealing. Though,
there are only small variations in electrical and optical
properties between the annealed samples, and the drop in
relative concentration of oxygen in the reference sample
can not completely explain the improvements in these
properties. This is also found for annealing of sol-gel
deposited ITO films by M. J. Alam et al. [5].
In case of as-deposited layers a high resistivity and
reflectance and a low transmittance is observed for the
hydrogenated sample. It is also observed that these
properties are significantly improved after annealing
when the SIMS characterization shows a reduction of a
factor 2 in relative concentration of hydrogen. This may
indicate that the hydrogenation has a large influence on
the optical and electrical properties of the ITO.
Annealing experiments done by Y. Hu et al. [4]
showed an increment in resistivity when annealed in air
at 300 °C and higher temperatures for ITO sputtered with
no other gases than Ar. That is contradictory to the
results shown here, where annealing at 300 °C in air led
to better optical and electrical properties for all the
different ITO layers. A low resistivity is found by M. J.
Alam et al. [4] for sol-gel deposition of ITO with 10at.%
Sn when annealed at 500°C independent of ambient
(oxygen, nitrogen or air).
The measurements of the optical and electrical
properties of the oxygenated sample are less affected by
the heat treatment, than the other two samples. This is
also shown in the SIMS measurements, where the only
observed change due to annealing is in the relative
concentration of hydrogen. This increase in hydrogen
may be due to in-diffusion of hydrogen from the H2O
vapour in the air.
The results show a strong correlation between the
optical and electrical properties of the ITO films. The
lower the transmittance the higher is the resistivity. As
this is not clearly explained by the elemental study by
SIMS measurements, this correlation may be addressed
to other film properties as structure and crystallinity.
5
CONCLUSION
The effects of annealing and presence of hydrogen
and oxygen in the working gas during sputtering are
presented. The results show a strong correlation between
the optical and electrical properties of the films,
independent of sputtering ambient. A heat treatment at
300 °C improves both the optical and electrical properties
of all the different sputtered ITO layers.
The oxygenated sample shows only small influences
of the annealing on the electrical and optical properties.
The only significant change measured is the relative
concentration of hydrogen.
The relative concentration of hydrogen in the
hydrogenated sample is reduced after 15 min annealing at
300°C. The measurements of the secondary 18O ions in
the reference sample may indicate a reduction of oxygen
due to the annealing treatment.
6
ACKNOWLEDGEMENT
Thanks to Lasse Vines at the Center for Materials
Science and Nanotechnology (University of Oslo) who
performed the SIMS measurements and assisted us with
helpful information.
7
REFERENCES
[1] M. Ando, Journal of Applied Physics 93
(2003), 1023.
[2] S. Bhagwat, R. P. Howson, Surface and coating
Technolog 111 (1999) 163-171
[3] M. Rottmann et al., Journal of Applied Physics 28
(1995), 1448-1453.
[4] Y. Hu et al., Vacuum 75 (2004) 183-188.
[5] M. J. Alam and D.C. Cameron, Thin Solid Films 420421 (2002) 76-82
Paper III
M. H. Rein, J. Mayandi, E. Monakhov and A. Holt. Annealing of ITO films sputtered
with argon and oxygen. In Proceedings of the 26th European Photovoltaic Solar
Energy Conference, pages 2608 - 2612, 2011.
97
ANNEALING OF ITO FILMS SPUTTERED WITH ARGON AND OXYGEN
M. H. Rein1, J. Mayandi2, E. Monakhov1,3 and A. Holt1
1 Department of Solar Energy, Institute for Energy Technology, Kjeller, Norway.
2 Department of Materials Science, School of Chemistry, Madurai Kamaraj University, Tamil Nadu, Madurai,
India.
3 Physical Electronics group, University of Oslo, Oslo, Norway.
ABSTRACT: Dc magnetron sputter deposited indium tin oxide (ITO) films for silicon based hetero-junction solar
cell application are studied. The films show suitable properties with transmittance as high as ~94% and resistivity as
low as 2.2 x 10-4 Ω cm. Oxygen is used as reactive gas and the sputtering gas is argon. The depositions are done at
room temperature, and subsequent annealing is carried out on a hot-plate at different temperatures (250 °C, 300°C
and 350°C) in air atmosphere. The O2/(Ar+O2) flow vol% into the sputtering chamber is varied from 0 to 25. An
improvement in optical and electrical properties is demonstrated when the films are annealed. The structure of the asdeposited films is shown to be dependent on the oxygen present during sputtering. When no oxygen is added, the
films exhibit an amorphous structure. When oxygen flow is introduced, the XRD patterns indicate crystalline films.
Besides, high flows of oxygen results in higher resistivity of the films. The resistivity can be considerably reduced by
a post-deposition heat treatment.
Keywords: ITO, annealing, sputtering
1
INTRODUCTION
Transparent conducting oxides (TCO) are materials
which have the property of transmitting light in the
visible and infrared range of the electromagnetic
spectrum, even though they are highly conducting. This
is due to a wide band gap and a high doping level.
Among the TCOs, indium tin oxide (ITO) is one of the
most promising material due to its low resistivity (~10-5
Ω cm) and high transparency (~90%) [1]. Due to these
excellent material properties, the application areas of ITO
are several. Some of these are gas sensors [2], light
emitting diodes [3], flat panel displays [4] and
photovoltaic cells [5].
Production of such devices demands different kinds
of deposition techniques. It is known that ITO is
deposited by methods such as the sol-gel process [6],
reactive thermal deposition technique [7] pulsed laser
deposition [8], electron beam evaporation [9, 10], r.f.
magnetron sputtering [11] and DC magnetron sputtering
[12]. Due to its capability of large scale production and
high throughput of uniform films of sufficient quality, dc
magnetron sputtering is one of the most efficient
deposition methods.
In order to produce films with sufficient properties,
the processing parameters are of importance. Parameters
such as substrate temperature, present gases and the
partial pressure of these during deposition, and postdeposition treatment are vital for determining whether the
film will obtain the properties requested or not [11, 12].
In order to lower the cost of photovoltaic cells, the
production demands high throughput and low cost
processes. One main parameter for these requirements is
low temperature production. The motivation for our
experiments was to optimize parameters for room
temperature deposition of ITO thin films by an industrial
designed dc magnetron sputter. The present study shows
the effect of various flow ratios of oxygen and argon
). In addition, the impact of annealing
( ⁄
treatment at different temperature on the films is studied
in terms of electrical, optical and structural properties.
This is done with the aim to make high quality ITO thin
films for silicon based hetero-junction solar cell
application.
2
EXPERIMENTAL
2.1 Sample preparation
Thin films of indium tin oxide were deposited at
room temperature by an industrial designed dc magnetron
sputter on glass substrates of size 72 mm x 26 mm x 1
mm. Prior deposition, all samples were cleaned 30 min
in ultra sound bath filled with de-ionized water and
subsequently dried with compressed air. A pre-heating of
the samples before deposition was done in load lock at
100 °C for 30 min, in order to evaporate water residuals.
Temperature measurements showed that the substrate
temperature reached during this heating was maximum
45 °C, i.e. we assume that the substrate temperature
during deposition was lower than 45 °C. The base
pressure in the sputtering chamber was in the order of
10-7 mbar and the working pressure was kept at 3.1 x 10-3
mbar for all experiments. The flow of argon was constant
at 180 sccm and the flow of oxygen was varied from 0 to
60 sccm. The different flow rates, in percentage oxygen
of total flow, were 0, 0.3, 1.6, 3.2, 6.3, 9.1, 14.3 and 25.0.
This will later be denoted as O2 flow with percentage as
unity. The weight ratio of the target (In2O3/SnO2) was
90/10%. The power density was held constant at 5.3
W/cm2 and the target to sample distance was 80 mm. The
sample holder was oscillated once in front of the target.
The oscillation velocity of the sample holder was kept at
0.7 m/min. Subsequent heat treatment was performed in
air on a hot-plate. The temperatures examined were 250
°C, 300 °C and 350 °C and the duration of the annealing
was 15 min.
2.2 Characterization
The transmittance of the films was measured as a
function of wavelength in the range 400-1000 nm by the
use of a spectral response. Film thickness was measured
using an alpha-step 200 profilometer (TENCOR
Instruments). Resistivity measurements were done by the
use of a four-point-probe set-up (ai-alessi). Structural
characterization was done by X-ray diffraction (XRD)
with Cu Kα (λ = 1.5418 Å) radiation using a BrukerSiemens D5000 diffractometer. Electronic properties
were examined by X-ray photoelectron spectroscopy
(XPS). The instrument used was a KRATOS AXIS
ULTRADLD. A monochromatic Al Kα radiation with
hν=1486.66eV was used at
a 15 kV and 10
0 mA. In orderr to
achieve coontact betweenn the film and the spectrometter,
all samplees were mounted to the samp
ple holder withh a
metallic cclips, i.e. the Fermi
F
levels off the film and tthe
spectromeeter are aligned.
3
ULTS
RESU
3.1 Opticaal properties
Figuree 1 shows the average
a
transmittance, taken ovver
the measuured wavelengthh range, versuss oxygen flow for
both as-deeposited and annealed
a
samp
ples. The loss in
transmittannce due to the glass substraate (a bare gl ass
substrate was measuredd) is added to
t the measurred
film/glass substrate transsmittance. The figure shows th
that
the transm
mittance is higghly dependent on the oxyggen
present duuring depositionn. It is shown that only a sm
mall
amount of added oxxygen (1.6%)) improves tthe
transmittannce of as-depoosited films from
m 80% (when no
oxygen is added) to 92%..
Indepeendent on the applied anneaaling temperatuure,
there is ann overall increm
ment in transmitttance of the film
ms,
except forr some flow ratte cases. The figure
fi
indicates an
optimized annealing tem
mperature at 300
0 °C. The highhest
transmittannce (~94%) of annealed films is achieved whhen
the films aare deposited with
w 1.6% oxygeen.
Figure 1. Transmittancee of as-depositted and anneaaled
ITO films as a function of
o oxygen flow.
3.2 Electriical properties
Figuree 2 shows the measured
m
resistiivity as a functiion
of oxygenn flow. The figuure shows that increasing
i
oxyggen
I is also observved
flow raises the resistivityy of the films. It
in the figuure that annealinng has a large impact on the IT
TO
films, espeecially on film
ms deposited wiith high O2 flow
ws.
In the cases of annealeed ITO deposiited with oxyggen
3
the figuree illustrate that tthe
flows highher than about 3.2%,
resistivity decreases to abbout 10-3 Ω cm, regardless of tthe
s
This iss a five orders of
annealing temperatures studied.
magnitudee decrement in resistivity
r
for th
he films depositted
with 25% O2 flow. For films deposited
d with lower thhan
3.2% oxyygen, there arre only small changes in tthe
resistivity due to annealinng and the anneealing temperatuure
is not siggnificant. The lowest resisttivity of the asdeposited samples was 3.4
3 x 10-4 Ω cm.
c This was tthe
resistivity of the film depposited with 1.6
6% oxygen. Whhen
this sampple was anneaaled at 250 °C the resistivvity
decreased to 2.2 x 10-4 Ω cm. This resistivity was aalso
measured on the sample deposited with
w
0.3% oxyggen
flow and aannealed at 300 °C.
nealed ITO
Fiigure 2. Resisttivity of as-depposited and ann
fillms as a function
n of O2 flow duuring deposition.
operties
3.3 Structural pro
The growth raate of the ITO ffilm is shown to
o be linearly
deependent on thee O2 flow (figurre 3). This impllies that the
th
hicknesses of thee deposited film
ms decrease with
h increasing
ox
xygen flow as the oscillationn velocity of the sample
ho
older was kept constant. Hence
ce, the thicknesss may have
an
n impact on the transmittancee shown in fig
gure 1. The
th
hickness varied between
b
30 and 90 nm.
Fiigure 3. Grow
wth rate of spuutter deposited ITO films
veersus oxygen flo
ow during sputttering.
Figure 4 show
ws the XRD pat
attern of as-depo
osited films
an
nd films anneaaled at 300 °C
C. The figure shows the
paattern for filmss deposited wiith 0%, 3.2% and 25 %
ox
xygen flow. The film depositeed without oxyg
gen present
du
uring sputtering
g shows only a small indication of a peak
at 2θ=30°, sugg
gesting that thee film has a dominating
morphous stru
ucture. For rreactive sputteering with
am
ox
xygen, the as-d
deposited filmss show a single oriented
crrystalline structture, independeent on oxygen flow. The
peeak shown att 2θ=30° corrresponds to the (222)
diffraction planee of In2O3, as reported in th
he literature
[13, 14]. In casee of 3.2 % oxyygen, small ind
dications of
5
otther peaks are seen at 2θ aboutt 21°, 41° and 51°.
When the films
f
are anneealed, the film
m deposited
without supply of oxygen shoows a randomlly oriented
crrystalline structture. The film
ms sputter depo
osited with
ox
xygen show an increased intennsity of the (222) intensity
peeak after annealling (clearly shhown in prelimiinary study,
no
ot presented).
Fiigure 5. Valence band spectrra of as-depositted (a) and
an
nnealed (at 300 °C) (b) ITO film
lms versus oxyg
gen flow.
4
Figure 4. XRD pattern of
o as-deposited
d (a) and anneal
aled
(at 300 °C
C) (b) ITO.
3.4 Electroonic properties
The vvalence band sppectra of the as-deposited
a
fillms
and anneaaled films at 3000 °C, are illusttrated in figuree 5.
The figure shows that the slope of the
t valence baand
ards
decreases and the valennce band maxima shift towar
n flow increasses.
lower binnding energies, as the oxygen
n the valence baand
After anneealing, the diffeerence between
spectra is aabsent.
DISCUSSION
4.1 Optical propeerties
The transmittance of the as--deposited sam
mples shows
a dependency off the O2 flow. W
When the transsmittance is
nalysed taking film
f
thickness iinto account, itt is obvious
an
th
hat transmittancce of films depposited with sm
mall flows,
i.ee. lower than 3..2% O2, is not ddependent on th
hickness as
th
here are small variations in thicckness of thesee films. The
difference in tran
nsmittance betw
ween these film
ms must be
related to other properties likke film structure. This is
in
ndicated by the XRD patternss. The XRD reesults show
th
hat the structuree of the film depposited with 25
5% O2 flow
is similar to the structure
s
of the film deposited with 3.2%.
Hence, the measured transmitttance of filmss deposited
with higher O2 flows is more likely influenced by the
fillm thickness ass the thickness is shown to bee dependent
off O2 flow.
nce of the
Heat treatmeent improves the transmittan
fillms. In our study, the as-depoosited film spu
uttered with
pu
ure argon is sh
hown to be am
morphous. From
m the XRD
results shown in figure 4, it is obvious that an
nnealing in
airr at 300 °C cry
ystallizes this fi
film. We thereffore assume
th
hat the improveement in transm
mittance of thiis film is a
co
onsequence of
o
incorporatition of oxy
ygen and
crrystallization du
uring annealing [9, 15].
Other films which showeed crystallinity when asdeeposited, also exhibit a hig
igher transmitttance after
an
nnealing. We suggest that thhis improvemeent can be
ad
ddressed to incrreased grain siize and a lowerr degree of
deefects after annealing. Increasee in grain size is
i indicated
by
y preliminary examination
e
off XRD peak width before
an
nd after anneaaling (results nnot presented).. Increased
traansmittance due to annealing is previously attributed
a
to
en
nlargement of crrystallite size [116].
operties
4.2 Electrical pro
It was shown
n that the resistiivity of the ITO
O thin films
was affected by the content oof oxygen pressent during
deeposition and presumably
p
duuring post-depo
osition heat
treeatment as welll. The film dep
eposited with 0% O2 flow
ex
xperienced a small
s
improvem
ement in resisttivity after
an
nnealing. We atttribute this to the transformaation of the
am
morphous film to a polycryystalline film. The films
deeposited with 1.6% and 3.22% O2 flow exhibit
e
low
resistivity as as--deposited andd heat treatmen
nt does not
afffect this resistiv
vity significantlly.
Films deposiited with highher O2 flows, i.e.
i >3.2%,
sh
how increased resistivity witth increased O2 flow. A
similar dependency is previously shown by others [15,
17]. González et al. [17] reported that this was due to
changes in number of the electrons as they assumed that
the mobility was independent on the oxygen present. At
high oxygen pressure they found that the electron
⁄
.This was
concentration was proportional to
consistent with the model of Frank and Köstlin [18] who
presented a theory of interstitial defects consisting of one
oxygen and two tin ions. The dissociation of these defects
would then give two free electrons according to the
equation [18]
∙
∙
2
⇆2
2
1⁄2
(1)
Thus, the decrease in resistivity due to annealing of
our films deposited with high flows of oxygen is
therefore attributed to the dissociation of such interstitial
defects and outdiffusion of the non-stoichiometric extra
oxygen.
4.3 Structural properties
As shown in figure 3, the thickness of the samples
decreases as the O2 flow increases. The most significant
changes in thickness are observed for flows higher than
3.2%. It is known from the literature that above a certain
percentage of oxygen present during sputtering, the
deposited film will not be able to absorb significant
amounts of oxygen as a fully stoichiometric structure is
formed. At this critical point the excess oxygen will start
to create an insulating surface on the sputtering target,
and hence, the sputtering yield is significantly reduced
[19]. The lower sputtering yield causes thinner ITO films
as the flow increases. The thickness variations can
explain variations in optical properties, as thinner films
exhibit higher transmittance.
The XRD results show that the structural properties
of the as-deposited films are dependent on the content of
oxygen present during sputtering at room temperature.
Without oxygen added during sputtering the as-deposited
film we achieved was amorphous. When the deposition
was done with oxygen the XRD patterns showed that the
films were crystalline.
Amorphous films are usually achieved when
deposition is performed at low substrate temperatures.
Sun et al. [8] reported amorphous films deposited by
excimer pulsed laser deposition at substrate temperatures
below 150 °C. This was found to be due to initial 3D
growth with nucleation separation distance dependent on
substrate temperature. Ma et al. [20] achieved crystalline
films at deposition temperature as low as 50 °C with a
facing target sputtering system. Nevertheless, other
deposition parameters than substrate temperature may
also affect the film structure. Song et al. [21] reported
that the morphology of ITO films deposited by DC
magnetron sputtering at room temperature was dependent
on the total pressure of the sputtering gas. They assigned
this dependency to the number of collisions the sputtered
particles were exposed to. The higher pressure the more
collisions and the lower kinetic energy of the sputtered
particles reaching the sample surface. Other deposition
parameters such as target to substrate distance may also
affect the surface mobility of the deposited particles and
hence, the structure of the sputter deposited films [22,
23]. We assume that the deposition parameters used in
our system are optimal for achieving deposition particles
with sufficient surface mobility and hence, crystalline
films.
In the case of the amorphous as-deposited film we
assume that the lack of oxygen present during deposition
induce an amorphous structure as the film becomes nonstoichiometric without the required amount of oxygen for
forming a crystalline film [21]. We also assume that
when the amorphous film is annealed, enough energy is
supplied for incorporation of the oxygen present in the
annealing ambient and a crystalline structure is formed.
4.4 Electronic properties
It is evident from figure 5 that the oxygen content
during deposition affects the valence band maxima.
Higher oxygen content induces lower binding energy of
the valence bands, i.e. the Fermi energy level lies closer
to the valence band edge. A lower Fermi level is assumed
to be a consequence of lower doping. This coincides with
the resistivity measurements, which showed that
increased oxygen flow led to increased resistivity. The
lower doping level is presumably an effect of the
formation of interstitial defects previously discussed [24].
As the films are annealed, the shift in binding energy
disappears and the resistivity drops. This can be
attributed to the dissociation of the defects caused by the
excess oxygen.
5
CONCLUSION
Both amorphous and crystalline ITO films were
deposited at room temperature by dc magnetron
sputtering. The film structure was affected by the oxygen
present during deposition. Adding oxygen to the
sputtering gas induced crystalline films. By adding a
small amount of oxygen, the films obtained high
transparency and conductivity. Large oxygen flows
increased the resistivity, which is attributed to formation
of oxygen-related interstitial defects. Annealing of these
films caused a decrease in resistivity, presumably due to
dissociation of the defects.
ACKNOWLEDGEMENT
Thanks to S. Diplas who carried out the XPS
measurements. The work is funded by REC Solar and the
RENERGI program by the Research Council in Norway.
REFERENCES
[1]
[2]
[3]
[4]
S. Ray, R. Banerjee, N. Basu, A. K. Batabyal,
and A. K. Barua, Journal of Applied Physics 54
(1983) 3497.
K. S. Yoo, S. H. Park, and J. H. Kang, Sensors
and Actuators B: Chemical 108 (2005) 159.
C. Suess, F. P. Wenzl, G. Jakopic, M. Wuchse,
S. Muellegger, N. Koch, A. Haase, K.
Lamprecht, M. Schatzmayr, C. Mitterbauer, F.
Hofer, and G. Leising, Surface Science 507510 (2002) 473.
U. Betz, M. Kharrazi Olsson, J. Marthy, M. F.
Escolá, and F. Atamny, Surface and Coatings
Technology 200 (2006) 5751.
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
F. C. Krebs, H. Spanggard, T. Kjær, M.
Biancardo, and J. Alstrup, Materials Science
and Engineering: B 138 (2007) 106.
M. J. Alam and D. C. Cameron, Thin Solid
Films 377-378 (2000) 455.
P. Thilakan and J. Kumar, Vacuum 48 (1997)
463.
X. W. Sun, H. C. Huang, and H. S. Kwok,
Appl. Phys. Lett. 68 (1996) 2663.
N. Mori, S. Ooki, N. Masubuchi, A. Tanaka,
M. Kogoma, and T. Ito, Thin Solid Films 411
(2002) 6.
R. X. Wang and et al., Journal of Physics D:
Applied Physics 38 (2005) 2000.
R. Das, K. Adhikary, and S. Ray, Applied
Surface Science 253 (2007) 6068.
S. Bhagwat and R. P. Howson, Surface and
Coatings Technology 111 (1999) 163.
H. Kobayashi, T. Ishida, K. Nakamura, Y.
Nakato, and H. Tsubomura, Journal of Applied
Physics 72 (1992) 5288.
Y.-C. Park, Y.-S. Kim, H.-K. Seo, S. G. Ansari,
and H.-S. Shin, Surface and Coatings
Technology 161 (2002) 62.
C. Guillén and J. Herrero, Vacuum 80 (2006)
615.
I. Hamberg and C. G. Granqvist, J. Appl. Phys.
59 (1986) 2950.
G. B. González, T. O. Mason, J. P. Quintana,
O. Warschkow, D. E. Ellis, J. H. Hwang, and J.
P. Hodges, J. Appl. Phys. 96 (2004) 3912.
G. Frank and H. Köstlin, Applied Physics A:
Materials Science &amp; Processing 27 (1982)
197.
K. Seshan, Handbook of Thin-Film Deposition
Processes and Techniques - Principles,
Methods, Equipment and Applications (2nd
Edition), William Andrew Publishing/Noyes,
2002.
H. Ma, J.-S. Cho, and C.-H. Park, Surface and
Coatings Technology 153 (2002) 131.
P. K. Song, Y. Shigesato, I. Yasui, C. W. OwYang, and D. C. Paine, Jpn. J. Appl. Phys. 37
(1998) 1870.
P. K. Song, Y. Shigesato, M. Kamei, and I.
Yasui, Jpn. J. Appl. Phys. 38 (1999) 2921.
A. Antony, M. Nisha, R. Manoj, and M. K.
Jayaraj, Applied Surface Science 225 (2004)
294.
Y. Gassenbauer and A. Klein, The Journal of
Physical Chemistry B 110 (2006) 4793.
104
Paper III
Paper IV
M. H. Rein, J. Mayandi, A. O. Holt and E. V. Monakhov. Optical electrical, chemical
and structural properties of nitrogen doped indium tin oxide thin films.
105
Optical, electrical, chemical and structural properties of nitrogen
doped indium tin oxide thin films
1*
2
1
M. H. Rein , J. Mayandi , A. O. Holt and E. V. Monakhov
1, 3
1 Department of Solar Energy, Institute for Energy Technology, Kjeller, Norway.
2 Department of Materials Science, School of Chemistry, Madurai Kamaraj University, Tamil Nadu, Madurai, India.
3 Physical Electronics group, University of Oslo, Oslo, Norway.
Abstract: Thin films of indium tin oxide have been deposited onto glass substrates at room temperature by the
use of dc magnetron sputtering in an atmosphere of argon mixed with nitrogen. Optical, electrical, chemical
and structural properties of the films have been studied in case of different flows of nitrogen. These properties
have also been examined after post deposition annealing of the films at 300 °C for 15 min in air. The N2/(Ar+N2)
flow vol% are examined in the range 0-25. The results show that transmittance of the films is above 70%
independent on the flows examined. Transmittance at about 90% is achieved when the film is deposited with
1.6% nitrogen and annealed. Among the as-deposited samples the films deposited with 3.2% nitrogen show the
best transmittance at about 87%. Electrical measurements show that the resistivity of the films is in the range
-4
6-14 x 10 Ω cm. Annealing decreases the resistivity to a certain extent. XPS measurements reveal that states
close to the valence band maximum are induced by the nitrogen supply. XRD measurements of the films
deposited with 3.2% nitrogen show some crystallinity in the as-deposited films and the crystallinity improves
significantly after annealing. XRD measurements of the samples deposited with 25% nitrogen show amorphous
structure of both as-deposited and annealed films.
Keywords: Indium tin oxide, nitrogen, sputtering.
*Corresponding author: Tel.: +4799045855; fax:+4763899964. E-mail address: [email protected] (M.
H. Rein). Postal address: Institute for Energy Technology, Pb. 40, 2027 Kjeller.
1
1.
Introduction
Since the first report on CdO as transparent conducting oxide (TCO) was published by Badeker in 1907 [1], the
number of work published on new materials picture well the interest and applicability of TCOs. Still, due to its
superior optical and electrical properties [2], the most suitable TCO for a number of applications is indium tin
oxide (ITO) [3–5]. Consequently, the number of published work on ITO is huge, both on fundamental research
and research for specific applications.
When depositing ITO films, it is well known from the literature that the deposition technique used and the
appurtenant processing parameters have a significant impact on the properties of the film as well as interfaces
[6–8]. By sputter deposition, one way to adjust the material properties is doping by introducing gases in
addition to argon. Kim et al. showed that a small quantity of hydrogen can improve electrical and optical
properties, and larger quantities has the effect of transforming the film structure into amorphous phase [9]. It
was also shown by Zhang et al. that hydrogen can affect the surface morphology of the ITO film [10]. The
addition of small quantities of oxygen generally improve the optical, electrical and structural properties
presumably due to improved stoichiometry by replacement of oxygen removed under low pressure conditions
[11–13]. Though, larger quantities may induce interstitial defects like Sn2O complexes [14]. The effect of adding
water vapour during reactive sputtering has been, among others, studied by Rottmannt et al. [15]. They
showed that H2O can improve the crystallinity and hence, the electrical properties. Nitrogen doping has not
been extensively studied. The effect of adding nitrogen is examined in TCOs like zinc oxide (ZnO) [16–18] and
tin oxide (SnO2), where p-type behaviour due to nitrogen is reported [19,20]. Some studies of the effect of
nitrogen in indium oxide (In2O3) and ITO as a photocatalyzing material for water splitting purpose are also
published [21,22]. Still, to our knowledge the literature on ITO doped with nitrogen is scarce.
For ITO applied in solar cells, the main film properties of interest are high transmittance and conductivity. The
focus of our work has been on how to achieve this by low temperature processing. This paper presents the
properties of ITO thin films doped with nitrogen, deposited onto glass substrates at room temperature by the
use of an industrial designed dc magnetron sputter. The effect of increased nitrogen content and postdeposition heat treatment on optical, electrical, structural and electronic properties is investigated.
2
2.
Experimental
Thin films (76-90 nm) of indium tin oxide were deposited at room temperature onto glass substrates (size
72x26x1 mm) by an industrial designed dc magnetron sputter (Leybold Optics). Prior deposition, all samples
were clean in DI-water (18 MΩ cm) in an ultra sound bath for 30 min and dried by compressed air. In order to
evaporate water residuals, a pre-heating of the samples was performed in load lock at 100 °C for 30 min. The
sputtering gas was argon and nitrogen. The argon flow was kept constant at 180 sccm, while the flow of
nitrogen was varied from 0 to 60 sccm. The N2/(Ar+N2) proportions were 0, 1.6, 3.2, 14.3 and 25 vol%, and the
samples are hereafter referred to as N0, N1, N2, N3 and N4, respectively. The power was 4 kW (power density
2
-3
of 5.3 W/cm ) and the pressure 3.1x10 mbar. The distance between target and sample holder was 80 mm.
The films were deposited by one oscillation in front of the target and the velocity of the sample holder was 0.7
m/min. The weight ratio of the target (In2O3/SnO2) was 90/10%. Some samples were subsequently annealed on
a hot plate in air at 300 °C for 15 min. Such annealing have previously been shown to improve the properties of
ITO films, and the effect of such heat treatments is well-documented (see, for instance, Refs. [13,23]).
The transmittance was measured in the wave length range 400 – 1000 nm by the use of a spectral response.
The film thickness was measured using an alpha-step profilometer (TENCOR Instruments). Structural
characterization was done by X-ray diffraction (XRD) with Cu Kα (λ = 1.5418 Å) radiation using a Siemens D5000
diffractometer. Electronic properties were examined by X-ray photoelectron spectroscopy (XPS) and the
instrument used was a KRATOS AXIS ULTRADLD. A monochromatic Al Kα radiation with hν=1486.6eV was used
at 15 kV and 10 mA. In order to achieve contact between the film and the spectrometer, all samples were
mounted to the sample holder with a metallic clips, i.e. the Fermi levels of the film and the spectrometer was
aligned.
Secondary Ion Mass Spectroscopy (SIMS) measurements were carried out on both as-deposited and annealed
samples in order to examine the effect of the nitrogen addition and heat treatment on the relative content of
nitrogen in the films. The instrument was a CAMECA ims 7f. The elements were ionized in negative mode with a
15 keV primary ion beam of Cs+, and the sputter rate was 2.2 Å/s. The depth of the sputtered craters was
measured with a Dektak 8 surface stylus profilometer.
3
3.
Results and discussion
3.1 Structural properties
Figure 1 shows the XRD pattern of as-deposited and annealed ITO films on sample N0, N2 and N4. One
observation in this figure is that 3.2% nitrogen enhances the crystallinity of the as-deposited film compared to
the reference film (sample N0) deposited without nitrogen. When the nitrogen constitute 25 vol% (sample N4),
the XRD pattern shows an amorphous structure. Annealing enhances the crystallinity of the N0 and N2 films,
but has no significant influence on the N4 film. The (222) peak exhibited by the annealed N2 sample,
corresponding to the (222) diffraction plane of In 2O3, increases by five times the peak of the as-deposited film.
The annealed N2 film shows small peaks at higher degrees than the (622) peak, which is not observed for the
N0 sample. Other peaks exhibited by the films are assigned to the (211), (400) and (440) diffraction plane of
In2O3.
Figure 1: XRD pattern of as-deposited and annealed ITO films sputter deposited with 0 (N0), 3.2 (N2) and 25
(N4) percentage nitrogen.
4
3.2 Secondary Ion Mass Spectroscopy
Figure 2 shows the relative concentrations of nitrogen in the as-deposited and annealed samples. The SIMS
measurements of nitrogen content have been performed by detecting
14 16
N O negative secondary ions. The
figure illustrates the increase in nitrogen content when nitrogen is added during sputtering. The SIMS data
illustrate a non-linear dependence of incorporated nitrogen on vol% of nitrogen in the sputtering chamber.
There is not a significant change in nitrogen content after heat treatment. The SIMS measurements of Sn and In
show no significant variation in the ratio Sn:In as a function of N flow.
14
16
Figure 2: Content of nitrogen shown by N O- ions measured by SIMS on as-deposited (Ad) and heat treated (Ht)
films.
3.3 X-ray Photoelectron spectroscopy
Figure 3 shows the valence band maximum of as-deposited and annealed films for deposition with different
percentage of nitrogen. From the measurement of the as-deposited films, we observe that the valence band
shifts upwards with introduction of nitrogen. This shift is interpreted as formation of nitrogen induced acceptor
states close to the valence band, and the number of states increases as the supply of nitrogen increases, up to
14.3 vol% (illustrated by the arrow). For a high nitrogen supply of 25 vol%, the amount of the induced states
does not seem to increase. In addition, the heat treatment does not eliminate the states. These results can be
5
seen in correlation to the non-linear dependence and the slight effect of annealing found by SIMS in Fig. 2. This
indicates that there is a solubility limit of nitrogen in ITO.
Figure 3: Valence band maximum of as-deposited and annealed films as a function of nitrogen flow.
3.4 Optical and electrical properties
Increasing the nitrogen flow has a minor impact on the growth rate of the films. The thickness decreased from
about 90 nm to 76 nm as the nitrogen flow was changed from 0 to 25 vol%. The average transmittance in the
wavelength range 400-1000 nm of the as-deposited and annealed films is shown in Figure 4A. The
transmittance shown is compensated for the loss due to the substrate. The figure shows that transmittance
improves due to introduction of nitrogen up to 3.2 vol%, but diminishes for the higher percentages of nitrogen.
Subsequent heat treatment improves the transmittance, except for the N2 film, and the best transmittance is
achieved with the N0 sample. As shown in Figure 4B, the resistivity of the as-deposited films has a minimum at
1.6 vol% nitrogen and increases as the nitrogen flow increases. The resistivity of the films decreases
considerably after annealing, although only limited decrease of the resistivity occurs for the N2 film.
As shown in Figure 3 the number of band gap states increased as the nitrogen flow increased. Hence, the
distance between the Fermi level and the valence band maximum decreased. This change as a function of
nitrogen supply is illustrated in Figure 4C. It shows that the energy difference (E F-EVBM) changes slightly after
heat treatment. The distance is measured as the difference between the Fermi level (defined at x=0) and the
position where the slope of the valence band intersects the x-axis. See for instance Ref. [24] and the references
therein.
6
From Fig. 4A, B and C, it is observed that the properties correlate with each other. Heat treatment has little or
no impact on any of the properties of the N2 sample, while the films deposited with other nitrogen flows
exhibit changes.
Figure 4: Transmittance, resistivity and EF-EVBM as a function of nitrogen flow for as-deposited and annealed
ITO films.
From the results presented, one may predict that the improvement in transmittance and conductivity of the
films are partly explained by structure and the formation of band gap states. The best transmittance of the asdeposited films is achieved with the N2 sample. This is also the sample which exhibits crystalline structure in
the XRD pattern. We assign the improved transmittance, compared with the reference sample (N0), to the
introduction of nitrogen and the enhanced crystallinity. When the nitrogen flow increases, the number of
states increases and the transmittance diminishes, as shown in Fig. 3 and 4A, respectively. This effect is
presumably due to an increased absorption. Similar results are previously presented by Reyes-Gil et al. on
nitrogen doped In2O3 [21], who reported that a red shift in absorption was due to a band gap narrowing. The
narrowing was larger if the nitrogen was at an interstitial site, rather than substitutional O sites. This correlated
well with the position of the band gap states; substitutional nitrogen forms states close to the valence band
while interstitial nitrogen forms states further into the band gap. The acceptor states observed in Fig. 3 indicate
that we have nitrogen at substiutional O sites in our films.
The improved properties of the reference sample (N0) are assigned to the change in structure, from
amorphous to crystalline, after annealing. The crystalline and amorphous structure exhibited by the N2 and N4
samples, respectively, respond differently to the heat treatment (Fig. 1). While the N2 sample exhibits an
7
enhanced crystallinity, the N4 sample was not affected, still showing an amorphous structure. Nevertheless,
the optical and electrical properties of the N2 sample did not change significantly, while the same properties of
the N4 sample clearly improve. Thus, from Fig. 1 and 4C, it is obvious that these features are more dependent
on changes of the valence band maximum, than structural properties.
From Fig. 4C, it is observed that increasing the nitrogen flow rate from 14.3 to 25 vol% does not increase the
number of states considerably. The SIMS results shown in Fig. 2 illustrates that the nitrogen content does not
increase significantly from 3.2 to 25 vol% nitrogen. XPS measurements of the N 1s peak (not presented)
confirm that the content of nitrogen does not increase as the supply of nitrogen increases from 14.3 to 25%.
These findings indicate that there is a solubility limit of the quantity of nitrogen which is possible to incorporate
into ITO.
4.
Conclusion
We have studied optical, electrical, chemical and structural properties of magnetron sputtered ITO films doped
with nitrogen. For ITO deposited at room temperature, it is found that a small addition of nitrogen to the
sputtering gas improves the optical and electrical properties, which can be ascribed to an enhanced
crystallinity. The study revealed that the nitrogen content in the films has a non-linear dependence on nitrogen
concentration in the sputtering plasma. However, our main conclusion is the observation of acceptor states in
ITO that are tentatively assigned to substitutional nitrogen on oxygen sites.
Acknowledgement: This study was partially funded via the Nanomat program by the Norwegian Research
Council and Renewable Energy Corporation.
[1]
K. Badeker, Annalen Der Physik 22 (1907) 749–766.
[2]
S. Ray, R. Banerjee, N. Basu, A.K. Batabyal, A.K. Barua, Journal of Applied Physics
54 (1983) 3497–3501.
8
[3]
J. Plá, M. Tamasi, R. Rizzoli, M. Losurdo, E. Centurioni, C. Summonte, F. Rubinelli,
Thin Solid Films 425 (2003) 185–192.
[4]
U. Betz, M. Kharrazi Olsson, J. Marthy, M.F. Escolá, F. Atamny, Surface and Coatings
Technology 200 (2006) 5751–5759.
[5]
F. Zhu, K. Zhang, E. Guenther, C.S. Jin, Thin Solid Films 363 (2000) 314–317.
[6]
M.J.U. Alam, D.C. Cameron, Thin Solid Films (2000) 455–459.
[7]
S. Bhagwat, R.P. Howson, Surface and Coatings Technology 111 (1999) 163–171.
[8]
A. Tho̸gersen, M. Rein, E. Monakhov, J. Mayandi, S. Diplas, Journal of Applied
Physics 109 (2011) 113532.
[9]
D. Kim, S. Lee, G. Lee, S. Kwon, Thin Solid Films 515 (2007) 6949–6952.
[10] K. Zhang, F. Zhu, C.H.A. Huan, A.T.S. Wee, Thin Solid Films (2000) 255–263.
[11] R. Das, K. Adhikary, S. Ray, Applied Surface Science 253 (2007) 6068–6073.
[12] H. Lee, O. Okpark, Vacuum 77 (2004) 69–77.
[13] M.H. Rein, J. Mayandi, E.V. Monakhov, A.O. Holt, Hamburg, Germany (2010) 2608–
2612.
[14] G. Frank, H. Köstlin, Appl. Phys. A 27 (1982) 197.
[15] M. Rottmannt, K. Heckner, Components 28 (1995) 1448–1453.
[16] A. Marzouki, A. Lusson, F. Jomard, A. Sayari, P. Galtier, M. Oueslati, V. Sallet,
Journal of Crystal Growth 312 (2010) 3063–3068.
[17] M. Sumiya, a. Tsukazaki, S. Fuke, a. Ohtomo, H. Koinuma, M. Kawasaki, Applied
Surface Science 223 (2004) 206–209.
[18] K. Iwata, P. Fons, a Yamada, K. Matsubara, S. Niki, Journal of Crystal Growth 209
(2000) 526–531.
[19] L.L. Kerr, X. Li, M. Canepa, A.J. Sommer, Thin Solid Films 515 (2007) 5282–5286.
[20] S.S. Pan, G.H. Li, L.B. Wang, Y.D. Shen, Y. Wang, T. Mei, X. Hu, Applied Physics
Letters 95 (2009) 222112–222113.
[21] K.R. Reyes-Gil, E.A. Reyes-Garcia, D. Raftery, Journal of Physical Chemistry C 111
(2007) 14579–14588.
[22] K.R. Wu, C.W. Yeh, C.H. Hung, L.H. Cheng, C.Y. Chung, Thin Solid Films 518
(2009) 1581–1584.
9
[23] F. Zhu, C.H.A. Huan, K. Zhang, A.T.S. Wee, Thin Solid Films 359 (2000) 244–250.
[24] A. Walsh, J. Da Silva, S.-H. Wei, C. Körber, a. Klein, L. Piper, A. DeMasi, K. Smith,
G. Panaccione, P. Torelli, D. Payne, a. Bourlange, R. Egdell, Physical Review Letters
100 (2008) 2–5.
10
Paper V
Annett Thøgersen, Margrethe Rein, Edouard Monakhov, Jeyanthinath Mayandi and
Spyros Diplas. Elemental distribution and oxygen deficiency of magnetron sputtered
indium tin oxide films. JOURNAL OF APPLIED PHYSICS 109, 113532 (2011)
117
JOURNAL OF APPLIED PHYSICS 109, 113532 (2011)
Elemental distribution and oxygen deficiency of magnetron sputtered
indium tin oxide films
Annett Thøgersen,1 Margrethe Rein,1 Edouard Monakhov,1 Jeyanthinath Mayandi,1
and Spyros Diplas2
1
Institute for Energy Technology, Department of Solar Energy, Instituttveien 18, 2008 Kjeller, Norway
SINTEF Materials and Chemistry, P.B 124 Blindern, N-0314 Oslo, Norway, and Centre for Material Science
and Nanotechnology, University of Oslo, Norway
2
(Received 26 December 2010; accepted 4 April 2011; published online 10 June 2011)
The atomic structure and composition of noninterfacial ITO and ITO-Si interfaces were studied
with transmission electron microscopy and x-ray photoelectron spectroscopy (XPS). The films
were deposited by dc magnetron sputtering on monocrystalline p-type (100) Si wafers. Both as
deposited and heat treated films consisted of crystalline ITO. The ITO/Si interface showed a more
complicated composition. A thin layer of SiOx was found at the ITO/Si interface together with In
and Sn nanoclusters, as well as highly oxygen deficient regions, as observed by XPS. High energy
electron exposure of this area crystallized the In nanoclusters and at the same time increased the
C 2011 American Institute of Physics. [doi:10.1063/1.3587174]
SiOx interface layer thickness. V
I. INTRODUCTION
Technological developments in photovoltaics such as
the use of transparent conducting oxides (TCOs) like indium
tin oxide (ITO) demand sufficient characterization of thin
films, surfaces, and interfaces. An In2 O3 thin film has an optical bandgap of more than 3.4 eV,1,2 with low resistivity,3,4
high transparency (87%) (Ref. 3) in the visible range, and
compatibility with fine patterning processes.5 ITO can be
used as electrodes for flat panel displays,6 including liquid
crystal displays,5 transparent electrodes for light-emitting
diodes,2 thin film gas sensors, solar cells,7 and as anodes in
organic light emitting diodes. The structure and composition
of ITO films has been the focus of research the last years.8–11
The through thickness composition in various types of ITO
films is an important issue, because of its impact on the device properties when used in photovoltaic applications.
In modern microelectronic systems consisting of thin
films deposited on substrates it is important to have a stable
film/substrate interface. Interface reactions may occur during
thin film deposition involving high energy incident species
in the substrate surface or during subsequent high-temperature post deposition processing.12 The deposition of ITO on
Si tends to create a thin layer of SiOx at the ITO/Si interface.13 This oxide is expected to change the properties of the
interface, making it unsuitable for ohmic contact applications. However, it might be utilized for in situ formation of
ITO-metallized gate oxides.12 Kobayashi et al.14 reported
formation of metallic In when they deposited ITO normal to
the Si substrate. The formation of metallic In in ITO films
changes their work function, which in turn may strongly
affect the I-V characteristics of the ITO/Si solar cells.14
In this work we attempt to gain a better understanding
of the composition of ITO films deposited by dc magnetron
sputtering on Si as well as the structure of the ITO-Si interface, high resolution transmission electron microscopy
(HRTEM), energy filtered transmission electron microscopy
(EFTEM), and x-ray photoelectron spectroscopy (XPS) were
0021-8979/2011/109(11)/113532/8/$30.00
used to study the composition of the noninterfacial ITO and
the ITO/Si interface.
II. EXPERIMENTAL
ITO films were deposited on monocrystalline p-type Si
(100) substrates by an industrially designed dc magnetron
sputtering equipment made by Leybold Optics. Before deposition, the samples were cleaned in an ultra sound bath with
DI water for 30 min and dried with compressed air. An additional HF dip for one minute and subsequent rinsing in DIwater for 2 min should provide an oxide free surface. The
samples were mounted on the sputtering carrier and loaded
into the load lock. The pressure in the load lock was 106
mbar when the samples were heated at 100 C for 30 min,
and the carrier was transferred from the load lock to the process chamber. The base pressure in the process chamber was
in the order of magnitude 107 mbar, and the working pressure was always kept at 3.1 103 mbar. The flow of argon
and oxygen was 180 sccm and 6 sccm (3.2% of the total
flow), respectively. The weight ratio of the indium-tin-oxide
(In2 O3 /SnO2 ) target was 90% versus 10%. The ramp up time
of the target was 15 min and the power density used was 5.3
W/cm2. There was only one passing of the carrier in front of
the target, and the carrier velocity was 0.7 m/min. The target
to substrate distance was 80 mm. The sputtering experiments
were performed without intentional substrate heating during
deposition. A maximum temperature of 80 C during deposition was measured by the use of Thermax thermostrips. After deposition, the samples were cut in two pieces, and one
half of each sample was annealed on a hot plate at 300 C
for 15 min in ambient air.
Cross-sectional TEM samples were prepared by ionmilling using a Gatan precision ion polishing system with 5
kV gun voltage. The samples were analyzed by HRTEM and
EFTEM in a 200 keV JEOL 2010 F microscope with a Gatan
imaging filter and detector. The spherical (Cs ) and chromatic
aberration (Cc ) coefficients of the objective lens were
109, 113532-1
C 2011 American Institute of Physics
V
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-2
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
FIG. 1. HRTEM image of the (A) as-deposited and (B)
heat treated ITO layer on a Si substrate.
0.5 and 1.1 mm, respectively. The point to point resolution
was 0.194 nm at Scherzer focus ( 42 nm). XPS was performed in a KRATOS AXIS ULTRADLD using monochromatic Al Ka radiation (hm ¼ 1486:6 eV) on plan-view
samples at zero angle of emission (vertical emission). The xray source was operated at 10 mA and 15 kV. The spectra
were peak fitted using Casa XPS (Ref. 15) after subtraction of
a Shirley type background. The Si-2p photoelectrons in ITO
have a mean free path (k) of 2.35 nm. This results in a photoelectron escape depth of 3k ¼ 7.05 nm. Assuming uniform
Arþ etching, an etching rate of about 0.04 nm/s was estimated
by measuring the ITO thickness on cross sectional TEM samples, observing the Si-2p peak from XPS survey spectra as
well as considering the photoelectron escape depth.
III. RESULTS AND DISCUSSION
OF NONINTERFACIAL ITO
Figure 1 shows TEM images of both the as deposited
and heat treated sample. Both samples are crystalline, with a
layer thickness of about 80 nm. Usually ITO is amorphous
when sputtered at room temperature (RT),16 but crystalline
ITO has also previously been shown to form upon deposition
at room temperature.17–20 Crystallization may be influenced
by the gas pressure and the target to sample distance. A suitable gas pressure has to be chosen so as to minimize collisions between the sputtering gas and the sputtered target
atoms, and the distance between target and the sample
should not be too large in order to preserve the kinetic
energy of the sputtered particles. By tuning these parameters
it is feasible to obtain a sufficient surface mobility of the
sputtered atoms and, hence, increase nucleation sites.17–19 It
is also established that crystallization of dc sputtered ITO at
room temperature (RT) depends on the thickness of the sputtered film. Moreover, it has been suggested that the substrate
type affects the critical thickness of the film required for
crystallization.20 It is established that the crystallization temperature of ITO is of the order of 150–160 C,21,22 while in
this work the measured sample temperature during deposition was 80 C.
Crystalline ITO may also form by plasma enhanced
crystallization. In this context the presence of crystallinity in
the as deposited samples could be explained as follows. The
distance between target and substrate is 8 cm. At this distance, the plasma surrounding the target also covers parts of
the wafer. It is therefore possible that the crystallization of
the as deposited ITO layer is enhanced by the plasma.
XPS was used to identify the composition and chemical
state of the elements in the ITO. The XPS high resolution
peaks of the O-1s, Sn-3d5=2 , and In-3d5=2 for the two samples
are shown in Fig. 2. The O-1s spectra have been fitted with
two peaks located at a binding energy of 530.5 and 531.8
eV. The Sn-3d5=2 spectra have been fitted with two peaks at
486.8 and 488.0 eV, and the In-3d5=2 at 444.9 and 446.0 eV;
see Table I. The peaks have been labeled as to distinguish
them during discussion. The origin of the Sn and In components labeled as InII , SnII , InIII , SnIII , and will be discussed in
detail in this chapter. The InI and SnI components will be
FIG. 2. XPS spectra of (A) O-1s, (B) Sn-3d5=2 , and (C) In-3d5=2 of the as deposited bulk sample.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-3
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
TABLE I. Binding energy values for noninterfacial ITO, including the
reference values for crystalline ITO, amorphous ITO, In(OH)x, pure In,
SnO2 , SnO, and pure Sn.
TABLE II. Atomic percentages and compositional ratios of both as
deposited and heat treated noninterfacial ITO.
Element
Composition
In-3d5=2 (eV)
InII
InIII
OI
OII
SnII
SnIII
ITO-cryst.
ITO-a
In(OH)x
In
SnO2
SnO
Sn
444.9
446.0
O-1s (eV)
Sn-3d5=2 (eV)
530.5
531.8
444.6 (Ref. 23)a
445.0 (Ref. 24)a
446.0 (Ref. 24)a
443.7 (Ref. 23)a
530.1 (Ref. 24)b
531.4 (Ref. 24)d
532.6 (Ref. 24)d
530.6 (Ref. 9)d
530.4 (Ref. 25)d
486.8
488.0
486.4c
487.1 (Ref. 24)e
486.2 (Ref. 24)e
485.0 (Ref. 26)e
a
Full width at half maximum (FWHM): 1.4 6 0.1 eV (Ref. 24).
FWHM: 1.3 6 0.1 eV (Ref. 24).
c
Reference (Ref. 27).
d
FWHM: 1.6 6 0.1 eV (Ref. 24).
e
FWHM: 1.5 6 0.1 eV (Ref. 24).
b
discussed in Sec. IV. The peak positions and reference values are presented in Table I.
The binding energies of InII and InIII are very close to
the reference values for amorphous ITO and In(OH)x,
respectively. However, since TEM images revealed that
most of the ITO was crystalline, the InII and SnII peaks are
most likely due to crystalline ITO. The two peaks have therefore been fitted with an asymmetry, due to the metallic character of ITO. The asymmetry of the peak was varied in order
to see if the InIII component could only be an expression of
the asymmetry of the InII component. However, fitting the
spectra with only one peak did not give satisfactory peak fitting. Kim et al.8 fitted the In-3d5=2 peak with two components, located at 444.08 and 445.24 eV (energy separation
1.16 6 0.1 eV), attributed to crystalline and amorphous ITO,
respectively. The energy separation between the two fitted
In-3d components in this work is 1.1 eV. This indicates that
the small In-3d5=2 peak at 446.0 eV may be attributed to
amorphous ITO. The InII /InIII relative fractions are 90 and
10 at. %, respectively. Since the SnII peak corresponds to
crystalline ITO, the SnIII may be attributed to Sn in amorphous ITO. The binding energies of the OI and OII peaks fit
also well with that of crystalline and amorphous ITO,
respectively.
In order to fully determine the chemical state of the oxide, the electroneutrality principle was used.9 The doubly
InII
InIII
OI
OII
SnII
SnIII
(OI þ OII )/(Intotal þ Sntotal )
OI /(InII þ SnII )
OII /(InIII þ SnIII )
OII /OI
As deposited
Heat treated
34.0
5.0
48.3
10.2
2.0
0.5
1.41
1.34
1.83
0.21
33.4
4.8
48.4
10.9
2.1
0.6
1.45
1.36
2.02
0.22
charged O2 and singly charged OH will have different
stoichiometry when combined with the lattice cation. Substitutional Sn in the In2 O3 oxide will have an (OI þ OII )/
(In þ Sn) theoretical ratio of 1.5, while Sn interstitial will be
1.55 when the oxygen is in the form of O2 .9 If the anions in
the oxide are in the form of OH instead of O2 , the ratio
will be 3.0.9 The atomic percentages of the different elements in ITO bulk as well as their calculated ratios for the as
deposited (asd) and heat treated (ht) samples are presented in
Table II. If OI , InII , and SnII correspond to crystalline ITO
(with the oxygen present as O2 ), the OI /(InII þ SnII ) ratio
would be about 1.5. The actual ratio for the as deposited
sample was 1.34 and for the heat treated sample 1.36. The
components that may correspond to amorphous ITO have a
ratio of OII /(InIII þ SnIII ) ¼ 1.83 and 2.02 for as deposited
and heat treated, respectively. This ratio is far higher than
what is expected for amorphous ITO (about 1.5). When
using the sum of all, the calculated ratios (OI þ OII )/
(InII þ InIII þ SnII þ SnIII ) 1.41 (asd) and 1.45 (ht), which are
closer to the value for crystalline ITO. This suggest that the
OI , OII , InII , InIII , SnII , and SnIII all result from crystalline
ITO, and the presence of In(OH)3 is also excluded.
Application of the electroneutrality principle contradicts
our initial assignment of the InIII , SnIII , and OII compounds.
However, crystalline ITO may have contributions from both
OI and OII peaks. Kim et al.9 showed that OI and OII may
and
originate from two different types of O2 ions (O2
I
2
O2
);
see
Fig.
3.
It
has
been
suggested
that
a
double
O
II
peak is common for oxides containing cations in multiple valence states.9,28 A higher binding energy of the O2
II ions
may be due to presence of oxygen in oxygen-deficient
regions. This means that these oxygen ions do not have
neighboring In atoms with full complement of six nearest
neighbor O2 ions;9,11 see Fig. 3. According to Fan et al.,11
FIG. 3. Two different kinds of O2 lattice ions located in different regions.
The higher binding energy O2
II arises
from the oxygen-deficient region. region.
Adapted from Kim et al. (Ref. 9).
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-4
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
FIG. 4. HRTEM images of the interface
between the ITO layer and the Si substrate of (A) the as deposited and (B) the
heat treated sample.
the O2
II peak has a about 1.5 eV higher binding energy than
.
The
chemical shift between the OI and OII peaks in Fig.
O2
I
2 is 1.3 6 0.1 eV. The OII peak may therefore also contain
contributions from oxygen deficient regions in the ITO. As a
consequence, the OII /OI ratio may be used to determine the
oxygen deficiency of the material.9 This ratio is 0.2 6 0.1
for the as deposited and 0.2 6 0.1 for the heat treated sample
(see Table II). The uncertainty in the calculated oxygen deficiency ratio is 0.1, assuming an uncertainty of 0.05 at. % in
the measured composition. This measured ratio is lower than
what was found for bulk ITO (1.06) by Kim et al.9 and is
closer to bulk ITO made by a combination of aquaregia treatment (HNO3 , HCl, and distilled water) and dry cleaning with
oxygen plasma,9 which was 0.8–0.9. A lower oxygen deficiency in the film may be a result of a decreased number of
oxygen vacancies,29 which may result in a decreased film
conductivity.30
IV. RESULTS AND DISCUSSION OF THE ITO/Si
INTERFACE
A detailed HRTEM, EFTEM, and XPS study of the
interface is presented in the next sections, together with the
effects of electron beam irradiation.
A. Elemental composition measured using XPS
Figure 4 shows HRTEM images of the ITO/Si interface.
A 2 nm amorphous layer is visible between the Si substrate
and the ITO film. This interface layer is present in both the
as deposited and the heat treated sample. Si readily oxidizes
when exposed to small amounts of O2 , and SiOx is expected
to form at the ITO/Si interface. When heat treated at high
temperatures, this oxide has been reported to grow12 (heated
at 785 C for 33 min). The interface oxide seen in our samples did not grow during heat treatment. However, our sample has only been heated at 300 C for 15 min, and this
temperature is probably too low to induce further oxidation.
Ramasse et al.13 have studied ITO deposited on Si using
chemical vapor deposition and re-annealed in vacuum for 30
s at 400 C. They identified the interfacial oxide as SiOx .
EFTEM imaging of the plasmon peak of pure Si (16 eV) and
SiOx (23 eV, which is very close to the plasmon peak of
SiO2 ) was performed to confirm this. Figure 5(A) is filtered
at 16 eV, which means that the bright areas in the image
result from pure Si. Figure 5(B) is filtered at 23 eV. The
amorphous region is bright when imaging with electrons at
23 eV. As expected, the interface oxide consists of SiOx .
FIG. 5. Two EFTEM images from the
as deposited sample where (A) has been
filtered at 16 eV showing the Si and (B)
at 23 eV showing SiOx .
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-5
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
FIG. 6. XPS depth profile spectra of the (a) O-1s peak, (b) Sn-3d5=2 , and (c) In-3d5=2 peak of the as deposited sample.
FIG. 7. XPS depth profile spectra of the (a) O-1s peak, (b) Sn-3d5=2 , and (c) In-3d5=2 peak of the heat treated sample.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-6
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
TABLE III. Binding energy values for ITO film at ITO/Si interface.
Composition
InI
InII
InIII
OI
OII
OII
SnII
SnII
SnIII
Binding energy (eV)
444.2
445.0
446.4
530.6
531.5
532.7
484.5
486.8
488.2
The interface composition of In, O, and Sn was examined using XPS. Figures 6 and 7 show XPS spectra of the (a)
O-1s peak, (b) Sn-3d peak, and (c) In-3d5=2 peak near the
ITO/Si-substrate interface of the as deposited and heat
treated sample, respectively, acquired upon depth profiling
using Arþ etching. The In-3d5=2 spectra have been fitted with
three components located at 444.2, 445.0, and 446.4 eV. The
O-1s spectrum is also fitted with three components at 530.6,
531.5, and 532.7 eV. The Sn-3d5=2 peak has been fitted with
three components located at a binding energy of 484.9,
486.8, and 488.2 eV. All binding energy values are presented
in Table III. The spectra from the as deposited and the heat
treated sample are very similar and show that heat treatment
did not have a large effect on the elemental and chemical
state distribution. From the spectra in Figs. 6 and 7 the
atomic percentages of the different oxidation states were
measured. These values are plotted in Fig. 8 together with
the percentage of Si0 and Six (spectra not shown).
Peaks InII , InIII , SnII , SnIII , OI , and OII are due to crystalline ITO, as discussed in Sec. III. The binding energies of InI
and SnI are similar to the literature values for pure In and Sn,
443.7 (Ref. 23) and 485 eV,26 respectively. The pure In and
Sn concentration increase toward the interface. If the InIII ,
SnIII , and OII components exist in crystalline ITO, they
should show the same trend (increase or decrease). However,
the intensity of the OII component increases toward the ITO/
Si-substrate interface where those of InIII and SnIII ones
decrease. The InIII and SnIII peaks could also not be in a
composition with the OIII peak, for the same reason.
As seen from the HRTEM and EFTEM images in Figs. 4
and 5, SiOx is present as a 2 nm layer near the interface. Oxygen in SiO2 has a reported binding energy of 533.05 eV.31 The
OIII peak has a binding energy of 532.7 eV and appears around
the same depth as the Si-2p peak (spectra not shown). This
suggests that the OIII peak results from oxygen in SiOx . The
peak is present at a distance of 8 nm from the interface. This
distance is more than what was identified as SiOx by TEM and
EFTEM. This is because the image resolution in the TEM
(0.19 nm) is far better than the depth resolution in XPS (which
is the photoelectron escape depth of about 7 nm).
The electroneutrality principle was again used to determine the chemical state of the oxide close to the interface,
for both the as deposited and heat treated sample. Since both
TEM and XPS data show no significant differences between
the two samples, only the data of the as deposited sample is
FIG. 8. A plot of the atomic percentage of Si, In, Sn, and O near the ITO/Sisubinterface in the as deposited and heat treated sample.
presented in Table IV. The (OI þ OII )/(Intotal þ Sntotal ) ratio
is within 1.4–1.6 for the first four spectra in Fig. 6. This
means that the oxide probably contains mostly O2 . Toward
the interface the ratio increases to around 4. This is considerably higher than what was observed for noninterfacial ITO
in Sec. III.
As discussed in Sec. III, the OII /OI ratio may be used to
determine the oxygen deficiency in the material.9 The
TABLE IV. Variation of atomic ratios in as deposited sample, with distance
from the ITO/Si interface.
Distance (nm)
OI þOII
Intotal þSntotal
OI
InII þSnII
OII
OI
11.5
10.0
8.5
7.0
5.5
3.0
1.5
0.0
1.5
1.45
1.45
1.45
1.57
1.62
2.16
3.72
4.41
3.28
1.36
1.28
1.22
1.28
1.11
0.92
0.20
0.25
0.27
0.32
0.63
2.61
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-7
Thøgersen et al.
J. Appl. Phys. 109, 113532 (2011)
FIG. 9. HRTEM images of the ITO/Si
interface, (A) before and (B) after electron beam exposure.
oxygen ratio is presented in Table IV for the as deposited
sample. The oxygen deficiency ratio of noninterfacial ITO as
presented in Sec. III was 0.2 and 0.21 for the as deposited
and the heat treated sample, respectively. The ratio of oxygen deficient regions near the interface is higher. At 11 nm
from the interface the ratio is the same as bulk for ITO film.
Moving toward the interface the ratio increases, and at the
interface the ratio is close to 2. These results show that there
is presence of pure In0 and Sn0 at the interface, together with
highly oxygen deficient regions.
The pure In and Sn at the interface have most likely
formed during deposition, when oxygen from ITO may have
reacted with the unsaturated Si at the wafer surface to make
SiOx .32 Pure In and Sn have therefore probably been a byproduct of this reaction.
B. Electron beam induced crystallization
In the previous section XPS analysis indicated that pure
In and Sn coexist with oxygen deficient regions near the
interface. Since conventional TEM bright field imaging did
not reveal presence of In clusters, one can assume that pure
In exists as amorphous nanoclusters near the ITO/Si interface. Upon exposure to high electron beam illumination, the
sample at the interface reacted with the electron beam. Figure 9 shows an HRTEM image of the as deposited sample
before (A) and after (B) electron beam exposure. During
electron beam illumination, small nanocrystals appear at the
ITO/Si interface.
Figure 10(A) shows HRTEM images of the ITO/Si interface after only seconds of electron beam irradiation. After a
short interval of electron irradiation, nanocrystals appear at
the ITO/Si substrate interface in the interfacial oxide layer.
The fast Fourier transformed (FFT) image inserted in the figure, shows the diffraction pattern of both the Si substrate and
the interface nanocrystal. Figure 10(B) shows the HRTEM
image after applying a circular mask with both the Si and the
interface nanocrystal diffraction pattern, while Fig. 10(C) has
resulted from applying a mask of only the diffraction patterns
of the interface nanocrystal. The inserted FFT patterns in the
image show the applied mask. After calibrating the diffraction
pattern using the Si diffraction pattern as a reference, the lattice parameters of the interface nanocrystals were found to be
0.270 and 0.243 nm. In has a space group of I4/mmm (No.
139), with a ¼ b ¼ 0:325 nm and c ¼ 0:495 nm.33 The measured lattice parameters fit well with the (101) and (002) plane
of pure In, which are 0.270 nm and 0.244 nm, respectively.
Therefore, the structure of the interface nanocrystal fits well
with the structure of pure In. The elemental composition near
the interface as discussed in the beginning of this section was
found to be composed of SiOx in addition to the presence of
In and Sn nanoclusters. The TEM results strengthen the argument that pure In is present as nanoclusters. During high
energy electron irradiation, the In clusters crystallize. In addition to this crystallization, the interface oxide increases in
thickness from 2 to 5 nm (see Fig. 10).
Pure In at the interface has previously been reported by
Ow-Yang et al.12 and Kobayashi et al.14 Kobayashi et al.14
FIG. 10. HRTEM images of the Si substrate and the interface nanocrystal, where (A) is the HRTEM image, (B) is the HRTEM image after applying a circular
mask, and (C) is after applying a mask of only the In diffraction pattern.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113532-8
Thøgersen et al.
found presence of metallic In by XRD when the deposition
of ITO occurred at 0 angle. Ow-Yang et al.12 studied the
ITO/Si interface by XRD and thermodynamics. Even though
no other reaction product was detected by TEM at the ITO/
Si interface, pure In was observed by XPS. This reaction was
also found to be thermodynamically stable at 1058 K, as
shown below:12
2In2 O3ðsÞ þ 3SiðsÞ ¼ 4InðsÞ þ 3SiO2ðsÞ ;
DG ¼ 981:3 kJ=mol:
The In crystallization could be attributed to localized heating
at the interface or to charging produced by the electron
beam. An in situ heating experiment of the sample showed
no nanocrystal formation at the ITO/Si interface. It is therefore likely that crystallization and growth of the In nanocrystals are due to direct electron beam induced splitting and
reconfiguration of atomic bonds. Since In is present as In3þ ,
the electron beam might attract additional In to the interface,
thereby resulting in larger In clusters. This may be assisted
by the oxygen depletion at the interfacial areas and subsequently by the presence of broken In–O bonds. The crystallization may naturally occur when the In cluster reaches its
critical radius. Nanocluster formation and e-beam induced
crystallization may have significant implications in manipulating the ITO/Si interface at the nanoscale via a combined
method of deposition and electron beam irradiation in the
early stages of deposition.
V. CONCLUSION
As deposited and heat treated ITO films made by sputter
deposition were studied in detail using TEM and XPS. The
samples were very similar and both crystalline. This was
attributed to plasma enhanced crystallization during deposition. The noninterfacial ITO consists only of crystalline ITO.
XPS depth profiling and TEM of the ITO/Si interface
revealed increasing amounts of In and Sn toward the interface, as well as the presence of an SiOx layer. This interface
oxide did not grow during heat treatment. Pure In was present as amorphous clusters. Close to the interface areas oxygen deficient regions were also found. During electron beam
exposure of the interface, the In nanoclusters crystallize and
grow. In addition, growth of the SiOx layer occurred.
1
C. Xirouchaki, G. Kiriakidis, T. Pedersen, and H. Fritzsche, J. Appl. Phys.
79, 9349 (1996).
2
S. H. Kim, N.-M. Park, T. Kim, and G. Sung, Thin Solid Films 475, 262
(2005).
J. Appl. Phys. 109, 113532 (2011)
3
M. Ando, E. Nishimura, K. Onisawa, and T. Minemura, J. Appl. Phys. 93,
1032 (2003).
4
K. Chopra, S. Major, and K. Pandya, Thin Solid Films 102, 1 (2000).
5
H. Ohta, M. Orita, M. Hirano, H. Tanji, H. Kawazoe, and H. Hosono,
Appl. Phys. Lett. 76, 2740 (2000).
6
R. Tahar, T. Ban, Y. Ohaya, and Y. Takahashi, J. Appl. Phys. 83, 2631
(1998).
7
H. Kobayashi, T. Ishida, Y. Nakato, and H. Tsubomura, J. Appl. Phys. 69,
1736 (1991).
8
Y.-S. Kim, Y.-C. Park, S. Ansari, B.-S. Lee, and H.-S. Shin, Thin Solid
Films 426, 124 (2003).
9
J. Kim, P. Ho, D. Thomas, R. Friend, F. Cacialli, and G. Bao, Chem. Phys.
Lett. 315, 307 (1999).
10
T. Ishida, H. Kobayashi, and Y. Nakato, J. Appl. Phys. 73, 4344
(1993).
11
J. Fan and J. Goodenough, J. Appl. Phys. 48, 3524 (1977).
12
C. W. Ow-Yang, Y. Shigesato, and D. C. Paine, J. Appl. Phys. 88, 3717
(2000).
13
Q. Ramasse, A. Anapolsky, C. Lazik, and M. J. A. A. A. Wang, J. Appl.
Phys. 105, 033716 (2009).
14
H. Kobayashi, T. Ishida, K. Nakamura, Y. Nakato, and H. Tsubomura, J.
Appl. Phys. 72, 5288 (1992).
15
http://www.casaxps.com/
16
S. Diplas, A. Ulyashin, K. Maknys, A. Gunnaes, S. Jørgensen, D.
Wright, J. Watts, A. Olsen, and T. Finstad, Thin Solid Films 515,
8539 (2007).
17
P. K. Song, Y. Shigesato, I. Yasui, C. W. Ow-Yang, and D. C. Paine, Jpn.
J. Appl. Phys. 37, 1870 (1998).
18
P. K. Song, Y. Shigesato, M. Kamei, and I. Yasui, Jpn. J. Appl. Phys. 38,
2921 (1999).
19
A. Antony, M. Nisha, R. Manoj, and M. K. Jayaraj, Appl. Surf. Sci. 225,
294 (2004).
20
G. Mei-Zhen, R. Job, X. De-Sheng, and W. R. Fahrner, Chin. Phys. Lett.
25, 1380 (2008).
21
P. K. Song, H. Akao, M. Kamei, Y. Shigesato, and I. Yasui, Jpn. J. Appl.
Phys. 38, 5224 (1999).
22
D. C. Paine, T. Whitson, D. Janiac, R. Beresford, and C. O. Yang, J. Appl.
Phys. 85, 8446 (1999).
23
F. Zhua, C. Huan, K. Zhang, and A. Wee, Thin Solid Films 359, 244
(2000).
24
C. N. de Carvalho, A. B. do Rego, A. Amaral, P. Brogueira, and G. Lavareda, Surf. Coatings Technol. 124, 70 (2000).
25
V. Jimenez, A. Fernandez, J. Espinos, and A. Gonzalez-Elipe, Surf. Sci.
350, 123 (1996).
26
F.-S. Ke, L. Huang, J.-S. Cai, and S.-G. Sun, Electrochim. Acta 52, 6741
(2007).
27
X. S. Peng, G. W. Meng, X. F. Wang, Y. W. Wang, J. Zhang, X. Liu, and
L. D. Zhang, Chem. Mater. 14, 4490 (2002).
28
F. H. Kichinosuke Hirokawa and M. Oku, J. Electron. Spectrosc. Relat.
Phenom. 6, 333 (1975).
29
W.-F. Wu, B.-S. Chiou, and S.-T. Hsieh, Semicond. Sci. Technol. 9, 1242
(1994).
30
H. Kim, C. M. Gilmore, A. Pique, J. S. Horwitz, H. Mattoussi, H. Murata,
Z. H. Kafafi, and D. B. Chrisey, J. Appl. Phys. 86, 6451 (1999).
31
G. Hollinger, Y. Jugnet, P. Pertosa, and T. M. Duc, Chem. Phys. Lett. 36,
441 (1975).
32
S. Diplas, O. M. Lovvik, H. Nordmark, D. Kepaptsoglou, J. M. Graff,
C. Ladam, F. Tyholdt, J. C. Walmsley, A. E. Gunnaes, and R. Fagerberg,
Surf. Inter. Anal. 42 (2010).
33
E. G. Moshopoulou, R. M. Ibberson, J. L. Sarrao, J. D. Thompson, and Z.
Fisk, Acta Crystallogr. B 39, 173 (2006).
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
Paper VI
M. H. Rein, M. V. Hohmann, A. Thøgersen, J. Mayandi, A. O. Holt, A. Klein, and E.
V. Monakhov. An in situ X-ray photoelectron spectroscopy study of the initial stages
of rf magnetron sputter deposition of indium tin oxide on p-type Si substrate.
APPLIED PHYSICS LETTERS 102, 021606 (2013)
127
APPLIED PHYSICS LETTERS 102, 021606 (2013)
An in situ x-ray photoelectron spectroscopy study of the initial stages of rf
magnetron sputter deposition of indium tin oxide on p-type Si substrate
M. H. Rein,1,a) M. V. Hohmann,2 A. Thïgersen,3 J. Mayandi,4 A. O. Holt,1 A. Klein,2
and E. V. Monakhov1,5
1
Institute for Energy Technology, Department of Solar Energy, Instituttveien 18, 2008 Kjeller, Norway
Technische Universit€
at Darmstadt, Institute of Materials Science, Surface Science Division,
Petersenstrasse 32, D-64287 Darmstadt, Germany
3
SINTEF-Materials and Chemistry Syntesis and Properties, Forskningsveien 1, Pb. 124 Blindern, 0314 Oslo,
Norway
4
Department of Materials Science, School of Chemistry, Madurai Kamaraj University, Tamil Nadu, Madurai,
India
5
Department of Physics/Centre for Materials Science and Nanotechnology, University of Oslo, 0316 Oslo,
Norway
2
(Received 27 October 2012; accepted 20 December 2012; published online 17 January 2013)
The interface between indium tin oxide and p-type silicon is studied by in situ X-ray photoelectron
spectroscopy (XPS). This is done by performing XPS without breaking vacuum after deposition of
ultrathin layers in sequences. Elemental tin and indium are shown to be present at the interface,
both after 2 and 10 s of deposition. In addition, the silicon oxide layer at the interface is shown to
C 2013 American Institute of
be composed of mainly silicon suboxides rather than silicon dioxide. V
Physics. [http://dx.doi.org/10.1063/1.4774404]
Despite the growing number of available transparent
conducting materials (TCMs), indium tin oxide (ITO) is still
the most preferable TCM for a number of applications due to
its excellent electrical and optical properties.1 When ITO is
deposited on a Si substrate a thin (1–3 nm) silicon oxide
(SiOx ) layer is established, due to the thermodynamics of
this formation.2 The properties of this interfacial layer may
influence the operation of a device, such as solar cells.3 For
instance, Goodnick et al. showed a thermal degradation of
ITO/p-Si solar cells due to growth of additional interfacial
SiOx .4
X-ray photoelectron spectroscopy (XPS) is a suitable
characterization method for the study of physical and chemical properties of the SiOx layer and the ITO/Si interface.
Chemical analysis of bonds between elements in an ITO/Si
interface is previously carried out by the use of ex situ XPS
of ultrathin ITO films on monocrystalline Si and XPS depth
profiling of ITO on monocrystalline silicon and amorphous
silicon.5–7
XPS is a characterization method which is sensitive to
surface contamination. For instance, carbon impurities are
previously proven to affect the work function value of ITO
measured by XPS.8 Thus, in order to avoid influence by any
contamination, a cleaning procedure prior the spectroscopy
is required. Cleaning the sample by ion bombardment will
remove most of the impurity elements, though, some atoms
will still remain at the surface. A drawback of ion bombardment is that it may induce side-effects such as chemical
changes in the material, as described in Ref. 9, and references therein. One type of chemical change can be depletion of
certain elements when the material is a compound. This can
either be due to different sputtering yield or due to preferential sputtering caused by higher volatility of one element. In
a)
Author to whom correspondence should be addressed. Electronic mail:
[email protected]
0003-6951/2013/102(2)/021606/4/$30.00
the case of a metal oxide, such depletion might induce a
reduction in the oxidation state of the metal. Other chemical
changes would be the effect of “knock on” (Ref. 9), which is
an incorporation of elements further into the material during
sputtering. Another side-effect of ion bombardment is the
topological changes at the sample surface, also described in
Ref. 9. For both sample cleaning and depth profiling XPS,
the possibility of chemical changes must be taken into
consideration.
A method which can overcome the influence of contamination and cleaning procedures is in situ XPS. Without
breaking vacuum after material deposition, it is possible to
carry out XPS studies of ultrathin layers step by step during
growth of a thicker film, not influenced by contamination.
This paper presents an in situ XPS study of ITO deposited by
rf magnetron sputtering onto a monocrystalline silicon substrate at room temperature. It investigates the initial stages of
the growth of the ITO film on a Si wafer, which includes the
formation of a SiOx interfacial layer and the presence of metallic indium and tin at the interface.
The ITO was deposited onto a polished p-type Czochralski Si substrate of size 2 2 cm. The orientation of the silicon crystal was h100i and the resistivity 1–3 X cm. Prior
deposition, the substrate was etched in hydrofluoric acid
(HF) (5 vol. %) in order to remove any native SiOx . The substrate was rinsed in deionized (DI) water (18 MX cm) and
blown dry with nitrogen. Subsequent to this treatment, the
substrate was mounted to the sample holder and transferred
quickly, in nitrogen atmosphere, to the load lock of DAISYMAT (DArmstadt Integrated SYstem for MATerial
research). This system is described in Ref. 10. XPS was carried out on the silicon substrate prior to the depositions. As
both the deposition chamber and the X-ray spectrometer are
connected to DAISY-MAT, the sample was transferred back
and forth from deposition to spectroscopy, without breaking
the vacuum. The ITO layers were deposited at room
102, 021606-1
C 2013 American Institute of Physics
V
021606-2
Rein et al.
temperature in four steps of 2, 8, 30, and 900 s, giving a total
deposition time of 940 s. The substrate to target distance
was 10 cm, the working power was 25 W (power density of
1:2 W=cm2 ) and the working pressure was 5 103 mbar.
The flow of argon during sputtering was 6 sccm and the
target composition was 90 wt. % In2 O3 and 10 wt. % SnO2 .
XPS was carried out using a Physical Electronics PHI
5700. Monochromatic Al Ka radiation (h ¼ 1486:6 eV) was
employed as excitation for XPS. XP spectra of In 3 d, Sn 3d,
O 1s, and Si 2p were recorded with a take-off angle of 45 .
The XPS results were peak fitted using the fitting program
CasaXPS. The background type was Shirley and the fitting
components were Guassian/Lorenzian (30%/70%) product
formulas. Some of the components were also modified by the
exponential blend, due to the metallic state of elements.
Cross-sectional transmission electron microscopy
(TEM) samples were prepared by ionmilling using a Gatan
precision ion polishing system with 5 kV gun voltage. The
samples were analyzed by high-resolution TEM (HRTEM)
and energy filtered TEM (EFTEM) in a 200 keV JEOL
2010 F microscope with a Gatan imaging filter and detector.
The spherical (Cs) and chromatic aberration (Cc) coefficients
of the objective lens were 0.5 and 1.1 mm, respectively. The
point to point resolution was 0.194 nm at Scherzer focus
(42 nm).
Appl. Phys. Lett. 102, 021606 (2013)
Secondary ion mass spectroscopy (SIMS) measurements
were carried out by the use of a CAMECA ims 7f instrument. The elements were ionized in negative mode with a
15 keV primary ion beam of Csþ, and the sputtering time
was 42 480 s. The depth of the sputtered craters was measured with a Dektak 8 surface stylus profilometer.
Figure 1 shows the XPS spectra of In 3d5=2 , Sn 3d5=2 , O
1s, and Si 2p carried out prior and subsequent to the depositions. The In 3d5=2 , Sn 3d5=2 , and O 1s spectra are fitted with
two or three components, while the Si 2p spectra from substrate and after 2, 10, and 40 s deposition are fitted with two,
five, six, and one Si3=2 and Si1=2 peaks, respectively. No In
or Sn peaks are observed prior deposition and no Si peaks after the last deposition. The thicknesses are estimated from
TEM images and calculations based on the integrated area of
the Si peaks in the XPS spectra.11
The predominant peaks in the In 3d5=2 and Sn 3d5=2
spectra after the first deposition for 2 s are attributed to elemental In and Sn, following a similar analysis procedure as
in our previous work.12 Pure In is also found after 10 s, while
pure Sn is present both after 10 and 940 s. A segregation of
tin to the film surface is previously reported by others.13 The
mentioned analysis is performed for the predominant InII
and SnII peaks after 10, 40, and 940 s as well, and these
peaks are assigned to In and Sn in crystalline ITO.
FIG. 1. The XPS spectra and peak fitting
of the In 3d5=2 , Sn 3d5=2 , O 1s, and Si 2p
peaks. The Si 2p spectra include SiOx
insets. The intensities of some spectra
are enhanced with the indicated factors.
021606-3
Rein et al.
Appl. Phys. Lett. 102, 021606 (2013)
FIG. 2. Energy filtered TEM images using the plasmon peak of (a) SiO2 at
23 eV, and (b) for Si at 16 eV.
In the previous work, we argued that In components at
444.9 eV and 446.0 eV (energy separation of 1.1 eV) could
be attributed to crystalline and amorphous ITO, respectively.12 By further examination, applying the electroneutrality principle, the presence of amorphous ITO has been
excluded.13 We can confirm this conclusion by analysis of
TEM images (Fig. 2) showing no diffuse circles in the diffraction pattern.
The origin of the minor peaks InIII (at 445.9, 445.7, and
446.0 eV after 2, 40, and 940 s, respectively) and SnIII (487.7
and 487.5 eV after 40 and 940 s, respectively) could be
related to hydrogen, as hydrogen was detected by SIMS in
both bulk and interfacial ITO (Fig. 3). The position of the
InIII component coincides with the InðOHÞ3 state (445.8 eV)
in Ref. 14. Similarly, the high binding energy peak SnIII
could be assigned to bonds between hydrogen and Sn.
Another explanation could be plasmon excitation. This is
previously discussed by Christou et al.15 and Gassenbauer
et al.16
The two O 1s components found at the Si wafer surface
(531.9 and 532.6 eV) are most likely due to water molecules
and contamination adsorbed at the substrate surface, originated from the DI water rinsing or exposure to air. The values coincide with the binding energies for hydroxides,
contamination, and oxygen due to air exposure (531:7 6 0:2
and 532:7 6 0:2 eV) reported by Pla et al.17 Hydroxides are
plausible as the substrate was cleaned in HF and rinsed in DI
water. OH groups at Si(100) wafers after HF etch and DI
water rinsing are previously detected using high resolution
electron energy loss spectroscopy (HREELS).18 Contamination due to air exposure is also plausible as traces of carbon
were detected at the substrate surface and after 2 s deposition
(not illustrated).
After 2 s of deposition, the predominant O 1s peak is at
532.6 eV (OIII ). The intensity of this peak increases slightly
after the next deposition (not illustrated in the figure) and
disappears after further deposition. Hence, this binding
energy is more likely due to other chemical states than found
at the substrate surface. An obvious explanation of the origin
would be compounds of oxygen and silicon as SiOx is
observed as a shoulder after 2 and 10 s in the Si 2p spectra
(discussed below). A binding energy of 532.6 eV is reported
by Wagner et al. of thermally oxidized silicon wafer.19 The
interfacial SiOx layer is also observed by TEM and SIMS in
Figs. 2 and 3, respectively.
FIG. 3. SIMS of the ITO/Si interface and SiOx interfacial layer.
The analysis of the OI and OII components found after
deposition is similar to our previous work.12 Both peaks are
attributed to crystalline ITO, the OII component related to
O2 ions at oxygen deficient sites.
The binding energy, full width at half maximum
(FWHM), and chemical shift of the Si 2p and SiOx peaks
were fitted as shown in our previous work.20 In addition to
the Si 2p peaks corresponding to the 2p1=2 and 2p3=2 spin
states, two small components with binding energy in the
range 99.8 - 100.6 eV are found. These small components
are attributed to silicon bonded to hydrogen, according to
Thïgersen et al.20 In that work, a peak with binding energy
at 99.7 eV was attributed to Si3 SiH and is close to our value
(99.8 eV). The two small components at 100.1 and 100.6 eV
after 10 s can be attributed to Si2 SiH2 , which has a binding
energy 0.57 eV higher than elemental Si.21
The SiOx shoulder with a peak position at 102.6 eV is
fitted by three and four components after 2 and 10 s deposition, respectively. The components in the decomposed SiOx
peak correspond to the Si2 O; SiO; Si2 O3 (suboxides) and
SiO2 states. No SiO2 is found after 2 s deposition and the
predominant peaks correspond to the Si2 O and Si2 O3 suboxide states. After 10 s deposition, the Si2 O3 peaks are predominant and small peaks corresponding to the SiO2 state are
present. These results show that SiO2 is not the dominating
oxide in the SiOx interfacial layer between ITO and Si.
From the literature, it is known that oxidation of suboxides requires heat at a certain level. He et al. found that suboxides oxidize into SiO2 when annealing at high temperatures
(>1000 K).22 Zhang et al. reported changes in concentration
of the different suboxides due to increased annealing temperature, though low annealing temperature (<400 C) did not
change the concentration of Si2 O and Si2 O3 states.23 According to this information, we assume that the presence of SiO2
in the SiOx layer is significantly low throughout the layer as
our sample did not undergo heat treatment.
The presence of suboxides may introduce another band
gap than SiO2 in the SiOx layer. A narrower band gap for
suboxides in a SiO2 /n-type Si(111) interface region is calculated by Yamashita et al.24 They have shown that the main
oxides in a thin interface region between a Si substrate and a
SiO2 film are Si2 O and Si2 O3 and that the band gap increases
021606-4
Rein et al.
as the oxidation state increases (Eg ðSi2 OÞ < Eg ðSi2 O3 Þ
< Eg ðSiO2 Þ). A different band gap alignment of the ITO/Si
interface than in the case of SiO2 as the main silicon oxide
and the presence of metallic In and Sn at the interface may
play a significant role in the carrier transport mechanism in
the ITO/Si junction. The prominent presence of suboxides at
the ITO/Si interface and the likely effect on the tunneling
probability through the SiOx layer has previously been put
forward by Kobayashi et al.25
In conclusion, this in situ XPS study qualitatively confirms previous ex situ XPS results. We have shown that both
elemental In and Sn are present at the ITO/Si interface. Our
study also reveals that suboxides are predominant in the
interfacial SiOx layer. This may have an impact on the carrier transport properties at the ITO/Si interface.
The authors would like to thank Lasse Vines for performing the SIMS characterization. The work was funded by
REC Solar, the Research Council in Norway, through the
Nanomat program, and the XPS work in Darmstadt has been
supported by the Deutsche Forschungsgemeinschaft (DFG)
within the collaborative research center SFB 595.
1
S. Ray, R. Banerjee, N. Basu, A. Batabyal, and A. Barua, J. Appl. Phys.
54, 3497 (1983).
2
C. Ow-Yang, Y. Shigesato, and D. Paine, J. Appl. Phys. 88, 3717 (2000).
3
J. Shewchun, J. Dubow, C. Wilmsen, R. Singh, D. Burk, and J. Wager,
J. Appl. Phys. 50, 2832 (1979).
4
S. Goodnick, J. Wager, and C. Wilmsen, J. Appl. Phys. 51, 527 (1980).
5
S. Diplas, O. M. Lovvik, H. Nordmark, D. M. Kepaptsoglou, J. M. Graff,
C. Ladam, F. Tyholdt, J. C. Walmsley, A. E. Gunnaes, R. Fagerberg, and
A. Ulyashin, Surf. Interface Anal. 42, 874 (2010).
6
A. Montesdeoca-Santana, E. Jimenez-Rodriguez, N. Marrero, B. Gonzalez-Diaz, D. Borchert, and R. Guerrero-Lemus, Nucl. Instrum. Methods
Phys. Res. B 268, 374 (2010).
Appl. Phys. Lett. 102, 021606 (2013)
7
S. Sheng, H. Hao, H. Diao, X. Zeng, Y. Xu, X. Liao, and T. L. Monchesky,
Appl. Surf. Sci. 253, 1677 (2006).
K. Sugiyama, H. Ishii, Y. Ouchi, and K. Seki, J. Appl. Phys. 87, 295
(2000).
9
J. C. Riviere, Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, edited by D. Briggs and M. P. Seah (Wiley, Chichester,
1983), pp. 17–85.
10
D. Ensling, A. Thissen, Y. Gassenbauer, A. Klein, and W. Jaegermann,
Adv. Eng. Mater. 7, 945 (2005).
11
M. P. Seah, Practical Surface Analysis by Auger and X-ray Photoelectron
Spectroscopy, edited by D. Briggs and M. P. Seah (Wiley, Chichester,
1983), pp. 181–216.
12
A. Thïgersen, M. Rein, E. Monakhov, J. Mayandi, and S. Diplas, J. Appl.
Phys. 109, 113532 (2011).
13
J. Kim, P. Ho, D. Thomas, R. Friend, F. Cacialli, G. Bao, and S. Li, Chem.
Phys. Lett. 315, 307 (1999).
14
C. D. Wagner, Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, edited by D. Briggs and M. P. Seah (Wiley, Chichester,
1983), pp. 477–509.
15
V. Christou, M. Etchells, O. Renault, P. J. Dobson, O. V. Salata, G. Beamson, and R. G. Egdell, J. Appl. Phys. 88, 5180 (2000).
16
Y. Gassenbauer, R. Schafranek, A. Klein, S. Zafeiratos, M. Havecker,
A. Knop-Gericke, and R. Schlogl, Phys. Rev. B 73, 245312 (2006).
17
J. Pla, M. Tamasi, R. Rizzoli, M. Losurdo, E. Centurioni, C. Summonte,
and F. Rubinelli, Thin Solid Films 425, 185 (2003).
18
D. Graf, M. Grundner, R. Schulz, and L. Muhlhoff, J. Appl. Phys. 68,
5155 (1990).
19
C. Wagner, D. Passoja, H. Hillery, T. Kinisky, H. Six, W. Jansen, and
J. Taylor, J. Vac. Sci. Technol. 21, 933 (1982).
20
A. Thïgersen, J. H. Selj, and E. S. Marstein, J. Electrochem. Soc. 159,
D276 (2012).
21
G. Cerofolini, C. Galati, and L. Renna, Surf. Interface Anal. 35, 968
(2003).
22
J. He, X. Xu, J. Corneille, and D. Goodman, Surf. Sci. 279, 119 (1992).
23
W. Zhang, S. Zhang, Y. Liu, and T. Chen, J. Cryst. Growth 311, 1296
(2009).
24
Y. Yamashita, S. Yamamoto, K. Mukai, J. Yoshinobu, Y. Harada,
T. Tokushima, T. Takeuchi, Y. Takata, S. Shin, K. Akagi, and S. Tsuneyuki,
Phys. Rev. B 73, 045336 (2006).
25
H. Kobayashi, T. Ishida, Y. Nakato, and H. Tsubomura, J. Appl. Phys. 69,
1736 (1991).
8
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement