New TCO for Use as
Transparent Front Contact in
Chalcopyrite Thin Film Solar Cells
vorgelegt von
M. Sc.
Darja Erfurt
geb. in Nowosibirsk, Russland
von der Fakultät IV - Elektrotechnik und Informatik
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktorin der Ingenieurwissenschaften
- Dr.-Ing. -
genehmigte Dissertation
Promotionsausschuss
Vorsitzender: Prof. Dr. Jürgen Bruns
Gutachter: Prof. Dr. Bernd Szyszka
Prof. Dr. Rutger Schlatmann
Prof. Dr. Günter Bräuer
Tag der wissenschaftlichen Aussprache: 10.01.2019
Berlin 2019
Table of Contents
Abstract iv
Zusammenfassung v
1 Introduction 1
2 Fundamentals 5
2.1 Transparent Conductive Oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Electrical Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Scattering mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.3 Optical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.4 Indium oxide based high mobility TCOs . . . . . . . . . . . . . . . . . . . 11
2.2 CIGS solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Basic Principles of a CIGS Solar Cell . . . . . . . . . . . . . . . . . . . . . 15
2.2.2 Solar Cell parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.3 Module structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Experimental Details 21
3.1 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.1 Indium Oxide Based Layer Preparation . . . . . . . . . . . . . . . . . . . 21
i
Table of Contents
3.1.2 CIGS Solar Cell Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Characterization Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.1 Material Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.2 Device Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4 Sputtered Hydrogen doped Indium Oxide 36
4.1 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Electrical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Optical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.5 Strategies to improve the electro-optical properties after annealing in air . . . . . 50
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5 Substrate Influences on Growth Mechanism and Properties 54
5.1 Sputtered Zn(O,S) and ZnO films on glass . . . . . . . . . . . . . . . . . . . . . . 54
5.2 Sol-Gel Indium - and Gallium Oxide Layers on glass . . . . . . . . . . . . . . . . 67
5.3 Polycrystalline CIGS and textured glass . . . . . . . . . . . . . . . . . . . . . . . 71
5.3.1 Detrimental effects on the electron mobility . . . . . . . . . . . . . . . . . 75
5.4
Strategies to improve the electron mobility of hydrogen doped indium oxide based
TCOs on CIGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4.1 Spin Coated Sol-Gel Layers . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4.2 Etching of the CIGS Surface . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.4.3 Influence of TCO thickness . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.5 Stability of IOH thin films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
ii
Table of Contents
6 Application of Indium Oxide based TCOs in CIGS solar cells 101
6.1 Basic Requirements for the Application as Front Contact . . . . . . . . . . . . . 102
6.1.1 Band alignment IOH/ZnO . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
6.1.2 Effect of Post Deposition Thermal Treatment on CIGS Solar Cells . . . . 106
6.2 Challenges in the Application as Front Contact in CIGS Modules . . . . . . . . . 115
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells 121
6.3.1 Spin Coated Sol-Gel Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.2 Etched CIGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.3.3 Increase of TCO thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7 Conclusions and Outlook 133
A Supplementary Information 138
A.1 General description of standard characterization methods . . . . . . . . . . . . . 138
A.2 Additional information concerning investigations of IOH layers on ZnO . . . . . . 142
A.3
Additional information concerning investigations of the low TCO electron mobility
on CIGS samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Symbols and Acronyms 151
Bibliography 153
List of Publications 171
Acknowledgement 174
iii
Abstract
In this thesis the applicability of high-mobility indium oxide based transparent conductive oxides
(TCOs) as front contact in Cu(In,Ga)(S,Se)
2
(CIGS) solar cells and modules was investigated.
The unique trade-off between optical and electrical properties promises improved short circuit
current densities compared to conventional CIGS device configurations, in particular in CIGS
modules. By doping indium oxide films with hydrogen an amorphous growth can be achieved. A
subsequent thermal treatment results in solid phase crystallization and high electron mobilities.
In order to process CIGS modules the deposition process of hydrogen doped indium oxide was
first scaled up using an in-line pulsed DC magnetron sputtering tool. The structural, electrical
and optical properties were studied in dependence on the deposition conditions water vapor and
oxygen partial pressure as well as on the annealing atmosphere.
Furthermore, influences of sputtered Zn(O,S) and ZnO as well as spin coated sol-gel In
x
O
y
and
Ga
x
O
y
sub-layers on the structural and electrical properties of In
2
O
3
:H were investigated. It was
found, that the poly-crystalline structure of ZnO promotes crystalline growth of In
2
O
3
:H films
during deposition and thus results in poor electrical properties. In contrast an amorphous sub-
layer, such as, here, Zn(O,S) or Ga
x
O
y
had no adverse effect on the properties of In
2
O
3
:H films.
The rough structure of CIGS layers, in particular spiky grain edges and sharp recessions induced
void formation in indium oxide TCOs deposited by sputtering or reactive plasma deposition,
regardless of their crystallinity fraction after deposition or the deposition conditions in general.
Due to the presence of voids the films exhibit low electron mobilities, which did not improve
sufficiently after annealing. The consequently increased sheet resistance resulted in a limited fill
factor when applied in CIGS modules. We therefore developed strategies to improve the electron
mobility of the indium oxide based TCOs when grown on rough CIGS samples and tested the
applicability in CIGS devices. Furthermore, the influence of the thermal treatment, as required
in order to crystallize the amorphous grown indium oxide films, on the solar cell performance
was studied.
The investigations point out challenges and provide appropriate strategies and possible solutions
for the successful implementation of TCOs with high mobility in CIGS devices in order to
demonstrate the potential for improving CIGS solar cell and module efficiencies.
iv
Zusammenfassung
In dieser Arbeit wurde die Anwendbarkeit von Indiumoxid-basierten transparenten leitfähigen
Oxiden (TCOs) mit hoher Elektronenbeweglichkeit als Frontkontakt in Cu(In,Ga)(S,Se)
2
(CIGS)
Solarzellen und -Modulen untersucht. Der spezifisch gute Kompromiss zwischen optischen und
elektrischen Eigenschaften verspricht im Vergleich zu herkömmlichen Konfigurationen, insbeson-
dere bei CIGS-Modulen, verbesserte Kurzschlussstromdichten. Durch Dotierung von Indiumoxid-
schichten mit Wasserstoff kann ein amorphes Wachstum erreicht werden. Eine anschließende
thermische Behandlung führt zu Festphasenkristallisation und hohen Elektronenbeweglichkeiten.
Zur Herstellung von CIGS-Modulen wurde der Abscheidungsprozess von wasserstoffdotiertem
Indiumoxid zunächst unter Verwendung einer in-line gepulsten DC-Magnetron-Sputteranlage
skaliert. Die strukturellen, elektrischen und optischen Eigenschaften wurden in Abhängigkeit
von den Abscheideparametern Wasserdampf- und Sauerstoffpartialdruck, sowie der Atmosphäre
der Wärmebehandlung untersucht.
Weiterhin wurden Einflüsse von gesputterten Zn(O,S) und ZnO sowie von spin - beschichteten
Sol-Gel In
x
O
y
und Ga
x
O
y
Unterschichten auf die strukturellen und elektrischen Eigenschaften
von In
2
O
3
:H untersucht. Es wurde festgestellt, dass die polykristalline Struktur von ZnO das
kristalline Wachstum von In
2
O
3
:H-Filmen schon während der Abscheidung fördert und somit zu
schlechten elektrischen Eigenschaften führt. Im Gegensatz dazu hatte eine amorphe Unterschicht,
wie hier Zn(O,S) oder Ga
x
O
y
keine negativen Auswirkungen auf die diesbezüglichen Eigenschaften
von In
2
O
3
:H-Filmen. Die raue Struktur der CIGS-Schichten, insbesondere scharfe Kornränder
und spitze Täler, induzieren die Bildung von Hohlräumen in den durch Sputtern oder reaktiver
Plasmaabscheidung abgeschieden Indiumoxid-TCOs, unabhängig von deren Kristallinität nach
der Abscheidung oder den Abscheidebedingungen im Allgemeinen. Durch das Auftreten der
Hohlräume wiesen die Schichten eine geringe Elektronenbeweglichkeit auf, die sich nach dem
Tempern nicht ausreichend verbesserte. Der damit verbundene erhöhte Schichtwiderstand
führte beim Einsatz in CIGS-Modulen zu einem begrenzten Füllfaktor. Wir entwickelten daher
Strategien zur Verbesserung der Elektronenbeweglichkeit der auf Indiumoxid basierenden TCOs
für den Einsatz in rauen CIGS-Proben und testeten deren Anwendbarkeit in Bauteilen. Weiterhin
wurde der Einfluss der thermischen Behandlung auf die Eigenschaften der Solarzelle untersucht.
Das Tempern ist erforderlich, um die amorph gewachsenen Indiumoxidschichten zu kristallisieren.
Die Untersuchungen weisen auf Probleme hin und liefern entsprechende Strategien und Lö-
sungsmöglichkeiten für die erfolgreiche Implementierung von TCOs mit hohen Beweglichkeiten in
CIGS-Bauteilen, um das Potenzial zur Verbesserung der CIGS-Solarzellen- und Modulwirkungs-
grade aufzuzeigen.
v
CHAPTER 1
Introduction
In the last decades the demand for energy has increased drastically, impacting the environment
by e.g. CO
2
emission and the corresponding increase in global temperature. In order to protect
the environment, a change in energy production and consumption is needed. In 2015 the historic
climate accord aspired a limit of the rise of the average global temperate to 2
°
C until 2050. A
key aspect for the fulfilment of this target is the development of renewable energies. In 2015
only 25 %of the electricity were generated by renewables. According to the Roadmap of the
International Renewable Energy Agency (IRENA) [1] an increase of up to 85 %has to be realized
to limit the temperature rise to 2
°
C. As the main sources for electricity generation wind (36 %)
and solar photovoltaic (22 %) are suggested. To satisfy this need, photovoltaic devices have to
be further developed to achieve higher efficiencies and to reduce the costs. Additionally new
application ranges, such as building integrated photovoltaic, need to evolve. One approach is the
application of thin film photovoltaic.
Cu(In,Ga)(S,Se)
2
(CIGS) is one of the most promising materials in this field due to high
efficiencies, which are comparable to Si-based solar cells, low energy consumption, short energy
payback time and minimized material consumption [2]. In 2017 SolarFrontier presented a record
cell efficiency of 22.9 %[3]. However, the commercial module efficiency is far behind this value,
in the range of 14 %to 15 %. Bermudez et al. [4] reviewed the challenges of the cell-to-module
efficiency gap in Cu(In,Ga)(S,Se)
2
. According to this study the main power losses in CIGS
modules with respect to CIGS solar cells are optical and caused by the front contact (transparent
conductive oxide (
TCO
), typically ZnO:Al). Thus the relative power losses due to the increased
optical absorption of the TCO in a module is stated to be
≈
40 %with regard to the application
in solar cells.
To counteract these losses new alternative TCOs are required, which have to satisfy several
demands. The films must have a wide bandgap of > 3 eV and high transmittance to ensure a high
1
Introduction
amount of photons absorbed by the CIGS layer. Furthermore the films must be highly conductive
with resistivity in the range of < 10
−3Ω
cm and typical sheet resistances of around 10
Ω
/Sq or
less due to the absence of a contact grid. To achieve such low sheet resistances, the TCO film
thickness consequently must be higher in modules compared to cells, where the lateral current
collection paths are shorter, resulting in additional optical losses. One approach to minimize
these losses is the implementation of new high-mobility TCOs. Due to their excellent trade-off
between optical and electrical properties they show high conductivity with low optical absorption
in the relevant range. Therefore the same sheet resistances can be reached as with ZnO, but the
optical losses can be minimized, resulting in higher photo current. Amorphous grown indium
oxide based films with various dopants, such as hydrogen, W, Zr, Mo or Ti showed electron
mobilities > 80 cm
2
/Vs [5
–
10] after solid phase crystallization, initiated by a post deposition
thermal treatment. These properties make these materials promising candidates as front contact
in CIGS modules. Therefore in this thesis we investigate the applicability of high mobility indium
oxide based TCOs, namely In
2
O
3
:H and In
2
O
3
:H,W and the feasibility of a large-scale deposition
process.
This thesis is structured as follows:
Chapter 2 provides basic fundamental knowledge concerning the materials and devices studied in
this thesis. This includes a review of the principles of transparent conductive oxides, the material
indium oxide in general, and CIGS solar cells and modules.
Chapter 3 provides detailed information about the experimental methods used in this thesis. First
the sample preparation is explained. Then, the characterization techniques used to determine
the material and device properties are described.
Chapter 4 presents the properties of hydrogen doped indium oxide, which was deposited by a
large scale in-line pulsed direct current magnetron sputtering tool. Here the influence of the
deposition conditions water vapor pressure and oxygen supply are identified. Furthermore the
influence of the annealing in different atmospheres is investigated.
Chapter 5 identifies influences of the substrate and sub-layers on the growth of indium oxide
based TCOs. This includes investigations of the growth on sputtered ZnO and Zn(O,S), as well
as on spin coated In
x
O
y
and Ga
x
O
y
layers on planar glass substrates. Furthermore the impact
of textured substrates, such as textured glass, and sub-layers, such as poly-crystalline CIGS,
is studied. Here the roughness and the local slope of the sub-layers was correlated with the
structure and the resulting electrical properties of the indium oxide films. Strategies to mitigate
these influences and possible adverse effects are presented. Lastly the stability of hydrogen doped
indium oxide on rough substrates is discussed.
2
Chapter 6 addresses the application of indium oxide based TCOs as front contact in CIGS solar
cells and modules. First general requirements for the implementation are discussed. This includes
investigations of the band alignment at the interface of the highly resistive layer and the TCO
and the impact of the post deposition thermal treatment on the solar cell performance. Annealing
is required in order to initiate solid phase crystallization of the amorphous phase in the indium
oxide based films. Based on these results, hydrogen doped indium oxide is implemented as a
front contact in CIGS modules. The impact of the growth and properties of the indium oxide
based films on the module performance is discussed in more detail. The approaches presented in
chapter 5 to overcome the adverse effect of the CIGS roughness are applied and the performance
of the corresponding CIGS solar cells is investigated.
Chapter 7 summarizes the findings of this thesis and presents suggestions for future research.
At the end of each section the main findings are summarized. The main approaches that will be
studied in this thesis can be summarized as followed:
•Development of a large-scale deposition process with a subsequent post depo-
sition thermal treatment feasible for industry-like production
The application
of high mobility TCOs require a large scale deposition and annealing process which is
feasible for industry-like production. We therefore investigated hydrogen doped indium
oxide films which were deposited by in-line pulsed direct current magnetron sputtering
and the influences of the deposition parameters water vapor, oxygen supply as well as the
annealing conditions.
•Growth on typical sub-layers
In CIGS solar cells the front contact is deposited on
different sub-layers, typically ZnO or in an alternative configuration Zn(O,S). These layers
can influence the growth of indium oxide based TCOs, such as In
2
O
3
:H. This influence is
investigated using model structures of sub-layers deposited directly on planar glass.
•Growth on textured sub-layers/substrates
As CIGS absorber have a specific rough-
ness, the influence on the growth is investigated in detail for In
2
O
3
:H and In
2
O
3
:H,W thin
films for different deposition techniques.
•Band-line up
The successful implementation of high mobility TCOs such as In
2
O
3
:H
requires a suitable band-alignment to the sub-layer in the CIGS cell, typically intrinsic
ZnO. Therefore the valence and conduction band offsets of In2O3:H and ZnO are studied.
•Impact of thermal treatment on CIGS solar cells
In order to initiate solid phase
crystallization of the amorphous phase of indium oxide based TCOs a post deposition
thermal treatment is required, which will also influence the CIGS solar cell.
3
Introduction
•Front contact in CIGS modules
The implementation of hydrogen doped indium oxide
as front contact in CIGS modules is investigated in order to improve the short circuit
current density and thus the overall efficiency of CIGS modules.
4
CHAPTER 2
Fundamentals
2.1 Transparent Conductive Oxide
Conductive materials (e.g. metals) typically have a low optical transparency while oxide insulators
(e.g. glass) are very transparent. Transparent conductive oxides, which are wide band gap
semiconductors (E
g
> 3 eV), combine both benefits by possessing high conductivity and high
transmittance in the visible (and near infrared) optical range at the same time. Due to this
unique trade-off of the opto-electrical properties transparent conductive oxides are of high interest
for a broad field of applications, such as flat panel displays, optical coatings or photovoltaic
devices [11,12]. The most common materials used are n-type aluminum doped zinc oxide (ZnO:Al
or AZO) and tin doped indium oxide (In
2
O
3
:Sn or ITO). Over the last years high-mobility TCOs
have attracted much attention and have been investigated by several research groups, as their
opto-electrical properties are even more promising. An alternative to TCOs is graphene, which
is a single layer of carbon and shows very high electron mobility, but has no band gap [13].
2.1.1 Electrical Conductivity
The high conductivity of n-type TCOs results from degenerate doping, as the fermi level is shifted
into the conduction band, due to the very high charge carrier density (
ne
), which can be treated
as a free electron gas. Therefore, no additional activation energy is required to excite electrons
in the conduction band. According to the Mott criteria [14,15]
n1/3
deg a∗
0≈0.25 (2.1)
5
Fundamentals
the required charge carrier density to induce degeneration (
ndeg
) can be estimated. Here
a0
is
the effective Bohr radius
a∗
0=h2M0
πm∗
ee2(2.2)
where
M
is the static dielectric constant of the host lattice,
0
the vacuum permittivity,
m∗
e
the
effective electron mass and
e
the elementary charge [16]. Hamberg et al. estimated for In
2
O
3
1
a∗
0≈
1.3 nm and degeneration when
ne
>
nc≈
6x10
18
cm
−3
. As typical charge carrier densities
in indium oxide based TCOs lie in range of 10
20
cm
−3
to 10
21
cm
−3
, the films are considered to
be degenerated. The free electrons can hence be described by the Drude model [17]. Within the
free electron gas the electrons diffuse and are accelerated by an external field with
−→
E=−→
exE0.(2.3)
The equation of motion describes the interactions of a single electron in the electric field:
−e−→
E=m∗
e
d2x
dt2+m∗
e
τ
dx
dt .(2.4)
Due to collisions, e.g. with impurities in the lattice, the electrons are scattered. For the stationary
case in a constant electric field the drift velocity −→
υdcan be calculated to
−→
υd=dx
dt =eτ
m∗
e
−→
E=µe−→
E(2.5)
where τis the average time between collisions and µethe electron mobility with
µe=eτ
m∗
e
.(2.6)
Hence the ensemble of free electrons results in the current density −→
jwith
−→
j=−nee−→
υd=neeµ−→
E=σ−→
E(2.7)
where σis defined as the conductivity with
σ=1
ρ=ne2τ
m∗
e
=neeµe.(2.8)
The inverse of the conductivity is defined as the resistivity
ρ
. In technical studies commonly the
sheet resistance RSq with
RSq =ρ
d(2.9)
is used to describe the films properties with
d
as the thickness of the film. Hence, an improved
film conductivity can be realized by an increased charge carrier density or electron mobility.
Additionally the sheet resistance of the films can be reduced by an increase of the film thickness
[18,19]. The electron mobility depends on the relaxation time
τ
. With increased scattering, the
electron mobility drops. Therefore scattering mechanism are described in the following.
1m∗
e= 0.35 mefor In2O3[16]; M=(0) = 8.9
6
2.1 Transparent Conductive Oxide
2.1.2 Scattering mechanisms
As mentioned above, low resistivities can be achieved with high carrier concentrations and high
mobilities. Heavy doping can result in several adverse effects, such as the formation of scattering
centers, e.g. ionized dopant atoms, which can reduce the electron mobility. Furthermore phase
separation can occur if the impurity concentration exceeds the solubility limit. Also a high
charge carrier concentration results in the absorption of light at longer wavelengths. This effect is
called free carrier absorption and will be described in detail later. To overcome these effects and
nevertheless reduce the resistivity the electron mobility can be increased. However, the electron
mobility can be limited by several effects. The most important scattering mechanisms reported
in literature are ionized impurity scattering (
µi
), neutral impurity scattering (
µn
), scattering
at grain boundaries (
µg
) and lattice vibration scattering (
µl
). These effects are desribed in the
following, based on the explanations stated in Luque et al. [20]. However, scattering due to
further effects (addressed here as
µx
) is also possible. These aspects are for example surface
roughness scattering, as shown e.g. for films applied in n-type metal-oxide semiconductor (NMOS)
transistors [21] or scattering at the interface of amorphous and crystalline phases, as desribed for
indium oxide by Buchholtz et al. [22].
The scattering effects on the electron mobility can therefore be described by the following
equation:
1
µe
=1
µi
+1
µl
+1
µg
+1
µn
+1
µx
(2.10)
Ionized impurity scattering
In heavily doped TCOs a large number of ionized impurities can occur, such as oxygen vacancies,
dopands or excess metal atoms. Such impurities act as strong scattering centers for electrons
in degenerated semiconductors, where typically the scattering increases with increased charge
carrier density. According to the Brooks-Herring-Dingle model [23,24] the contribution to the
electron mobility µican be described with [20]
µi=3(0r)2h3
Z2(m∗
e)2e3
ne
ni
1
Fi(ξ)(2.11)
where
0
and
r
are the vacuum and relative static permittivity, respectively,
h
is Planck’s
constant, Zand niare the charge and concentration of the impurities, respectively and
Fi(ξ) = ln(1+ξ)−ξ
1+ξ(2.12)
7
Fundamentals
and
ξ= (3π2)1/30rh2n1/3
e
m∗
ee2.(2.13)
In an uncompensated semiconductor with fully ionized charge carriers the charge carrier density
is equal to the impurity density (ne=ni).
Neutral impurity scattering
Neutral impurities, such as atoms, induce scattering. The corresponding contribution to the
mobility µncan be written as
µn=m∗
ee3
200r}3nN
(2.14)
where
}
=
h
2π
and
nN
is the density of neutral centers [25,26]. In heavily doped TCOs neutral
impurity scattering is assumed to have a minor impact.
Grain boundary scattering
Grain boundaries are present in poly-crystalline materials and can be described as a quasi-two
dimensional disruption of the atomic structure which contain interface states. These states can
trap charges and result in a potential barrier that hinders electrons to pass between neighboring
grains. This becomes typically crucial, when the grain size is comparable to the mean free path
of the electrons λmfp, which is given by [19]:
λmfp =}µe
e(3π2ne)(1/3)(2.15)
and is valid for values
ne
> 1-2x10
19
cm
−3
. In TCOs these values typically are in the range
of a few nm and thus much smaller than typical grain sizes. However, for electron traps, a
depletion region forms on the sides of the grain boundary. The current transport over the barrier
can be described by thermionic emission [27]. Petritz [28] developed a model to describe the
mobility which results from grain boundary scattering. The model was extended by Seto [29]
and Baccarani et al. [30] and can be written as
µg=eL
p2πm∗
ekT exp(−φb
kT )(2.16)
where
L
is the grain size, k the Boltzmann constant, T the temperature and
φb
is the grain
boundary potential (barrier height)
φb=e2N2
T
80rN,forL ∗N > NT(2.17)
8
2.1 Transparent Conductive Oxide
φb=e2L2N
80r
,forL ∗N < NT(2.18)
where
NT
is the surface trap density. In general intergrain properties can be identified by hall
mobilities, as here the influence of the grain boundaries is included. According to the Drude
theory the electro-optical properties are connected [17]. But, however, optical mobilities do only
provide information on the intragrain properties. The amplitude of oscillating electrons under
the influence of an electromagnetic field is about 10
−7
nm for each volt per meter at visible and
near infrared frequencies, thus much smaller than typical grain sizes of TCO films. Therefore
grain boundaries are typically not encountered by the oscillating electrons. Grain boundary
scattering has typically only a minor influence in TCOs, but must still be considered.
Lattice vibration scattering
Lattice vibration scattering generally increases at higher temperatures, as the lattice vibration
increases and the lattice may deform. Thus the electron mobility can be limited by acoustic
phonon scattering. The mobility can be described by the following equation
µl=2√2πe}4Cl
2(m∗
e)5/2E2
d(kT )3/2(2.19)
with Clas the longitudinal elastic constant and Edis the deformation potential constant in eV.
Furthermore also optical phonons can lead to scattering [19].
2.1.3 Optical properties
The interaction of an electro-magnetic field (i.e., light) and matter can lead to three fundamental
wavelength (
λ
) dependent phenomena, typically taking place simultaneously: transmittance (T),
reflection (R) and absorption (R). The relation can be expressed as
1=A(λ) + T(λ) + R(λ).(2.20)
Figure 2.1 (a) shows as an example the transmittance, reflectance and absorption spectra of
a typical TCO, deposited on glass. The spectra can be categorized in 3 significant areas: I)
fundamental absorption due to excitation of electrons from the valence into the conduction band,
dependent on the optical band gap; II) optical window, typical area for use in several applications,
as the absorption is very low (typically < 10 %); III) free carrier absorption due to collective
oscillations of the free carriers.
9
Fundamentals
100
80
60
40
20
0
Transmittance, Reflection, Absorption / %
200015001000
500
Wavelength / nm
Transmittance
Absorption
Reflection
Plasma
frequency
I II III
(a)
E
k
Eg
∆Eg
BM
Eg
o
(b)
Figure 2.1: (a) Transmittance, reflectance and absorption as example of a hydrogen doped indium
oxide film after deposition on glass in dependence of the wavelength; three specific areas are marked:
I) fundamental absorption, II) optical window, III) free carrier absorption; (b) Schematic explanation of
the Burstein Moss shift
The absorption of light in a material is described by Lambert-Beer’s Law
Aint(d,λ) = 1−e−α(λ)d(2.21)
where
d
is the distance passed by the light within the medium and
α
is the specific absorption
coefficient. In degenerate semiconductors the states at the bottom of the conduction band are
filled with electrons. Thus the lowest empty states of the conduction band shift towards higher
energies. The difference of the conduction band minimum and the energy of the lowest empty
states, that can be filled with electrons from the valence band is called Burstein-Moss shift
∆EBM
g[31]:
∆EBM
g=Eo
g−Eg=}2
2m∗
e
(3π2ne)2/3(2.22)
with Eo
gas the enlarged band gap.
The interaction of matter with light can further be described by the the complex dielectric
constant
(ω) = 1(ω) + i2(ω) = (n+ik)2.(2.23)
where
ω
is the frequency,
n
as the refractive index and
k
as the extinction coefficient, which is
related to the absorption coefficient αwith
α=2kω
c.(2.24)
10
2.1 Transparent Conductive Oxide
with
c
as the speed of light. According to Pflug et al. [32] the dielectric constant in TCOs can be
expressed as
T CO(ω) = ∞+χBG +χF C (2.25)
where
∞
is a constant, the high-frequency permittivity limit.
χBG
is the contribution of the
band gap and χF C the contribution from the free carriers, which can be expressed as
χF C ≈ω2
p
ω2+iωωτ
.(2.26)
Here ωpis the plasma frequency
ωp=se2ne
0m∗(2.27)
and ωτthe damping frequency, which in first approximation can be expressed as
ωτ=1
τ=e
m∗µe
.(2.28)
To illustrate the correlation of the electrical and optical properties of TCOs, we estimated the
absorption of two TCO films with electron mobilities of 20 cm
2
/Vs and 100 cm
2
/Vs with varied
charge carrier densities according to the calculation procedure described by Luque et al. [20]. The
absorption was correlated with
EAM1.5G(λ)
the AM1.5 global solar spectrum (ASTM standard
G173-03 [33]) by the following equation [34]:
Aint.AM1.5G=R1100nm
400nm A(λ)EAM1.5G(λ)dλ
R1100nm
400nm EAM1.5G(λ)dλ
.(2.29)
The evaluation of the estimated integral Absorption
Aint.AM1.5G
over the corresponding sheet
resistance of the two TCO films is shown in Figure 2.2.
Due to the low electron mobililies (here 20 cm
2
/Vs) higher charge carrier densities are required
to achieve similar sheet resistances compared to films with higher electron mobilities (here
100 cm
2
/Vs), according to equations (2.8) and (2.9). The increased charge carrier density results
in an increased free carrier absorption in the NIR region and thus to higher integral absorption
values. The graph demonstrates the importance of high electron mobilities in transparent
conductive oxides to minimize parasitic absorption.
2.1.4 Indium oxide based high mobility TCOs
As shown in Figure 2.2 high electron mobilities are required to achieve low sheet resistances and
low optical absorption at the same time. According to equation (2.6) the electron mobility can
be improved by increasing
τ
, the time between collisions of electrons, or by reducing the effective
11
Fundamentals
Figure 2.2: Development of the integral absorption
Aint.AM1.5G
in the range 400 nm to 1100 nm over the
calculated sheet resistance for TCO films with 300 nm thickness and electron mobilities of
µe
= 20 cm
2
/Vs
and µe= 100 cm2/Vs, respectively; the carrier density was varied to achieve different sheet resistances
mass
m∗
e
. An increased
τ
can be achieved when the material exhibit a low defect density, e.g.
low carrier density (ionized impurities), few grain boundaries or few neutral defects. Reduction
of the effective mass requires semiconductors with a widely dispersed conduction band [35].
One of the most common transparent conductive oxides is ZnO:Al (AZO) due to its availability
and adequate trade-off between optical and electrical properties. However, typical electron
mobilities of sputtered ZnO:Al films with substrate temperatures below 300
°
C are in the range
of 10 cm
2
/Vs to 40 cm
2
/Vs [36
–
41]. Another common TCO is In
2
O
3
:Sn (ITO). Typical electron
mobilities are in the range of 20 cm
2
/Vs to 50 cm
2
/Vs [42
–
45] for sputtered ITO films with
substrate temperatures below 300
°
C. Higher electron mobilities were also reported for both
materials, but for special conditions, such as high deposition temperatures or specific substrates.
In recent years researchers reported about high mobility TCOs with electron mobilities above
80 cm
2
/Vs by doping indium oxide with H [46
–
48] and/or metals, such as Zr [49], W [5,7], Mo [6],
Ce [5,50] or Ti [6]. The deposition and post deposition thermal treatment, if required, were
conducted at temperatures below 300
°
C. Thus, low resistivities with low absorption can be
achieved. These materials show therefore a high potential for the application in photovoltaic
devices, such as CIGS solar cells.
In this thesis indium oxide based transparent conductive oxides are studied, in particular hydrogen
doped indium oxide (In
2
O
3
:H, or IOH) and indium oxide co-doped with hydrogen and tungsten
(In
2
O
3
:H,W or IWO:H). Thus, in the following first fundamental characteristics of undoped
In2O3are presented before the properties of In2O3:H and In2O3:H,W films are discussed.
12
2.1 Transparent Conductive Oxide
In2O3
In general In
2
O
3
crystallizes in three main polymorphic structures: (i) a high-temperature
rhombohedral corundum structure (space group R
3
c), (ii) a cubic bixbyite structure (space group
I2
1
3) and (iii) the body centered cubic (bcc) bixbyite structure (space group Ia
3
) [51]. The latter
structure is the most common and stable under standard condition. The lattice parameter is
a
=
10.117 Å [52,53]). Crystalline indium oxide films within this thesis also exhibit the bcc bixbyite
structure, which is therefore explained in more detail. Each cubic unit cell consists of 80 atoms,
which form 16 formula units. The unit cell can be derived from a 2
×
2
×
2 supercell structure of
calcium fluoride. To retain an ordered structure 25 %of the oxygen atoms are removed from the
fluorite structure, thus 48 equivalent oxygen atoms fill the edge positions Wycoff 48e. The indium
atoms are located at two different lattice positions, 8 indium atoms occupy In-b (Wycoff 8a)
while 24 indium atoms occupy In-d (Wycoff 24d) sites. The atomic positions in bcc structure are
shown in Table 2.1. The In-b position is located at the body diagonal of two oxygen vacancies,
the bonding length In-O is 2.160 Å. The In-d position is located at the face diagonal of the cube
and is surrounded by two oxygen vacancies V
O
. This results in 3 different In-O bond lengths:
2.133 Å, 2.187 Å, and 2.248 Å [53].
Table 2.1: Atomic positions in body centered cubic bixbyite structure of indium oxide [53]
Atom Site X Y Z
In-b 8a 0.2500 0.2500 0.2500
In-d 24d 0.4668 0.0000 0.2500
O 48e 0.3905 0.1529 0.3832
The fundamental (direct) band gap of In
2
O
3
is in the range of 2.6 eV to 2.9 eV and was topic
of several studies [54
–
58], as optical measurements revealed significantly higher band gaps.
Walsh et al. [59] revealed that electrical dipole transitions between the valence and conduction
bands at the
Γ
point are prohibited up to 0.8 eV below the valence band maximum. Optical
measurements therefore overestimate the band gap. The upper limit of the fundamental band
gap was determined as 2.9 eV. The optical band is in the range of 3.75 eV. The effective electron
mass m∗
eof indium oxide is 0.35me[16].
In2O3:H and In2O3:H,W
In 2007 Koida et al. [46] reported first that electron mobilities of up to 130 cm
2
/Vs can be
reached with hydrogen doped indium oxide deposited by RF sputtering. During the deposition
water vapor is introduced into the sputter chamber where water molecules H
2
O can be easily
decomposed by plasma into H and OH [60]. H and OH can form In-OH bonds at the growing
13
Fundamentals
surface with an oxygen dangling bond and an In dangling bond, respectively, thus preventing the
construction of In-O-In bond networks [61]. Films deposited at lower
pH2O
show an increased
density of crystalline nuclei compared to films deposited at increased pH2O.
A post deposition thermal treatment leads to solid phase crystallization of the film in the structure
of bcc bixbyite In
2
O
3
[46]. The presence of abundant crystalline nuclei within the as-deposited
film can lead to full crystallization at lower temperatures. Consequently the films exhibit smaller
grains [5].
During annealing structural rearrangement eliminates oxygen vacancies V
++
O
and releases H
+
.
The charge carrier density of the crystallized films is in the range of n
e
= 1-2x10
20
cm
−3
.
Limpijumnong et al. [62] reported that both interstitial hydrogen (H
+
i
) and substitutional
hydrogen (H
+
O
) act as shallow donors in In
2
O
3
and are energetically more favorable than doubly
charged oxygen vacancies V
++
O
, which are deep donors in In
2
O
3
. Similar findings were obtained
for metal dopants, such as W. The metal dopant substitutes the indium atom, resulting in
charged donors (e.g. W
+++
In
) [5,10,46]. These findings were supported by results of several
studies [63
–
65]. In fact it was found that only a minor amount of hydrogen serves as donors.
According to Koida et al. [65] free carrier generation is estimated to occur from 8 %, 17 %and
24 %of hydrogen in the post-annealed films grown at 200
°
C at
pH2O
of 5x10
−5
Pa, 1x10
−4
Pa
and 1x10
−3
Pa, respectively. This is similar to the findings of Macco et al. [64], who observed
that only
≈
4%of the hydrogen atoms in crystallized In
2
O
3
:H were active dopants. It was
found, that the dopant W did not act as an electron donor in the amorphous phase, but got
activated as donor by the crystallization process, even in the presence of hydrogen [5].
The high electron mobility of up to 130 cm
2
/Vs of crystallized films results from the suppressed
scattering of doubly charged ionized impurities and the reduced carrier scattering by H-doping.
It was found that H atoms that do not contribute to doping are not ionized or neutral impurity
scattering centers [46,64,65]. Wardenga et al. [66] further proposed hydrogen passivation at
grain boundaries as a further origin for the high electron mobility. After crystallization phonon
scattering becomes dominant and limits the electron mobility of hydrogen doped indium oxide
films [61]. Additionally tungsten doping in In
2
O
3
:H was found to produce additional carrier
scattering centers. The point defects were associated with W and H or OH species within the
In2O3:H,W films.
Following the work of Koida et al. [46] several studies reported about high mobility In
2
O
3
:H
films deposited by Atomic Layer Deposition [67,68], Reactive Plasma Deposition [5], radio
frequency [48,68
–
71] or direct current [72] magnetron sputtering. High mobility In
2
O
3
:H,W films
could be deposited by Reactive Plasma Deposition [5,7,73].
14
2.2 CIGS solar cells
2.2 CIGS solar cells
2.2.1 Basic Principles of a CIGS Solar Cell
The aim of a solar cell is to convert light into electrical energy. Its configuration can be categorized
generally into three parts: a back contact (metal layer), the main absorber layer (semiconductor)
and the window layer (several secmiconductors). In this thesis CIGS solar cells were used, the
layer configuration is sketched in Figure 2.3 (a). Typically on top of the TCO a metal contact
grid is evaporated mostly on cells for improved current collection, but, however, also causing
shading. In a CIGS solar cell these layers are typically molybdenum as the back contact and
Cu(In,Ga)(S,Se)
2
(or CIGS) as the absorber layer. The window layer comprises the buffer layer,
typically CdS, a highly resistive layer, typically intrinsic ZnO, and the front contact, which
is the transparent conductive oxide, typically ZnO:Al, or other materials, as discussed in this
thesis. The semiconductors are charged oppositely in each sides. The absorber layer is a p-type
semiconductor, thus holes are majority charge carriers and electrons are minority charge carriers.
The semiconducting components of the window layer are n-type, thus electrons are the majority
charge carriers and holes the minority charge carriers. An electrical voltage is generated when the
electrons and holes are separated and migrate to the contacts (back and front contact). When
connected to an electrical circuit, electrons flow from one contact through the circuit to the other
contact. This electron flow is the electrical current generated by a solar cell.
Absorber -
Cu(In,Ga)(S,Se)2
Glass substrate
Back contact - Mo
Buffer - CdS
HR-layer: i-ZnO
Front contact - ZnO:Al ↔ In O :H
2 3
Light
(a)
EV
EC
CdS
i-ZnO
TCO
Cu(In,Ga)(S,Se)2Mo
EF
Eg,CIGS
Eg,TCO
-
+
Space Charge Region Quasi Neutral Region
E
x
(b)
Figure 2.3: (a) Structure of a typical CIGS solar cell; (b) Simplified energy band diagram of a CIGS solar
cell in equilibrium
15
Fundamentals
To characterize a solar cell generally a voltage is applied. Operation of the solar cell in the
forward direction corresponds to a polarity of the voltage with a negative potential at the front
contact (n-range) and a positive potential at the back contact (p-range).
When the solar cell is exposed to light, photons with energy
hυ
penetrate the window layer. Pho-
tons with energies lower than the band gap of the window (
Eg
(ZnO) = 3.3 eV;
Eg
(CdS) = 2.4 eV)
are transmitted to the CIGS absorber layer. Depending on the composition of the layer, the
band gap can vary between about 1.0 eV and 1.7 eV. When the energy of a photon is absorbed,
an electron is excited from the valence band (
EV
) into the conduction band (
EC
), creating a hole
in the valence band (electron hole pair). If the energy of the photon is higher than the band gap,
the excited electron will fall back from a higher energy level to the conduction band minimum
while the excess energy (
∆E
=
hυ
-
Eg
) is converted to heat (thermalization). Typically a CIGS
absorber is therefore graded in its composition for optimized photon absorption. The electrons
and holes are separated within the space charge region, which leads to the generation of current,
otherwise the electron-hole pairs recombine. The space charge region is typically located within
the absorber layer, close to the interface window/Cu(In,Ga)(S,Se)
2
and can reach several nm.
Photons that are absorbed in the front contact layer typically do not contribute to photo-current
collection, but are recombination losses. However, it was shown, that a part of the photons
with 3.2 >
hυ
> 2.4 eV can contribute to photo-current collection, as the thin CdS layer does
not absorb all the photons. Furthermore a part of the electron hole pairs created in the buffer
layer can still contribute to the photo-current [74]. However, the main part originates from
electron hole pairs, which were photogenerated in the absorber layer [75]. The application of
wide band gap materials as window layers is therefore of great importance. Figure 2.3 (b) shows
a simplified band diagram of a CIGS solar cell as an example. The Fermi level
EF
indicates the
occupation of energy states of the bands with charge carriers in the equilibrium. However, when
the semiconductors are depicted under non-equilibrium conditions (e.g. under illumination),
electrons and holes have their own quasi Fermi level, EF n and EF p, respectively [76].
2.2.2 Solar Cell parameters
A solar cell can be in general described by a one-diode model. The equivalent circuit model is
shown in Figure 2.4. The current source indicates the photo-generated current density
jP h
, the
diode represents the dark characteristics of the junction with the corresponding current density
jDark
. The series resistance
RS
indicates internal and external ohmic losses while the parallel or
shunt resistance
RSh
indicate ohmic shunts. Ideally the series resistance is as low as possible
while the shunt resistance is as high as possible to reduce ohmic losses.
16
2.2 CIGS solar cells
The corresponding j-V characteristic can hence be written as
j=jP h −j0[exp(q(V−jRS)
AkT )−1]−V−jRS
RSh
| {z }
jDark
.(2.30)
where
j0
is the reverse saturation current density and A the diode quality factor, which is in the
easiest case
A
= 1 for back surface and neutral zone recombination, as well as for recombination
at the buffer/absorber interface and
A
= 2 for sphace charge recombination [75]. However, for
real solar cells the one-diode model might be too simple, as different recombination processes
can occur. Therefore a second diode is typically introduced parallel to the first diode (two diode
model). The two diode model was used in this thesis to simulate the obtained
j
-
V
curves and
to calculate the corresponding solar cell parameters. Another approach to describe non-ideal
solar cells is given by Scheer et al. [76]. Here in series to the diode, shunt resistance and current
source a back-contact diode (reverse direction of the main diode), a series resistance and a so
called space charge current limitor are connected. Furthermore inhomogeneities within a solar
cell can be described by a multi-diode model [77]. A solar cell can be considered as a network
of sub-cells with different electronic qualities, which are represented by the subcells’ electronic
characteristics with local inhomogeneous material qualities [78].
D
j0
jP h
RSh
jSh
RSj-
V
+
Figure 2.4: Equivalent circuit of the one-diode model of a solar cell
Figure 2.5 presents typical j-V curves. The characteristics of a solar cells are the open circuit
voltage (
Voc
), short circuit current density (
jsc
), fill factor (
FF
) and efficiency (
η
). The maximum
power point mmp is the largest product of the voltage and current and represents the point,
where the most power is produced by the solar cell.
The short circuit current density jsc is the photo-current when no voltage is applied:
jsc =jP h,V =0(2.31)
The
open circuit voltage Voc
results when the current is equal to zero. The
Voc
can be
calculated with
Voc =AkT
qln(jP h
j0
+1)(2.32)
17
Fundamentals
-40
-30
-20
-10
0
10
20
30
40
j / mA cm
-2
0.60.40.20.0
U / V
illuminated
dark
j
sc
j
MPP
V
MPP
V
oc
Figure 2.5: j-V curves of a typical CIGS solar cell (dark and illuminated); characteristic values are marked
The fill factor F F describes the quality of the hetero-junction and can be calculated with
FF =Pmpp
Vocjsc
(2.33)
It is limited by the diode quality factor, the series and shunt resistance and voltage-dependent
collection. The dependence on the series resistance (Rs) can be described with [79]
FF =FF0(1−rs)(2.34)
where FF0is the ideal fill factor without parasitic resistance and
rs=Rs
Voc/Isc
(2.35)
The
efficiency
is the ratio of the generated power (
Pel
) to the irradiation power (
Pin
) and can
be determined with
η=Pel
Pin
=jscVocF F
Pin
(2.36)
In standard test conditions the measurements are conducted at a temperature of 25
°
C and
irradiation power density of 1000 Wm−2with the global spectrum AM1.5g [33].
The external quantum efficiency represents the ratio of the number of generated carriers per
incident photon. With respect to the spectral response SR the QE is defined as
QE(λ) = SR(λ)hc
λe (2.37)
18
2.2 CIGS solar cells
where
h
is the Planck constant,
c
the speed of light and
e
is the elementary charge of an electron.
The external quantum efficiency (
EQE
) is referred when the total photon flux, which is impinging
the solar cell is taken into account. The internal quantum efficiency (
IQE
) is referred when only
the photon flux that is absorbed in the cell is taken into account and can be calculated as
IQE(λ) = EQE(λ)
1−R(λ)(2.38)
2.2.3 Module structure
In CIGS modules several solar cells are interconnected in series, realized by three scribes in the
solar cell structure (P1, P2, P3). First the molybdenum back contact layer, which is typically
sputtered on a glass substrate, is patterned by a laser, creating the P1. After the subsequent
deposition of the CIGS absorber, buffer and highly resistive layer the second scribe (P2) is
processed mechanically next to the P1. After deposition of the TCO front contact the isolation
scribe (P3) is done mechanically close to the P2. A sketch of the interconnect region of a CIGS
module with a simplified equivalent circuit is presented in Figure 2.6.
For improved visualization, the buffer and the highly resistive layer have been omitted. The solar
cells, which are connected in series (cell n and cell n+1), were assumed to be each composed
of several sub-cells, denoted with (m), to represent inhomogeneity. The P1 disconnects the
molybdenum stripes, thus a high resistance is assumed in between, represented by R
GP1
. The
connection of the front and back contact is realized by the P2, the contact resistance is noted as
R
w,BC
. The area of the three scribes does not contribute to current generation and is therefore
called "dead area". In parallel, the remaining area of the cell is addressed as "active area".
Typically a cell of the module has a width of about 5 mm. To transport the current within
the TCO without significant losses, the resistance of this layer must be very low, generally
R
T CO
< 10
Ω
/Sq. In a conventional solar cell a metal grid is evaporated on the front contact,
minimizing the distance the generated current has to flow through the TCO to the contact grid.
Therefore the sheet resistance of the TCO layer in a single solar cell can be significantly higher
(up to about 70
Ω
/Sq) than in a module. Large TCO sheet resistances in a module result in
high series resistances and can thus limit the fill factor, as discussed in section 6.2. The results
of modules obtained in this thesis are presented as cell equivalent, for improved comparison to
results of solar cells.
19
Fundamentals
RTCO j
RBC
RTCO j
RBC
j
RTCO j
j
RBC
j
RGP1
jRw,BC
Cell n Cell n+1P1 P2 P3
Active area Dead area
D
jD
jPh
Rsh
j
D
jD
jPh
Rsh
j
D
jD
jPh
Rsh
j
Subcelln, m Subcelln, m+1 Subcelln+1, m=1
TCO
CIGS
Back
contact RBC
j
Figure 2.6: Sketch of the interconnection of 2 cells of a CIGS module with the corresponding simplified
equivalent circuit
20
CHAPTER 3
Experimental Details
3.1 Sample Preparation
3.1.1 Indium Oxide Based Layer Preparation
Deposition by Magnetron Sputtering
Sputtering is a physical vapor deposition process in vacuum conditions. A sputter gas, typically
Ar due to its inertness and availability, is admitted into the sputter chamber between two
electrodes, resulting in total pressures of 0.1 Pa - 1 Pa. Reactive gases (e.g. O
2
) can be added
for a reactive process. The application of a voltage between the anode (typically chamber walls
and substrate carrier) and cathode (target) leads to ignition of a plasma. The required voltage
is in the range of a few hundred volts. The target consists of the the material that is going to
be deposited or a corresponding composition for reactive sputter processes. Due to the applied
electric field positive Ar-ions are accelerated towards the cathode, which can result in ionization
of additional Ar-atoms. When the highly energetic particles hit the surface of the sputter target,
they can eject particles from the sputter material. These particles condensate at the chamber
walls and on the substrate, which is located opposite the target. The particles can react with
ions from the reactive gas, when admitted. Apart from this, the ions from the reactive gas
can also react at the target surface, changing the surface conditions. In case of magnetron
sputtering permanent magnets are located behind the target, causing an overlay of the electric
field. Consequently electrons are forced on cycloid trajectories, increasing the ionization rate and
thereby also the sputter rate. Non-conductive or semi-conductive target materials will charge
during a direct current sputtering process, resulting in an opposing electric field regarding the
applied electric field. Thus, the sputtering process is disturbed.
21
Experimental Details
Pulsed Direct Current (DC) Magnetron Sputtering
During pulsed direct current magnetron sputtering the voltage is periodically set to zero or to a
few positive volts for a few
µ
s, also called pulse off time (
τoff
) within one complete cycle. The
pulse duration when a negative voltage is applied is called pulse on time (
τon
). The duration
of one complete cycle is set by the pulse frequency, typically in the range of a a few 10 kHz
to 100 kHz. The duty cycle represents the ratio of the pulse on time and the duration of one
complete cycle. During the pulse off time electrostatic charges are neutralized. In this thesis a
typically applied pulse frequency was 65 kHz and the pulse off time
τoff
= 3.2
µ
s, resulting in a
duty cycle of 79 %. The pulse parameters have a high impact on the deposition condition and
the process stability and must be given special attention when determining the working point.
During deposition heating of the deposited layer can be initiated controlled by a heating unit or
by process related effects, such as radiative heating from the target or bombardment of highly
energetic secondary electrons. The last mentioned effects are hard to control and can adversely
affect the growth of the corresponding films.
In this work hydrogen doped indium oxide was deposited without intentional heating by a pulsed
DC magnetron sputtering process in a large inline sputtering tool (Model A600V7) from Leybold
Optics. Indium oxide planar ceramic targets with dimensions of 12.5 cm x 60 cm, produced from
several target manufacturers were used. The power was adjusted for changing the deposition
rates, power densities of 0.8 W/cm
2
to 5.3 W/cm
2
could be applied. The total pressure was
varied in the range of 0.26 Pa to 0.51 Pa. Argon was used as the main sputtering gas. An
argon/oxygen gas mixture and water vapor, introduced by a needle valve to the sputter chamber
served as reactive gases. The corresponding partial pressures were monitored by a residual
gas analyzer (
RGA
) before and during deposition. However, cleaning of the chamber walls
and change of the
RGA
filament resulted in different water vapor pressures though the needle
valve positions remained unchanged. The substrate carrier oscillated through the plasma in
front of the target. Due to process instabilities, caused by low sputter power densities, low
sputter pressures and unfavorable pulse parameters, readjusting of the deposition parameters
was required. Therefore deposition parameters are mentioned in the corresponding sections
separately. Nevertheless a homogeneous process over a substrate size of 30 cm x 60 cm was
demonstrated. On one 30 cm x 30 cm glass substrate the deviation in film thickness was
±
2.1 %,
fitted from optical spectra, deviations of the charge carrier density and electron mobility of the
as grown films were
±
4.2 %and
±
2.1 %, respectively. After annealing the deviations decreased
to
±
2.4 %and
±
1.2 %, respectively. Over a 30 cm x 30 cm glass substrates average electron
mobilities of
≈
124 cm
2
/Vs could be achieved. In this thesis the major sputter technique was
pulsed DC magnetron sputtering. Only hydrogen doped indium oxide films, which are discussed
in section 6.1.1 were deposited by radio frequency (RF) magnetron sputtering.
22
3.1 Sample Preparation
Radio Frequency (RF) Magnetron Sputtering
During radio frequency (RF) magnetron sputtering the ions and electrons are accelerated by a
radio-frequency electric field. The electrons oscillate in the plasma and thus sustain the discharge
of the target. As the impedance is reduced with increasing frequency, any kind of material
can be sputtered by applying appropriate alternating voltages. The most common frequency is
13.56 MHz and was also used for the experiment carried out in this thesis.
In this thesis the 150 nm thick hydrogen doped indium oxide layer in section 6.1.1 was deposited
by
RF
magnetron sputtering on Mo-coated glass substrates using a 3 inch ceramic In
2
O
3
target.
The deposition process was based on the study of Steigert et al. [71].
Deposition by Reactive Plasma Deposition (RPD)
Reactive plasma deposition is considered to be a physical vapor deposition process using a pressure-
slope type plasma ion gun. Advantages are low-ion damage, low deposition temperatures, large
area deposition and high growth rates [80]. Following the work of Koida et al. [5], we deposited
hydrogen doped or hydrogen and tungsten co-doped indium oxide thin films by in-line reactive
plasma deposition (Sumitomo Heavy Industries, URT-IP2). Ar gas was injected through the
plasma ion gun into the process chamber. A beam controller focused the plasma towards the
indium oxide and 1 wt.%tungsten doped indium oxide tablets. Target material is ejected
and condensates on the moving substrate and on the chamber walls. O
2
and H
2
O were used
as reactive gases. Depending on
pH2O
the base pressure was in the range of 0.5x10
−4
Pa to
2x10
−4
Pa and the sputter pressure around 0.3 Pa, measured by quadrupole mass spectrometry
(Inficon, Transpector XPR3). The depositions were conducted at low
pH2O≈
1x10
−5
Pa or high
pH2O≈
1x10
−4
Pa without intentional heating. The temperature of the substrates did not exceed
60 °C during depositions, as determined with temperature stickers on glass and Si-substrates.
Post-Deposition Thermal Treatment
The post-deposition thermal treatment was carried out in order to initiate solid phase crystal-
lization of the amorphous phase in the indium oxide based TCOs. Annealing was performed in
three different atmospheres with different conditions:
•
The annealing was conducted in ambient air in a preheated oven from Binder. For
annealing the samples were placed on glass substrates which served as carriers. The
annealing temperature was set to 220
°
C, the annealing duration was varied between 15 min
23
Experimental Details
and 120 min. After annealing the samples were taken out and cooled down in air on a
metallic plate.
•The annealing was conducted in vacuum in different conditions.
–
The annealing was carried out at a base pressure of 1x10
−3
Pa using infrared radiant
heaters. The temperature was set by adjustment of the relative power of the heaters.
The relative power of Heaters in the middle (m) was set lower than for heaters located
at the edges (e) to yield a homogeneous temperature distribution over the whole sample
(30 cm x 30 cm area). The relative power ratio is addressed as "m/e", respectively. The
temperature profile was measured on glass substrates with the temperature logging
system SuperM.O.L.E.
r
on different positions. Two different annealing conditions
were used. (i) The relative powers were set to 60 %/ 50 %for 7 min, resulting in
an average temperature of (196
±
9)
°
C before they were reduced to 40 %/ 10 %
for 30 min for a approximately constant temperature. The measured temperature
profile revealed (221
±
0.6)
°
C after 23 min in the relevant range. (ii) The relative
powers were set to 60 %/ 50 %for 4 min, resulting in a temperature of approx.
(114
±
6)
°
C. Reduction to 30 %/ 8 %for 56 min resulted in a steady increase of
the temperature up to (211
±
1)
°
C. After annealing the chamber was floated with
nitrogen and ambient air to initiate cooling of the samples.
–
Annealing was carried out at a pressure of about 5x10
−5
Pa by placing the samples on
a hot plate. Before annealing the samples were placed on a Si-wafer, which served as
carrier. The wafer was installed in the load lock of a sputter tool from Roth&Rau and
the chamber was evacuated. The heaters in the sputter chamber were preheated and
the wafer placed on the heaters. This resulted in a constant temperature of approx.
180
°
C over the annealing duration of 1h, as indicated by temperature stickers. After
annealing the carrier transferred the Si-wafer back to the load lock, which was vented
with nitrogen to cool the samples.
–
The annealing was carried out in ultra high vacuum at 250
°
C for 40 min. The samples
cooled down in vacuum. This method was applied for the
RF
sputtered In
2
O
3
:H film
presented in section 6.1.1.
•
Film deposited by Reactive Plasma Deposition (
RPD
) were annealed in nitrogen atmosphere.
Before annealing the chamber was evacuated and vented with nitrogen up to a pressure
of
p
= 7x10
4
Pa. The samples were placed on a carbon plate, which was heated during
annealings, the temperatures were monitored by thermocouples. The temperature profile
was programmed, the heating rate was set to 20 K/min. The annealing temperature was
varied between 150
°
C and 250
°
C. The annealing duration at the target temperature was
between 30 min and 60 min. After annealing the heaters were switched off and the samples
were cooled down to room temperature inside the chamber.
24
3.1 Sample Preparation
3.1.2 CIGS Solar Cell Fabrication
CIGS Absorber
Cu(In,Ga)(S,Se)
2
(CIGS) films were used as sub-layers and as the absorber layer of photovoltaic
devices analyzed within this study. Typically the layers were grown by co-evaporation or by
a sequential process on Mo-coated soda lime glass substrates and had a thicknesses between
1.1
µ
m and 3.1
µ
m. For evaluation of the growth of amorphous indium oxide layers on rough
substrates, CIGS film were deposited directly on glass substrates without the Mo coating. The
two main fabrication processes are described in more detail.
•
CIGS layers were deposited by multi-source evaporation, also referred to as the 3-stage-
process. In different compositions all elements are evaporated separately from crucibles.
Therefore the composition and growth of the CIGS layer is controlled directly during the
evaporation process. Co-evaporated films used in this study were fabricated at Helmholtz-
Zentrum Berlin für Materialien und Energie (
HZB
), Berlin, Germany and National Institute
of Advanced Industrial Science and Technology (
AIST
), Tsukuba, Japan. A detailed
description of the 3-stage process, as well as a description of the deposition chamber at
HZB were reported by Heinemann [81]. Key aspects are the following. The Cu(In,Ga)Se
2
layer grows Cu-poor until shortly before the end of the process, where a Cu-rich growth is
required. When the composition passes the stochiometric point re-crystallization is induced,
resulting in reduced stress and crystallographic disorder. During the Cu-rich growth CuSe
x
forms at the surface, which is transformed into CIGS with supply of Ga-In-Se in the
3
rd
stage. After consumption of CuSe
x
, new nucleation rather than continued growth of
existing grains can occur, resulting in a surface layer with poorer film quality compared to
the re-crystallized grains in the bulk. Beneficial aspects of the deposition technique are the
ability to control the film composition e.g. the Ga-gradient. A post deposition treatment
(PDT) can be applied [82
–
84]. Typically alkali fluorides, such as NaF and KF are thermally
evaporated on the CIGS surface, which is aimed to improve the doping concentration of
the absorber layer. The typical thickness of the CIGS layer is about 2
µ
m. For thickness
variations of the CIGS layer the durations of the corresponding stages were reduced.
•
CIGS layers were fabricated by a sequential process. First the metallic precusors are
deposited on the Mo-coated glass substrates, typically by sputtering. This step defines
the overall composition of Cu/In/Ga of the later grown CIGS layer. Selenization and
sulfurization of the metallic precursors are conducted in a specific oven with several reaction
chambers. H
2
S or evaporated Se can be in operation as reactive gases. For Cu(In,Ga)Se
2
films the selenium supply, referred to as Se partial pressure, was found to be crucial for
the resulting elemental depth profile of the CIGS layer and the corresponding solar cell
25
Experimental Details
performance [85]. Advantages of this process are short cycle times and the possibility to
process at atmospheric pressure. At
HZB
such Cu(In,Ga)Se
2
absorber were fabricated in
an oven from Smit Thermal Solution (previously Smit Ovens) at soaking temperatures of
580
°
C. The precursor layers were sputtered. A detailed description of the process can be
found elsewhere [86]. Cu(In,Ga)(S,Se)
2
samples were also provided by Avancis GmbH &
Co KG (Avancis), a detailed description of the process can be found elsewhere [87].
Window layer
In this section the deposition of the window layers (buffer, highly resistive layer and ZnO:Al front
contact are presented. Buffer layers are aimed minimizing recombination at the absorber/window
interface by improving the band alignment at the absorber/buffer layer and interface properties.
In this thesis different buffer layers were applied.
•
CdS buffer layers were deposited by chemical bath deposition (
CBD
). A general description
of the process was summarized by Kaufmann [88]. During the CBD-process at HZB CIGS
absorbers were placed in a solution of NH
3
, Cd-acetat and Thiourea, which was heated
up to temperatures of 45
°
C for multi-source-evaporated CIGS absorbers, and 65
°
C for
sequentially processed CIGS absorbers, both processed at HZB. The resulting CdS thickness
was
≈
60 nm. CIGS samples provided by
AIST
and CIGS absorbers, which are referred
as "etched", as in sections 5.4.2 and 6.3.2, were processed with a CBD-CdS buffer layer,
following, however, a similar recipe.
•
Zn(O,S) buffer layer were deposited by Atomic Layer Deposition (
ALD
) or RF magnetron
sputtering.
–
The
ALD
deposition was conducted in a Beneq TFS500 reactor at 130
°
C. As precursors
diethylzinc (DEZ), H
2
S and H
2
O were used for zinc, sulfur and oxygen, respectively.
N
2
served as purging gas. After 9 cycles of ZnO layer depositions 1 ZnS cycle was
processed. This procedure resulted in a S/(S+O) ratio of about 25 %[89]. The target
Zn(O,S) film thickness was 50 nm.
–
The RF-sputtering process of Zn(O,S) was conducted in the in-inline sputtering
tool Von Ardenne Anlagentechnik VISS300 from a ceramic target with a ZnO:ZnS
composition of 75:25 at%. The films were deposited without intentional heating at
a total pressure of 0.5 Pa and target power density of 2 W/cm
2
for a film thickness
of 60 nm. The described sputtering process was used for CIGS absorber fabricated
at HZB. Zn(O,S) buffer layer used in cobination with provided CIGS samples from
Avancis were fabricated by sputtering in the Munich R&D pilot line from Avancis [87].
26
3.1 Sample Preparation
The following materials were applied as highly resistive (
HR
) layer, typically between the buffer
and front contact layer. However, some CIGS devices were produced without a HR layer.
•
Intrinsic zinc oxide (i-ZnO) films were used as
HR
layer and deposited by sputtering or ALD
from ceramic targets. Sputtering was operated in RF-mode. Three different deposition
tools were used. Typically the films were deposited without intentional heating at a pressure
of 0.8 Pa at the Von Ardenne Anlagentechnik VISS300. In section 5.1 ZnO films were
additionally deposited with substrate heating and at a reduced total pressure of 0.2 Pa. The
target power density was set to 2 W/cm
2
, Ar was used as the purging gas. The thickness
was varied between 40 nm and 200 nm. For indium oxide thin films that were deposited
by
RPD
, another in-line sputtering tool, located at
AIST
was used. In this the films were
deposited at approx. 140
°
C and a O
2
/(Ar + O
2
) flow ratio of 1 %[90]. The ZnO layer
discussed in section 6.1.1 was deposited using a stationary sputtering tool with a target
diameter of 3 inch. The substrate was not intentionally heated. Further ZnO films were
deposited in the ALD reactor Beneq TFS500. As for the Zn(O,S) deposition, the precursors
diethylzinc and H
2
O were used for zinc and oxygen, respectively. Also here nitrogen was
used as the purging gas. For CIGS cell a film thickness of 75 nm, for CIGS modules a
thickness of 220 nm was targeted.
•
Amorphous In-Ga-Zn-O layer were deposited without intentional heating by
RF
magnetron
sputtering from a ceramic target (In:Zn:Ga = 1:1:1). Ar and O
2
were used as sputtering
gases, the total pressure was 0.5 Pa, the power density 3.3 W/cm2[91].
•
Sol-gel Ga
x
O
y
and In
x
O
y
films were deposited by spin coating. In and Ga precursors
were prepared following the work of Zhou et al. [92]. One deposition cycle is processed as
following: substrates were coated with a few hundred
µ
l of the precursors solutions and
spin coated with a specific amount of resolutions per minute (rpm) for 1 min. Then the
substrates were transferred to a preheated hot plate and heated for 2 min at a specific
temperature. For a sample that is referred as e.g. "6xGaO
x
this deposition procedure
was repeated 6 times with Ga precursor. For depositions on glass substrates with sizes of
5 cm x 5 cm, 200
µ
l of the precursors were used. Ga precursors were spin coated with
300 rpm, In precursors with 1500 rpm. Afterwards the sol-gel layers were annealed at
150
°
C to 300
°
C. For the deposition on CdS/buffered CIGS samples 400
µ
l of the Ga
precursors were spin coated with 500 rpm and annealed at 200 °C.
Aluminum doped zinc oxide (AZO) was used as the reference front contact. The layers were
deposited by DC sputtering at different deposition conditions and sputtering tools.
27
Experimental Details
•
The films were deposited at
TS≈
160
°
C at the Von Ardenne Anlagentechnik VISS300.
Ceramic ZnO:Al
2
O
3
targets with 1.5 wt.%Al
2
O
3
were used. Typical thicknesses were
240 nm (for cells) and 865 nm (for modules).
•
The films were deposited without intentional heating in a large inline sputtering tool (Model
A600V7) from Leybold Optics from a rotatable target with 1.0 wt.%Al
2
O
3
. The films had
a thickness of approx. 240 nm.
•
The ZnO:Al front contact prepared by Avancis was deposited by sputtering and serves as a
reference [87].
Metallization, structuring and light soak
For the fabrication of solar cells a Ni-Al-Ni layered contact grid with thicknesses of 25 nm /
3000 nm / 25 nm was deposited on top of the TCO by electron beam evaporation in the system
Creamet
r
400 by Creavac. Solar cells processed at HZB had typically an area of about 1 cm
2
and were structured manually, mechanically. Solar cells processed at
AIST
had a Ni/Al contact
grid and were scribed mechanically in an automated setup for solar cell areas of 0.52 cm
2
[90].
The cells discussed in section 6.3.2 had an area of 0.35 cm2.
Modules were structured with a P1, P2 and P3, as described in section 2.2. The P1 was done by
laser scribing, P2 and P3 scribes were done mechanically. The same applies for samples produced
for transmission line measurements.
Transmission line structures were fabricated following the work of Marinkovic et al. [93]. The
samples were processed with increasing cell widths in steps of 0.2 cm from 0.5 cm to 1.5 cm. The
cell length was 1.0 cm.
Light soak of CIGS devices is common practice to evaluate the stability of samples under
continued irradiation (typically under standard test conditions). In this thesis light soak was
carried out for modules described in section 6.3.1 and 6.3.3.
3.2 Characterization Techniques
This section provides information of the used characterization methods. A short general descrip-
tion of the used methods can be found in the appendix A.1.
28
3.2 Characterization Techniques
3.2.1 Material Characterization
Film thickness
The film thickness is an important parameter to evaluate the film properties (e.g. charge carrier
density, optical absorption) and the deposition conditions (e.g. deposition rate). In this thesis
the thickness was determined by two methods. Thicknesses of sputtered hydrogen based indium
oxide films were evaluated by profilometry (DektakXT, Bruker). In this thesis the glass substrate
was marked with edding prior to the TCO deposition and removed afterwards with aceton to
create a step in the film profile. The thickness of films, which were prepared with reactive plasma
deposition, was evaluated by spectral ellipsometry. Details concerning this method can be found
in the following subsection.
Optical Characterization
Ultraviolet–Visible-Near Infrared (UV-Vis-NIR) spectroscopy
Optical properties are a key characteristic of transparent conductive oxides. In this study
the reflection (R) and transmittance (T) of thin films on glass substrates were evaluated by
UV-Vis-NIR spectroscopy, typically in the range of 250 nm to 2450 nm. The measurements were
performed using a Perkin Elmer spectrophotometer (Lambda 1050) equipped with deuterium
and halogen lamps, a monochromator and an integrating sphere. Optical absorption (A) was
calculated with
A=1−R−T(3.1)
The optical band gap of hydrogen doped indium oxide films was estimated from the relation [94]:
(αhν)2=A02(hν −Eg)(3.2)
where
α
is the absorption coefficient,
h
is Planck’s constant,
ν
is the photon’s frequency,
Eg
is
the optical band gap and A0is a constant.
Spectral Ellipsometry (SE)
In this thesis the method was used to assess the thickness of films deposited by reactive plasma
deposition on glass substrates. The measurements were done using the spectral ellipsometer
J.A.Woolam, M-2000. Prior to fitting the data, the measurements were corrected due to reflection
of the back site of the glass substrate As model SLG/TCO/EMA/Air was set. EMA represents
29
Experimental Details
the effective medium approximation, a composition of TCO 50 %: 50 %Air was assumed. The
thickness of the film was set as the sum of the TCO thickness and half of the EMA thickness.
Electrical Characterization
4 point probe
In this thesis 4 point probe measurements were done using the Model RM3-AR by Jandel. The
sheet resistance can be obtained by measurements in the von der Pauw geometry.
Hall effect measurements
In this thesis Hall measurements were conducted in van der Pauw geometry with the system
HMS-3000, Ecopia (at
HZB
) and ResiTest8300, Toyo (at
AIST
). Multi-layered stacks of the
transparent conductive oxide (n-type) and sub-layers (n and p type) with a much higher resistance
were measured. The simplifying assumption has been made that due to the big difference in
conductivity, current is only transported through the transparent conductive oxide and that the
influence of the sub-layers is negligible.
Transmission line measurements
Transmission line structures were processed to evaluate the sheet resistance of TCOs grown on
CIGS samples, as discussed in section 6.2. The procedure is described elsewhere [93].
Structural Characterization
Glow Discharge Optical Emission Spectroscopy (GDOES)
The elemental profile was measured by GDOES. Measurements were conducted with a GDA650
analyzer from Spectruma Analytik GmbH.
Transmission Electron Microscopy (TEM)
In this thesis two procedures were used. Hydrogen doped indium oxide films deposited on
ZnO/glass, as discussed in section 5.1 were prepared conventionally and analyzed with a Zeiss
LIBRA 200FE. CIGS samples, as discussed in section 5.3 were coated with 10 nm thick carbon
for improved conductivity during SEM-focused ion beam (
FIB
) and with 1
µ
m thick Pt for
protection during mechanical dimpling. A
≈
30
µ
m x 20
µ
m x 10
µ
m "trench" was cut for an
observation window. The cross-section of the CIGS sample was ion polished with decreased ion
current of 1 nA, 0.5 nA, 0.1 nA and 50
µ
A. Scanning Transmission Electron Microscopy (
STEM
)
measurements were performed with a JEOL JEM-ARM 200CF electron microscope. In order to
30
3.2 Characterization Techniques
see grain boundaries clearly, an annular dark field (ADF) detector was used to collect medium
angle annular dark field (MAADF) and annular bright field (ABF) signal for an improved defect
signal.
Scanning Electron Microscopy (SEM)
Scanning electron Microscopy measurements were conducted at acceleration voltages of 5 kV to
10 kV at a LEO Gemini 1530 or Hitachi S-4300.
Electron Backscatter Diffraction (EBSD)
Electron backscatter diffraction measurements were carried out with EDAX-TSL equipment
[95,96]
X-Ray Diffraction (XRD)
X-ray diffraction is a wide used technique for phase identification and evaluation of the structure
of (poly-)crystalline materials. Within this thesis XRD measurements were an important
characterization method to evalulate the chrystalline structure of the films. Thus the method
is described in more detail. Monochromatic X-rays are directed towards the sample, where
they interact with the sample and are scattered from a series of lattice planes. Constructive
interference occurs when Braggs Law is satisfied [95]:
nλ =2dhklsinθ (3.3)
Figure 3.1: Schematic representation of the Bragg equation; waves are reflected from lattice planes
where
n
is an integer,
λ
is the characteristic wavelength of the X-rays, which impinge the
crystallite,
dhkl
is the interplanar spacing between lattice planes and
θ
is the angle of the X-ray
beam with respect to the planes. A simplified scheme can be seen in Figure 3.1. To collect all
scattered X-ray intensities, the angle
θ
is changed continuously in a specific range (a scan is
performed). The intensities are measured with a detector, which is typically moving around the
sample. The maximum intensity of the reflected rays results when the reflected X-rays are in
phase for an specific angle
θ
(Bragg angle). The
θ/
2
θ
diffractometer is an often-used instrument
31
Experimental Details
to measure Bragg reflections. In this configuration the angles of the incoming and exiting beam
are continously varied, but remain equal throughout the whole scan (
θin =θout
). The exit angle is
2
θ
with respect to the extended incoming beam. The crystallographic lattice planes contributing
to the scattering of X-rays are all parallel to the substrate. In this thesis this measurement set-up
is referred as Bragg Brentano X-ray diffraction (
BB-XRD
). An other set-up used in this study is
grazing incidence X-ray diffraction (
GI-XRD
). In this configuration the angle of the incoming
beam is fixed, typically at very small angles and is referred as
α
, while the detecor is moving
along the 2
θ
circle. Due to this configuration the path of the x-rays within the thin film can
be maximized, leading to larger intensities than during
θ/
2
θ
scans. Further the lattice planes
that cause scattering of X-rays are not necessarily parallel to the substrate. Therefore a random
orientation of the crystallites, can be measured. However, due to the changed configuration
the Bragg reflections are slightly shifted compared to a symmetrically measured pattern. In
amorphous phases atoms are arranged in a random way and thus do not cause sharp Bragg peaks
but result in a broad feature [97]. Thus by fitting the crystalline part and/or the amorphous part
of XRD pattern the crystalline fraction XCof, here, In2O3:H films can be determined by [98]:
XC(In2O3:H) = AC(In2O3:H)
AC(In2O3:H) + AA
(3.4)
where
AC
(In
2
O
3
:H) is the sum of all integrated peaks of
In2O3
that were observed in the XRD
pattern (here
GI-XRD
) and
AA
is the total area under the amorphous components in the XRD
pattern measurements, namely the broad feature.
Other evaluation procedures of X-ray diffraction patterns are summarized in the following.
Cu K
α
(
λ
= 0.154 nm) radiation was used as the source of the X-rays. Evaluation of the
crystalline fraction and the texture coefficient was done based on measurements in
GI-XRD
configuration. Calculation of structural parameters, i.e., the lattice constants of ZnO and
In
2
O
3
:H thin films, crystallites size and strain were conducted based on measurement in
BB-XRD
configuration. Crystalline ZnO films exhibit a hexagonal wurtzite structure. The ZnO lattice
parameter c of the (002) reflex was calculated:
c=d·l=2d(3.5)
with the corresponding Miller indices
l
= 2. Crystalline indium oxide has a body centered cubic
byxbite structure with lattice parameter
a
= 10.117 Å [52,53], thus the following relationship
applies:
dhkl =a
√h2+k2+l2(3.6)
The plot of
dhkl
against the reciprocal of
√h2+k2+l2
is used to calculate the lattice constant
aof the crystallites. The slope of the linear fit represents the average value of a.
32
3.2 Characterization Techniques
The instrumental broadening was estimated by a LaB
6
standard and the integral breadth
β
was
corrected using the relation [99]:
βLf =βLh −βLg (3.7)
for the Lorentian (L) component and
β2
Gf =β2
Gh −β2
Gg (3.8)
for the Gaussian (G) component of the Voigt profile. The denotations
f,h
and
g
represent the
components of the integral breadth of intrinsic profile, experimental profile and instrumental
profile. Assuming a Lorentzian size-broadened profile (
βLf =βS
), the volume-weighted crystallite
size hDiVcan be calculated as:
hDiV=λ
βScosθ (3.9)
The weighted average strain εcan be calculated as:
˜ε=1
4βDcotθ (3.10)
if a Gaussian strain-broadened profile (βGf =βD) is assumed [99]1.
Photoelectron Spectroscopy
In this study X-ray photoelectron spectroscopy (
XPS
) and ultraviolet photoelectron spectroscopy
(
UPS
) measurements were conducted to obtain the band alignment of hydrogend doped indium
oxide and intrinsic zinc oxide.
UPS
/
XPS
measurements of gold were conducted for each sample
configuration and serve as reference for calibration. XPS spectra were corrected for background
and corrected corresponding to the Au core level 4f
7/2
at 84 eV. XPS spectra were fitted using
a linear background and voigt functions. Valence band maximum
VBM
measured by
UPS
(He I; 21.2 eV) were corrected corresponding to the Au fermi level
EF
, which was set to 0 eV,
for each sample configuration. The valence band maximum (
VBM
)
EV BM
was determined by
linear extrapolation of the leading edge towards the extended base line of the VB spectra [101],
measured at two positions of the sample.
Atomic Force Microscopy
The surface topography of sub-layers was found be be a key aspect for successful growth of high
mobility amorphous indium oxide based TCOs. The topography was evaluated by atomic force
microscopy (
AFM
) measurements. The sample is scanned with a nano-scaled tip, attached to
the end of a cantilever. The measurement signal is controlled by a laser beam, which is reflected
at the cantilever. Measurements conducted within this thesis were done using tapping mode
1
Part of the calculations were adapted from mathematical derivations published in the manuscript and supportion
infromation of Erfurt et al. [100]
33
Experimental Details
AFM. Two systems were used: 1) Park Systems XE70 with the cantilever type PPP-NCHR;
2) SII NanoNavi E-Sweep with the cantilever type SI-DF40. Measurements were done with a
resolution of 512 pixel x 512 pixel, the image size varied, in dependene of the sample and required
information, from 2
µ
m x 2
µ
m to 40
µ
m x 40
µ
m. Data was evaluated with the software
Gwyddion [102]. The grain size was estimated by the watershed method. The median local slope
was assessed by applying the integral transformation "local slope".
Fourier Transform Infrared Spectroscopy (FTIR)
Fourier-transform infrared spectroscopy (
FTIR
) was used to evaluate the amount of residual
water in spin coated Ga
x
O
y
sol-gel layers. The layers were deposited on Si-wafers. Transmission
spectra were measured under nitrogen atmosphere at room temperature from 370 cm
−1
to
7500 cm
−1
using a Bruker Tensor 27. The measured spectra were evaluated using the software
OPUS 7 [103]. The following procedures were applied: correction for atmospheric absorption,
baseline correction and transformation of the transmission spectra to absorption spectra.
Accelerated Aging
The stability of the electrical properties of the TCOs was investigated by damp heat tests, where
the relative humidity was set to 85 %and the temperature to 85
°
C. The electrical properties
were studied by hall effect measurements before and after 24 h, 48 h, 120 h, 288 h, 500 h and
1000 h of damp heat. This settings fulfill the requirements of the standard IEC 61646 (10.13).
The damp heat tests were carried out in the climate chamber WK 11 - 600/40 from Weiss
Umwelttechnik GmbH.
3.2.2 Device Characterization
Current-Voltage Characterization
The main characteristics of the solar cells (
jsc
,
Voc
,
FF
,
η
) are determined by current density -
voltage measurement (
j
-
V
) under standard test conditions (AM1.5g spectrum [33], 1000 W/m
2
,
25
°
C). To determine the diode quality factor, saturation current density, shunt and sheet
resistance the j-V curve of a cell was fitted with a two-diode model with the software IGOR
Pro [104] and the evaluation procedure PV-Evaluate, developed by Roland Mainz, HZB.
34
3.2 Characterization Techniques
Quantum Efficiency
The quantum efficiency (QE) of the solar cells was measured in the range of 300 nm to 1400 nm
using a homemade set-up with a monochromator. A LED-based white light (spectrum and
intensity close to AM1.5) generates normal operation conditions, the current is measured for
every wavelength. The EQE measurements were performed without applied bias voltage.
Capacitance Voltage Characterization
Capacitance-Voltage measurements were performed to evaluate the ionized acceptor concentration
profile of the cells. Two different setups were used: (i) a LCR meter (Agilent E4980) at a frequency
of 100 kHz with bias voltage from 0.6 to (-3.0) V; (ii) a homemade setup using an Agilent 4284A
LCR meter at a frequency of 100 kHz with voltages between 0.5 V and (-0.5) V.
35
CHAPTER 4
Sputtered Hydrogen doped Indium Oxide
Hydrogen doped indium oxide and other related compounds are promising materials due to
their high electron mobility and low absorption. However, the fabrication of e.g. hydrogen
doped indium oxide (IOH) is limited to a laboratory scale, yet. To transfer the fabrication
to the industry, a feasible process has to be demonstrated, thus e.g. by large scale inline DC
magneteron sputtering. In this chapter
1
we therefore investigate IOH films which were deposited
by such a process, namely by in-line pulsed DC magnetron sputtering with a possible deposition
are of 30 cm x 60 cm, as described in section 3.1.1 on page 22. We present the film properties
after annealing in vacuum or ambient air for 30 min at a temperature of 220
°
C, respectively,
and compare both annealing procedures, as a post deposition thermal treatment is required to
initiate solid phase crystallization of the amorphous phase of the films and results in high electron
mobilities. If not mentioned otherwise, the films had a thickness of
≈
200 nm. In particular we
investigate the structural properties in section 4.1, the electrical properties in section 4.2 and
the optical properties in section 4.3. Additionally we studied the stability of the annealed film,
as this is another critical key aspect for the successful transfer to the industry production. We
observed, that the air annealed films showed lower charge carrier densities than the films annealed
in vacuum while the optical absorption of the films were comparable. However, an annealing in
air might be more feasible for a large scale production line. We therefore developed a strategy to
improve the charge carrier density and therefore the conductivity of air annealed films without
adversely affect the electron mobility or optical absorption. The strategy is presented in section
4.5. Finally we summarize the findings in section 4.6.
1
This chapter is based on the paper Darja Erfurt, Marc D. Heinemann, Stefan Körner, Bernd Szyszka,
Reiner Klenk, Rutger Schlatmann; Improved electrical properties of pulsed DC magnetron sputtered hydrogen
doped indium oxide after annealing in air;Materials Science in Semiconductor Processing 89 (2019) 170–175;
doi: 10.1016/j.mssp.2018.09.012 [105], licensed under a Creative Commons Attribution 4.0 International License
(CC-BY)
36
4.1 Structural Properties
The investigations presented in the following are based on preliminary studies for the evaluation
of an suitable working point. In the preliminary studies a wide range of combinations of the
deposition parameter water vapor and oxygen partial pressure were processed. Values for the total
pressure and the pulse parameter were adapted from an already established In
2
O
3
:Sn deposition
process. The deposition condition of the In
2
O
3
:H film, which showed the highest electron
mobilities after annealing, were defined as the suitable working point.In this point the water
vapor pressure was
pH2O
= 0.33x10
−3
Pa and the oxygen partial pressure
pO2
= 0.88x10
−3
Pa.
In the following the structural, electrical and optical properties of films deposited around this
point are evaluated. In particular, two sets of samples were deposited: (i) the water vapor
pressure was varied while the oxygen partial pressure was held constant, (ii) the oxygen partial
pressure was varied while the water vapor pressure was held constant. Furthermore the influence
of the annealing atmosphere was studied. The detailed parameter are summarized in Table 4.1.
Furthermore an additional experiment was carried out, which will be shortly discussed later2.
Table 4.1: Deposition and annealing parameter of hydrogen doped indium oxide films, grown by a pusled
DC magnetron sputtering process on bare glass substrates
Parameter Variation
Temperature deposited without intentional heating
Power density 0.8 W/cm2
Total pressure 0.31 Pa
pH2Ovaried from 0.2x10−3Pa to 0.68x10−3Pa, constant pO2= 0.88x10−3Pa
pO2varied from 0.64x10−3Pa to 1.11x10−3Pa, constant pH2O= 0.33x10−3Pa
q(Ar &O2) 28 sccm to 48 sccm
annealing in vacuum or ambient air for 30 min at ≈220 °C
pulse f= 40 kHz; τoff = 1 µs
4.1 Structural Properties
First the structure of as grown films, deposited at a constant
pH2O
= 0.33x10
−3
Pa but varied
pO2
is presented in Figure 4.1 (a). The films all exhibit a mainly amorphous structure, indicated
by the broad feature in the range of 28
°≤
2
θ≤
35
°
. Only a small (222) peak at 2
θ≈
30.5
°
can
be assumed, indicating the presence of a few small crystallites inside the amorphous matrix. No
change of the structure can be observed due to the change of oxygen partial pressure during
2
Two films were deposited with varied water and oxygen supply: (i) very high
pH2O
of 20x10
−3
Pa without
additional oxygen supply q(Ar &O
2
) = 0 sccm; (ii)
pH2O
of 3.7x10
−3
at q(Ar &O
2
) = 25 sccm; note that the
films were deposited at a reduced total pressure of
p
= 0.26 Pa and changed configurations of the residual gas
analyzer for the quantification of pH2O, which probably resulted in an overestimation of the values
37
Sputtered Hydrogen doped Indium Oxide
deposition. However, the water vapor pressure is known to be the most crucial parameter during
sputtering for the deposition of an amorphous indium oxide film. During the deposition we varied
the water vapor pressure in the range of 0.2x10
−3
Pa
≤pH2O≤
0.68x10
−3
Pa, the oxygen partial
pressure was constant at
pO2
= 0.88x10
−3
Pa. At
pH2O
= 0.2x10
−3
Pa partial film crystallinity
was observed already after deposition by
GI-XRD
measurements, as shown in Figure 4.1 (b). In
contrast, high water vapor pressures in this range led already to fully amorphous films when
deposited by RF sputtering [65]. By increasing the water vapor pressure during pulsed DC
sputtering to
pH2O≥
0.33x10
−3
Pa the crystalline fraction can be decreased, leading to a mainly
amorphous structure of the film. The overall amourphous growth indicates the incorporation of
hydrogen during the deposition [46].
(a) pH2O= 0.33x10−3Pa (b) pO2= 0.88x10−3Pa
Figure 4.1: X-ray diffraction patterns of hydrogen doped indium oxide films in the as grown state deposited
by pulsed DC magnetron sputtering with (a) constant
pH2O
and varied
pO2
and (b) constant
pO2
and
varied
pH2O
; the In
2
O
3
reference pattern was taken from PDF 00-006-0416, patterns were shifted vertically
for improved clarity
A post deposition thermal treatment leads to solid phase crystallization of the amorphous films.
The crystallization process can be influenced by its temperature, duration and atmosphere but is
also dependent on the initial film structure. Annealing in both vacuum and air at 220
°
C for
a duration of 30 min result in solid phase crystallization of the amorphous film. In Figure 4.2
X-ray diffraction patterns of the air and vacuum annealed films, which were deposited at
pH2O
= 0.33x10
−3
Pa and
pO2
= 0.88x10
−3
Pa, are shown as examples. The observed peaks can
be assigned to cubic bixbyite In
2
O
3
structure. The peak intensity ratios of (222)/(400) and
(222)/(440) are 3.4 and 3.7 for the air annealed film and 3.4 and 3.4 for the vacuum annealed
film, respectively. The peak intensity ratio of (222)/(400) is close to the reference value of 3.3 for
38
4.1 Structural Properties
Figure 4.2: X-ray diffraction patterns of hydrogen doped indium oxide films annealed in vacuum or air for
30 min at 220
°
C; the In
2
O
3
reference pattern was taken from PDF 00-006-0416, patterns were shifted
vertically for improved clarity
both films while the peak intensity ratio of (222)/(440) is higher than the reference value of 2.9,
indicating slightly less crystal growth in (440) orientation.
Figure 4.3 shows EBSD images of air annealed samples sputtered at
pH2O
= 0.33x10
−3
Pa and
(a)
pO2
= 0.64x10
−3
Pa and (b)
pO2
= 1.11x10
−3
Pa, respectively. A randomly oriented grain
structure can be observed for both films by EBSD. The domains of the film sputtered at lower
oxygen content reach sizes up to 600 nm and only a few small domains are located in between.
In contrast the air annealed film sputtered at higher
pO2
shows a significantly higher amount of
small domains. Further pixelated areas can be observed, which have a share of almost half of the
total area.
The finding provides evidence that the film morphology of the pulsed DC magnetron films
depends strongly on the deposition parameters. As an amorphous growth could be observed
for films sputtered at
pH2O≥
0.33x10
−3
Pa, we assume that during deposition hydrogen is
incorporated in the film structure. A high oxygen flow during the deposition can lead to an
increased nano-crystallinity in sputtered IOH films [106]. The nuclei do not coalesce during
solid phase crystallization and thus prevent the formation of large grains. The pixelated areas
observed in the EBSD images can be caused by either amorphous regions or domains smaller
39
Sputtered Hydrogen doped Indium Oxide
(a) pO2= 0.64x10−3Pa (b) pO2= 1.11x10−3Pa
(c) inverse pole figure
Figure 4.3: EBSD measurements of air annealed IOH layers sputtered at constant
pH2O
= 0.33x10
−3
Pa
and varied oxygen partial pressure (a)
pO2
= 0.64x10
−3
Pa and (b)
pO2
= 1.11x10
−3
Pa; (c) inverse pole
figure; adapted from Erfurt et al. [105]
than the resolution minimum of the measurement setup. If this were caused by an amorphous
structure, however, a clear difference between the films would have to be observed by GIXRD. In
fact the measurements showed an even higher amount of crystallinity for the film sputtered at
higher oxygen supply, as shown in Figure 4.4. No amorphous hump can be observed. Thus it
can be assumed that these areas are caused by nanocrystals and presumably defect rich areas,
which form during the solid phase crystallization.
The main findings can be summarized as follows:
•
The process window for the deposition of amorphous In
2
O
3
:H seems to be more narrow for
pulsed DC magnetron sputtering than RF. magnetron sputtering, regarding the parameter
"water vapor pressure"
40
4.2 Electrical Properties
3000
2000
1000
0
Intensity / arb. units
524844403632282420
2
/ °
3000
2000
1000
0
In
2
O
3
Reference
p
O2
= 1.11 x 10
-3
Pa
p
O2
= 0.64 x 10
-3
Pa
(211)
(222)
(400)
(440)
(411)
(332)
(431)
Figure 4.4: GIXRD diffraction pattern for air annealed films deposited at
pO2
= 0.64x10
−3
Pa and
pO2
= 1.11x10
−3
Pa, respectively; Reference diffraction pattern of In
2
O
3
(Ref. 00-006-0416); adapted
from Erfurt et al. [105]
•
Films sputtered at increased oxygen supply showed a more porous surface morphology
after the annealing in air than films deposited at low oxygen content.
4.2 Electrical Properties
In the following first the electrical properties of the sputtered films before annealing are discussed.
Based on this results the properties after annealing in vacuum or air are compared.
In Figure 4.5 the charge carrier density, the electron mobility and the resistivity of the as grown
films are shown in dependence of the oxygen partial pressure (a) and water vapor pressure (b).
With increasing oxygen supply the charge carrier density was found to decrease, presumably as
a higher amount of oxygen atoms is incorporated into the films. This consequently decreases
the amount of oxygen vacancies V
++
O
, which are known to act as doubly charged donors [62].
At the same time the electron mobility increases from 41 cm
2
/Vs up to 54 cm
2
/Vs due to the
reduced impurity scattering by doubly charged oxygen vacancies. These results are consistent
with results found in literature [106]. In combination, the dominating decrease of the charge
carrier density results in an increased resistivity with higher pO2.
41
Sputtered Hydrogen doped Indium Oxide
500
450
400
350
300
/ µ
W
cm
1.2x10
-3
1.11.00.90.80.70.6
p(O
2
)
/ Pa
60
50
40
30
µ
e
/ cm
2
V
-1
s
-1
5.0x10
20
4.0
3.0
2.0
n
e
/ cm
-3
(a) pH2O= 0.33x10−3Pa
500
450
400
350
300
/ µ
W
cm
0.7x10
-3
0.60.50.40.30.2
p(H
2
O)
/ Pa
60
50
40
30
µ
e
/ cm
2
V
-1
s
-1
5.0x10
20
4.0
3.0
2.0
n
e
/ cm
-3
(b) pO2= 0.88x10−3Pa
Figure 4.5: Development of the electrical properties of as grown In
2
O
3
:H films deposited at (a) constant
pH2O
= 0.33x10
−3
Pa and varied
pO2
and (b) constant
pO2
= 0.88x10
−3
Pa and varied
pH2O
; samples
that were deposited at the same conditions are circled
With increasing
pH2O
from 0.26x10
−3
Pa to 0.68x10
−3
Pa a slight decrease of the charge carrier
density from 3.62x10
20
cm
−3
to 3.07x10
20
cm
−3
can be observed. In this range the films grow
amorphous. In contrast, the electron mobility was found to increase from 40 cm
2
/Vs to 50 cm
2
/Vs
with increase of the water vapor pressure. Consequently the resistivity reaches a minimum of
360 µΩcm at pH2O= 0.33x10−3Pa.
The electrical properties of the crystallized films are strongly correlated with the film structure,
which can be influenced by the deposition parameters oxygen partial pressure and water vapor
pressure and the subsequent annealing. For all films the charge carrier density, electron mobility
and resistivity were investigated, the results are shown in Figure 4.6 (a) and (b), respectively.
With increasing Oxygen partial pressure (
pO2
) the carrier density decreases for both annealing
atmospheres, as already observed for the as deposited films, shown in Figure 4.5 (a). However the
charge carrier density of layers annealed in air is consistently lower than that of vacuum annealed
films and the spread increases from
∆ne
= 0.12x10
20
cm
−3
to
∆ne
= 0.58x10
20
cm
−3
when
increasing
pO2
from 0.64x10
−3
Pa to 1.11x10
−3
Pa, respectively. In parallel to the decreasing
42
4.2 Electrical Properties
10
2
2
4
10
3
2
4
10
4
/ µ
W
cm
1.2x10
-3
1.11.00.90.80.70.6
p(O
2
)
/ Pa
annealed in...
air
vacuum
120
100
80
60
40
20
µ
e
/ cm
2
V
-1
s
-1
2.5x10
20
2.0
1.5
1.0
0.5
0.0
n
e
/ cm
-3
(a) pH2O= 0.33x10−3Pa
120
100
80
60
40
20
µ
e
/ cm
2
V
-1
s
-1
2.5x10
20
2.0
1.5
1.0
0.5
0.0
n
e
/ cm
-3
10
2
2
4
10
3
2
4
10
4
/ µ
W
cm
0.7x10
-3
0.60.50.40.30.2
p(H
2
O)
/ Pa
annealed in...
air
vacuum
(b) pO2= 0.88x10−3Pa
Figure 4.6: Development of the electrical properties of In
2
O
3
:H films deposited at (a) constant
pH2O
= 0.33x10
−3
Pa and varied
pO2
and (b) constant
pO2
= 0.88x10
−3
Pa and varied
pH2O
after
annealing in vacuum or air, respectively, at 220
°
C for 30 min; samples that were deposited at the same
conditions are circled; adapted from Erfurt et al. [105]
carrier density the mobility of vacuum annealed films increases slightly up to 113 cm
2
/Vs. An
increase of the electron mobility with increasing
pH2O
was already observed for the as grown
films. For films annealed in air this could not be observed. Electron mobilities of 113 cm
2
/Vs
were determined only at
pO2
= 0.88x10
−3
Pa for an annealing duration of 30 min. Here, the
electron mobility of films sputtered at lower
pO2
increased by a prolonged thermal treatment
while no improvement is observed for films sputtered at higher
pO2
for shorter or longer annealing
durations, respectively. Additionally the electron mobility of all air annealed films dropped when
prolonging the annealing time further (not shown here).
As seen in Figure 4.6 (b), the film properties are also sensitive to the water partial pressure
within the pulsed DC sputtering process. Electron mobilities of more than 100 cm
2
/Vs could
only be achieved in the range of 0.25x10
−3
Pa <
pH2O
< 0.45x10
−3
Pa after annealing in vacuum
and only for
pH2O
= 0.33x10
−3
Pa after annealing in air. For
pH2O
> 0.33x10
−3
Pa the electron
43
Sputtered Hydrogen doped Indium Oxide
mobility starts to decrease, regardless of the annealing atmospheres but with a stronger decline
for the layers annealed in air. In analogy to the as deposited films the charge carrier density of
vacuum annealed films decreased slightly with increasing
pH2O
, but dropped drastically for air
annealed films.
The findings of the present study show that the electrical properties can be influenced by the
deposition parameter water vapor and oxygen partial pressure as well as the annealing atmosphere.
By increasing the water vapor pressure during deposition it is likely that a higher amount of
In-OH and In(OH)
3
is incorporated in the film eliminating more doubly charged oxygen vacancies.
This can lead to the decreased charge carrier density, as observed in Figure 4.5 (b).
In fact an additional experiment showed that films sputtered with a very high
pH2O≈
20x10
−3
Pa
and without any additional oxygen supply (q(Ar/O
2
) = 0 sccm) had a reduced charge car-
rier density of
ne
= 3.1x10
20
cm
−3
in the as grown state. In contrast films deposited at
pH2O≈
3.7x10
−3
Pa with additional oxygen supply (q(Ar/O
2
) = 25 sccm) showed an increased
charge carrier density of
ne
= 4.2x10
20
cm
−3
. The electron mobility of both as grown films was
µe≈
43 cm
2
/Vs. Note that both films were sputtered at otherwise equal conditions, but still
different compared to the films further discussed in this chapter. This result indicates that during
deposition oxygen vacancies are eliminated and that the increased amount of hydrogen does not
necessarily act as donor. We assume that the increase of electron mobility in Figure 4.5 (b) can
be explained by the increasingly amorphous structure of the layers, as shown in Figure 4.1 (b),
as scattering at grain boundaries or at the interface of the crystalline and amorphous phase [22]
is reduced. Thus the low
µe
of films sputtered at
pH2O
= 0.2x10
−3
Pa could be explained by the
films not being completely amorphous after the deposition and before annealing.
The results suggests that during the post deposition thermal treatment in air atmospheric species
like O
2
, H
2
O or CO
2
can diffuse into the layer and act as charge carrier traps or bond the dopant
hydrogen [107] and thus lower the charge carrier density of air annealed films compared to films
annealed in vacuum. The stronger decrease of
ne
with increased oxygen partial pressure during
desposition is attributed to the differences in film morphology. The nanocrystalline structure of
layers sputtered at higher
pO2
is assumed to provide a higher amount of percolation paths at
the grain boundaries. Consequently more atmospheric species may diffuse into the layer and
thus promote the decrease of the charge carrier density. The diffused atmospheric species may
also act as impurity scattering centers or lead to an increase of the electron barrier at the grain
boundaries due to chemical modifications [107]. Both effects may explain the decreased electron
mobility of air annealed films which were sputtered at higher pO2.
Furthermore a dependence of the electrical properties on the water vapor pressure during
deposition was observed. We assume that this can be correlated with the resulting film structure.
A higher amount of H or OH can be incorporated in the films when increasing
pH2O
[46]. During
44
4.3 Optical Properties
crystallization these species can diffuse to the grain boundaries, especially interstitial hydrogen
is highly mobile [108], forming a void rich structure, as shown by Koida et al [109], or an overall
reduced film crystallinity. This leads to increased impurity scattering and decreased electron
mobility. When annealing these films in air these structures provide diffusion paths for different
atmospheric species, as described before, leading to the lower mobility and carrier density. Also
Scherg-Kurmes et al [110] predicted that a porous film structure, which forms at sputter pressures
as high as 0.6 Pa, is harmful for the electrical properties of IOH after annealing in air.
The main findings can be summarized as follows:
•
IOH films obtained lower charge carrier densities after annealing in air compared to films,
which were annealed in vacuum, presumably due to diffusion of atmospheric species during
annealing, acting as charge carrier traps.
•
The films showed similar high electron mobilities after annealing in vacuum and air only
for a specific set of deposition parameters, otherwise electron mobilities of film annealed in
vacuum were typically higher than the ones of air annealed films.
4.3 Optical Properties
In this section we compare the optical properties of a film, which showed reasonable high electron
mobilites after annealing. First the transmittance, reflectance and absorption of the film before
and after annealing in vacuum or air are compared. Second the optical band gaps of the films
are estimated and the results discussed.
Figure 4.7 presents the reflectance, transmittance and absorption spectra of the as deposited,
air or vacuum annealed
IOH
films, which were deposited at an oxygen partial pressure of
pO2
= 0.88x10
−3
Pa and a water vapor pressure of
pH2O
= 0.33x10
−3
Pa, respectively. From this
data we can see, that the transmittance increases after annealing while the absorption decreases
over the whole wavelength range. The absorption of the vacuum annealed film is higher in the
NIR
than that of the air annealed film. In the visible region no difference can be observed
between the samples. Additionally the graph shows the absorption spectra of an AZO thin film
with
d
= 550 nm,
ne
= 2.8x10
20
cm
−3
,
µe
= 25 cm
2
/Vs and
RSq
= 16
Ω
/Sq. The vacuum
annealed film showed a comparable sheet resistance of
RSq
= 14
Ω
/Sq but had a much lower
film thickness of only
d=
200
nm
. Due to its high electron mobility of
µe
= 112 cm
2
/Vs a lower
charge carrier density of only
ne
= 2.0x10
20
cm
−3
is sufficient to achieve a similar sheet resistance
at the decreased film thickness. This combination leads to the lower absorption of the vacuum
annealed IOH film compared to the AZO film over the whole spectra.
45
Sputtered Hydrogen doped Indium Oxide
Figure 4.8 illustrates the dependence between
(αhν)2
and the photon energy
hν
of the air and
vacuum annealed films. The extrapolation of the linear part of the curves onto the energy axis
indicates the optical band gap of the vacuum annealed film to be 3.79 eV and 3.73 eV for the air
annealed sample. In contrast the as grown film had a optical band gap of only 3.47 eV.
100
80
60
40
20
0
R, T, A / %
140012001000
800600400
Wavelength / nm
As deposited
Air
Vacuum
AZO
Reflection (R)
Transmittance (T)
Absorption
(A)
Figure 4.7: Reflection (R), transmittance (T) and absorption (A) spectra of
IOH
, deposited at
pO2
= 0.88x10
−3
Pa and
pH2O
= 0.33x10
−3
Pa, after annealing in vacuum or air; for comparison
the absorption spectra of an AZO film is shown
These findings highlight the beneficial aspect of the annealing on the optical properties. The
shift of the absorption edge towards lower wavelengths indicates the crystallization process. The
slightly higher absorption of the vacuum annealed films in the
NIR
can be related to the free
carrier absorption, as the charge carrier density of the vacuum annealed film is higher compared
to the air annealed film. The relationship of the optical band gap and the photon energy indicates
that the electronic transitions are direct transitions across the band gap of the films [111]. The
slightly higher optical band gap of the vacuum annealed (
∆Eg
= 0.06 eV) film can be explained
by the higher charge carrier concentration and filling of electronic states in the conduction band
(Burstein-Moss shift) [112,113]. According to the Burstein Moss equation the difference was
estimated to be 0.07 eV, which is in good agreement with the measured value.
The main findings can be summarized as follows:
•
Due to the lower charge carrier density of air annealed IOH films the parasitic absorption
in the NIR is lower than that of vacuum annealed films.
•
Due to the lower charge carrier density of air annealed IOH films the optical band gap is
lower than that of vacuum annealed films, according to the Burstein-Moss-shift.
46
4.4 Stability
Figure 4.8: Tauc plot of
IOH
film, deposited at
pO2
= 0.88x10
−3
Pa and
pH2O
= 0.33x10
−3
Pa, after
annealing in vacuum or air
4.4 Stability
The implementation of high mobility TCOs in e.g. solar cells require a certain stability of their
properties over several years and can be estimated by accelerated aging during damp heat tests.
We investigated the stability of the electrical properties of vacuum and air annealed films, which
were sputtered at high and low water vapor pressures and high/low oxygen partial pressures,
respectively. The results are presented in the following.
The changes of the electrical properties over time can be found in Figure 4.9 (a) and (b),
respectively. After 1000 h of damp heat the resistivity of the vacuum and air annealed films
sputtered at
pO2
= 0.64x10
−3
Pa increased from 243
µΩ
cm to 280
µΩ
cm (+ 15 %) and from
283
µΩ
cm to 358
µΩ
cm (+ 27 %), respectively. In contrast, the resistivity of the annealed films
sputtered at high oxygen partial pressure increased significantly by 223 %from 325
µΩ
cm to
1048 µΩcm and by 345 %from 654 µΩcm to 2913 µΩcm, respectively. The main cause is the
strong decrease of the electron mobility. For films sputtered at
pO2
= 1.11x10
−3
Pa the decrease
was observed already after 24 h. In contrast, annealed films sputtered at
pO2
= 0.64x10
−3
Pa
were stable up to 500 h of damp heat, regardless of the annealing atmosphere. The charge
carrier density was overall stable and decreased only slightly for the air annealed film which was
sputtered at pO2= 0.64x10−3Pa.
This study indicates that the differences in the film morphology of air annealed films observed
in the EBSD images can explain the corresponding poor damp heat stability of films sputtered
at higher
pO2
. A higher amount of percolation paths would promote diffusion of atmospheric
47
Sputtered Hydrogen doped Indium Oxide
10
2
10
3
10
4
10
5
10
6
/ µ
W
cm
10
2 4 6 8
100
2 4 6 8
1000
2
Time / h
0
Vacuum
Air
p(O
2
)
0.64 x10
-3
Pa
1.11 x10
-3
Pa
120
80
40
0
µ
e
/ cm
2
V
-1
s
-1
3.0x10
20
2.0
1.0
0.0
n
e
/ cm
-3
(a) pH2O= 0.33x10−3Pa
120
80
40
0
µ
e
/ cm
2
V
-1
s
-1
3.0x10
20
2.0
1.0
0.0
n
e
/ cm
-3
10
2
10
3
10
4
10
5
10
6
/ µ
W
cm
10
2 4 6 8
100
2 4 6 8
1000
2
Time / h
0
Vacuum
Air
p(H
2
O)
0.20 x10
-3
Pa
0.68 x10
-3
Pa
(b) pO2= 0.88x10−3Pa
Figure 4.9: Change of the electrical properties during damp heat tests of films annealed in vacuum or air,
deposited at (a) constant
pH2O
= 0.33x10
−3
Pa (adapted from Erfurt et al. [105]) and (b) constant
pO2
= 0.88x10−3Pa
species during the damp heat into the bulk. In contrast to the processes taking place during
air annealing, no further drop of the carrier density is observed. Nevertheless, these species can
bond hydrogen. Tohsophon et al [114] proposed the removal of H atoms at grain boundaries by
adsorbed OH-radicals which form H
2
O molecules. The formation and desorption of H
2
O would
lead to a loss of passivation at grain boundaries and to a drop of electron mobility [66,114].
Atmospheric species which already diffused into the films during the annealing in air can accelerate
this process and therefore lead to a more pronounced loss of electron mobility, as observed here
for films annealed in air compared to the films annealed in vacuum.
Similarly, a more distinct
µe
decrease was also found for the annealed films which were deposited
at higher
pH2O
, as can be seen in Figure 4.9 (b). This is consistent with the finding of Jost et
al. [107] and Koida et al. [109]. It was shown that a porous grain boundary structure, which
forms when the films contain a high amount of H, accelerate the degradation of
µe
, as they would
48
4.4 Stability
have indium hydroxide or weakly bonded structures incorporating large amounts of OH species
around the grain boundary [109]. However, this might not be the main cause for the different
degradation of films sputtered at low and high oxygen partial pressure, as here the water vapor
pressure was held constant during deposition. Thus the incorporated amount of hydrogen is
assumed to be equal.
To evaluate the stability of IOH films the results are compared to the stability of an AZO
film of
≈
290 nm thickness. The initial electrical properties before the damp heat tests were
ne
= 2.9x10
20
cm
−3
and
µe
= 18.5 cm
2
/Vs resulting in
ρ
= 1177
µΩ
cm. After 1000 h of damp
heat the charge carrier density and electron mobility decreased by 23 %and 21 %respectively,
resulting in an increase of resistivity of 65 %to 1944
µΩ
cm. The finding is consistent with
findings of past studies by Greiner et al [115], who also reported an increase in resistivity of 70 %
for 420 nm thin films, which is similar to the film thickness of the AZO film in this study. The
study of the IOH stability revealed that IOH thin films are more stable (
ρ
: + 15 %) if prepared at
low water vapor pressure and oxygen partial pressure, respectively and annealed in vacuum (here:
pH2O
= 0.33x10
−3
Pa and
pO2
= 0.64x10
−3
Pa) and still have a lower resistivity of
ρ
= 310
µΩ
cm
after 1000 h of damp heat than the investigated AZO film. All films were deposited on a smooth
glass substrate. However, it is known that the substrate roughness can influence the stability of
TCOs. As highlighted by Greiner et al [115,116] an accelerated degradation of the conductivity
of AZO thin films was found when deposited on rough substrates. Similar to these findings, also
for IOH thin films such an influence could be observed. The results are discussed in section 5.3.
The main findings can be summarized as follows:
•
Air annealed films were slightly less stable during damp heat tests than films annealed
in vacuum, presumably due to the presence of incorporated atmospheric already after
annealing.
•
Annealed films deposited at respectively high water or oxygen partial pressured showed
pronounced degradation during damp heat tests, most likely due to a porous structure.
49
Sputtered Hydrogen doped Indium Oxide
4.5 Strategies to improve the electro-optical properties after an-
nealing in air
For successful transfer to an industrial production annealing in air may be advantageous compared
to annealing in vacuum. However, the sheet resistance of the air annealed films is higher compared
to vacuum annealed films at similar weighted optical absorption values (400 nm to 1100 nm), as
can be seen in Figure 4.10 (a). This is caused by the lower carrier density of air annealed films, as
discussed before. To increase the carrier density after annealing in air we developed a bi-layered
IOH
film structure. The approx. 160 nm thick bulk-layer was sputtered at
pH2O
= 0.33x10
−3
Pa
and
pO2
= 0.88x10
−3
Pa, as here the highest mobilities were achieved after 30 min of annealing
in air. At these conditions an amorphous growth can be realized, as seen in 4.1 (b). The approx.
30 nm thick cap-layer was deposited at a reduced water vapor pressure of
pH2O
= 0.2x10
−3
Pa,
which led to a partly crystalline growth. The combined bi-layer film showed an amorphous
structure in XRD, as the influence of the partly crystalline thin cap-layer is very low compared
to the amorphous structure of the 160 nm thick bulk-layer underneath. In Figure 4.11
GI-XRD
measurements of the bi-layer film and 200 nm thick reference layers are shown. The X-ray
diffraction patterns of the reference layers were equal to the ones of the corresponding layers
from Figure 4.1.
(a) (b)
Figure 4.10: Weighted optical absorption in dependence of the sheet resistance of (a) IOH films annealed
in vacuum or air with varied oxygen supply during sputtering; (b) uncapped (solid symbols) and capped
(empty symbols) IOH films before and after annealing; the weighted optical absorption of the glass
substrate was calculated to 1.01; adapted from Erfurt et al. [105]
50
4.5 Strategies to improve the electro-optical properties after annealing in air
Intensity / arb. units
524844403632282420
2
/ °
bulk layer
In
2
O
3
Reference
bilayer
cap reference layer
(211)
(222)
(400)
(440)
Figure 4.11: X-ray diffraction patterns of hydrogen doped indium oxide films in the as grown state as a
single bulk layer, bi-layer and capped layer, respectively; the In
2
O
3
reference pattern was taken from PDF
00-006-0416, patterns are shifted for improved differentiation; adapted from Erfurt et al. [105]
Table 4.2: Electrical film properties and integral absorption (A
int,400nm−1100nm
) of capped and uncapped
IOH before and after annealing in vacuum or air; partly adapted from Erfurt et al. [105]
Annealing Sample neµeρ RSq Aint,400 nm−1100 nm
(cm−3) (cm2/Vs) (µΩcm) (Ω/Sq) (a.u.)
as deposited reference 3.7x1020 47 362 18.5 3.95
as deposited bi-layer 3.6x1020 48 368 19.3 3.74
vacuum reference 2.0x1020 115 276 14.1 2.00
vacuum bi-layer 2.0x1020 113 281 14.7 2.06
air reference 1.4x1020 109 404 20.6 1.77
air bi-layer 1.8x1020 108 320 16.8 1.92
In Figure 4.10 (b) the integral optical absorption (400 nm to 1100 nm) and sheet resistance
of the approx. 190 nm bulk reference and the bi-layer sample before and after annealing are
shown, the electrical properties and weighted absorption values are summarized in Table 4.2.
Before annealing and after annealing in vacuum, the properties of capped and uncapped films,
respectively, are very similar. However, in case of air annealing, the bi-layer film shows an
increased charge carrier density of
ne
= 1.8x10
20
cm
−3
compared to the charge carrier density of
the bulk reference film with
ne
= 1.4x10
20
cm
−3
. The electron mobilities were similar for the
bi-layer and the bulk reference film, around
µe≈
108 cm
2
/Vs to 109 cm
2
/Vs. This leads to a
significant improvement of the sheet resistance of ∆RSq = 3.8 Ω/Sq for the capped film.
51
Sputtered Hydrogen doped Indium Oxide
The results suggest that interactions of the film with atmospheric species during the annealing
in air can be counteracted by a bi-layered film structure with a partly crystalline cap layer.
An amorphous structure (uncapped bulk reference) is rich in lattice defects, distortions and
vacancies, which can also act as percolation paths [107]. By growing a partly crystalline layer
on the sample surface we can reduce the amount of percolation paths and thus the diffusion of
atmospheric species into the bulk material already prior to and during annealing. Due to the
higher amount of crystalline nuclei in the cap layer the film surface can be fully crystallized more
easily during the annealing, while the bulk can still contain an increased amount of amorphous
phase. This may lead to the observed higher charge carrier density for the bi-layer sample. As
the partly crystalline cap layer does not harm the film quality significantly, electron mobilities
over 100 cm
2
/Vs can still be reached easily. The slightly increased weighted absorption is caused
by the increased free carrier absorption in the
NIR
. Due to the still low defect density in the
film, the absorption in the visible range does not increase significantly. As interactions with
atmospheric species are negligible before annealing and after annealing in vacuum, the properties
of capped and uncapped films, respectively, are very similar. Consequently, films crystallized in
air with an increased conductivity and similar absorption values were realized.
The main findings can be summarized as follows:
•
The decreased charge carrier density of air annealed films results in higher sheet resistances
but comparable optical absorption with regard to vacuum annealed films.
•
Due to implementation of a thin cap-layer with a higher crystalline fraction after deposition
the charge carrier density of air annealed films was improved without significant losses in
optical absorption or electron mobility.
4.6 Conclusion
In this chapter we showed how the deposition conditions water and oxygen partial pressure as well
as the annealing atmosphere influence the properties of hydrogen doped indium oxide films, which
were deposited by a pulsed DC magnetron sputtering process. We observed that the process
window for the deposition of amorphous IOH is more narrow than for a RF sputtering process.
For a water vapor pressure of
pH2O
= 0.2x10
−3
Pa during deposition, already a pronounced
crystalline growth was observed within this study, while such water vapor pressures were shown
to result in mainly amorphous grown films, when deposited by RF sputtering [46]. When the
films were deposited at respectively high oxygen partial pressures, a more porous structure was
observed after annealing. We assume that this is similar for films deposited at high water vapor
pressures. The annealing in vacuum or air initiates solid phase crystallization. During annealing
52
4.6 Conclusion
in air atmospheric species might diffuse into the films and act as charge carrier traps, thus
reducing the charge carrier density of air annealed films. These films showed reduced parasitic
absorption in the near infrared region and a reduced optical band gap due to the Burstein-Moss
effect. However, a porous structure after annealing and further the presence of incorporated
atmospheric species led both to accelerated degradation during damp heat, presumably due
to enhanced diffusion of -OH at grain boundaries. The charge carrier density of air annealed
films was improved by the implementation of a thin partly crystalline cap layer on a standard
bulk film. Such a cap layer presumably counteracts the diffusion of atmospheric species during
annealing in air, as a dense crystalline layer forms accelerated at the surface.
53
CHAPTER 5
Substrate Influences on Growth Mechanism and Properties
As front contact in CIGS solar cells indium oxide based TCOs can be deposited on different
sublayers, dependent on the CIGS configuration. These layers as well as the topography of the
CIGS samples can influence the growth of indium oxide based TCOs. In this chapter these
influences are discussed separately. First the influence of layers such as sputtered Zn(O,S) and
intrinsic zinc oxide (
i-ZnO
) on planar glass substrates on the properties of IOH is identified
in section 5.1, as such layers are widely used in CIGS solar cells [117
–
119]. Alternatives to
these layers may be indium gallium oxide thin films [120]. In section 5.2 their influence on the
properties of IOH thin films is discussed when deposited on flat glass. Additionally the substrate
morphology and topography of the CIGS absorber influence the growth of the indium oxide based
TCOs. Therefore in section 5.3 influences of rough CIGS samples and textured glass substrates
are evaluated. A limitation of the electron mobility of indium oxide based TCOs was found
for several sub-layer conditions, mainly due to rough surfaces. Thus in section 5.4 strategies to
improve the electron mobility on CIGS samples are presented. Furthermore the stability of IOH
films on rough substrates is discussed in section 5.5. Lastly we summarize the findings in a chart,
see section 5.6.
5.1 Sputtered Zn(O,S) and ZnO films on glass
ZnO is a common material for the highly resistive layer in CIGS solar cells. In the last years also
alternative materials, such as Zn(O,S) or (Zn,Mg)O are getting more attention. These materials
can be applied as highly resistive or as the buffer layer. The zinc oxide based layers would the
the template for the deposition of hydrogen doped indium oxide based TCOs, when applied in
CIGS solar cells. Therefore it is important to study the combined layer stack and to identify
possible influences of the sub-layer on the TCO. As example we investigated how sputtered ZnO
54
5.1 Sputtered Zn(O,S) and ZnO films on glass
and Zn(O,S) layers affect the growth of sputtered In
2
O
3
:H. To exclude any further effects, the
films were deposited on planar glass substrates. The roughness of CIGS is a further influence
factor, the combination of both effects will be considered, in section 5.3.1. First the crystalline
structure of the Zn(O,S) and several ZnO films is studied. Based on this we investigate the
crystalline structure of In
2
O
3
:H and correlate the findings with the electrical properties of the
films. Finally we discuss the obtained results1.
For the investigations 6 different films were deposited, one Zn(O,S) film and 5 ZnO films, deposited
at varied conditions, as shown in Table 5.1. The sample ID consists of the material and the
film thickness in nm for the ZnO films (40 nm, 130 nm or 200 nm). Additionally to the ZnO
thickness variation, 130 nm films were deposited with intentional substrate heating (marked with
"T"), or with a reduced total pressure (marked with "p"). The structure of the Zn(O,S) and ZnO
films was investigated by X-ray diffraction measurements in the
θ
- 2
θ
and grazing incidence
configuration. In Figure 5.1
GI-XRD
measurements of In
2
O
3
:H on glass, Zn(O,S) and i-ZnO
are shown. For the Zn(O,S) film no assignable peak was observed, thus the film was assumed
to be amorphous. In contrast, the ZnO films were crystalline in the wurtzite type structure.
In particular, the ZnO (002) diffraction peak is much stronger than other ZnO corresponding
diffraction peaks, indicating oriented growth with the crystallographic c axis perpendicular to the
substrate surface. To evaluate the film structure of the ZnO films, the lattice parameters c and
the average domain sizes L were calculated, the results are shown in Table 5.1. The calculation
is explained in section 3.2.1 on page 31. Furthermore the lateral grain size was estimated by
AFM measurements (hDiAF M ). The calculated values are listed in Table 5.1.
Table 5.1: Deposition parameter and structural properties of sputtered Zn(O,S) and ZnO films; adapted
from Erfurt et al. [100]
Sample film thickness ptot Heating chDiVhDiAF M ˜ε
(nm) (Pa) (a.u.) (Å) (nm) (nm) (a.u.)
Zn(O,S) 60 0.5 no - - - -
ZnO40 40 0.8 no 5.2235 22.0 30.5 3.47 x10−3
ZnO130 130 0.8 no 5.2146 29.0 43 2.34 x10−3
ZnO200 200 0.8 no 5.2132 33.4 47 2.20 x10−3
ZnO130T 130 0.8 yes 5.2109 28.1 49 2.74 x10−3
ZnO130p 130 0.2 no 5.2307 21.7 33.4 3.66 x10−3
1
Reproduced in part with permission from Darja Erfurt, Marc D. Heinemann, Sebastian S. Schmidt, Stefan
Körner, Bernd Szyszka, Reiner Klenk, Rutger Schlatmann; Substrate influence on the growth of hydrogen doped
indium oxide;ACS Appl. Energy Mater. 2018, 1, 5490-5499; doi: 10.1021/acsaem.8b01039, Copyright 2018
American Chemical Society.; https://pubs.acs.org/articlesonrequest/AOR-vSC6y2YnvNhDtSbYrpi3 [100], licensed
under a Creative Commons Attribution 4.0 International License (CC-BY)
55
Substrate Influences on Growth Mechanism and Properties
Figure 5.1: X-ray diffraction patterns of as grown In
2
O
3
:H films grown on bare, Zn(O,S)-coated and
ZnO-coated planar glass substrates; Sample names and descriptions can be found in Table 5.1; Reference
pattern of In
2
O
3
and ZnO were taken from PDF 00-006-0416 and PDF 01-070-8070, respectively; patterns
were shifted vertically for improved clarity; adapted from Erfurt et al. [100]
The increase in film thickness for samples ZnO40, ZnO130 and ZnO200 results in an increase of
the (002) volume-weighted crystallite size
hDiV
and AFM grain size
hDiAF M
and slight decrease
of the lattice parameter
c
and weighted average strain
˜ε
respectively. When the substrate was
heated, the film (ZnO130T) exhibit smaller
hDiV
and lattice constant, but a larger lateral grain
size
hDiAF M
and strain
˜ε
. A decrease of the total pressure from 0.8 Pa to 0.2 Pa results in
decreased lateral and vertical grain size, but increased lattice constant and strain. These values
were furthermore the smallest/largest within this series. For all ZnO films the calculated lattice
parameter
c
is larger than the ZnO reference value
c
= 5.2066 Å [121]. The results suggest
the unit cell is elongated along the
c
axis and the films are presumably under uniform stress
with compressive components parallel to the substrate. This is in good agreement with other
reported values for sputtered ZnO films [122
–
125]. The lattice constant
a
could not be calculated
from the diffraction patterns, as the peak intensities of the corresponding peaks were too small
to be evaluated. An increase of the ZnO crystallite size perpendicular to the substrate (
hDiV
,
obtained by BB-XRD) typically correlates with an increase of the lateral grain size at the surface
56
5.1 Sputtered Zn(O,S) and ZnO films on glass
(
hDiAF M
, obtained by AFM). However, the lateral grain size is overall larger than the calculated
volume-weighted (vertical) crystallite size. For further investigations
hDiV
is used, since
hDiAF M
does not contain information on the crystalline orientation of the films, which might be an
important parameter. The weighted average strain
˜ε
was found to increase with decreased
hDiV
.
100
80
60
40
20
0
Crystalline fraction
X
c
of as deposited In
2
O
3
:H films / %
3432302826242220
ZnO (002) volume-weighted crystallite size
<D>
V
/ nm
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
(a) (b)
Figure 5.2: Dependence of crystallinity of as grown In
2
O
3
:H films on (a) ZnO (002) volume-weighted
crystallite size and (b) ZnO (002) weighted average strain; dashed line serves as a guide to the eye; adapted
from Erfurt et al. [100]
The
GI-XRD
measurements in Figure 5.1 revealed a mainly amorphous structure of In
2
O
3
:H
films grown on amorphous Zn(O,S) and glass substrates, indicated by the low In
2
O
3
peak
intensities and the broad feature in the range 28
°
< 2
θ
< 36
°
. In contrast, films grown on
poly-crystalline ZnO showed significantly higher In
2
O
3
peak intensities, indicating a higher
crystallinity. Furthermore changes in the preferred orientation were observed, which will be
discussed later. Note that the deposition conditions of the In
2
O
3
:H layers were identical and that
the sample were not annealed, yet. These results indicate that the crystalline structure of ZnO
promotes crystalline growth of In
2
O
3
:H. Generally, the formation of crystalline bonds during the
deposition of In
2
O
3
:H is suppressed by the incorporation of hydrogen, as suggested by Koida et
al. [46]. The crystalline fraction of as grown hydrogen doped indium oxide
XC(In2O3:H)
was
estimated using equation (3.4), as described in section 3.2.1. As the amorphous components in
the
GI-XRD
measurements the broad feature in the range 28
°
< 2
θ
< 36
°
was considered. Here,
it is assumed that the ZnO films are fully crystalline and have no amorphous components. The
crystallinity of In
2
O
3
:H films was found to decrease with increasing ZnO (002) volume-weighted
crystallite size and weighted average strain, as can be seen in Figure 5.2 a) and b), respectively.
In
2
O
3
:H films grown on ZnO films with the smallest
hDiV≈
22 nm and
˜ε≈
3.6x10
−3
respectively,
are almost fully crystalline as they show a crystalline fraction of approximately 90 %. The
crystalline fraction of In
2
O
3
:H was found to decrease to 50 %with increasing crystallite size of
the ZnO sub-layers. A similar trend was observed for the dependence of the crystalline fraction
57
Substrate Influences on Growth Mechanism and Properties
on the lateral grain size, as shown in Figure A.1 in the appendix. In contrast, the crystallinity of
In
2
O
3
:H films grown on amorphous Zn(O,S) was calculated to be only 10 %. A post deposition
thermal treatment led to solid phase crystallization, as shown in Figure A.2.
0.25
0.20
0.15
0.10
0.05
0.00
Texture coefficient
T
c(hkl)
/ arb. units
35302520
ZnO (002) volume-weighted crystallite size
<D>
V
/ nm
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
(222)
(400)
(440)
(a)
0.25
0.20
0.15
0.10
0.05
0.00
Texture coefficient
T
c(hkl)
/ arb. units
100
90807060504030
Crystalline fraction
X
c
of as deposited In
2
O
3
:H films / %
(222)
(440)
(400)
(b)
Figure 5.3: Texture coefficient of In
2
O
3
:H (222), (400) and (440) planes in dependence of (a) ZnO (002)
volume-weighted crystallite size and (b) crystalline fraction of In
2
O
3
:H after deposition; dashed line serves
as a guide to the eye; adapted from Erfurt et al. [100]
The degree of the preferred orientation of In
2
O
3
:H films grown on different ZnO sub-layers was
estimated by calculating the texture coefficient for In
2
O
3
planes (222), (400) and (440) by the
following equation [126]:
TC(hkl)=I(hkl)/I0(hkl)
1/N[PNI(hkl)/I0(hkl)](5.1)
where
TC(hkl)
is the texture coefficient of the plane specified by Miller Indices
(hkl)
,
I(hkl)
is
the integrated measured peak intensity,
I0(hkl)
is the corresponding reference intensity of PDF
00-006-0416 (i.e., the measured peak intensities were normalized to the corresponding reference
peak intensity) and
N
is the number of diffraction peaks. For the calculations all observed peaks,
correlated to In
2
O
3
:H were taken into account. The texture coefficient was found to depend on
the ZnO (002) volume-weighted crystallite size, as shown in Figure 5.3 (a). A decreasing ZnO
(002)
hDiV
results in a decreased texture coefficient of the (222) plane, but an increased texture
coefficient of the (400) plane. In fact, a change of the preferred orientation from (222) to (400)
was determined for the In
2
O
3
:H film grown on the ZnO film with the smallest
hDiV
(
≈
22 nm).
No significant change of
TC(440)
was observed. A similar correlation was found in dependence
on the lateral grain size estimated by AFM, as shown in Figure A.3. Assuming that the ZnO
(002) volume-weighted crystallite size and grain size of the ZnO films increase equally, these
results suggest that the growth mechanism of In
2
O
3
:H differs on ZnO grain boundaries and
grains, as the amount of grain boundaries decreases with increased grain sizes. Furthermore the
change of the texture coefficient was correlated with the crystalline fraction of the as deposited
In
2
O
3
:H films, as shown in Figure 5.3 (b). The increase in crystallinity can be attributed to
58
5.1 Sputtered Zn(O,S) and ZnO films on glass
pronounced (400) orientation of the In
2
O
3
:H crystallites, while films with lower crystallinity
show a pronounced (222) orientation. These results are consistent with the findings presented in
Figure 5.2 (a), where an increased crystalline fraction was correlated with the ZnO crystallite
size.
To evaluate the growth mechanism of indium oxide crystallites on ZnO films TEM measurements
of a cross section of sample ZnO200 (coated with In
2
O
3
:H, as deposited) were carried out. Figure
5.4 (a) presents an overview of the In
2
O
3
:H/ZnO interface. Additional images can be found in
Figure A.4 in the Supplementary. From this data we can see that ZnO grains have a columnar
structure while the In
2
O
3
:H film shows a strucutre with stacked grains, starting directly from
the ZnO surface. Figure 5.4 (b) shows two grains which are located in the In
2
O
3
:H film growing
on top of the ZnO surface at the interface of two ZnO grains, respectively. The marked area is
illustrated in Figure 5.4 (c).
The area on the lower left hand site of the image corresponds to the ZnO film and is marked
in red, the area on the upper right hand site corresponds to the In
2
O
3
:H film and is marked in
yellow, respectively. The colored boxes represent the area used for
FFT
analysis to determine the
inter-planar distances, which are labeled in white, respectively. The corresponding determined
lattice plane is written next to the box.
In the ZnO film inter-planar distances with d = (2.674
±
0.107) Å and d = (2.608
±
0.104) Å
were measured. This values are close to the reference value of the ZnO (002) lattice plane with
d
= 2.6024 Å. The (002) planes are approximately parallel to the substrate and the film surface.
Thus, the observed grains exhibit (002) orientation. Additionally a lower inter-planar distance
d
= (2.372
±
0.095) Å was measured in an area located in the ZnO film. As the measured
area is close to the surface, additional measurement spots located deeper in the ZnO film were
evaluated. The images of the evaluated region can be found in Figure A.5 in the Supplementary.
The determined inter-planar distances were
d
= (2.378
±
0.095) Å and
d
= (2.395
±
0.096) Å
are in good correlation with the value determined close to the surface. Note that the inter-planar
distance increased with larger distance to the interface. This suggests that the determined value
close to the interface corresponds to the ZnO (101) lattice plane with
d
= 2.4751 Å as this is the
most reasonable ZnO reference value.
The large deviation of the measured and reference value, which increases towards the interface,
indicates that the lattice is highly strained. In the In
2
O
3
:H film two different inter-planar distances
were determined. In the lower region of the image an inter-planar distance of
d
= (2.950
±
0.118) Å
was measured, which is close to the reference value of the In
2
O
3
:H (222) lattice plane with
d
= 2.921 Å. Furthermore the (222) lattice planes are approximately parallel to the ZnO (002)
lattice planes and the glass substrate below. Accordingly, this grain exhibits (222) orientation.
The second determined inter-planar spacing in the In
2
O
3
:H film was
d
= (2.463
±
0.099) Å. The
59
Substrate Influences on Growth Mechanism and Properties
most reasonable In
2
O
3
reference value is
d
= 2.529 Å of the In
2
O
3
(400) lattice plane. The large
deviation indicates that the lattice is strained. From Figure 5.4 (b) it becomes apparent, that the
corresponding crystallite is rather small and forms at the interface of the two ZnO (002) grains.
Additionally the grain seems to be overgrown by the indium oxide crystal with (222) orientation.
Crystalline indium oxide has a body centered cubic structure with lattice parameter
a
= 10.117 Å
[52,53]. The calculation of the lattice parameter for the presented samples is explained in section
3.2.1. For the as grown films, the lattice parameter is representative for the crystallites included in
the amorphous phase. Here only films deposited on ZnO were evaluated as the peak intensities of
films deposited on Zn(O,S) and glass were too low to be evaluated due to the mainly amorphous
structure. As the thermal treatment of the films led to solid phase crystallization of the amorphous
phase in In
2
O
3
:H, the lattice parameter of the annealed films was determined for In
2
O
3
:H films
deposited on ZnO, Zn(O,S) and bare glass due to the increased peak intensities observed for all
films. Figure 5.5 shows the dependence of the lattice constant
a
on the crystalline fraction of as
grown In
2
O
3
:H films. The determined lattice parameters of the as grown films are overall larger
than the reference value and larger for films with an increased crystalline fraction after deposition.
This indicates, that the lattice is more strained when the films exhibit a crystalline growth. The
post deposition thermal treatment led to relaxation of the lattice and to an overall reduction of
the lattice parameter. However, films with a higher crystalline fraction after deposition retained
an increased lattice parameter also after annealing. Here, the calculated lattice parameter is
lower than the reference value, indicating compressive stress.
The crystallinity of as grown In
2
O
3
:H films has a major influence on the electrical properties,
as can be seen in Figure 5.6. Here the charge carrier density (
ne
), electron mobility (
µe
) and
resistivity (
ρ
) of as grown as well as annealed In
2
O
3
:H films on bare glass, Zn(O,S) and ZnO
layers are shown in dependence on the corresponding In
2
O
3
:H crystallinity after deposition. The
increase of the crystalline fraction of as grown In
2
O
3
:H from 10 %to 90 %can be correlated
with a loss in charge carrier density from 4.7x10
20
cm
−3
to 1.7x10
20
cm
−3
and a loss in electron
mobility from 36 cm
2
/Vs to 24 cm
2
/Vs. Thus the specific resistivity increases fourfold. After the
annealing and solid phase crystallization the charge carrier density decreased while the electron
mobility increased. For the mainly amorphous films after deposition, the charge carrier density
was found to decrease to half of the initial value. In contrast, crystalline grown films showed a
less pronounced drop in charge carrier density after annealing. The electron mobilities of the
annealed films were in the range of 80 cm
2
/Vs to 90 cm
2
/Vs. Here, a dependence on the lattice
parameter of the annealed films was found. The results are shown in Figure 5.7. A higher (lower)
electron mobility results for films with lower (higher) lattice parameter.
First, the structure of the ZnO films is discussed. It was shown in literature that ZnO film growth
starts with a bottom layer of small grains. The grains coalesce at increased thicknesses, when
60
5.1 Sputtered Zn(O,S) and ZnO films on glass
ZnO
In O :H
2 3
glass
(a)
ZnO
In O :H
2 3
glass↓
(b)
5 nm
→
→
→
→
→
→
→
→
→
→
In O :H
2 3
ZnO
(400)
(222)
(002)
(002) (101)
2.372Å
2.608Å
2.674Å
2.950Å
2.463Å
↓
glass
(c)
Figure 5.4: TEM images of the In
2
O
3
:H /ZnO interface of sample ZnO200 in the as grown state;
(a) overview of the sample structure; (b) grains located in the In
2
O
3
:H film are observed on the surface of
ZnO grains; (c) zoom in of the white marked area in (b) with measured inter-planar distances and the
corresponding crystalline assignment, colored boxes represent the area taken for FFT analysis; the glass
substrate is indicated in white; adapted from Erfurt et al. [100]
61
Substrate Influences on Growth Mechanism and Properties
10.35
10.30
10.25
10.20
10.15
10.10
10.05
Lattice parameter
a
of In
2
O
3
:H / Å
100
80604020
0
Crystalline fraction
X
c
of as deposited In
2
O
3
:H films / %
glass
ZnO
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
bcc - In
2
O
3
Reference
as deposited
annealed
Figure 5.5: Dependence of the lattice constant of as grown and annealed In
2
O
3
:H crystallites, respectively
on the crystallinity of the films in the as grown state; as the crystalline fraction of In
2
O
3
:H films grown
on Zn(O,S) and bare glass was too low in the as grown state, the lattice parameter of these films could
not be calculated; adapted from Erfurt et al. [100]
the surface temperature is high enough, leading to increased grain sizes (taperwise growth) [127].
Also for the films described in this study we observed an increase of the vertical and lateral grain
size with increased film thickness, indicating taperwise growth. The coalescence of grains seems
to results in a reduction of strain. An increase of the substrate temperature during deposition
improves the adatom diffusion at the surface. This might lead to enhanced coalescence at lower
thicknesses and thus larger lateral and smaller vertical grain sizes at the bottom of the film
towards the glass substrates, as
hDiV
is an average value. In literature a lower sputter pressure
is predicted to result in denser films, since the arriving particles on the substrate have a higher
energy and thus a higher surface mobility with the possibility the form larger grains [128,129].
However, Bachari et al. [123] showed, that pressures as low as 0.25 Pa result in a decreased
grain size compared to films deposited at 1 Pa. The authors suggest, that at low total pressures,
combined with low oxygen pressure and high RF sputter power, the high energy particles produce
recoil implantation when arriving at the substrate surface, resulting in higher compressive stress.
These findings might explain the results observed for ZnO films sputtered at a reduced total
pressure of 0.2 Pa. Here a reduced lateral and vertical grain size was determined. At the same
time the films exhibit the highest compressive strain parallel to the substrate. The strain was
estimated from the lattice parameter c of the films, which was higher than the reference value. It
is unlikely that this difference is caused by a high error during the BB-XRD measurements, but
cannot be excluded completely.
Next, we will discuss the results of In
2
O
3
:H films grown on bare, Zn(O,S) and ZnO - coated
planar glass substrates, respectively. It is known that the amorphous growth of In
2
O
3
:H is based
62
5.1 Sputtered Zn(O,S) and ZnO films on glass
1600
1200
800
400
/ µ
W
cm
100
80604020
0
Crystalline fraction
X
c
of as deposited In
2
O
3
:H films / %
glass
Zn(O,S)
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
as deposited
annealed
100
80
60
40
20
0
µ
e
/ cm
2
V
-1
s
-1
5.0x10
20
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
n
e
/ cm
-3
Figure 5.6: Electrical properties of as grown and annealed In
2
O
3
:H films in dependence on the crystalline
fraction after deposition; adapted from Erfurt et al. [100]
on the incorporation of hydrogen during the film deposition [46]. Koida et al. suggested that
oxygen-hydrogen bonds (O-H) disturb the formation of (O-In-O) bonds and thus the formation
of the crystalline network [61]. As a mainly amorphous growth was observed on substrates
with an amorphous structure like glass and Zn(O,S), the amount of hydrogen offered during the
deposition process was in principle sufficient to suppress crystalline growth on these substrates.
Furthermore the water vapor pressure is in good agreement with values found in literature
which resulted in a mainly amorphous growth of In
2
O
3
:H [5,46]. Nevertheless, an increased
film crystallinity occured for as grown In
2
O
3
:H films deposited on poly-crystalline ZnO. Thus
the growth of In
2
O
3
:H films was influenced by the sub-layer. The findings highly indicate that
the ZnO structure determines the structure of In
2
O
3
:H. We assume, that no or only negligable
amounts of hydroged diffused through grain boundaries into the ZnO during In
2
O
3
:H deposition.
It was found that during In
2
O
3
:H deposition ZnO with (002) preferred orientation promotes the
63
Substrate Influences on Growth Mechanism and Properties
90
88
86
84
82
80
µ
e
of annealed In
2
O
3
:H / cm
2
V
-1
s
-1
10.1610.1410.1210.1010.08
Lattice parameter
a
of annealed In
2
O
3
:H / Å
glass
Zn(O,S)
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
Figure 5.7: Influence of the lattice parameter
a
of annealed In
2
O
3
:H films on the resulting electron
mobility; adapted from Erfurt et al. [100]
growth of In
2
O
3
:H crystallites with a (222) preferred orientation, especially when the ZnO grain
size increases. Thus we suggest that the two structures intergrow along the hexagonal c-axis
direction, as proposed by Cannard and Tilley [130]. They showed that two
{
111
}
planes of the
cubic In
2
O
3
structure form on
{
001
}
ZnO planes in ZnO-In
2
O
3
systems. Each of the two
{
111
}
planes would thus have a composition of InO
1.5
. The planes most densely packed with oxygen
atoms are parallel to (111)
In2O3
and (001)
ZnO
and the average oxygen-oxygen distance weighted
by number/unit for (111)
In2O3
was calculated to be 0.33534 nm while it was 0.32498 nm for
(001)ZnO, resulting in a misfit of 3 %[131].
Thus (002)-textured ZnO films can provide a suitable template for the growth of (222) oriented
In
2
O
3
films. This finding was further confirmed by other research groups [132
–
134]. However,
we observed a change of the preferred orientation of as grown In
2
O
3
:H from (222) to (400) when
grown on ZnO with smaller crystallites and (002) preferred orientation. We assume that the
decreased crystallite size results from an increased density of ZnO grain boundaries. Grain
boundaries are regions which can be distorted, contain different coordination of atoms and a
higher density of lattice defects. This may lead to change in the average oxygen-oxygen distance
of (001)
ZnO
which may promote In
2
O
3
:H crystal growth in other orientations like (400), as
observed by
GI-XRD
. In literature a more pronounced (400) layer growth could be shown for
low oxygen conditions during deposition [134
–
137]. In contrast, Keller et al. [47] did not observe
a correlation of the oxygen content and texture of films deposited by
ALD
. Nevertheless an
increased (400) orientation was demonstrated for increased deposition temperatures. The authors
suggest that the films nucleate in (222) orientation, but that the orientation changes to (400)
with increased film thickness. This might occur, when the grains get in touch. In regard of the
observed findings in this work, such an effect could enhance the formation of (400) crystallites
64
5.1 Sputtered Zn(O,S) and ZnO films on glass
during In
2
O
3
:H deposition at increased thicknesses, if we assume that a higher amount of (222)
nuclei form at the interface to ZnO films with smaller grains, resulting in a higher density of
grains which get in touch and change the orientation. To overcome the crystalline growth of
In
2
O
3
:H on poly crystalline ZnO layers, the amount of hydrogen during the deposition must be
increased to facilitate the incorporation during the growth of the interlayer at the ZnO surface,
as observed in additional experiments (not shown here). Also an increased deposition rate may
be generally beneficial to prevent crystalline growth.
Hydrogen was proposed to act as a shallow donor in In
2
O
3
:H [62,63,138] and thus to increase
the carrier density compared with undoped films. Further, previous studies have suggested that
that the source of doping is attributed to interstitial hydrogen (H
+
i
) or substitutional hydrogen
(H
+
O
) rather than to doubly charged oxygen vacancies (V
++
O
) [63
–
65], as discussed in section 2.1.4.
Wu et al. [139] reported that in In
2
O
3
:H films that contained both, amorphous and crystalline
parts, a higher amount of hydrogen was incorporated in the amorphous region compared to the
crystalline region. We therefore assume that after deposition a lower amount of hydrogen is
incorporated in the crystalline grown In
2
O
3
:H films. The charge carrier density of these films
must consequently be lower compared to the amorphous grown film, if only H contributes to
doping and the doping efficiency remains unchanged. This is supported by the measurement
results of the charge carrier density, as shown in Figure 5.6. Furthermore, the electron mobility
of as grown and annealed In
2
O
3
:H films was found to decrease with increased crystalline fraction
of the In
2
O
3
:H films before annealing. In transparent conductive oxides, the electron mobility is
limited by scattering processes which can be induced by phonons, ionized or neutral impurities
and in polycrystalline material also by grain boundaries. In In
2
O
3
:H compounds that contain
both, amorphous and crystalline phases, the incoherent boundaries in-between can additionally
act as scattering centers [22]. When first crystallites form in a mainly amorphous phase, they can
act as scattering centers and thus lower the electron mobility. An increased amount of crystallites
will lead to further decrease of the electron mobility until the crystalline phase becomes the
dominating phase. The results presented within this thesis show that the electron mobility of as
grown In
2
O
3
:H decreases with increased crystallinity, similar to the model proposed by Buchholz
et al. [22]. However, a reduced electron mobility was still observed in films with crystallinity
as high as 90 %. The estimated lattice constant of these films was significantly higher than
for films with lower crystallinity after deposition and higher than the In
2
O
3
reference value.
This suggests, that especially the crystalline grown films are strained and the crystals may
contain defects and impurities which serve as scattering centers resulting in a decreased electron
mobility [46,65]. Additional scattering may occur at the grain boundaries. In crystalline In
2
O
3
:H
films, hydrogen that does not contribute to doping is proposed to passivate grain boundaries
thus lowering the transport barrier for electrons [66] or passivating microscopic and macroscopic
defects, respectively [65]. Both effects result in general in an increased electron mobility. We
65
Substrate Influences on Growth Mechanism and Properties
assume that less hydrogen is incorporated in the crystalline grown In
2
O
3
:H films. In this case
the scattering at grain boundaries is most likely even more pronounced.
A thermal treatment led to solid phase crystallization of the amorphous phase in In
2
O
3
:H. The
determined lattice constant after annealing was found to be lower than that of the crystallites
present after deposition. Thus we assume that strain in the crystalline grown films could be
reduced during annealing, but is still higher than in amorphous grown films, as they exhibit a
larger lattice parameter a. This determines the electron mobility, which decreases with increased
lattice parameter. Therefore we assume that the remaining strain of the initially crystalline
grown films limits the electron mobility of annealed films. However, electron mobilities of over
80 cm
2
/Vs could still be achieved for films grown on ZnO. Nevertheless the reduced charge carrier
density of these films leads to a twofold increased resistivity.
These findings enhance our understanding of the growth of hydrogen doped indium oxide on
crystalline ZnO and amorphous Zn(O,S) layers. The results of this investigation indicate the
need of an amorphous layer underneath hydrogen doped indium oxide for improved electrical
properties.
The main findings can be summarized as follows:
•
Poly-crystalline ZnO promotes crystalline growth of hydrogen doped indium oxide; in
contrast films grown on amorphous Zn(O,S) were mainly amorphous.
•
The In
2
O
3
:H films had an higher crystalline fraction (up to 90 %) when grown on small
and strained ZnO grains.
•
The preferred orientation of In
2
O
3
:H changed from (222) to (400) in films that were
deposited on small ZnO grains.
•
Crystalline grown In
2
O
3
:H films were strained and showed poor electrical properties before
and after annealing compared to the mainly amorphous grown films.
•
The crystalline growth can be suppressed by an increased hydrogen supply during the
deposition of the first In2O3:H layers at the ZnO interface.
66
5.2 Sol-Gel Indium - and Gallium Oxide Layers on glass
5.2 Sol-Gel Indium - and Gallium Oxide Layers on glass
As alternative to the high resistant intrinsic ZnO layer, sol-gel indium and gallium oxide layers
were studied. Researchers showed that such layers can be used as buffer layers in CIGS solar cells.
Koida et al. [120] analyzed CIGS solar cells in substrate coordination with sputtered amorphous
indium gallium oxide (
a-In2−2xGa2xO3
) as buffer layer. The solar cells with x = 0.9 and x = 1
exhibited V
oc
values comparable with that of a reference cell with standard i-ZnO/CdS layer stack.
In addition, Heinemann et al. [140] showed that low temperature deposited amorphous Ga
x
O
y
and alloys with amorphous In
x
O
y
were suitable buffer layer materials for CIGS solar cells in
superstrate coordination. Related to this thesis, these amorphous materials could give a suitable
template to achieve amorphous grown IOH thin films with high mobilities after post-annealing.
Additionally the materials exhibit a higher band gap than the conventional i-ZnO or CdS layer,
thus showing a high potential for reduced parasitic absorption when applied in the window layer
of solar cells. It was shown, that a raise of the short circuit current density could be achieved
when substituting a low band gap material (e.g. CdS) with a higher band gap material (e.g.
Zn(O,S)) [141]. Therefore in this section we investigate the properties of pulsed DC magnetron
sputtered IOH films grown on indium oxide and gallium oxide layers which were deposited by a
sol-gel spin coating process, as described in section 3.1.2 on page 27. As discussed in section 5.4
such a deposition process can be beneficial for the electron mobility of IOH on CIGS samples.
For an improved understanding the sol-gel layers were first investigated in dependence of the
spin coat deposition conditions on planar glass substrates. In this case 15 repetitions of the spin
coat process were done. Based on this studies IOH thin films were deposited on several sol-gel
layers and their structure and electrical properties were investigated before and after annealing.
The structure of the spin coated sol-gel films could be highly influenced by the annealing
temperature during the deposition routine. In Figure 5.8 X-ray diffraction patterns are shown
for sol-gel indium oxide (left) and gallium oxide (right) layers which were annealed at different
temperatures, respectively. From this data we can see that the crystallinity of the indium
oxide films increases with higher annealing temperatures. At 300
°
C the films were found to
be crystalline in the bcc In
x
O
y
bixbyte structure. The films which were annealed at 200
°
C
and 250
°
C, respectively, showed a hump in the range of 27
°≤
2
θ≤
37
°
, indicating increased
nano-crystallinity. In contrast no peak was observed for sol-gel gallium oxide films. Thus these
films were amorphous regardless of the annealing temperature. The broad peak in the range
around 15°≤2θ≤30°is assumed to be caused by the quartz glass substrate.
Due to the sol-gel deposition procedure water is incorporated into the indium and gallium oxide
layers. During annealing H
2
O desorbs, leading to a reduced film thickness. This process can be
accelerated by higher temperature. To investigate the residual amount of water in the sol-gel
layers
FTIR
measurements were done on sol-gel Ga
x
O
y
layers, which were deopsited on a Si-wafer
67
Substrate Influences on Growth Mechanism and Properties
(a) (b)
Figure 5.8: X-ray diffraction patterns of indium oxide (a) and gallium oxide (b) layers deposited by a
sol-gel procedure and annealed at different temperatures; the In
x
O
y
reference pattern was taken from
PDF 00-006-0416; patterns were shifted for improved differentiation
(5 repetitions of the spin coat process) and annealed at 150
°
C and 250
°
C, respectively. A bare
Si-wafer served as reference. The results are shown in Figure 5.9. The insert represents a close
up of the marked area. In the range 3200 cm
−1
to 3650 cm
−1
a broad peak can be observed for
the Ga
x
O
y
layers. The peak intensity is significantly higher for the film which was annealed at
150
°
C than for the film annealed at 250
°
C. This peak can be correlated to stretching modes of
hydrogen bonded OH groups [142]. In contrast in the mentioned region no peak was observed
for the bare silicon wafer. Consequently the sol-gel layers contain residual water and the amount
decreases with increase in annealing temperature.
The purpose of this study was to evaluate a suitable sol-gel layer as a template for an amorphous
grown IOH film with reasonable electrical properties. Therefore the growth of
≈
370 nm thick
sputtered IOH films was investigated by X-ray diffraction, as shown in Figure 5.10. The films
were deposited on sol-gel gallium oxide and indium oxide layers (5 repetitions of the spin coat
process), which were annealed at 150
°
C and 200
°
C, respectively. For these sol-gel deposition
conditions a mainly amorphous film structure was found. Additionally an IOH reference layer
was deposited on quartz glass. While the IOH films grown on gallium oxide and quartz glass
were overall amorphous in XRD, an increased crystallinity was observed for IOH films grown on
sol-gel indium oxide layers. The crystalline fraction of IOH increased with annealing temperature
of the indium oxide sol-gel layers.
Figure 5.11 presents the change of the charge carrier density and electron mobility of IOH films
before and after 1 h and 2 h of annealing at 160
°
C in vacuum, respectively. The IOH layers were
deposited on glass substrates, sol-gel Ga
x
O
y
and In
x
O
y
layers, which were annealed at 150
°
C
68
5.2 Sol-Gel Indium - and Gallium Oxide Layers on glass
2.0x10
-2
1.5
1.0
0.5
0.0
Absorption / arb. units
3800 3600 3400 3200 3000
Wavenumber / cm
-1
0.25
0.20
0.15
0.10
0.05
0.00
Absorption / arb. units
4000 3500 3000 2500 2000 1500 1000 500
Wavenumber / cm
-1
Ga
x
O
y
- 150°C
Ga
x
O
y
- 250°C
Si-wafer
Figure 5.9: FTIR spectra of sol-gel Ga
x
O
y
layers which were annealed at 150
°
C and 200
°
C, respectively
and a reference Si-wafer; the insert represents a close up of the marked area
and 200
°
C, respectively. Before annealing (
t
= 0 h) the charge carrier density was in the range of
3.0x10
20
cm
−3
to 3.3x10
20
cm
−3
, except for the sample deposited on In
x
O
y
- 200
°
C. Here a lower
charge carrier density of
ne
= 2.2x10
20
cm
−3
was observed. A similar trend was found for the
electron mobility. It was 32.9 cm
2
/Vs for the IOH layer deposited on In
x
O
y
- 200
°
C, significantly
lower than for the other IOH layers with 40 cm
2
/Vs
≤µe≤
44.5 cm
2
/Vs. During annealing the
charge carrier density decreases for all samples. After 2 h of annealing the charge carrier density
of IOH films deposited on both sol-gel In
x
O
y
layers and on Ga
x
O
y
- 150
°
C was in the order of
magnitude of 10
19
cm
−3
. In contrast the IOH films deposited on Ga
x
O
y
- 200
°
C and bare glass
substrate had higher charge carrier densities of
ne
= 1.5x10
20
cm
−3
and
ne
= 2.4x10
20
cm
−3
,
respectively. Here, a similar trend can be observed for the electron mobility. A significant drop
to values of 3 cm2/Vs to 9 cm2/Vs was found after 2 h of annealing for IOH films deposited on
both sol-gel In
x
O
y
layers and on Ga
x
O
y
- 150
°
C. In contrast, the electron mobility of IOH films
deposited on Ga
x
O
y
- 200
°
C and bare glass substrate was found to increase to 78 cm
2
/Vs and
85 cm2/Vs, respectively.
The discussion of the results starts with the structure of the sol-gel layers. According to the
XRD measurements shown in Figure 5.8, we assume that crystalline nuclei form in the indium
oxide sol-gel layers already at temperatures as low as 150
°
C and 200
°
C and that their growth
is accelerated by an increased annealing temperature. When depositing IOH onto these films,
the nuclei seem to induce crystalline growth, leading to increased IOH crystallinity as shown in
Figure 5.10. The increased crystallinity after deposition results in a lower charge carrier density
and electron mobility, as observed in Figure 5.11 and as already discussed in general in sections
4.2 and 5.1. The drop in charge carrier density and electron mobility of IOH films deposited on
69
Substrate Influences on Growth Mechanism and Properties
Figure 5.10: X-ray diffraction patterns of as grown IOH films deposited on quartz glass, indium oxide and
gallium oxide sol-gel layers which were annealed at 150
°
C and 200
°
C, respectively; the In
2
O
3
reference
pattern was taken from PDF 00-006-0416; patterns were shifted for improved differentiation
the sol-gel layers which were annealed at 150
°
C may be explained by the structure of these films.
As observed by
FTIR
measurements, Ga
x
O
y
- 150
°
C sol-gel layers still contain a high amount of
water which decreases with higher annealing temperatures during the sol-gel deposition process.
This can also be assumed for sol-gel In
x
O
y
layers. It is possible that H
2
O or -OH molecules
from the sol-gel layers diffuse into the IOH layers during deposition or even through the IOH
film to the surface during annealing. Both effects might lead to deteriorated film structures and
electrical properties. Similarly, such a significant decrease of charge carrier density and electron
mobility after annealing could be observed for IOH films which were deposited at high water
vapor pressures, as discussed in section 4.2.
This study has found that generally sol-gel layers can be suitable templates for the growth of
IOH thin films, but certain conditions must be fulfilled. The sol-gel layers must be amorphous
and free of crystalline nuclei after the deposition process, as they otherwise might support a
crystalline growth of IOH films. Furthermore a high amount of residual water in the sol-gel
layers is a disadvantage as it limits the electrical properties of IOH films after annealing. These
findings will serve as a base for future studies within this thesis.
The main findings can be summarized as follows:
70
5.3 Polycrystalline CIGS and textured glass
100
80
60
40
20
0
µ
e
/ cm
2
V
-1
s
-1
4x10
20
3
2
1
0
n
e
/ cm
-3
10
2
10
3
10
4
10
5
/ µ
W
cm
2.01.51.00.50.0
Annealing duration
t
@ 160°C / h
Glass
Ga
x
O
y
150°C
Ga
x
O
y
200°C
In
x
O
y
150°C
In
x
O
y
200°C
Figure 5.11: Change of the charge carrier density and electron mobility of IOH films during annealing at
160
°
C in vacuum; IOH films were deposited on glass substrates, sol-gel Ga
x
O
y
and In
x
O
y
layers, which
were annealed at 150 °C and 200 °C, respectively
•
Amorphous spin coated Ga
x
O
y
layers are suitable templates for the deposition of In
2
O
3
:H
layers, when the amount of residual water is reasonable low.
•
Spin coated sol-gel In
x
O
y
layers induce crystalline growth of In
2
O
3
:H during deposition,
resulting in poor electrical properties.
•
Similar electrical properties of as grown and annealed In
2
O
3
:H are were reached on glass
and GaxOylayers, which were annealed at 200 °C.
5.3 Polycrystalline CIGS and textured glass
The aim to implement IOH as front contact in CIGS modules requires a sheet resistance
RSq
< 10
Ω
/Sq for a CIGS module cell width of approx. 5 mm cell width to avoid increase of
the module series resistance and a low fill factor. Thus, the electrical properties of the TCO
grown on CIGS samples are decisive. In this section we show how a CIGS sub-layer affect the
71
Substrate Influences on Growth Mechanism and Properties
growth and the resulting properties. For this, the electrical properties of films deposited at varied
conditions, before and after annealing, and the film structure were investigated.
In this study a significant detrimental effect of the CIGS topography on the electrical properties,
eminently the electron mobility of amorphous indium oxide based TCOs was observed. As
example serve hydrogen and tungsten co-doped indium oxide (
In2O3:H,W
) (or IWO:H) thin films
deposited by
RPD
(see section 3.1.1 on page 23) on planar soda lime glass substrates and CIGS
samples (IGZO/CdS/CIGS/Mo/glass), respectively. The CIGS samples are expected to have a
RMS roughness value of 80 nm to 90 nm due to the deposition method. For the deposition of
In2O3:H,W
the oxygen supply was varied between 30 ml/min and 60 ml/min at "high" (
≈
1x10
−3
Pa) and "low" (
≈
1x10
−4
Pa) water vapor pressures. Figure 5.12 shows the corresponding charge
carrier density, electron mobility and resistivity of the deposited IWO:H thin films. While the
charge carrier density is similar for films on SLG and CIGS, the electron mobility of the IWO:H
films is overall lower for the films deposited on CIGS. Furthermore the electron mobility of
IWO:H/CIGS does not exceed the value of 30 cm
2
/Vs while the electron mobility of IWO:H/SLG
increases from 29.8 cm
2
/Vs to 42.4 cm
2
/Vs and from 37.0 cm
2
/Vs to 49.5 cm
2
/Vs with increased
oxygen flow for films deposited at "low" and "high" water vapor pressures, respectively. The
decreased electron mobility results in a higher resistivity of the IOH/CIGS samples with an
increased spread for oxygen rich deposition conditions. An additional experiment (not shown)
revealed, that no beneficial effect on the electron mobility resulted from an increased deposition
rate, realized by an increased DC current during deposition.
Figure 5.13 shows as example an SEM cross section image of an IWO:H layer in the as deposited
state grown on IGZO/CdS/CIGS. In the middle of the SEM image disconnected parts of the
IWO:H layer film were observed. On the right side the cross section shows a smooth IWO:H film
structure. In contrast on the left side specific structures occur. These structures are found to be
present over the whole cross section towards the interface of the sub-layers.
A post deposition thermal treatment in nitrogen atmosphere at
p
= 7x10
4
Pa of samples deposited
at
q(O2)
= 60 ml/min led to solid phase crystallization resulting in a decrease of the charge
carrier density and an increase of the electron mobility. In Figure 5.14 the charge carrier density
and electron mobility of IWO:H thin films grown on SLG or CIGS samples and deposited at
(a) "high" (
pH2O≈
1x10
−3
Pa) and (a) "low" (
(pH2O≈
0.7-4x10
−4
Pa) water vapor pressure are
shown as a function of the annealing temperature. The corresponding hall measurements were
done after 30 min of annealing of each of the temperatures. In general an increased pH2Oleads
to a highly amorphous film structure and only few crystalline nuclei. Films that were deposited
at reduced
pH2O
are known to have a higher density of crystalline nuclei, which results in a
decreased crystallization temperature, as shown by Koida et al [5]. This also applies for films
presented in this study. Thus the change of the electrical properties due to crystallization can be
72
5.3 Polycrystalline CIGS and textured glass
50
40
30
20
µ
e
/ cm
2
V
-1
s
-1
SLG
CIGS
1200
1000
800
600
400
/ µ
W
cm
60555045403530
Oxygen flow / ml min
-1
high p
H2O
low p
H2O
5x10
20
4
3
2
n
e
/ cm
-3
Figure 5.12: Comparison of the electrical properties of IWO:H thin films on flat soda lime glass and rough
CIGS samples in dependence of their deposition conditions
observed at 220
°
C to 230
°
C and 150
°
C to 160
°
C for films deposited at "high" and "low"
pH2O
,
respectively. In fact, IWO:H-CIGS samples deposited at "high"
pH2O
showed a decrease in
ne
and increase in
µe
at lower temperatures in comparison to the corresponding film on SLG. Thus
additional crystalline IWO:H nuclei may form at the interface to the CIGS sample, promoting
the solid phase crystallization of the film. Although also the electron mobility of IWO:H films
deposited on CIGS samples increased due to the solid phase crystallization, the values remain
significantly lower than that of the film on SLG. In fact, the spread
∆µe
=
µe(IW O :H−SLG)
-
µe(IW O :H−CIGS)
increases from
∆µe
= 14.9 cm
2
/Vs to
∆µe
= 28.3 cm
2
/Vs and from
∆µe
= 20.8 cm
2
/Vs to
∆µe
= 29.4 cm
2
/Vs after annealing up to 250
°
C for samples deposited at
"high" and "low" pH2O, respectively.
The results show that the electrical properties of indium oxide based TCOs, here IWO:H, change
dramatically when grown on CIGS samples instead of planar glass. The conductivity is limited
by the electron mobility, independent of the deposition condition when the films are deposited by
RPD. The SEM image in Figure 5.13 shows disconnected parts in the films, as observed in the
middle of the image. On the left side the cross section film appearance is cauliflowers-like. Such
a structure is usually observed at the surface of films deposited by sputtering or other deposition
73
Substrate Influences on Growth Mechanism and Properties
Figure 5.13: SEM image of an IWO:H layer on IGZO/CdS/CIGS
techniques [143
–
145]. This indicates, that prior to the measurement the sample broke along such
a disconnected part, which becomes visible in the cross section view. The appearance of the
IWO:H structure on the right side of the SEM is smooth and indicates breakage through the
bulk of the amorphous film. We assume that the disconnected parts, such as cracks or voids, are
the origin of the limited electron mobility. Research by Keller et al. [68] also points towards the
formation of cracks and voids as the origin of the increased sheet resistance in, here, sputtered
IOH films on CIGS samples. As the electron mobility of the films is still low after crystallization,
we assume that the cracks do not coalesce during the thermal treatment. Thus, to achieve
high electron mobilities in indium oxide based TCOs on CIGS samples it is required to further
investigate the limiting effects. The electron mobility of the films grown on rough CIGS samples
has to be improved already in the as deposited state, since solid phase crystallization did not
lead to sufficient improvement, compared to films deposited on planar glass substrates.
The main findings can be summarized as follows:
•
Regardless of the deposition conditions significantly low electron mobilities evolved, as
demonstrated with In
2
O
3
:H,W films, which were grown on rough CIGS samples compared
to films grown on planar glass substrates.
•No sufficient improvement of the electron mobiliy occurred after annealing.
74
5.3 Polycrystalline CIGS and textured glass
1600
1400
1200
1000
800
600
400
/ µ
W
cm
300250200150100
50
0
Temperature / °C
100
80
60
40
20
0
µ
e
/ cm
2
V
-1
s
-1
SLG
CIGS
3.0x10
20
2.5
2.0
1.5
1.0
n
e
/ cm
-3
high p
H2O
(a)
1600
1400
1200
1000
800
600
400
/ µ
W
cm
300250200150100
50
0
Temperature / °C
100
80
60
40
20
0
µ
e
/ cm
2
V
-1
s
-1
SLG
CIGS
3.0x10
20
2.5
2.0
1.5
1.0
n
e
/ cm
-3
low p
H2O
(b)
Figure 5.14: Dependence of the electronic properties of IWO:H films grown on planar SLG and rough
CIGS samples on the annealing conditions; the IWO:H films were deposited with
q(O2)
= 60 ml/min at
high (a) and low (b) water vapor pressures; Hall measurements were done after 30 min of annealing, each
•Indium oxide based TCOs deposited on CIGS samples have much higher resistances.
5.3.1 Detrimental effects on the electron mobility
CIGS samples are known to show a specific roughness which is usually described by the root
mean square (
RMS
) roughness value [146
–
149] and can be influenced by growth conditions
and treatments after deposition. In this section
2
we therefore investigate the influence of the
surface morphology on the electrical properties, i.e., the electron mobility, of In
2
O
3
:H and
In
2
O
3
:H,W films. The investigations were conducted twofold. First (i) we study the morphology
2
Reproduced in part with permission from Darja Erfurt, Marc D. Heinemann, Sebastian S. Schmidt, Stefan
Körner, Bernd Szyszka, Reiner Klenk, Rutger Schlatmann; Substrate influence on the growth of hydrogen doped
indium oxide;ACS Appl. Energy Mater. 2018, 1, 5490-5499; doi: 10.1021/acsaem.8b01039, Copyright 2018
American Chemical Society.; https://pubs.acs.org/articlesonrequest/AOR-vSC6y2YnvNhDtSbYrpi3 [100], licensed
under a Creative Commons Attribution 4.0 International License (CC-BY)
75
Substrate Influences on Growth Mechanism and Properties
of CIGS samples, grown by multi-source evaporation with different deposition durations by AFM
measurements. The films showed different roughness, but the main structure type remained the
same. After this we correlate the structure and electrical properties of In
2
O
3
:H in dependence
on different topographic values. Second (ii) we compare the morphology of multi-source CIGS
samples, grown using different recipes and study the impact on the structure and electron
mobilities of In
2
O
3
:H and In
2
O
3
:H,W films, deposited by either sputtering or
RPD
. Finally we
discuss the obtained results.
The roughness of CIGS absorber, deposited by multi-source-evaporation at HZB, was adjusted
by varying the process duration, resulting in changes of the absorber thickness on bare and
Mo-coated glass substrates. Additionally smooth and textured glass substrates were used for
investigations of the influence of the substrate roughness on the electrical properties of indium
oxide based TCOs. The samples were coated with 60 nm Zn(O,S) or 60 nm Zn(O,S) and 130 nm
i-ZnO, respectively, prior to the IOH deposition by pulsed dc magnetron sputtering. The IOH
films had a thickness of
≈
300 nm. Figure 5.15 illustrates the rise of the RMS roughness of
CIGS/Zn(O,S)/i-ZnO samples in dependence of the CIGS process time. The figure showed that
the RMS roughness increases more significantly when the films were deposited on Mo-coated
glass substrates. The main cause for the increased RMS values is an increase of the CIGS grain
size, which was determined by AFM measurements and the Watershed method of the program
Gwyddion [102] (see Figure A.6 on page 146 in the appendix). For comparison AZO thin films
were deposited on bare and i-ZnO coated flat glass and on i-ZnO/Zn(O,S) coated CIGS samples.
The total sample assignment can be found in Table A.1 in the appendix.
140
120
100
80
60
40
20
0
RMS roughness / nm
120100
80604020
0
CIGS process duration / min
Mo/CIGS/Zn(O,S)/ZnO
glass/CIGS/Zn(O,S)/ZnO
Figure 5.15: Change of the RMS roughness of CIGS/Zn(O,S)/i-ZnO samples deposited on bare and
Mo-coated glass substrates in dependence of the CIGS process duration
In Figure 5.16 the profile height, local slope and AFM topography images with the corresponding
measurement line as example of 3 different sample types. Figure 5.16 (left) shows a rather
76
5.3 Polycrystalline CIGS and textured glass
1000
800
600
400
200
0
Profile height / nm
i-ZnO/Zn(O,S)/CIGS/Glass
RMS = 40 nm
4
3
2
1
0
Slope / arb. units
201510
50
Position / µm
Median Local Slope (MLS) = 0.261
1000
800
600
400
200
0
Profile height / nm
i-ZnO/Zn(O,S)/CIGS/Mo/Glass
RMS = 118 nm
4
3
2
1
0
Slope / arb. units
201510
50
Position / µm
Median Local Slope (MLS) = 0.497
1000
800
600
400
200
0
Profile height / nm
i-ZnO/Zn(O,S)/textured glass
RMS = 174 nm
4
3
2
1
0
Slope / arb. units
201510
50
Position / µm
Median Local Slope (MLS) = 0.276
Figure 5.16: Profile height, local slope and corresponding AFM topography images with profile line of
(left) a rather smooth CIGS sample (ID 4-3998-4-2), (middle) a rather rough CIGS sample (ID 4-3995-1-2)
and (right) a textured glass sample (ID 5-15-2-83) before IOH deposition; the sample configuration can be
found in Table A.1 in the appendix
smooth CIGS sample (glass/CIGS-46 min/Zn(O,S)/ZnO) with RMS = 40 nm; (middle) a rather
rough CIGS sample (Mo/CIGS/Zn(O,S)/ZnO) with RMS = 118 nm; (right) a textured glass
sample (textured glass/Zn(O,S)/ZnO) with RMS = 174 nm. The samples will be addressed as
smooth CIGS sample,rough CIGS sample and textured glass sample, respectively, from now on.
Values of the RMS roughness and median local slope (
MLS
) were added to the Figures. The
rough CIGS sample shows significantly larger grains, higher profile heights and local slopes than
the smooth CIGS sample. In contrast the textured glass sample shows even larger structures
and an increased RMS roughness, but a median local slope (
MLS
= 0.276) comparable to the
one of the smooth CIGS sample (
MLS
= 0.261). Both the profile height and local slope of the
textured glass sample show less fluctuation over the measured range than both CIGS samples.
Within one sample the relatively highest slopes were found at the edges of grains on the CIGS
samples or structures on the textured glass sample. Thus not only the median local slope, but
also its frequency has to be taken into account. Therefore we calculated the median local slope
per grain/structure size (MLS/GS), determining the grain size by AFM measurements using
the Watershed method. It was found, that the textured glass sample had a lower MLS/GS value
(
MLS/GS
= 0.29x10
−3
nm
−1
) than the smooth CIGS sample (
MLS/GS
= 0.75x10
−3
nm
−1
)
77
Substrate Influences on Growth Mechanism and Properties
or the rough CIGS sample (
MLS/GS
= 1.24x10
−3
nm
−1
), due to the low median local slope
but large structures of the textured glass sample
The samples mentioned in Table A.1 were coated with IOH within the same deposition run
as the samples discussed in section 5.1, where a pronounced crystalline growth of IOH was
observed when deposited on ZnO layers. Similar to these results also on rough samples a more
pronounced crystalline growth occurred when the films were deposited on ZnO-coated samples.
IOH films which were deposited on Zn(O,S)-coated samples were mainly amorphous, similar to
the results in section 5.1. In Figure 5.17 (a)
GI-XRD
measurements of glass/CIGS/Zn(O,S)/IOH
and glass/CIGS/Zn(O,S)/ZnO/IOH in the as deposited state are shown, the RMS value of the
samples before IOH deposition was approx. 65 nm. Intensities of the In
2
O
3
peaks were clearly
higher for the films which were deposited directly on ZnO than on Zn(O,S). In Figure 5.17 (b)
patterns of
GI-XRD
measurements of IOH/ZnO/Zn(O,S)-coated samples with different roughness
are shown. For all samples an increased In
2
O
3
peak intensity can be observed, independent from
the RMS roughness value. Thus, no influence of the CIGS roughness on the crystalline structure
of IOH can be observed, when the growth conditions are determined by the IOH sub-layers, as
here ZnO and Zn(O,S).
Intensity / arb. units
6560555045403530252015
2
/ °
In
2
O
3
:H/Zn(O,S)/CIGS/glass
In
2
O
3
:H/ZnO130/Zn(O,S)/CIGS/glass
In
2
O
3
(222)
ZnO
(002)
In
2
O
3
(400)
In
2
O
3
(211)
In
2
O
3
(440)
In
2
O
3
(622)
CISe
CISe
CISe
In
2
O
3
(200)
In
2
O
3
(411)
ZnO
(101)
(a)
Intensity / arb. units
6560555045403530252015
2
/ °
textured glass; RMS = 178 nm
glass/CIGS; RMS = 65 nm
In
2
O
3
(222)
ZnO
(002)
In
2
O
3
(400)
In
2
O
3
(211)
In
2
O
3
(440)
In
2
O
3
(622)
CISe CISe CISe
smooth glass; RMS = 0.4 nm
glass/CIGS; RMS = 40 nm
ZnO
(101)
In
2
O
3
(200)
(b)
Figure 5.17: X-ray diffraction patterns of (a) IOH grown on ZnO/Zn(O,S)- and Zn(O,S)-coated CIGS
samples with RMS = 65 nm (adapted from Erfurt et al. [100]) and (b) IOH grown on ZnO/Zn(O,S)-coated
CIGS and glass samples with different RMS roughness values
Figure 5.18 shows the change of the IOH and AZO sheet resistance over the RMS roughness
of the samples described in Table A.1 in the as deposited state. For the AZO coated samples
no influence of the roughness on the sheet resistance can be observed, it remains unchanged at
≈
8.2
Ω
/Sq on smooth glass and rough CIGS samples. In contrast the IOH sheet resistance
increases linearly when deposited on rough CIGS samples. In fact, the slope is more than 3 times
larger for the crystalline grown IOH films, which were deposited on ZnO/Zn(O,S)-coated CIGS
78
5.3 Polycrystalline CIGS and textured glass
Figure 5.18: Comparison of the change of the sheet resistance of IOH and AZO when grown on different
substrates and layers
samples compared to IOH films which were grown on Zn(O,S)-coated CIGS samples, as they
showed a mainly amorphous structure. However, no or only a slight
RSq
increase was observed
when the films were grown on Zn(O,S)- and ZnO/Zn(O,S)-coated textured glass substrates,
although a significantly higher RMS value was measured. From this we conclude that the RMS
roughness value is not a sufficiently meaningful value to describe the substrate surface topography
and the cause of the increase in sheet resistance of IOH if the morphology of the samples changes
significantly. Within one structure type, as here CIGS, the RMS value gives an adequately good
characterization value. For quantitative comparison a linear expression is used to empirically
describe the sheet resistance increase in this and similar graphs.
The increased sheet resistance can be caused by a decreased charge carrier density and/or a
decreased electron mobility of the IOH films. In Figure 5.19 the dependence of these values on the
RMS roughness (left), the median local slope (middle) and median local slope per grain/structure
size (right) is shown. The charge carrier density and electron mobility were determined by Hall
effect measurements for IOH films grown on ZnO/Zn(O,S)- and Zn(O,S)-coated smooth and
textured glass substrates and glass/CIGS samples, respectively. No dependence of the charge
carrier density can be observed for IOH films grown on Zn(O,S) on the surface morphology in
general and for IOH films grown on ZnO/Zn(O,S)/textured glass. In contrast for IOH films
grown on ZnO/Zn(O,S)/CIGS the charge carrier density decreases with increased RMS, median
local slope and median local slope per grain size. In total IOH films grown on ZnO have a lower
charge carrier density than IOH films grown on Zn(O,S). This offset can be explained by the
crystalline growth of IOH on ZnO, as discussed in section 5.1. This also applies to the electron
79
Substrate Influences on Growth Mechanism and Properties
5x10
20
4
3
2
1
0
n
e
/ cm
-3
0.50.40.30.20.10.0
Median Local Slope (MLS) / arb. units
{ µ
e
} = - 43.1 { MLS } + 36.3
{ µ
e
} = - 41.0 { MLS } + 27.9
R
2
= 0.984
R
2
= 0.973
1.2x10
-3
1.00.80.60.40.20.0
MLS/Grain Size / nm
-1
{ µ
e
} = - 18414 { MLS/GS } + 39.2
{ µ
e
} = - 15203 { MLS/GS } + 28.3
R
2
= 0.930
R
2
= 0.987
40
35
30
25
20
15
10
5
0
µ
e
/ cm
2
V
-1
s
-1
200150100
50
0
RMS roughness / nm
{ µ
e
} = - 0.24 { RMS } + 27.5
{ µ
e
} = - 0.26 { RMS } + 36.0
R
2
= 0.994
R
2
= 0.970
as grown on...
... glass/CIGS
IOH/i-ZnO/Zn(O,S)
IOH/Zn(O,S)
... glass
IOH/i-ZnO/Zn(O,S)
IOH/Zn(O,S)
Figure 5.19: Comparison of the impact of the RMS roughness, median local slope and median local slope
per grain/structure size of the substrate/sub-layers on the charge carrier density and electron mobility of
sputtered IOH films deposited on ZnO/Zn(O,S)- and Zn(O,S)-coated samples; for the linear fits the films
on the textured glass sample were only taken into account for the fit on the dependence on the median
local slope per grain size
mobility of IOH films grown on Zn(O,S) and ZnO layers. The electron mobility of IOH films
decreased linearly with increased RMS value when the films were grown on CIGS samples. The
slope of the decrease is similar for amorphous grown IOH films on Zn(O,S) and partly crystalline
grown IOH films on ZnO. In spite of the high RMS value of the textured glass sample, the
IOH electron mobility is significantly higher than that of films grown on CIGS samples with
lower RMS value. In comparison the strongest dependence of the IOH electron mobility was
observed on the median local slope per grain size. Here a linear decrease of
µe
(IOH) can be
observed for IOH films grown on textured glass as well as CIGS samples. A similar dependence
was observed after annealing, as shown in Figure A.7 in the appendix. However, the slope of the
linear regression was higher compared to the fit before annealing. In fact, the slope increased
1.3 times for the mainly amorphous grown films on Zn(O,S)/CIGS, while it increased 2.3 times
for the pronounced crystalline grown films on i-ZnO/Zn(O,S)/CIGS. These rises apply for the
dependence on the RMS, the median local slope as well as the median local slope / grain size.
In general we can say, that the increase of the IOH sheet resistance grown on rough CIGS samples
is mainly caused by a decreased electron mobility. The more pronounced increase of
RSq
for
films grown on ZnO can be explained by additional decrease of the charge carrier density due to
a more pronounced crystalline structure after deposition.
80
5.3 Polycrystalline CIGS and textured glass
Figure 5.20: SEM images of samples a) 5-15-2-83 (IOH/ZnO/Zn(O,S)/textured glass), b) 4-3995-1-2
(IOH/ZnO/Zn(O,S)/CIGS/Mo/glass), c) 4-3998-2-2 (IOH/ZnO/Zn(O,S)/CIGS/glass) and d) 4-3998-4-2
(IOH/ZnO/Zn(O,S)/CIGS/glass)
Figure 5.20 shows SEM cross section images of the coated textured glass sample (a), and
coated CIGS samples with different thicknesses (b) - (d). All samples were coated with an
IOH/ZnO/Zn(O,S) window layer. As already observed by AFM measurements the textured
glass substrate shows large but smooth structures. In contrast the CIGS layer consists of smaller
grains and has an increased surface height fluctuation. This value increases for thicker CIGS
absorber (Figure 5.20b). The CIGS surface becomes smoother with decreased CIGS thickness
(Figure 5.20(c) and (d). The IOH layer which was deposited on the textured glass sample seems
to be closed and to cover the whole sample without void formation. In contrast darker areas
are observed inside the IOH layers which were deposited on CIGS samples. The dark areas
are assumed to be voids which separate the material. Similar findings were shown by Jäger et
al. [150]. The amount and size of the voids inside the IOH layer increase with increased CIGS
thickness, surface roughness and height fluctuation. From the images it seems that the voids
mainly appear above CIGS grain boundaries.
These findings are consistent with the following study of In
2
O
3
:H and In
2
O
3
:H,W films, deposited
by Reactive Plasma Deposition and In
2
O
3
:H films, deposited by pulsed DC magnetron sputtering.
81
Substrate Influences on Growth Mechanism and Properties
Koida et al. [5] studied the material properties of the RPD - indium oxide based TCOs on flat
glass, the results served as basis for this study. The CIGS layers were fabricated by co-evaporation
by two different recipes and institutes,
HZB
in Berlin, Germany and
AIST
, Tsukuba, Japan,
resulting in different CIGS morphologies. In Figure 5.21 SEM top view images of both CIGS types
after CdS deposition are shown. The presented samples were produced without a post deposition
treatment (
PDT
). The HZB sample has sharper edges, rather triangle shaped grains and seems
rougher than the sample fabricated at AIST. Here the surface shows flatter, rounder shaped
grains.
AFM
measurements confirmed the rougher structure of the HZB sample (RMS = 85 nm)
in comparison to the AIST sample (RMS = 39 nm). The corresponding topography images can
be found in Figure 5.22.
Note that a CIGS
PDT
can influence the surface topography of the absorber and lead to diffusion
of alkali metals into the TCO layer. In fact, GDOES measurements confirmed that sodium,
supplied by a NaF
PDT
, diffused into the used IOH layer. However, no correlation between the
sodium amount in the TCO layer and the electron mobility was observed. Additional information
can be found in the Appendix A.3. To evaluate the electronic properties of the indium oxide
based TCOs the films were deposited on four types of CIGS samples, 2 fabricated by each
institute with and without a
PDT
. An overview of the the topographic values of the CdS coated
CIGS samples can be found in Table 5.2. The values were calculated from 10
µ
m x 10
µ
m AFM
topography images.
(a) HZB (b) AIST
Figure 5.21: SEM top view images of CdS coated CIGS samples fabricated at (a) HZB and (b) AIST
Table 5.3 summarizes the electronic properties of the three TCO films on flat soda lime glass
and the four rough CIGS samples. The data shows that the charge carrier density of the TCO
layers does not change systematically when deposited on samples with different topographies,
similar to the results presented in Figure 5.19 on page 80. Another similarity is the decrease of
82
5.3 Polycrystalline CIGS and textured glass
(a) HZB (b) AIST
Figure 5.22: AFM topographic images of CdS coated CIGS samples fabricated at (a) HZB and (b) AIST
Table 5.2: Overview of the topographic values rout mean square (RMS) roughness, median local slope
(MLS) and median local slope per grain size (MLS/GS) of four types of CIGS samples
Sample PDT RMS MLS MLS/GS
(nm) (arb. units) (nm−1)
flat SLG 0.3 - -
HZB-1 none 84.9 0.62 3.41x10−3
HZB-2 NaF 81.2 0.66 4.52x10−3
AIST-1 NaF, KF 55.7 0.47 2.83x10−3
AIST-2 none 39 0.39 2.35x10−3
Table 5.3: Overview of the electrical properties charge carrier density (
ne
) and electron mobility (
µe
) of
IWO:H, IOH thin films grown by
RPD
and sputtered IOH which were deposited on four types of CIGS
samples
Film IWO:H - RPD IOH - RPD IOH - sputtered
Property neµeneµeneµe
(cm−3) (cm2/Vs) (cm−3) (cm2/Vs) (cm−3) (cm2/Vs)
flat SLG 3.70x1020 37.0 3.80x1020 43.5 3.56x1020 39.6
HZB-1 3.10x1020 27.1 3.64x1020 28.9 3.21x1020 16.2
HZB-2 3.24x1020 28.5 3.64x1020 30.9 3.18x1020 18.8
AIST-1 3.25x1020 33.9 3.55x1020 37.3 3.26x1020 26.6
AIST-2 3.04x1020 35.0 3.67x1020 40.8 3.28x1020 27.9
83
Substrate Influences on Growth Mechanism and Properties
50
45
40
35
30
25
20
15
10
5
0
µ
e
/ cm
2
V
-1
s
-1
100
80604020
0
RMS roughness / nm
R
2
= 0.873
In
2
O
3
:H,W - RPD
In
2
O
3
:H - RPD
In
2
O
3
:H - sputtered
R
2
= 0.982
R
2
= 0.911
{ µ
e
} = - 0.12 { RMS } + 38.3
{ µ
e
} = - 0.17 { RMS } + 45.3
{ µ
e
} = - 0.26 { RMS } + 39.5
Figure 5.23: Dependence of the electron mobility of In
2
O
3
:H, In
2
O
3
:H,W, deposited by RPD and sputtered
In
2
O
3
:H on the substrate’s RMS roughness value; The films were grown on samples described in Table 5.2
the electron mobility of all three TCO layers with increased RMS roughness. The drop of the
electron mobility can be described by a linear fit as well. The strongest decline was observed for
the sputtered IOH films. Note that the slope is approximately the same as for the films grown
on CIGS samples with different thicknesses, as shown in Figure 5.19. The lowest decline was
found for the hydrogen and tungsten co-doped indium oxide layer grown by RPD. Compared to
the sputtered hydrogen doped indium oxide films, it becomes apparent that the layers deposited
by RPD show higher electron mobilities and a 34 %lower drop of µewith increased RMS.
The drop in electron mobility can be correlated with the appearance of voids in the TCO layer.
The structure of the as grown sputtered In
2
O
3
:H film, which was deposited on a CIGS sample
with RMS
≈
85 nm, was investigated by STEM measurements. Images of two significant regions
of the sample In
2
O
3
:H/i-ZnO/CdS/CIGS are shown in Figure 5.24(a) and (b). Red arrows point
towards the voids in the In
2
O
3
:H. In Figure 5.24(a) two voids in the In
2
O
3
:H layer with a width
of
≈
15 nm can be clearly observed. Figure 5.24(b) presents another void with a width of
≈
5 nm.
In Figures A.12 and A.13 in the appendix further examples are shown. The voids start in the
In
2
O
3
:H layer, close to the ZnO interface. From Figure 5.24(b) it is apparent, that the void
forms after
≈
16 nm of In
2
O
3
:H were deposited. This value is lower than the estimated deposited
amount during one path of the in-line sputtering process, which is
≈
25 nm. This results indicate,
that the cracks are formed due to geometry effects. The TEM measurements revealed that the
CdS lattice follows the lattice of CIGS with many twins and defects. Furthermore some coherent
lattice relationship between ZnO and CdS was observed. In contrast, no such relationship was
found for In
2
O
3
:H and i-ZnO, thus the formation of crystalline nuclei on the surface of ZnO
84
5.3 Polycrystalline CIGS and textured glass
must have been successfully suppressed by the increased water vapor supply and consequently
hydrogen supply during deposition
3
. The results indicate that the voids form not due to change
of the atomic structure across CIGS grain boundaries, but rather due to a geometry effect.
Furthermore it can be seen that the voids form at triangle shaped regressions, which can be
considered as a sharp transition. Figure A.11 presents images of the elemental compositions of
the region showed in Figure 5.24(a), measured by EELS. The measurements confirm, that in the
area of the voids no indium or other material was detected. In general it was found, that the
majority of the voids could be aligned with CIGS grain boundaries.
CIGS
ZnO/
CdS
In O :H
2 3
Pt/C
(a)
CIGS
CdS/i-ZnO
In O :H
2 3
(b)
Figure 5.24: STEM measurements of as grown sputtered In
2
O
3
:H grown on i-ZnO/CdS/CIGS; Voids are
clearly visible in the In2O3:H layer (dark areas) and are indicated by red arrows
In the following the obtained findings are discussed. The results show how the electron mobility
of indium oxide based TCOs is influenced when the films are grown on rough samples. The
topography of the CIGS samples can be described in a first approximation by the root mean
square (RMS) roughness, if it can be assumed, that the CIGS topography type is similar. We
found by two independently taken studies that the electron mobility of indium oxide based TCOs
such as IWO:H and IOH decreases with increased RMS of the sub-layer. This applies for films
deposited by sputtering as well as
RPD
. The good electron mobility on textured glass, which
showed smooth transitions on the surface, and STEM measurements confirmed that high local
slopes and sharp transitions are detrimental. Such structures mainly appear along the grains, at
3here PH2O= 5.5x10−3Pa; for the experiment shown in section 5.1 and corresponding films grown on CIGS
samples as shown in Figure 5.17 on page 78: PH2O≈2x10−4Pa
85
Substrate Influences on Growth Mechanism and Properties
grain boundaries. The topographic structure of CIGS has therefore a high impact on the electrical
properties of indium oxide based TCOs, as it was shown for In
2
O
3
:H and In
2
O
3
:H,W. Thus we
conclude, that a specific amount of sharp transitions resuls in a defined amount of voids in the
TCO layer. This also explains the limited mobility of
≈
30 cm
2
/Vs for In
2
O
3
:H,W films on CIGS
samples, regardless of the improving electron mobility of the films deposited on glass substrates
with increased oxygen supply during deposition, as observed in Figure 5.12. We conclude that
the film quality in general improves also in the film deposited on CIGS. However, the amount
of voids is assumed to be approximately the same in all of the films grown on CIGS. Therefore
the fixed amount limits the electron mobility to
≈
30 cm
2
/Vs in general for all films, although
the quality of the material itself may differ. Also after a thermal treatment the amount of voids
most likely does not change significantly, resulting in a similar effect also for the crystallized
films. Moreover the voids might promote effusion of hydrogen due to the increased surface area.
This could explain the more pronounced decrease of mobility with increased RMS values after
annealing, as mentioned on page 80 (compare Figure 5.19 and Figure A.7). The dependency of
electron mobility of the as grown films on the RMS of the sub-layer was described by a linear
progression with coefficients of determination higher than 0.87. However, the observed linear
dependency can not be valid for very high RMS values, as the electron mobility can not drop
below zero. It was found that the decline of
µe
of In
2
O
3
:H,W and In
2
O
3
:H films deposited by
RPD
is not as strong as for sputtered In
2
O
3
:H films. As a significant difference between the two
deposition techniques the energy of the deposition particles is suggested. In
RPD
it is less than
40 eV during ITO deposition. In contrast in sputtering methods particles such as backscattered
argon and negative oxygen ions are considered to have high energies above 100 eV. The lower
energies and higher ionization rates of the depositing particles are considered to be the origins of
higher quality ITO films grown by
RPD
[151] . In this study the
In2O3:H
films grown by
RPD
also are of higher quality than the sputtered films. The lowest drop in electron mobility was
found for tungsten and hydrogen co-doped indium oxide. Therefore we assume that tungsten
reduces the formation of voids/cracks or promote their coalesce by changing the surface energy
of the films.
In contrast to the indium oxide based TCOs the sheet resistance of aluminum doped zinc oxide
films did not increase, as shown in Figure 5.18. Sputtered AZO thin films are known to grow
crystalline [152
–
155]. In contrast the deposited indium oxide films are mainly amorphous, or can
contain both, crystalline and amorphous fractions when ZnO is used as sub-layer. We assume that
the crystalline AZO films, even though some voids might form, coalesce at very low thicknesses.
We assume the origin in the preferred (002) columnar structure and the (partly) tilted growth on
rough samples. In contrast the amorphous fractions of the indium oxide based films might lower
the merging effect of separated film parts. Moreover we observed that the substrate roughness
has no significant impact on the crystalline growth of the indium oxide based films, here IOH, as
it is highly dependent on its sub-layer and in general the deposition conditions. The crystalline
86
5.3 Polycrystalline CIGS and textured glass
structure of the as deposited IOH films was quite similar for films deposited on flat and rough
substrates onto the sub-layers ZnO (pronounced crystallinity of IOH) and on Zn(O,S) (mainly
amorphous IOH).
The surface of the sub-layers was also described by the median local slope and median local slope
per grain size, which correlate the MLS and the structure or grain size estimated by the watershed
method. A linear dependence of the decreased electron mobility on these values was detected as
well. Comparing the MLS/GS values from Figure 5.19 and Table 5.2 it becomes apparent that
the values show large differences, however, the indium oxide layers had similar electron mobilities.
The origin of the differences lies in the estimated grain and structure size. For the samples
described in Figure 5.19 AFM topography images of 40
µ
m x 40
µ
m were obtained, in contrast
for the samples described in Table 5.2 10
µ
m x 10
µ
m AFM topography images were taken.
Consequently more details are taken into account by the calculation routine of the grain size
using the watershed method for the latter samples. Thus, the determined grain/structure size is
lower than the one calculated for the CIGS samples with varying thicknesses from 40
µ
m x 40
µ
m
images. The calculated MLS/GS value is consequently larger. Thus, this routine and the followed
evaluation by MLS/GS can only be applied for topographic images of the same size.
In the following strategies for improved electron mobility of indium oxide based TCOs in the as
deposited state grown on CIGS samples are presented. This is considered a key requirement to
achieve high mobility after crystallization on CIGS.
The main findings can be summarized as follows:
•
Formation of voids/cracks in the indium oxide layer when the films were deposited on
rough CIGS samples.
•Voids/cracks mainly are located at sharp structures (e.g. cliffs at CIGS grain boundaries,
sharp regressions).
•
Voids and cracks in the indium oxide layers are the main cause for limited electron mobility;
deficit is more pronounced for annealed films.
•
In films grown on CIGS samples the electron mobility seems to decrease linearly with
increase of the RMS roughness value.
•
No difference for mainly amorphous grown and pronounced crystalline grown films was
observed.
•
No limitation of electrical properties in poly-crystalline AZO films grown on rough CIGS
samples observed.
87
Substrate Influences on Growth Mechanism and Properties
5.4 Strategies to improve the electron mobility of hydrogen
doped indium oxide based TCOs on CIGS
As described in section 5.3 rough CIGS samples with sharp edges are the main cause of the
increase of electron mobiliy of indium oxide based TCOs when the films contain amorphous
regions. This section presents strategies which can be implemented in an already existing CIGS
solar cell process line to improve the electron mobility of
In2O3:H,W
and hydrogen doped indium
oxide (
In2O3:H
) thin films. One possibility is to smooth the CIGS surface by the deposition
of a spin-coated sol-gel layer or by etching. Furthermore the impact of the TCO thickness is
discussed.
5.4.1 Spin Coated Sol-Gel Layers
As presented in section 5.2 spin coated gallium oxide sol-gel layers can serve as sub-layers for IOH
thin films without deterioration of the electrical properties before and after annealing. Based on
these results Ga
x
O
y
sol-gel layers were spin coated on CdS-coated CIGS samples prior to the
IOH sputter deposition. The deposition process is described in section 3.1.2 on page 27. For this
study the deposition routine was processed 1, 3 and 6 times, respectively. Figure 5.25 shows AFM
topography images of the CIGS/CdS samples after 1, 3 and 6 Ga
x
O
y
sol-gel depositions. The
RMS value of the samples can be reduced significantly from 76 nm to 50 nm after 6 depositions.
Here, the structure of the CIGS grains is still present, but the grains have a rounder shape.
Table 5.4 summarizes the RMS value of the samples prior to IOH deposition and the resulting
electrical properties of the IOH layers. Additionally a flat SLG sample was processed as reference.
While the charge carrier density of the film remains unchanged, the electron mobility increases
with increased deposition amount, as the surface becomes smoother. Note that after 6 Ga
x
O
y
depositions the electron mobility is equal, even slighly higher than the electron mobility on SLG,
even though the CIGS sample has a RMS value of 50 nm.
Table 5.4: Influence of spin coated sol-gel layers on the CIGS sample RMS roughness and the electrical
properties of sputtered In2O3:H thin films
Sample RMS neµeρ
(nm) (cm−3) (cm2/Vs) (µΩcm)
Glass 0.3 2.23x1020 42.3 662
1xGaxOy76 2.23x1020 24.8 1130
3xGaxOy59 2.44x1020 35.9 712
6xGaxOy50 2.37x1020 43.7 602
88
5.4 Strategies to improve the electron mobility of hydrogen doped indium oxide based TCOs on
CIGS
Figure 5.25: AFM topography images of CdS/CIGS samples which were coated with sol-gel Ga
x
O
y
layers;
the amount of deposition repetitions and the determined RMS value are added
Figure 5.26 illustrates the change in
µe
over RMS. For comparison data of the sputtered IOH
film on different CIGS absorber types, as shown in Figure 5.23, is added to the graph. The
graph shows a linear dependency of the electron mobility on the RMS value also for the films
deposited on Ga
x
O
y
/CdS/CIGS. The CIGS samples with RMS values of around 80 nm were
deposited following the same recipe as the CIGS absorber for the Ga
x
O
y
depositions. Prolonging
the linear fit also leads to good agreement for these CIGS samples, which were fabricated with
an i-ZnO layer instead of the Ga
x
O
y
layer. Thus, already one spin coating deposition led to a
lower RMS value compared to the samples with i-ZnO. Further, it becomes apparent that the
modification of the surface with spin-coated sol-gel layers is much more beneficial on the IOH
electron mobility than the chance of the CIGS absorber type.
The data shows that the spin coated Ga
x
O
y
layers smoothed the CIGS samples with increased
deposition amount. We assume that the spin coating process round sharp grain structures and
thus leads to smoother transitions between the CIGS grains on the CIGS surface. Note that the
measured RMS value after 6 Ga
x
O
y
depositions was 50 nm. High local slopes, as they appear at
sharp grain edges, were found to be detrimental, as discussed in section 5.3. Thus, the impact
of the surface becomes similar to the one of the textured glass as described in section 5.3. On
the textured glass sample high IOH electron mobilities were achieved although the textured
glass substrate had a RMS roughness of approx. 170 nm. When the surface showed smooth
transitions, void formation in the IOH layer can be avoided and mobilites as high as on flat SLG
are reached. These results serve as the basis for further annealing treatments to crystallize the
amorphous IOH films. Despite of the improved electron mobility the implementation of such a
deposition process might be challenging for an industry-like CIGS solar cell fabrication, as spin
coating is not well scalable.
89
Substrate Influences on Growth Mechanism and Properties
50
45
40
35
30
25
20
15
10
5
0
µ
e
/ cm
2
V
-1
s
-1
100
80604020
0
RMS roughness / nm
In
2
O
3
:H on different absorber types
In
2
O
3
:H after GaO
x
sol-gel
GaO
X
deposition amount
changed CIGS type
same
CIGS
deposition
recipe
1x
3x
6x
Figure 5.26: Electron mobility of sputtered
In2O3:H
in dependence of the substrate RMS roughness of
CIGS samples of different process types and after Ga
x
O
y
spin coat deposition; values at RMS
≈
0 nm are
referred as flat SLG references
The main findings can be summarized as follows:
•
Smoothing of CIGS samples by deposition of spin coated Ga
x
O
y
layers possible; reduction
of sharp structures at the surface.
•
Restored electron mobility in IOH films deposited on smoothed CIGS samples with thick
spin coated GaxOylayer (RMS = 50 nm).
5.4.2 Etching of the CIGS Surface
Another possibility to smooth the CIGS surface is etching in an acid bromine solution [146,147].
By etching with HBr solution CIGS is removed leading to thinner and smoother samples. In
this study a set of four CIGS samples was prepared. The samples were etched for 0 s, 15 s,
30 s and 60 s in an acid bromine solution, the etching rate was estimated to 10 nm/min. The
sample which is referred with an etch duration of 0 s was not etched with HBr and serves as a
reference sample. During etching the solution was homogenized by a magnetic stirrers. Prior to
the etching a previously deposited CdS layer was removed from all four samples in a 5 mol%
HCl solution. After etching the CIGS samples in the HBr solution, a KCN treatment was done
followed by the CdS and i-ZnO deposition.
90
5.4 Strategies to improve the electron mobility of hydrogen doped indium oxide based TCOs on
CIGS
Figure 5.27 illustrates the topographic images taken by
AFM
of the non-etched reference sample
and samples after 15 s and 60 s of etching, respectively. The images were taken after CdS
deposition. After 15 s of etching it can be observed, that the grain size does not change
significantly, furthermore the RMS valule is equal to the one of the non-etched reference sample,
but that the surface of the grains was smoothed, less small particles are visible. The 60 s etched
sample has an approx. 20 nm lower RMS value of 46 nm. The grains become rounder shaped.
Figure 5.28 shows SEM cross section images of the surface near region of the CIGS samples.
The results confirm the smoothing and rounding effect of the etching. After 60 s of etching the
surface shows less structures with sharp slopes compared to the non-etched sample.
Figure 5.27: AFM topography images of etched CIGS sample after CdS deposition; the etch duration and
the determined RMS value are insert; etch duration of 0 s represents the non-etched reference sample
Figure 5.28: AFM topography images of etched CIGS sample after CdS deposition; the etch duration and
the determined RMS value are insert; etch duration of 0 s represents the non-etched reference sample
Table 5.5 summarizes the RMS roughness of the CIGS samples in dependence of the etching
duration and the electrical properties of RPD -
In2O3:H
thin films. Additionally a flat SLG
sample was processed as reference. The data shows that the RMS value decreases with increased
etching duration. Moreover, it becomes apparent that the charge carrier density of the films on
CIGS samples is slighlty lower than on the SLG reference and slightly increases with increased
etch duration. As the calculation of the charge carrier density is dependent on the estimated
91
Substrate Influences on Growth Mechanism and Properties
film thickness, it can not be excluded that this effect is due to uncertainties of the film thickness.
With increased etching duration the electron mobility of the
In2O3:H
films increases. This effect
is first determined after 15s of etching, although the RMS value remains unchanged. After 30s of
etching the electron mobility does not further increase significantly.
Table 5.5: Influence of the CIGS etching on the roughness and electrical properties of RPD -
In2O3:H
thin films
Sample RMS neµeρ
(nm) (cm−3) (cm2/Vs) (µΩcm)
Glass 0.3 3.37x1020 50.0 370
0 s 66 3.13x1020 36.2 551
15 s 67 3.15x1020 41.2 480
30 s 55 3.20x1020 44.1 443
60 s 46 3.24x1020 44.4 433
60
50
40
30
20
10
0
µ
e
/ cm
2
V
-1
s
-1
100
80604020
0
RMS roughness / nm
In
2
O
3
:H on different absorber types
In
2
O
3
:H on etched samples
{ µ
e
} = - 0.17 { RMS } + 45.3
{ µ
e
} = - 0.16 { RMS } + 50.8
etch duration
0 s
15 s
30 s
60 s
Figure 5.29: Electron mobility of RPD -
In2O3:H
in dependence of the substrate RMS roughness of CIGS
samples of different process types and of etched CIGS samples; values at RMS
≈
0 nm are referred as flat
SLG references
Figure 5.29 shows the dependency of the electron mobility on the substrate RMS roughness of
etched CIGS samples and of CIGS samples of different process types, taken from Figure 5.23
as comparison. Additionally the values of the SLG reference samples are added. It becomes
apparent that the electron mobility increases with decreased RMS value. The slope was found to
be equal to the slope of the films deposited on different types of CIGS. The electron mobility of
the non-etched reference can be described by the
µe
- RMS dependency as found for the films
92
5.4 Strategies to improve the electron mobility of hydrogen doped indium oxide based TCOs on
CIGS
deposited on different CIGS types, although the electron mobility of the SLG "etch" - reference
film is higher. After 15s of etching the electron mobility increases while the RMS value remains
unchanged. Further etching smooths the CIGS surface and leads to a decrease of the RMS value.
The electron mobility increases according to the observed
µe
- RMS dependency. However, the
electron mobility does not reach the SLG reference value.
Etching in an acid bromine solution was carried out to smooth the CIGS surface and to reduce
the amount of sharp structures by rounding the CIGS grains. These modifications led to an
improved electron mobility. We assume that due to the smoother structure with more round
grains less voids form in the
In2O3:H
layers. This reduced amount is beneficial for the electron
mobility. As the used etching rate of 10 nm/min is quite low, the CIGS absorber were not
reduced in thickness significantly, as the expected typical absorber thickness is in the range of
1.5
µ
m to 2
µ
m. Nevertheless, implementing this etching procedure into the CIGS solar cell
fabrication might be unfavorable, as an additional wet chemical process is required.
The main findings can be summarized as follows:
•
Etching of CIGS with acid bromine solution results in a smoothed surface with less sharp
structures.
•Improved electron mobility of IOH films on etched, smoothed CIGS samples.
5.4.3 Influence of TCO thickness
It is known for AZO thin films, that an increased film thickness can have a beneficial effect on the
electrical properties such as the electron mobility [36,156]. This effect is caused by the increased
crystallite size for increased thicknesses. Although an amorphous growth of the indium oxide
based TCOs described in this study is required, the film thickness was found to also have an
influence on the electron mobility of the films deposited on CIGS samples.
Figure 5.30 illustrates the charge carrier density and electron mobility of
RPD
- In
2
O
3
:H,W,
In
2
O
3
:H and sputtered In
2
O
3
:H films on soda lime glass and CIGS samples with different
RMS roughness values, respectively, in dependence on the film thickness. While the charge
carrier density of
RPD
- In
2
O
3
:H,W films is overall stable in the range of
≈
3x10
20
cm
−3
to
4x10
20
cm
−3
on SLG and CIGS samples, a decrease for increased film thickness was observed
for
RPD
- In
2
O
3
:H and sputtered In
2
O
3
:H thin films on both SLG and CIGS samples. In case
of In
2
O
3
:H film deposited by
RPD
the overall lowest charge carrier density of
≈
1.1x10
20
cm
−3
to 1.4x1020 cm−3was observed for a film thickness of around 525 nm. The electron mobility of
the indium oxide films deposited by
RPD
increased for increased film thickness when grown on
CIGS samples. While the electron mobility of In
2
O
3
:H,W grown on CIGS with RMS = 85 nm
93
Substrate Influences on Growth Mechanism and Properties
1000
800600400200
0
Thickness / nm
In
2
O
3
:H - RPD
SLG; RMS = 1 nm
CIGS; RMS = 39 nm
CIGS; RMS = 85 nm
1000
800600400200
0
Thickness / nm
In
2
O
3
:H - sputtered
SLG; RMS = 1 nm
CIGS; RMS = 56 nm
CIGS; RMS = 85 nm
60
50
40
30
20
10
0
µ
e
/ cm
2
V
-1
s
-1
1000
800600400200
0
Thickness / nm
In
2
O
3
:H,W - RPD
SLG; RMS = 1 nm
CIGS; RMS = 39 nm
CIGS; RMS = 85 nm
5.0x10
20
4.0
3.0
2.0
1.0
0.0
n
e
/ cm
-3
Figure 5.30: Charge carrier density and electron mobility of as-grown In
2
O
3
:H,W, In
2
O
3
:H deposited
by
RPD
and as-grown sputtered In
2
O
3
:H with different thicknesses grown on soda lime glass and CIGS
samples with different RMS roughness values
rise up to 91 %of the
µe
value of the SLG reference at a TCO film thickness of around 621 nm,
the electron mobility of the In2O3:H film even exceeds the value on SLG at around 525 nm.
The results show that the electron mobility of films deposited on CIGS samples with RMS = 85 nm
by
RPD
increases with increased film thickness and that consequently the difference between the
values on SLG and CIGS samples is reduced. This indicates that the isolated film parts merge at
increased film thickness, similar to crystalline AZO films. In contrast to the films deposited by
RPD
the electron mobility of the sputtered In
2
O
3
:H drops with increased film thickness. This
applies for films grown on SLG and CIGS samples with two different RMS roughness values.
If coalescence of isolated film parts at increased film thicknesses is the origin of improvement
of the electron mobility on CIGS samples, it can be assumed that this does not occur during
the deposition by pulsed DC sputtering. It is possible that the unfavorable conditions during
sputtering, as described on page 86, obstruct the coalescence of the films grown on CIGS samples.
This effect seems to be less pronounced in the RPD layers.
For the
RPD
- In
2
O
3
:H layer a strong decrease of the carrier density at increased thicknesses
was observed. This might be caused by irregularities during the film deposition or an increased
film crystallinity in the as grown state, as also for the sputtered In
2
O
3
:H film a decreased carrier
density at higher thicknesses was observed. As only the tungsten doped sample does not show
94
5.5 Stability of IOH thin films
such a decrease, we assume that the tungsten prevents the formation of crystalline nuclei at
increased thicknesses.
A disadvantage of thick indium oxide layers might be the consequently increased optical absorption
and increased costs. Thus the beneficial effect of high mobility indium oxide based TCOs can be
reduced or even eliminated compared to cheap zinc oxide based TCOs such as AZO.
The main findings can be summarized as follows:
•
Improved electron mobility with higher film thicknesses possible (here, for films deposited
by RPD).
•Material might merge at increased thicknesses, leading to a less voids/cracks.
5.5 Stability of IOH thin films
As already mentioned in section 4.4, the substrate morphology also influences the stability of
IOH thin films. For evaluation IOH thin films with thicknesses of 270 nm were deposited on
glass substrates with different morphologies and roughnesses, determined by AFM measurements
with a measurement area of 40
µ
m x 40
µ
m. The corresponding topography images can be found
in Figure 5.31. The textures of the glasses were produced by etching. The RMS roughness values
of the glasses A, B and C were 1 nm, 76 nm and 122 nm, respectively. Thus, glass
A
can be
considered to be smooth. Glass
B
showed a fine-meshed structure while glass
C
showed larger
structures, but with similar heights as glassB. These two glasses are considered as rough.
(a) glassA(b) glassB(c) glassC
Figure 5.31: AFM topography images of glass substrates with different morphologies used for stability
tests; left: glass
A
, RMS = 1 nm; middle: glass
B
, RMS = 76 nm; right: glass
C
, RMS = 122 nm; note that
the scale of glassAdiffers from glassBand glassC
The results of the IOH stability tests carried out on these substrates are presented in Figure
5.32. Here, the change of the charge carrier density, electron mobility and resistivity of vacuum
95
Substrate Influences on Growth Mechanism and Properties
600
500
400
300
/ µ
W
cm
10
2 4 6 8
100
2 4 6 8
1000
2
time / h
0
120
80
40
0
µ
e
/ cm
2
V
-1
s
-1
Glass
A
; RMS = 1 nm
Glass
B
; RMS = 76 nm
Glass
C
; RMS = 122 nm
3.0x10
20
2.5
2.0
1.5
1.0
n
e
/ cm
-3
Figure 5.32: Change of the electrical properties of annealed IOH films during damp heat in dependence of
the glass substrate RMS roughness
annealed IOH films during damp heat is shown. In the initial state the electrical properties of all
samples were quite similar. It becomes apparent that IOH films degrade faster when deposited
on rough glass. Here, after 1000 h of damp heat the resistivity of IOH deposited on glass
B
and
glass
C
was found to increase by 74 %and 66 %, respectively. In contrast for the IOH film on
glass
A
only an increase of 28 %was observed. Thus the degradation is more than two times
faster on rough than on smooth glass substrates. Interestingly no large difference was found
between the two rough glass substrates. According to the measured RMS roughness value a more
pronounced degradation process was expected for the film deposited on glass
C
. The main cause
for the accelerated degradation is the larger decrease in electron mobility. No differences in the
change of charge carrier density were observed between the samples.
Additionally the stability of an approx. 340 nm thin IOH film on a CIGS sample was evaluated by
changes of the sheet resistance, measured by 4 point probe. The architecture of the CIGS sample
was glass/Mo/CIGS/InS/i-ZnO/IOH, after IOH deposition the sample was annealed in vacuum.
The RMS roughness was calculated to be 123 nm. Note, that in this case the AFM measurement
was conducted after IOH deposition with a reduced measurement area of 20
µ
m x 20
µ
m. The
96
5.5 Stability of IOH thin films
Figure 5.33: AFM topography image of the sample glass/Mo/CIGS/InS/i-ZnO/IOH before damp heat
corresponding AFM topography image is shown in Figure 5.33. As a reference an annealed IOH
film of the same deposition run grown on a smooth glass substrate was evaluated. The results are
shown in Figure 5.34. For comparison the change of
RSq
of the 270 nm thin IOH films deposited
on glass
A
and glass
B
are added to the graph. The sheet resistance of IOH films deposited on
glass substrates was calculated from the Hall data. The results show that the sheet resistance of
IOH which was deposited on the CIGS sample increases more pronounced (+ 111 %) than that of
the reference IOH thin film deposited on a smooth glass substrate ( + 29 %). Although the RMS
value of the CIGS sample is comparable to the one of glass
C
, a significantly lower
RSq
increase of
the IOH film on the rough glass substrate (+ 66 %) was found. No significant difference in the
degradation of
RSq
was observed between the two IOH films which were deposited on smooth
glass substrates.
The findings suggest that the morphology of the substrate has a large influence on the IOH
stability. Similar results can be found in literature for AZO. While studies from Greiner et
al. [115] suggested that grain boundaries are as such not strongly affected by damp heat for
AZO thin films grown on smooth substrates, the accelerated degradation of conductivity of AZO
grown on rough glass substrates was explained by the presence of local perturbations (extended
grain boundaries (
eGB
)). These do not merely increase the penetration of water vapor into the
film, but themselves block the current transport in the degraded state after damp heat. It is
conceivable that this model can also be applied to IOH layers grown on rough substrates. The
results suggest that the amount of
eGB
varies due to the substrate morphology and does not
necessarily rely on the substrate roughness but on the specific morphology and structures. Thus,
the resistivity of IOH thin films deposited on glass
C
, which had a high RMS, but showed rather
large structures, decreased approximately as fast as that of IOH films deposited on glass
B
, which
had a lower RMS, but a fine-meshed structure. We assume that
eGB
may be induced at rather
sharp edges. As discussed before, IOH thin films which were deposited on CIGS samples showed
presumably a high density of voids and cracks. This could additionally accelerate diffusion of
97
Substrate Influences on Growth Mechanism and Properties
Figure 5.34: Comparison of the change of the sheet resistance of annealed IOH thin films during damp
heat when deposited on smooth and rough glass substrates and a CIGS sample
water vapor into the films and enhance the degradation of IOH films. Therefore the pronounced
RSq
increase during damp heat of the IOH film grown on CIGS samples is probably caused by
both effects.
The main findings can be summarized as follows:
•
Accelerated degradation of electrical properties of IOH grown on rough substrates during
damp heat tests.
•
Morphology of the substrate and thus the resulting void density in the IOH film is crucial.
5.6 Conclusion
The findings obtained in this chapter are summarized in the chart diagram (Figure 5.35). In
this study the influences of the substrate and sub-layers on the growth and properties of indium
oxide based TCOs, i.e., In
2
O
3
:H and In
2
O
3
:H,W were investigated and grouped in two aspects:
(i) the structure and (ii) the roughness of the substrate/sub-layer. Two structural aspects of
the sub-layers were studied. In the first case the sub-layer contained residual water, due to the
deposition process. This led to degradation of the IOH films during annealing, as water from the
sub-layer might interfere the crystallization process and diffuse into the IOH layer. Consequently
the films exhibit poor electrical properties after annealing. In the second case the sub-layers
98
5.6 Conclusion
were poly-crystalline (ZnO). This promoted a crystalline growth of the deposited IOH layers
and result to poor electrical properties before and after annealing. In additional experiments
the hydrogen supply during the deposition of the first IOH layers at the ZnO interface were
increased, resulting in a decreased crystalline fraction of the IOH films and improved electrical
properties. The second main aspect was the roughness, i.e., the topography of the sub-layers or
of the substrate. In a first approximation the topography can be categorized in two structures:
(i) rounder shapes and grains on the surface with smooth transitions and (ii) sharp edged grains,
sharp transitions and sinks. When the surface showed rather smooth transitions the deposition of
a closed IOH films was possible. No adverse effect could be observed on the electrical properties,
as these were similar to films deposited on planar glass substrates. In contrast in films that were
deposited on sub-layers with sharp structures at the surface (in particular CIGS films) formation
of voids and cracks was observed. The voids were located mainly at CIGS boundaries, as here
the highest local slopes were observed. As the type of the topography was similar within the
CIGS samples, the surface could be well described by the RMS roughness value. However, no
significant influence of the deposition conditions or the crystalline fraction of the IOH films on
the void formation was found. The disrupted film structure resulted in accelerated degradation
of the electrical properties during damp heat and overall poor electron mobilities after deposition.
This applied for In
2
O
3
:H deposited by sputtering or RPD and for In
2
O
3
:H,W films deposited by
RPD. No sufficient improvement could be observed after annealing and solid phase crystallization
of the indium oxide films. To improve the electron mobility of the indium oxide based TCOs on
rough CIGS samples three different strategies were presented: (i) Sharp structures on the CIGS
surface were etched in an acid bromine solution. This resulted in rounder shaped grains. Thus
the deposited IOH films shows less voids and consequently higher electron mobilities. (ii) By spin
coating Ga
x
O
y
sol-gel layers sharp edged grains and sinks on the surface of the CIGS sample
were smoothed. The rounder shaped structures again led to formation of less voids and improved
electron mobilities. (iii) Furthermore it was observed, that films with increased thicknesses
also exhibit higher electron mobilities on rough CIGS samples. Therefore we assume, that the
material might coalesce at increased thicknesses and thus led to the sealing of voids.
The findings show, that the substrate/sub-layers of the indium oxide based TCOs have a major
effect on the growth and properties of the TCO films. Thus the application of such films in
photovoltaic applications in not straightforward. The different effects need to be considered to
achieve high-mobility indium oxide films.
99
Substrate Influences on Growth Mechanism and Properties
Substrate Influences on Growth Mechanism and Properties
Structure
Roughness
Residual water in sub-layer
Degradation of IOH films during
annealing
poor electrical properties after
annealing
Crystalline material (ZnO)
pronounced crystalline growth of
IOH
poor electrical properties before
and after annealing
can be avoided by increased H
supply during deposition of first
IOH interlayers
Round shapes / smooth transitions
Deposition of (almost) closed films
good electron mobilities before
and after annealing achievable
Sharp edges, transitions, sinks
Formation of voids / cracks in
In O :H and In layers
2 3 2 3
O :H,W
Not significantly affected by
deposition conditions or film
crystallinity after growth
Accelarated degradation of
electrical properties in damp
heat
poor electron mobility of as-
grown films
No sufficient improvement
after annealing
Removal of sharp structures by
etching with HBr
Covering sharp structures with spin
coated layer (GaOx)
Grains might coalesce at increased
thicknesses
Substrate
TCO
On CIGS, void density can be
correlated with RMS
roughness of the sub-layer
improved electrical properties
possible
Figure 5.35: Main findings concerning the influence of the substrate and sub-layers on the growth and
properties of indium oxide based TCO (i.e., In2O3:H, In2O3:H,W)
100
CHAPTER 6
Application of Indium Oxide based TCOs in CIGS solar cells
Due to their low optical absorption and high conductivity, as described in chapter 4, indium
oxides doped with hydrogen or metals (such as W) are promising candidates as front contacts in
solar cells. In chapter 5 influences of the sub-layers on the growth and properties of indium oxide
based TCOs were identified. When applying the films as front contact in CIGS cells or modules,
these influences have to be taken into account. In this we investigate indium oxide based TCO as
front contact. Some basic requirements need to be fulfilled for a successful implementation. These
are addressed in section 6.1. One requirement is a suitable band line up between the TCO and
the rest of the device, such as the highly resistive layer. A large ’cliff’ (negative conduction band
offset) or ’spike’ (positive conduction band offset) at the interface highly resistive layer/TCO
would result in deterioration of the solar cell parameters, here, the fill factor. Similar effects were
shown for the absorber/buffer interface [157,158]. The band line up of crystalline hydrogen doped
indium oxide and intrinsic zinc oxide (as the typical highly resistive layer between the buffer and
TCO, as explained in section 2.2) is therefore studied in section 6.1.1. Another basic requirement
is the thermal stability of the CIGS solar cells during the thermal treatment to induce solid phase
crystallization of the amorphous phase of the indium oxide based films. Here, temperatures
above 150
°
C are applied. In section 6.1.2 the effect of thermal treatments of up to
≈
210
°
C on
the solar cell performance of CdS buffered CIGS solar cells is studied. Additionally, alternative
configurations are suggested to avoid or reduce an adverse effect. Only when these requirements
are fulfilled, successful implementation of indium oxide based TCOs as front contact in CIGS
solar modules might be achieved. However, additionally, loss-free current transport through the
TCO on larger lateral distances than in a solar cell is required. Increased sheet resistance of TCO
layers on CIGS samples will therefore limit the module characteristics. Thus, the implementation
of hydrogen doped indium oxide as front contact in CIGS modules is discussed in section 6.2. In
section 5.4 strategies for improved TCO sheet resistances on CIGS samples were suggested. The
applicability of these approaches is tested in section 6.3. Note that the CIGS films presented
101
Application of Indium Oxide based TCOs in CIGS solar cells
in this chapter were deposited at different points in time, by different techniques and recipes,
resulting in CIGS absorbers with different properties and solar cell efficiency, as described in
section 3.1.2. Therefore, only the experiments within one series can be compared directly.
6.1 Basic Requirements for the Application as Front Contact
The successful application of indium oxide based TCOs, as described in this thesis, depends on
several requirements. One of two key aspects is a suitable band alignment of the front contact and
the highly resistive layer, e.g. intrinsic zinc oxide. In the following section we therefore investigate
the band line up of annealed hydrogen-doped indium oxide and intrinsic zinc oxide, the most
common highly resistive layer. The second key aspect is the thermal stability of the solar cells,
as temperatures above 150
°
C are required for the initiation of the solid phase crystallization,
depending on the TCO deposition conditions. The influence of the annealing treatment on the
solar cell performance is therefore studied in section 6.1.2.
6.1.1 Band alignment IOH/ZnO
The band line up of the IOH/ZnO heterostructure can affect the solar cell properties, e.g. the fill
factor. Kaspar et al. [159] investigated the band line-up of ZnO and ITO (In
2
O
3
:Sn), determining
a conduction band offset (
CBO
)
∆ECBM
= 0.6 eV (ZnO conduction band minimum (
CBM
)
above ITO
VBM
). For the calculations an indium oxide band gap of 2.9 eV was assumed, the
valence band offset (
VBO
) was determined to 0.05 eV to 0.25 eV (ZnO
VBM
above the ITO
VBM
). Such a large
CBO
of ZnO/IOH would lead to a significant deterioration of the FF in
CIGS solar cells. We numerically calculated the FF for standard CIGS/CdS/i-ZnO/IOH solar
cells for different
CBO
s. The results are shown in Figure 6.1 (a). For a fixed charge carrier
density of i-ZnO (here
ne(i−ZnO)
= 1x10
16
cm
−3
) the FF drops with increased conduction
band offset. For a conduction band offset
≥
0.4 eV a more pronounced FF loss was calculated for
decreased ZnO charge carrier density. The calculated band diagram of CIGS/CdS/i-ZnO/IOH
with
∆ECBM
= 0.5 eV and
ne(i−ZnO)
= 1x10
16
cm
−3
is shown in Figure 6.1 (b). To investigate
the conduction band offset of ZnO/IOH we determined the band line-up of ZnO/annealed IOH
by
UPS
and
XPS
measurements. For practical reasons the layer sequence was inverse to the one
in solar cells and it is assumed here that this does not change the band line-up. We evaluated a
bare annealed IOH film, IOH/ZnO heterostructures (ZnO was sputtered on annealed IOH in
total for about 1 s, 2 s, 3 s) and a thick ZnO films (sputtered for 79 s).
Figure 6.2 shown the core levels Zn 2p (left), In 3d (middle), In 4d and Zn 3d (right) for different
ZnO sputter duration. With increase of the sputter time the intensity of the In core peaks
102
6.1 Basic Requirements for the Application as Front Contact
80
75
70
65
60
FF / %
-0.6 -0.4 -0.2
0.0 0.2 0.4 0.6
E
CBM
/ eV
(a)
-5.0
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Energy / eV
2.82.62.42.22.0
Distance to back contact / µm
E
CBM
= - 0.5 eV
CIGSe
CdS
i-ZnO
TCO
E
CBM
E
VBM
E
Fn
E
Fp
(b)
Figure 6.1: (a) Calculated FF (by SCAPS [160]) in dependence on the conduction band offset i-ZnO/TCO;
(b) band diagram of CIGS/CdS/i-ZnO/TCO; E
CBM
and E
V BM
(black solid lines) are the conduction
band minimum and valence band maximum, respectively, E
F n
and E
F p
(red dashed line) represent the
quasi fermi levels of electrons and holes [76], the negative conduction band offset at the i-ZnO/TCO
interface indicate a ’cliff’
decreases while the Zn core peaks becomes more intense, indicating an increase of the ZnO film
thickness. After 79 s of sputtering only the Zn core peaks are measurable, indicating that a
sufficiently thick ZnO film was deposited. Measurements at the IOH/ZnO interface (1 s, 2 s, 3 s
of ZnO sputtering) revealed that In core levels In 4d as well as In 3d 5/2 shifted towards lower
binding energies compared to the uncoated film. The core level shifts are assumed to be due to
band bending (rather than chemical shift) and are therefore used in the calculation of the band
edges (see below). In contrast the Zn core levels Zn 2p and Zn 3d of the thin ZnO films (1 s,
2 s, 3 s) showed higher binding energies compared to the thick ZnO film. The measured valence
band of the samples is shown in Figure 6.3. The determined average
VBM EV BM
and binding
energies of the core levels In 4d, In 3d 5/2, Zn 2p and Zn 3d are presented in Table 6.1, the error
was calculated with the standard deviation of the linear regression.
The valence band maximum of the bare annealed In
2
O
3
:H film was observed at
≈
2.89 eV below
EF
, consistent with values found in literature for In
2
O
3
[161,162]. For bulk ZnO a valence
band maximum of
EZnO
V BM ≈
3.35 eV below
EF
was obtained, which is also consistent with
literature [163
–
165]. The valence band offset
∆EV BM
was determined according to the method
developed by Kraut et al. [166], which is well established [121,167–170]:
∆EV BM = (EZnO
CL −EZnO
V BM )−(EIOH
CL −EIOH
V BM )−∆ECL (6.1)
103
Application of Indium Oxide based TCOs in CIGS solar cells
20
15
10
5
0
Intensity / arb. units
1026 1024 1022 1020 1018
Binding Energy / eV
IOH + 1s ZnO
IOH + 2s ZnO
IOH + 3s ZnO
IOH + 79s ZnO
Zn 2p
20
15
10
5
0
Intensity / arb. units
455 450 445 440
Binding Energy / eV
IOH
IOH + 1s ZnO
IOH + 2s ZnO
IOH + 3s ZnO
IOH + 79s ZnO
In 3d 5/2
In 3d 3/2
20
15
10
5
0
Intensity / arb. units
20 15 10
5 0
Binding Energy / eV
IOH
IOH + 1s ZnO
IOH + 2s ZnO
IOH + 3s ZnO
IOH + 79s ZnO
In 4d
Zn 3d
ZnO sputter
duration
Figure 6.2: Core levels Zn2p (left), In3d (middle) and In4d &Zn3d (right) in dependence of the layer
configuration with increasing ZnO thickness
Table 6.1: Valence band maximum
EV BM
and binding energies of the core levels In 4d, In 3d 5/2, Zn 2p
and Zn 3d measured by UPS and XPS, respectively, for different IOH/i-ZnO film configurations and ZnO
sputter durations
Sample EV BM Error ECL(In4d)ECL(In3d5/2)ECL(Zn2p)ECL(Zn3d)
(eV) (eV) (eV) (eV)
IOH 2.887 0.006 18.22 444.79 - -
IOH + 1s ZnO 3.204 0.010 18.11 444.68 1022.2 10.803
IOH + 2s ZnO 3.333 0.008 18.13 444.69 1022.2 10.879
IOH + 3s ZnO 3.341 0.003 18.12 444.66 1022.3 10.902
IOH + 79s ZnO 3.351 0.006 - - 1022.1 10.570
∆ECL =EZnO
CL (i)−EIOH
CL (i)(6.2)
where
EZnO
CL
and
EIOH
CL
are the binding energies of the core level of thick ZnO and bare IOH,
respectively, (i) indicates values at the interface.
EZnO
V BM
and
EIOH
V BM
represent the corresponding
valence band maximum of thick ZnO and bare IOH, respectively. Calculations were conducted
with the average
EV BM
of the materials, as shown in Table 6.1. For better statistics, the average
values of the core levels at the interface (1 s, 2 s, 3 s) were determined for
EZnO
CL
(i) and
EIOH
CL
(i),
respectively. The
∆EV BM
was calculated for all 4 combinations of both In and Zn core levels. The
resulting valence band offset was determined to
∆EV BM
= (-0.784
±
0.057) eV . The conduction
band offset ∆ECBM can be estimated with the following equation:
∆ECBM =∆EV BM +EZnO
g−EIOH
g(6.3)
104
6.1 Basic Requirements for the Application as Front Contact
300
250
200
150
100
50
0
Intensity / arb. units
5 4 3 2 1 0
Binding Energy / eV
IOH
IOH + 1s ZnO
IOH + 2s ZnO
IOH + 3s ZnO
IOH + 79s ZnO
Figure 6.3: Valence band of bare and ZnO coated IOH after different ZnO sputter duration
with
EZnO
g
and
EIOH
g
as the band gaps of ZnO and In
2
O
3
:H. The band gap of ZnO is well known
as
EZnO
g
= 3.3 eV [171]. Previous studies have reported that In
2
O
3
has a fundamental band
gap of 2.6 eV - 2.9 eV [54
–
58], Walsh et al. [59] determined the upper limit of the fundamental
band gap to 2.9 eV. Therefore we assumed
EIOH
g
= 2.75
±
0.15) eV. We calculated the
CBO
for IOH/ZnO to
∆ECBM
= (-0.234
±
0.207) eV (
CBM
of In
2
O
3
:H below the
CBM
of ZnO).
The large error is due to the large inaccuracy of the assumed indium oxide band gap. However,
within this range we expect no adverse effect on the FF, as shown in Figure 6.1 (a).
These results contradict those of Kaspar et al. [159] who considered a cliff between the ZnO
CBM
and In
2
O
3
:Sn
CBM
. This can be attributed to the different (opposed) determined valence
band offsets, as the authors concluded that the ZnO
VBM
is above the In
2
O
3
:Sn
VBM
. Similar
findings were obtained by Kamiya et al. [172] by ultraviolet and inverse photoemission spectra.
In contrast we observed the ZnO
VBM
below the In
2
O
3
:H
VBM
and the ZnO
CBM
below the
In
2
O
3
:H
CBM
. These results are supported by the findings of Song et al. [173], who also reported
a ZnO
VBM
below the In
2
O
3
:H
VBM
(
∆EV BM
= 0.49
±
0.11) eV). Assuming
EZnO
g
= 3.37 eV
and EIn2O3
g= 2.93 eV a conduction band offset of (-0.05 ±0.26) eV was calculated.
We investigated the band line up by depositing ZnO on annealed IOH films. However, conventional
CIGS solar cells are fabricated in the substrate configuration. In this case, IOH is deposited
onto ZnO. Furthermore, the as grown IOH film is considered to be amorphous after growth and
crystalline after the annealing. Therefore the band line up of ZnO/IOH might slightly differ
when the materials are applied in CIGS solar cells. However, we found no evidence for an adverse
effect on the fill factor by the band line up of ZnO/IOH.
105
Application of Indium Oxide based TCOs in CIGS solar cells
The main findings can be summarized as follows:
•
The conduction band offset of crystallized In
2
O
3
:H (
EIOH
g
= 2.75
±
0.15) eV) and intrinsic
ZnO (
EZnO
g
= 3.3 eV) was measured to be
ECBM
= (-0.234
±
0.207) eV. For this offset,
no adverse effect on the fill factor is expected. We therefore conclude that In
2
O
3
:H and
i-ZnO have a suitable band line up.
6.1.2 Effect of Post Deposition Thermal Treatment on CIGS Solar Cells
As determined in the previous section the band alignment of crystalline In
2
O
3
:H and intrinsic
ZnO is considered suitable for the application of In
2
O
3
:H as front contact in CIGS solar cells.
Therefore one of the key requirements, as listed on page 102 is satisfied. A further key aspect is
the thermal stability of the solar cells which will be investigated in this section.
Several groups studied the usability of hydrogen doped indium oxide films as front contact in
CIGS solar cells [68,71,90,91,150,174,175]. Koida et al. [90] studied the impact of IOH front
contacts in combination with different highly resistive layers on the performance of CIGS solar
cells in relaxed and metastable states. Witte et al. [175] compared the cell performance for
different window layer configurations. Similar or higher efficiencies were demonstrated with
an as grown IOH front contact compared to an AZO front contact when a Zn(O,S) buffer was
used. The CdS buffer cells showed in contrast lower efficiency due to reduced
jsc
and
FF
.
However, Keller et al. [68] showed that the short circuit current density of CdS buffered CIGS
cells improved due to implementation of an annealed RF sputtered IOH front contract by
≈
3 mA/cm
2
. The AZO reference cell had a
jsc ≈
31 mA/cm
2
, the cell with annealed IOH an
improved
jsc
of
≈
34 mA/cm
2
. Additionally an improved open circuit voltage but decreased fill
factor were observed. In total, the cell with the annealed sputtered IOH front contact showed a
higher average efficiency of ≈15.4 %than the reference with AZO with η≈14.7 %.
However, we observed that the implementation of an annealed IOH layer into a standard CIGS
sample with CdS/i-ZnO window is not straightforward. Solid phase crystallization of amorphous
grown hydrogen doped indium oxide films requires temperatures of 150
°
C to 220
°
C, depending
on the deposition conditions [5]. When applied as a front contact, the annealing procedure will
also affect the CIGS cell. We therefore investigated the properties of CIGS solar cells with IOH
front contact before and after annealing. A set of CIGS samples, fabricated by a sequential
process, were coated with CdS buffer, i-ZnO as highly resistive layer and sputtered IOH front
contact. The samples were annealed for 10 min, 20 min, 30 min and 60 min in vacuum with
infrared heaters. The obtained average temperatures and electrical properties of the reference
IOH films grown on glass substrates are shown in Table 6.2. The results indicate the initiation
of solid phase crystallization already after 10 min of annealing at 157
°
C, as a slight decrease in
106
6.1 Basic Requirements for the Application as Front Contact
ne
and increase in
µe
was measurable. However, only after 60 min and temperatures of 211
°
C
an electron mobility over 100 cm2/Vs was achieved.
Table 6.2: Electrical properties of reference IOH films deposited on glass substrates after annealing under
varied conditions; for annealing different specimens of one glass substrate were used
t T neµeρ
(min) (°C) (cm−3) (cm2/Vs) (µΩcm)
0 23 ±0 3.9x1020 43.2 374
10 157 ±4 3.7x1020 47.1 362
20 179 ±2 3.3x1020 53.3 358
30 190 ±2 2.8x1020 60.6 368
60 211 ±1 2.1x1020 106.1 283
0.55
0.50
0.45
0.40
0.35
V
OC
/ V
40
36
32
28
24
j
SC
/ mA cm
-2
70
60
50
40
30
20
FF / %
AZO BL
AZO RT
AZO 60
IOH RT
IOH 10
IOH 20
IOH 30
IOH 60
14
12
10
8
6
4
2
0
η
/ %
AZO BL
AZO RT
AZO 60
IOH RT
IOH 10
IOH 20
IOH 30
IOH 60
Figure 6.4: Box plots of
jsc
,
Voc
,
FF
and
η
of CIGS solar cells (60 cells each) with AZO or IOH front
contact; the AZO baseline (BL) film was sputtered at
≈
165
°
C, RT: deposited at room temperature
without intentional heating; numbers indicate the duration of annealing in minutes
The j-V characteristics of the corresponding CIGS solar cells (60 cells per sample) are shown in
Figure 6.4. Further CIGS cells with AZO front contact were prepared in three configuration:
1)sputtered at a substrate temperature of
≈
165
°
C, 1.5 wt.%Al
2
O
3
; 2) sputtered without
intentional heating, 1.0 wt.%Al
2
O
3
; 3)sputtered without intentional heating with subsequent
annealing for 60 min, 1.0 wt.%Al
2
O
3
. The j-V characteristics were added to Figure 6.4 for
comparison. In the following the average values of the cells serve for evaluation. For all cells the
shunt resistance was found to be in the range of 1 k
Ω
cm
2
or higher, thus can be excluded to
adversely affect the fill factor. Compared to the reference AZO BL sample, the non-annealed cells
107
Application of Indium Oxide based TCOs in CIGS solar cells
with AZO and IOH front contact, sputtered without intentional heating, showed higher
jsc
and
Voc
, but a decreased fill factor by
≈
5%to 8 %absolute. After 10 min of annealing the IOH-cell
showed a slightly improved fill factor, while the
Voc
dropped in average by 25 mV. No significant
change was observed for
jsc
. However, further annealing up to 30 min results in an
jsc
loss of
almost 6 mA/cm
2
, steadily decreased
Voc
and fill factor by
≈
20 mV and 1.6 %, respectively.
Note, that despite of the observed losses, no significant change of series resistance, ideality factor
or saturation current density was determined. Nevertheless, after 60 min of annealing all three
values increased significantly, indicating changes of the charge carrier recombination mechanism.
Additionally the
Voc
and
FF
decreased further. In contrast an improvement in
jsc
by 3 mA/cm
2
could be observed, still the determined value was in average 2.6 mA/cm
2
lower than for the non
annealed IOH sample. A similar effect was assessed for the cells with AZO front contact before
and after annealing. Also here degradation of the
jsc
,
Voc
and
FF
occurred after annealing, even
more pronounced than for the cells with IOH. The samples which were annealed for 60 min both
showed similar series resistances, ideality factors and saturation current densities. The j-V curves
showed no kink effect.
Table 6.3: Average j-V characteristics of CIGS solar cells with different front contacts (AZO and IOH); the
AZO baseline (BL) film was sputtered at
≈
165
°
C, RT: deposited at room temperature without intentional
heating; numbers indicate the duration of annealing in minutes; series resistance, shunt resistance, ideality
factor and saturation current density were determined by fits of the dark curves (2 diode model) of cells
with j-V characteristics close to the average values
.
sample jsc Voc FF η RSRSh AJ0
(mA/cm2) (V) (%) (%) (Ωcm2) (kΩcm2) - (mA/cm2)
AZO BL 37.5 0.492 69.2 12.8 0.45 3.01 1.4 4.9x10−8
AZO RT 39.1 0.503 61.6 12.1 0.92 1.94 2.1 1.5x10−6
AZO 60 28.6 0.405 24.9 2.9 1.02 2.67 2.4 2.9x10−6
IOH RT 38.1 0.511 64.6 12.6 0.91 1.82 1.9 8.0x10−7
IOH 10 38.3 0.486 66.7 12.4 0.47 3.24 1.5 7.6x10−8
IOH 20 32.6 0.473 65.9 10.2 0.40 0.97 1.4 5.6x10−8
IOH 30 32.5 0.466 65.1 9.9 0.44 1.37 1.4 4.3x10−8
IOH 60 35.5 0.457 51.5 8.4 1.06 1.21 2.2 2.0x10−6
Furthermore
EQE
of the AZO baseline and the IOH cells was evaluated. The results are shown
in Figure 6.5 (a). It is apparent, that the
EQE
of cells with AZO and as-grown IOH are very
similar, for these cells no benefit due to the IOH layer can be observed. The sample annealed for
10 min shows an increased
EQE
in the near infrared region, indicating an increased transmittance
due to decreased charge carrier density or improved electron mobility. However, in the short
wavelength region the
EQE
decreased, resulting in a similar
jsc,EQE
as the non-annealed IOH
cell. For prolonged annealing duration the
EQE
drops over the whole, but more pronounced in
108
6.1 Basic Requirements for the Application as Front Contact
the short wavelength region. Similar to the results obtained by j-V measurements the
EQE
of
the sample annealed for 60 min revealed a higher
EQE
and thus a higher
jsc,EQE
compared to
the samples annealed for 20 min or 30 min. However, for the short wavelength region still a loss
was observed. Figure 6.5 (b) illustrates the doping profile of the CIGS samples with IOH front
contact before and after several annealing durations, as determined by C-V measurements. The
data revealed a decrease charge carrier density and a shift towards higher depletion widths with
increased annealing time. After 60 min of annealing
NC
decreased from
≈
2.5x10
15
cm
−3
to
values as low as ≈7.0x1014 cm−3at the minimum.
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
EQE
140012001000
800600400
Wavelength / nm
AZO BL;
Jsc=39.37 mA/cm
2
IOH RT;
Jsc=39.41 mA/cm
2
IOH 10;
Jsc=39.53 mA/cm
2
IOH 20;
Jsc=34.09 mA/cm
2
IOH 30;
Jsc=33.66 mA/cm
2
IOH 60;
Jsc=38.2 mA/cm
2
(a)
10
14
2
4
6
10
15
2
4
6
10
16
2
4
6
10
17
N
C
/ cm
-3
1.51.00.50.0
Depletion width x / µm
as grown
10 min
20 min 30 min
60 min
(b)
Figure 6.5: (a) External quantum efficiency of CIGS samples with baseline-AZO front contact (AZO
BL), sputtered at
≈
165
°
C or IOH front contact, deposited without intentional heating (IOH RT) and
after several annealing durations, indicated by numbers; (b) Charge carrier densities in dependence of
the depleation width of CIGS solar cells with IOH front contact in the as-grown state and after several
annealing durations, determined by C-V measurements; values at V = 0 V are marked with diamonds;
corresponding temperatures can be found in Table 6.2
The results demonstrate degradation of the CIGS solar cells at increased temperatures, induced
for the solid phase crystallization of the amorphous phase in IOH layers, applied as the front
contact. With increased annealing temperature a steady drop of the
Voc
from 0.511 V to 0.457 V
was observed, which most likely results by the steady decrease of the charge carrier denisity, as
determined by C-V measurements. The observed drop correlates well with calculated
Voc
losses
1
.
1
4VOC =AkT
qln(N1
N2
)(6.4)
with A as diode quality factor, k as the Boltzmann constant, T as the temperature, q as elemental charge and
N1
,
N2as the charge carrier concentration of the absorber before and after annealing
109
Application of Indium Oxide based TCOs in CIGS solar cells
The annealed cells showed an inclined
EQE
at small wavelengths, already after 10 min. Such
an incline in combination with the observed decreased charge carrier density indicate Shockley-
Read-Hall (
SRH
) bulk recombination, at least for the samples annealed up to 30 min. With
decreased charge carrier density and increased space charge region (SCR) the SCR recombination
area shift towards the CIGS bulk, resulting in a higher amount of recombined electrons and holes
close the the window layer. As in this region mainly photons with higher energies are absorbed,
the loss in photo-current is more pronounced at small wavelengths. The overall low
EQE
of
the films annealed for 20 and 30 min, respectively, indicates bulk recombination over the whole
absorber region. After 60 min of annealing (211
°
C) the recombination mechanism seems to
change significantly. Surprisingly the
jsc
increased by 3 mA/cm
2
. Furthermore the FF dropped
drastically while the ideality factor increased to values > 2, thus exceeding the limiting value for
SRH
recombination of 2.0. This also applies for cells with AZO front contact, which showed a
similar drop of the
Voc
and
FF
. As the j-V characteristics after 60 min of annealing were similar
to the ones obtained for samples with IOH front contact, similar recombination mechanism are
suggested. However, in order to confirm this surprising recovery of the
jsc
in cells with IOH front
contact the experiments must be reproduced.
These results are supported by the findings of Wi et al. [176] who reported degradation of CdS
buffered CIS solar cells after annealing at temperatures above 200
°
C. The authors attributed
the degradation to Cd diffusion from the CdS buffer into the CIS absorber. Additionally it
was demonstrated that CIS cells with Zn(O,S) buffer showed a better thermal stability (up to
300
°
C). It was shown, that after annealing Zn atoms diffused not as far as Cd atoms into the
CIGS absorber. The findings are consistent with the work of Park et al. [177] who reported Cd
diffusion from the CdS buffer layer towards the CIGS absorber after annealing at 200
°
C. Here
the CIGS/CdS/ZnO interface was monitored in situ during annealing using TEM apparatus
equipped with an in situ heating stage. Similar results were obtained by Kijima et al [178],
but for temperatures above 360
°
C. However, also in this study the ZnS(O,OH) showed a
better thermal stability for temperatures up to 400
°
C. Based on these findings we suggest
that Cd diffusion might initiate the observed degradation of the solar cells annealed at higher
temperatures, i.e., above 180
°
C (20 min annealing). Furthermore the degradation might be
due to additional processes, such as sodium diffusion. However, to study the degradation and
recombination mechanism in more detail, further investigation, such as temperature dependent
j-V measurements are necessary.
The findings show that the thermal treatment required for the solid phase crystallization of
amorphous grown hydrogen doped indium oxide lead to degradation of the CdS buffered CIGS
solar cells. The results are contrary to the findings reported by Keller et al. [68], where no
degradation of the CdS buffered CIGS solar cell occurred after annealing in vacuum at 200
°
C
for 1 h. We therefore assume that the degradation mechanism highly depends on the condition
110
6.1 Basic Requirements for the Application as Front Contact
of the CdS/CIGS interface. To prevent cell degradation the interface must therefore be improved.
Other approaches, such as the change of the TCO deposition conditions and change of the buffer
layer are discussed in the following.
Stategies for Improved Cell Performance After Annealing
As shown in the previous section degradation of the solar cell performance might occur after
annealing of CIGS solar cells with CdS buffer and i-ZnO/IOH window. Notwithstanding possible
improvement of the CdS/CIGS interface, which is not subject of this thesis, we suggest two
approaches to prevent CIGS solar cell degradation caused by the post-deposition thermal
treatment required to initialize solid phase crystallization of hydrogen doped indium oxide:
1. Change of the deposition conditions of IOH thin films to increase the nucleation
density in the amorphous matrix of as grown films. This will lead to reduction of the
required crystallization temperature [5] and the related heating of the CIGS solar
cell. This will consequently minimize degradation.
2. Change of the applied buffer layer for improved thermal stability, e.g. Zn(O,S),
as reported in literature [176,178]. Steigert et al. [71] reported that solar cells with
sputtered Zn(O,S) buffer layer and IOH front contact showed a slightly improved
efficiency after annealing (180
°
C, 10 min in ambient air) due to an improved
jsc
by 0.3 cm
2
/Vs. However, still a slight decrease of the
Voc
by 8 mV was observed.
Overall the efficiencies of the cells with IOH were similar to the one with AZO as
front contact.
Both approaches are briefly presented in the following.
Change of the TCO deposition conditions
Reduction of the water partial pressure during deposition of amorphous indium oxide based
TCOs will result in an increased nucleation density in the amorphous matrix and thus reduce
the crystallization temperature, as shown by Koida et al. [5]. However, a too low water partial
pressure can cause crystalline grown, strained films and therefore lead to poor electrical properties
(see section 4.2). This has to be considered for the application in CIGS. Following the work
of Koida et al. [5] we deposited indium oxide films co-doped with hydrogen and tungsten by
reactive plasma deposition on (multi-source) CIGS/CdS/IGZO (indium gallium zinc oxide)
2
at
low (pH2O≈1x10−5Pa) and high (pH2O≈1x10−4Pa) water vapor pressures, the oxygen flow
2The IGZO in this case serves as an amorphous highly resistive layer replacing the standard i-ZnO
111
Application of Indium Oxide based TCOs in CIGS solar cells
0.66
0.65
0.64
0.63
0.62
0.61
0.60
V
OC
/ V
17.0
16.0
15.0
14.0
13.0
η
/ %
low p(H2O); as grown
low p(H2O); annealed 150°C
high p(H2O); as grown
high p(H2O); annealed 220°C
0.78
0.76
0.74
0.72
0.70
0.68
0.66
0.64
FF / %
low p(H2O); as grown
low p(H2O); annealed 150°C
high p(H2O); as grown
high p(H2O); annealed 220°C
35.5
35.0
34.5
34.0
33.5
33.0
j
SC
/ mA/cm
2
Figure 6.6: Box plots of
jsc
,
Voc
,
FF
and
η
of CIGS solar cells (8 cells each) with different window layers
was set to 60 ml/min. The experiments were carried out at AIST, where among other things
IGZO is used as the highly resistive layer. The annealing conditions were optimized (as shown in
section 5.3, i.e. in Figure 5.14) and set for low
pH2O
to 150
°
C for 60 min and for high
pH2O
to 220
°
C for 30 min in N
2
atmosphere at a pressure of 7x10
4
Pa. The main results obtained
by j-V measurements are shown in Figure 6.6 and reported in the following. Cells annealed at
low temperatures (150
°
C) showed improved
jsc
,
Voc
and
FF
after annealing. Consequently the
efficiency improved in average from 14.7 %to 15.9 %. In contrast, cells that were annealed at
higher temperatures (220
°
C) degraded after annealing, although the
jsc
improved in average by
0.5 mA/cm
2
. The
Voc
and
FF
decreased causing an efficiency drop from 16.1 %to 14.8 %. The
high Voc of the sample before annealing might resulted from inhomogeneities of the absorber.
Change of the buffer layer
As discussed above Zn(O,S) might be a suitable substitutional buffer layer in CIGS solar cell, as
it is suggested to be thermally more stable and therefore the more suitable buffer layer in CIGS
solar cells with annealed indium oxide based TCOs. We therefore investigated the performance of
(sequentially processed) CIGS solar cells with IOH front contact in combination with a Zn(O,S)
buffer layer before and after an annealing in vacuum at 180
°
C for 1 h. Note that at such
temperatures a degradation of the CdS buffered solar cells was observed. We studied the impact
of 280 nm thick AZO, as-grown and annealed IOH (120 nm) as well as the impact of the Zn(O,S)
buffer layer, deposited by ALD. Although the IOH films are approx. half as thick as the AZO
film, a sheet resistance of less than 30
Ω
/Sq on glass was achieved in the as grown state. The
112
6.1 Basic Requirements for the Application as Front Contact
0.50
0.45
0.40
0.35
V
OC
/ V
42
40
38
36
34
32
j
SC
/ mA cm
-2
70
65
60
55
50
45
FF / %
CdS/i-ZnO/AZO
Zn(O,S)/i-ZnO/AZO
Zn(O,S)/i-ZnO/IOH as grown
Zn(O,S)/i-ZnO/IOH annealed
13
12
11
10
9
8
7
η
/ %
CdS/i-ZnO/AZO
Zn(O,S)/i-ZnO/AZO
Zn(O,S)/i-ZnO/IOH as grown
Zn(O,S)/i-ZnO/IOH annealed
Figure 6.7: Box plots of
jsc
,
Voc
,
FF
and
η
of CIGS solar cells (60 cells each) with different window layers
AZO film was deposited in the sputtering tool VISS300 following the baseline recipe. These
layers typically show sheet resistances of
≈
38
Ω
/Sq on glass. Figure 6.7 presents the results
of
jsc
,
Voc
,
FF
and
η
of CIGS solar cells with different window layer configurations in box
plots, obtained from j-V measurements. For each configuration 60 solar cells were processed.
Table 6.4 summarizes the median solar cell parameters. By substituting the CdS layer with
Zn(O,S) an increase in the short circuit current density of 4.4 mA/cm
2
is observed. Merdes et
al. [141] predicted that the exchange of the CdS buffer with Zn(O,S) can result in an improved
jsc
of 1.2 mA/cm
2
, due to the wider band gap of Zn(O,S) compared to CdS. Consequently the
optical absorption loss in the short wavelength region decrease [71,179]. The large increase in
jsc
observed here can therefore not be attributed exclusively to the exchanged buffer layer, but
might result also from more favorable interference fringes in the external quantum efficiency. The
open circuit voltage and fill factor were found to decrease, which led to a lower median efficiency
of the cells with Zn(O,S)/AZO. Replacing AZO with as grown IOH led to further improvement
of the
jsc
by 0.9 mA/cm
2
, but also to additional losses in
Voc
, FF and efficiency. Note, that the
AZO layer was deposited at
≈
165
°
C, while the IOH layer was deposited without intentional
heating. The heat during AZO deposition might have improved the CIGS/Zn(O,S) interface
quality. The annealed IOH sample showed improved
Voc
,
FF
as well as
jsc
. Consequently the
efficiency of the annealed IOH sample was in median higher compared to the cells with as grown
IOH. Also compared to the cells with AZO front contact the median efficiency improved by 0.8 %
for Zn(O,S) and by 0.3 %for CdS buffered samples. The results show that no adverse effect can
113
Application of Indium Oxide based TCOs in CIGS solar cells
be observed for the thermal treatment carried out at 150
°
C, as even the solar cell parameters
Voc
and
FF
improved. However, already in the as grown state the cells with IOH front contact
had in average an improved
jsc
comprated to the cells with AZO front contact. The findings
demonstrate, that the thin (
≈
120 nm) IOH layer was sufficient for current collection and that
in average a similar fill factors as for the cells with AZO front contact can be achieved.
In Figure 6.8 the internal quantum efficiencies of the best cells with Zn(O,S) buffer and AZO or
annealed IOH front contact, respectively, are shown. The IQE spectra illustrates the beneficial
effect of the IOH front contact over the whole wavelength region of the solar cell, especially in
the near infrared region (800 nm to 1200 nm) compared to the standard AZO front contact. This
results from the lower free carrier absorption of the IOH layer. It is further apparent, that the
IQE is close to 1 in the range of approx. 500 nm to 900 nm. This result suggest that the current
gain would be even higher than the obtained
jsc
, when combining the sample stack with an anti
reflective coating, such as MgF. However, already without an anti reflective coating the efficiency
was improved in average by 0.8 %by replacing AZO with annealed IOH.
Table 6.4: Median parameters of CIGS solar cells with different window layers (60 cells per sample)
buffer HR-layer TCO jsc Voc FF η
(mA/cm2) (V) (%) (%)
CdS i-ZnO AZO 33.9 0.484 66.6 10.8
Zn(O,S) i-ZnO AZO 38.3 0.440 62.3 10.3
Zn(O,S) i-ZnO IOH as grown 39.2 0.434 58.0 9.9
Zn(O,S) i-ZnO IOH annealed 39.3 0.460 62.2 11.1
1.0
0.8
0.6
0.4
0.2
0.0
IQE
140012001000800600400
Wavelength / nm
AZO
IOH annealed
Figure 6.8: Internal quantum efficiency (IQE) of CIGS solar cells with Zn(O,S)/i-ZnO/AZO and Zn(O,S)/i-
ZnO/IOH (annealed) window layers, respectively
114
6.2 Challenges in the Application as Front Contact in CIGS Modules
Additional studies (not shown here) confirmed the thermal stability of CIGS cells with Zn(O,S)
buffer (deposited by sputtering or
ALD
) and IOH front contact compared to CdS buffered Cells.
For the same annealing conditions a degradation of the
jsc
,
Voc
and
FF
was determined for the
CdS buffered cells. In contrast, cells with the Zn(O,S) buffer layer improved in average in all j-V
parameter after annealing. Consequently the efficiency could be improved by 1 %with stack
ALD-Zn(O,S)/i-ZnO/IOH annealed compared to a reference CdS/i-ZnO/AZO stacked sample.
Furthermore the results confirmed the beneficial effect of IOH as front contact in CIGS solar
cells.
The main findings can be summarized as follows:
•
The post deposition thermal treatment (T above
≈
160
°
C) can degrade CdS buffered solar
cells when the absorber/buffer interface is of poor quality.
•
Change of the In
2
O
3
:X deposition conditions for increased nuclei density in the amorphous
matrix results in reduction of the crystallization temperature and thermal impact on the
solar cell. Degradation can be avoided. However, this might lead to increased TCO sheet
resistances.
•
By substitution of the CdS buffer by a thermally more stable buffer layer (e.g. Zn(O,S))
degradation of the solar cells can be avoided.
6.2 Challenges in the Application as Front Contact in CIGS
Modules
As shown in the previous section the basic requirements for the application of a high mobility
indium based TCO, a suitable band line up and thermal stability can be satisfied. Therefore in
this chapter we investigate the applicability of In
2
O
3
:H as front contact in CIGS modules (CIGS
deposition by a sequential process). We study the structure and electrical properties of In
2
O
3
:H
films before and after annealing, when deposited on the coated CIGS samples. Additionally we
determine the module characteristics by j-V measurements and correlate them with the properties
of the TCO layers.
Implementing In
2
O
3
:H or related high-mobility indium based TCOs as front contact in CIGS
modules is not straightforward. As discussed in section 5.3 an increase of the sheet resistance was
observed when the films were grown on rough CIGS samples. This holds for mainly amorphous as
well as for pronounced crystalline growth. TCO film resistance is, however, a decisive parameter.
In CIGS modules a sheet resistance of approx. 10
Ω
/Sq or less is required to avoid fill factor
losses caused by an increased series resistance. To evaluate the impact of an (
≈
300 nm) IOH
115
Application of Indium Oxide based TCOs in CIGS solar cells
front contact compared to AZO (
≈
865 nm), CIGS modules with several buffer layers were
fabricated. The overview of the samples is shown in Table 6.5.
Table 6.5: Sample assignment of CIGS modules with different buffer and TCO layers;
Ab,a
: annealing
before/after TCO deposition, respectively, conducted in vacuum at ≈180 °C for 1 h
ID buffer HR-layer AbTCO Aa
M1 CBD CdS sputter i-ZnO - AZO -
M2 CBD CdS sputter i-ZnO - IOH -
M3 sputter Zn(O,S) sputter i-ZnO - IOH -
M4 sputter Zn(O,S) sputter i-ZnO - IOH +
M5 ALD Zn(O,S) ALD i-ZnO - IOH -
M6 ALD Zn(O,S) ALD i-ZnO - IOH +
M7 ALD Zn(O,S) ALD i-ZnO + IOH -
M8 ALD Zn(O,S) ALD i-ZnO + IOH +
M9 soda lime glass - IOH -
M10 soda lime glass - IOH +
The crystalline structure of the samples was evaluated by
GI-XRD
measurements. The different
deposition techniques of the Zn(O,S)/i-ZnO layers result in ZnO films with different orientation.
The ALD deposited ZnO films showed preferential orientation of crystallites in (100) and (101)
direction of the surface normal while the sputtered ZnO films only show (002) peaks. The
measurements also revealed a higher crystalline fraction of the as grown IOH film deposited on
sputtered i-ZnO than the film on glass, in accordance to the findings described in sections 5.1 and
5.3. The corresponding X-ray diffraction patterns are shown in Figure 6.9 (a). The crystalline
fraction of IOH was lower when the films were grown on i-ZnO layers deposited by ALD and
decreased further when the ALD layers were annealed prior the IOH deposition. Annealing the
samples in vacuum at 180
°
C to 200
°
C for 70 min led to solid phase crystallization of the IOH
film, as revealed by X-ray diffraction measurements, shown in Figure 6.9 (b).
The sheet resistance of the TCO layers, as shown in Table 6.6, was determined by transmission
line (TLM) and 4-point measurements. The determined values showed the same trend, though
the values measured by TLM were slightly higher in average than the values determined by
4-point measurement, as illustrated in Figure 6.10. This systematic error might result from the
different measurement set up, as for the 4 point measurements it is assumed, that the sample
size is significantly larger than the distance of the probes.
Table 6.6 presents the module parameters
jsc
,
Voc
,
FF
and
η
, determined by j-V measurements
as cell equivalent and the series and shunt resistance of the module, estimated by fits of the dark
curves. Modules M1 and M2 were produced with a AZO and IOH (as grown) front contact,
116
6.2 Challenges in the Application as Front Contact in CIGS Modules
(a) (b)
Figure 6.9: X-ray diffraction patterns of CIGS module samples with different window layer before (a)
and after (b) annealing of the IOH films; IOH/glass is used as reference; reference patterns of In
2
O
3
were
taken from PDF 00-006-0416, of ZnO from PDF 01-070-8070; patterns were shifted vertically for improved
differentiation
Figure 6.10: Correlation of the TCO sheet resistance on CIGS samples determined by transmisstion line
(TLM) and by 4-point measurements
respectively, in otherwise the same layer configuration. By applying the as grown IOH, a gain
of almost 3 mA/cm
2
in
jsc
was achieved while the open circuit voltage was not influenced. A
major disadvantage of the modules with IOH was the significantly lower FF and high series
resistance which was affecting the efficiency. As described in section 2.2, the series resistance
depends on the TCO sheet resistance. Following the work of Hoppe et al. [180] we defined an
effective TCO sheet resistance
Reff,T CO
, which determines the contribution of the TCO sheet
resistance RSheet,T CO to the total series resistance:
Reff,T CO =RSheet,T CO ·l
3w(6.5)
117
Application of Indium Oxide based TCOs in CIGS solar cells
Table 6.6: Overview over the module properties determined by j-V measurements; series resistance of the
module was determined by fits of the dark curves (2 diode model) and the sheet resistance of the samples,
determined by TLM and 4 point measurements
Sample jsc Voc FF η RS,Module RSq,T LM RSq,4−point
(mA/cm2) (V/cell) (%) (%) (Ω) (Ω/Sq) (Ω/Sq)
M1 32.7 0.492 59.5 9.6 8.6 8.8 8.5
M2 35.5 0.490 35.2 6.1 29.4 - -
M3 24.4 0.299 27.9 2.0 50.2 57 46
M4 34.4 0.466 33.2 5.3 30.1 79 70
M5 37.1 0.435 42.3 6.8 12.4 53 37
M6 36.1 0.405 33.3 4.9 17.9 47 42
M7 38.4 0.495 44.7 8.5 12.3 36 28
M8 31.9 0.459 42.2 6.2 17.0 32 47
M9 - - - - - - 12
M10 - - - - - - 10
where
l
is the cell length and
w
the cell width. To calculate the total contribution of the TCO
RS,T CO
to the series resistance in a module
RS,Module
the effective sheet resistance
Reff
is
multiplied by the amount of cells connected in series Ns:
RS,T CO =Ns·Reff,T CO (6.6)
Figure 6.11 (a) presents the change of the modules series resistance
RS,Module
over the determined
contribution of the TCO
RS,T CO
for modules with IOH and AZO front contact. For this
calculation the sheet resistance determined by TLM was used. The dashed line represents equal
values of
RS,Module
and
RS,T CO
for improved visualization. The lowest
RS,T CO
and
RS,Module
were determined for the AZO reference module. The results indicate, that the increased
RS,Module
results almost exclusively from the increased
RS,T CO
. An exception occurs for module M3 (non-
annealed IOH with sputtered Zn(O,S) buffer), here a significantly higher
RS,Module
is observed.
Therefore an additional major contribution to
RS,Module
is assumed for this module. Additionally
a significantly low open circuit voltage and short circuit current density were measured for this
module. Note that other contributions to the module series resistance, e.g. contact resistance,
Mo-resistivity, have to be taken into account. Therefore, the real module series resistance is > 0
even for negligible TCO sheet resistance (
RS,T CO
= 0). The fill factor
FF
is known to depend on
the series resistance, as described in equation (2.34) in section 2.2. Thus, with increased
RS,T CO
also a drop of the fill factor occurs, as presented in Figure 6.11 (b). The dashed line estimates
the calculated FF loss caused by the TCO resistance, according to equations (2.34) and (2.35).
118
6.2 Challenges in the Application as Front Contact in CIGS Modules
The AZO module was taken as reference. It is apparent, that the experimental FF values are
lower than the calculated values. Thus additional losses can be assumed, i.e., in the modules
with the ID M3,M6 and M8 (see Table 6.5 and 6.6). The contact resistance, determined by
TLM measurements, of modules with IOH and AZO were quite similar. The shunt resistance
of the modules with IOH were approx. in the same range as the shunt resistance of the AZO
module. Thus the IOH front contact had no influence on the shunt resistance.
(a) (b)
Figure 6.11: Correlation of the measured module series resistance (a) and fill factor (b) with the contribution
of the TCO sheet resistance to the total module series resistance; the dashed line in (b) indicates the
calculated FF loss caused by increased RS,T CO
The results support the findings discussed in sections 5.1 and 5.3, as again a dependency of
the crystalline fraction of as grown IOH on the sublayer, here ZnO, was found. The IOH film
grown on sputtered ZnO showed a higher crystalline fraction than the film on bare glass. The
crystalline fraction of IOH films decreased when the films were deposited on ALD - ZnO. We
assume that this is due to the different structure of the ZnO films, as the deposition technique
of the ZnO layer influences its growth, structure and preferred orientation. Surprisingly, the
crystalline fraction of IOH further decreased, when deposited on a pre annealed ALD - ZnO
sample. We assume that the annealing may improve the grain quality of the ZnO films and
lead to an increase of the grain size on the film surface. For the deposition by ALD H
2
O is
used a the precursor for the oxygen component. Therefore it is likely, that hydrogen or water
molecules are incorporated in the films. The annealing may mobilize hydrogen and lead to its
diffusion on the film surface, where it supports amorphous growth of the IOH film. Also in other
experiments a lower crystalline fraction of as grown IOH and a consequently lower resistivity
was found for IOH films which were grown on annealed ALD ZnO films compared to films grown
on non-annealed ALD ZnO films. However, to fully understand this effect, further investigations
are needed. The increased crystallinity of the as grown IOH films resulted in an increased sheet
resistance, confirmed by both TLM and 4-point measurements. By the 4-point probe method
119
Application of Indium Oxide based TCOs in CIGS solar cells
the resistance of the whole multilayer compound is measured, therefore an influence of the IOH
sub-layers can not be excluded. However, as the conductivity of the IOH thin film is significantly
higher than that of intrinsic ZnO or other sub-layers, we assume that the current mainly flows
through the TCO layer rather than the sub-layers. Therefore we assume that the effect of the
sub-layers during 4 point probe measurements is negligible. This is supported by the fact, that
the determined values are similar to those measured by TLM. The increased sheet resistance
resulted in an increased module series resistance and decreased fill factor. Although a mainly
amorphous structure of the as grown IOH was found for module M7, the IOH sheet resistance
was still higher than that on glass, which showed a comparable mainly amorphous structure. This
raise in sheet resistance can be likely explained by the formation of voids/cracks, as described in
section 5.3. Annealing of the samples after IOH deposition led to solid phase crystallization and
to changes of the electronic properties of the IOH thin films. According to the 4-point probe
measurements the IOH sheet resistance was higher after annealing when the films were deposited
on CIGS samples. In contrast, a lower sheet resistance was determined when the films were
deposited on glass substrates. Measurements by TLM revealed contrary results. Here a decrease
of the sheet resistance after annealing was found for the IOH films which were deposited on
ALD ZnO layers. However, the improvement was not sufficient, the poor efficiency was still
determined by the low fill factor caused by the increased series resistance due to the high IOH
sheet resistance.
Furthermore, the corrspondingly low performance of module M3 can not only be attributed to
a high TCO sheet resistance. For this module additionally to the low
FF
a major drop in
jsc
and
Voc
occurs. We relate this to a high acceptor defect density at the CIGS/Zn(O,S) interface,
which might be caused by sputter damage [181]. Annealing of the sample after TCO deposition
seems to reduce the acceptor defect density at the interface, resulting in a improved module
performance.
The module efficiency was limited mainly be the high IOH sheet resistance, caused by an
pronounced crystalline growth and/or presumably crack formation in the IOH layer. As discussed
in section 5.3 voids or cracks in the IOH layer result in a poor electron mobility and thus a high
sheet resistance. Therefore, to increase the IOH module efficiency, the electrical properties of
the IOH layer, especially the electron mobility need to be improved. In section 5.4 strategies for
improved IOH electron mobility on CIGS samples are presented. In the following these strategies
are applied in CIGS solar cells and modules to evaluate their usability.
The main findings can be summarized as follows:
•
A high In
2
O
3
:H sheet resistance limits the fill factor and therefore the efficiency of CIGS
solar modules. The increased sheet resistance was caused by pronounced crystalline growth
120
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells
on poly-crystalline ZnO sub-layers as well as by a high void/crack density when grown on
CIGS with sharp structures.
6.3 Application of strategies for improved TCO electron mobil-
ity in CIGS solar cells
The last section pointed out, that unfavorable growing conditions of In
2
O
3
:H directly on the CIGS
samples can result in high sheet resistances, high series resistances and therefore in limitation of
the fill factor and module efficiency. The main goal is consequently to improve the sheet resistance
of the TCO when grown on the CIGS samples. In section 5.4 three possibilities were presented,
i.e., deposition of a spin coated sub-layer, etching of the CIGS absorber and significant increase
of the film thickness. In this section we now investigate the applicability of these strategies into
working devices.
6.3.1 Spin Coated Sol-Gel Layer
A spin coated sol-gel layer deposited on CdS buffered CIGS samples results in a smoother
surface and an improved IOH electron mobility, as described in section 5.4.1. However, this
procedure is only reasonable, if it does not adversely affect the solar cell properties. Figure 6.12
shows box plots of the characteristic j-V values of 8 CIGS solar cells (CIGS by multi-source
process), each with different highly resistive layers: i-ZnO as reference or spin coated Ga
x
O
y
layers, deposited multiple times, as described in section 3.1.2. The j-V curves of the best cells
are presented in Figure 6.13, the corresponding solar cell parameters are summarized in Table
6.7. The Ga
x
O
y
/CdS/CIGS samples are identical to the ones described in section 6.3.1. For all
samples as-grown IOH (approx. 500 nm thick) was used as front contact. The observed trend
of the cell properties is the same for the best cell and the average value, thus the properties
of the best cells are discussed. The highest efficiency was achieved with the i-ZnO reference
sample. Note that the CIGS deposition process is slightly inhomogeneous, the sample taken
from the central position was found to show the highest
Voc
and efficiency. The i-ZnO reference
CIGS sample was fabricated on such a position. Replacing the i-ZnO layer with Ga
x
O
y
leads
to deterioration of the solar cells with increasing amount of Ga
x
O
y
depositions and Ga
x
O
y
thickness, respectively, as a decrease in
jsc
,
Voc
and
FF
was observed. With increased Ga
x
O
y
amount the series resistance increases significantly, while the parallel resistance drops. The
j
-
V
curve of the sample "6xGa
x
O
y
" revealed a roll over effect, indicating a barrier for the diode
current [76], caused by acceptor states or a positive conduction band offset at buffer/window
interface, respectively. However, the sample processed with 1xGa
x
O
y
showed properties similar
121
Application of Indium Oxide based TCOs in CIGS solar cells
to the reference cell. It can not be excluded, that the improved performance of the i-ZnO cell is
due to the most favorable position during CIGS deposition.
40
35
30
25
j
SC
/ mA cm
-2
0.70
0.65
0.60
0.55
0.50
V
OC
/ V
20
15
10
5
0
80
60
40
20
FF / %
i-ZnO
1xGaOx 3xGaOx 6xGaOx
η
/ %
i-ZnO
1xGaOx 3xGaOx 6xGaOx
Figure 6.12: Box plots of characteristic properties of solar cells with different highly resistive layers: spin
coated Ga
x
O
y
, deposited several times and i-ZnO in combination with as grown IOH as front contact on
CdS buffered CIGS samples; 8 solar cells each
80
60
40
20
0
-20
-40
j / mA cm
-2
1.00.80.60.40.20.0
-0.2-0.4
U / V
i-ZnO
1xGaO
x
3xGaO
x
6xGaO
x
dark
illuminated
Figure 6.13: j-V curves of solar cells with spin coated Ga
x
O
y
, deposited several times and i-ZnO as highly
resistive (HR) layers
Based on these results CIGS modules with CdS/1xGa
x
O
y
/IOH were produced and compared with
a CdS/i-ZnO/AZO reference module, before and after 2 h of light soaking. The
≈
580 nm thick
IOH film was deposited without intentional heating, after deposition the module was annealed
in vacuum at
≈
180
°
C for 1 h. The
≈
865 nm AZO film was deposited at
≈
165
°
C substrate
temperature. The
j
-
V
curves of the modules are shown in Figure 6.14, the characteristic values
of the modules as cell equivalent are summarized in Table 6.8.
122
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells
Table 6.7: Parameters of the best CIGS solar cell with spin coated Ga
x
O
y
, deposited several times and
i-ZnO as highly resistive (HR) layers and IOH as front contact; fits were conducted for dark curves with a
two diode model (see section 2.2.2)
HR-layer TCO jsc Voc F F η RsRp
(mA/cm2) (V) (%) (%) (Ωcm2) (kΩcm2)
i-ZnO IOH 37.6 0.649 74.3 18.1 0.46 11.0
1xGaxOyIOH 37.0 0.639 74.1 17.5 0.53 32.1
3xGaxOyIOH 33.2 0.620 64.5 13.3 15.9 73.4
6xGaxOyIOH 33.2 0.602 28.0 5.6 6998 133.0
By implementing Ga
x
O
y
/IOH instead of i-ZnO/AZO as window layer, a gain in
jsc
of
≈
2 mA/cm
2
was realized. In addition, the sample exhibits a 23 mV higher open circuit voltage befor light
soak. After light soak
3
the voltage of the i-ZnO/AZO module dropped surprisingly by 43 mV
while the
Voc
of the Ga
x
O
y
/IOH module decreases only by 4 mV. Also the FF of the i-ZnO/AZO
module decreased after light soaking, while it increased in the Ga
x
O
y
/IOH module. The short
circuit current density, however, slightly dropped in both modules after light soak. Nevertheless,
these effects result in a higher efficiency of the module with Ga
x
O
y
/IOH of
∆η
= 1.1 %before
light soak and
∆η
= 2.7 %after light soak. Note that the series resistance of the i-ZnO/AZO
is 2.3
Ω
before and after LS while the series resistance of the Ga
x
O
y
/IOH module decreased
from 4.4
Ω
to 2.7
Ω
. Thus after LS similar series resistances for both module configurations
were achieved.
80
60
40
20
0
-20
-40
j / mA cm
-2
1.00.80.60.40.20.0
-0.2-0.4
U / V
i-ZnO / AZO before LS
i-ZnO / AZO after LS
GaO
x
/ IOH before LS
GaO
x
/ IOH after LS
values as cell equivalent
dark
illuminated
Figure 6.14: j-V curves of modules with CdS/i-ZnO/AZO and CdS/Ga
x
O
y
/IOH window, respectively,
before and after 2 h light soak (average values per cell)
3
as mentioned in section 3.1.2 light soaking was is a common procedure for CIGS devices and was performed
for several modules within this thesis
123
Application of Indium Oxide based TCOs in CIGS solar cells
Table 6.8: j-V parameter of modules with CdS/i-ZnO/AZO and CdS/Ga
x
O
y
/IOH window, respectively,
before and after 2 h light soak; Rswas fitted with a 2 diode model for the dark curves
LS HR-layer/TCO jsc Voc FF η Rs
(mA/cm2) (V/cell) (%) (%) (Ω)
before i-ZnO/AZO 32.0 0.591 58.7 11.1 2.3
after i-ZnO/AZO 31.9 0.548 55.7 9.7 2.3
before GaxOy/IOH 34.1 0.614 58.1 12.2 4.4
after GaxOy/IOH 33.7 0.610 60.2 12.4 2.7
The results show that application of a thin Ga
x
O
y
film (1 spin coat deposition) as the highly
resistive layer leads to working solar cells and modules. Thicker Ga
x
O
y
layers smoothen the
surface and gradually improve the electron mobility of IOH as reported in section 5.4.1, but lead
to the formation of a barrier in the solar cells, which could be reduced after light soaking. We
assume that the light soak did not affect the IOH layer or i-ZnO/AZO layers, as here no effect on
the series resistance due to light soak was observed. Gallium oxide has a high optical band gap
over 4 eV, depending on the O/Ga ratio [120,182]. This is higher than the band gap of 3.3 eV
for ZnO. Thus the Ga
x
O
y
layer has furthermore the potential to transmitt more photons with
higher energies to the absorber layer and consequently to reduce parasitic absorption in the blue
optical region. Heinemann et al. [182] reported, that oxygen vacancy V
O
donor states in Ga
x
O
y
cause defect bands below the conduction band maximum. Thus, a low (high) oxygen deficiency
resulted in a high (low) optical band gap, correlated with a low (high) electron affinity. A low
electron affinity can lead to a high barrier in the conduction band maximum at the Ga
x
O
y
/CdS
interface. We assume that the deposited Ga
x
O
y
films have a low amount of V
O
, as the films
exhibit a very low conductivity, presumably resulting from a low charge carrier density. Moreover
we assume that if a high density of V
O
were present in the film, the charge carrier density and
thus the conductivity would be higher. Additionally no optical band gap in the range below 5 eV
was observed in spin coated Ga
x
O
y
films deposited on quartz glass (not shown). We assume
that the 1xGa
x
O
y
layer is thin enough to allow tunneling or that the CIGS sample is not fully
covered, yet. With increased deposition amount the layer thickness increases, resulting in a
barrier for the passing current and the observed increase of the series resistance. By choosing a
more conductive material with a suitable electron affinity or by induce V
O
deep donor bands
in Ga
x
O
y
the barrier might be reduced leading to working solar cells even with thicker spin
coated sol-gel layers and without a the need of a light soak. However, the CIGS absorber might
be damaged by the annealing process in air (200
degree
C) of the sol-gel deposition procedure.
Nevertheless, the combination of a thin (1 deposition) Ga
x
O
y
layer, which slightly smoothed the
CIGS sample surface, and a thick (500 nm) annealed IOH layer resulted in a working module with
an efficiency higher than that of the i-ZnO/AZO reference module. We assume that the increased
IOH thickness was beneficial, since voids might coalesce at higher thicknesses, as discussed in
124
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells
section 5.4.3. Therefore a respectively low series resistance and high fill factor was achieved with
Ga
x
O
y
/IOH. A further approach might be the modification of the sol-gel layer, as described
above, to realize working solar cells with thicker sol-gel layers, which additionally provide a
smooth sample surface. This configuration allows the deposition of IOH with reasonable electron
mobilities, without the formation of voids and cracks in the layer, as discussed in section 5.4.1.
This further improvement of the IOH layer might enable a reduction of the film thickness to
achieve reasonable sheet resistances < 10 Ω/Sq of IOH films on CIGS samples.
The main findings can be summarized as follows:
•
Working devices can be realized with thin spin coated sol-gel Ga
x
O
y
layers. For the
In
2
O
3
:H/Ga
x
O
y
configuration an improved CIGS module efficiency compared to the
standard i-ZnO/AZO configuration was achieved.
6.3.2 Etched CIGS
As described in section 5.4.2 the CIGS surface can be smoothed by etching in a acid bromine
solution which results in an improved electron mobility of hydrogen doped indium oxide grown
on i-ZnO/CdS/etched CIGS. However, the etching in acid bromine solution followed by a KCN
treatment changes the surface properties of the CIGS film. This might lead to poor solar cell
performance. Therefore we investigate the potential of the etched CIGS samples described in
section 5.4.2 as solar cells (CIGS deposited by multi-source process).
Figure 6.15 shows the j-V curves (a) and the external quantum efficiency (b) of the samples.
Table 6.9 summarizes the determined solar cell parameters. The improved electrical properties of
the In
2
O
3
:H layer with increased etching duration (see section 5.4.2) result in a decreased series
resistance, estimated by a two-diode fit of the dark j-V curve. The current - voltage measurements
revealed further that the short circuit current density slightly decreases from 32.6 cm
2
/Vs to
31.1 cm
2
/Vs with increased etch duration. This was confirmed by EQE measurements where
the overall values were slightly lower. The loss occurs mainly in the range of around 700 nm to
1000 nm. The open circuit voltage was found to increase with increased etching duration. After
15 s of etching an increase of almost 30 mV could be observed. For a prolonged etch duration up
to 60 s the
Voc
increased only slightly by another 10 mV. Also the fill factor was found to improve
from 73 %to 75 %after 60 s of etching. As a result the efficiency improved from 14.9 %for the
non-etched reference up to 15.7 %for all etched samples. Due to the compensation of decreased
jsc
but increased
Voc
for increased etch duration the efficiency within the etched samples did not
change.
125
Application of Indium Oxide based TCOs in CIGS solar cells
We studied this unexpected effect of etching on cell performance, in particular on the
Voc
, in
more detail using CV to evaluate the doping profile of the etched samples. Figure 6.16 shows the
doping profile of the etched samples after CdS deposition. The carrier concentration at V = 0 V
of the non etched reference is approx. 8.0x10
14
cm
−3
. After 15 s of etching a decrease down to
5.8x10
14
cm
−3
occurs. However, a longer etching time improves the carrier concentration up to
1.7x1015 cm−3.
-40
-30
-20
-10
0
10
20
30
40
j / mA cm
-2
1.00.80.60.40.20.0
-0.2-0.4
U / V
0s
15s
30s
60s
(a)
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
EQE
140012001000800600400
Wavelength / nm
0s
15s
30s
60s
etch duration
(b)
Figure 6.15: j-V (a) and EQE (b) curves of the etched CIGS samples with CdS/i-ZnO/IOH window layer
in the as grown state
Table 6.9: Solar cell parameters of etched CIGS samples with as-grown IOH front contact;
Rs
and
Rp
were fitted with a 2 diode model for the dark curves
Etch jsc Voc F F η RsRp
(s) (mA/cm2) (V) (%) (%) (Ωcm2) (kΩcm2)
0 32.6 0.628 73 14.9 1.03 25.660
15 32.3 0.656 74 15.7 0.62 18.99
30 31.8 0.660 75 15.7 0.42 12.31
60 31.1 0.668 75 15.7 0.36 9.06
The data shows that the solar cell performance was improved by the etching procedure, mainly
by an increase of the
Voc
up to 40 mV. It is known from literature that increased net acceptor
density in polycrystalline Cu(In,Ga)Se
2
can improve the open circuit voltage [183]. However,
the large increase of
≈
30 mV after 15 s of etching can not be explained by this, as the carrier
concentration was found to slightly decrease. According to the CIGS 3-stage deposition process
the top layer, which forms in the 3rd stage, might be of poor quality compared the the film,
which forms in the second stage. Here the film re-crystallizes, resulting in reduced stress and
crystallographic disorder (see section 3.1.2). By etching the CIGS absorber in a bromine solution
126
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells
10
14
2
4
6
8
10
15
2
4
6
8
10
16
2
4
6
8
10
17
N
C
/ cm
-3
2.01.51.00.50.0
Depletion width x / µm
0V
0s
15s
30s
60s
Figure 6.16: Doping profile (charge carrier concentration
NC
versus depletion width
x
) as calculated form
CV-measurements from the etched CIGS samples
the low quality layer might be removed, resulting in an improved interface. This might be the
cause of the improved open circuit voltage. The observed decreased short circuit current density
of the bromine etched samples is likely due to reflection losses caused by smoothing of the surface,
similar to the findings by Deprédurand et al. [149]. The actual reflection could not measured,
because the cell area was too small.
The results of this investigation show that by etching CIGS samples in an acid bromine solution
in combination with a KCN treatment a capping layer with poor quality, according to the 3-stage
deposition procedure, can be removed, which improved the solar cell performance of the samples.
Additionally in section 5.4.2 it is shown that this process smoothed the surface and results in an
improved electron mobility of the hydrogen doped indium oxide layer which were deposited on
the ZnO-coated CdS/CIGS samples. However, no improvement of the
jsc
was obtained due to
the CIGS etching.
The main findings can be summarized as follows:
•
Etching of CIGS absorber with an acid bromine solution improved the solar cell efficiency.
This results most likely by the removal of absorber material of poor quality from the
absorber surface.
•
Improved electrical properties of In
2
O
3
:H on etched samples (see section 5.4.2) led to
improved series resistances in the solar cells.
127
Application of Indium Oxide based TCOs in CIGS solar cells
6.3.3 Increase of TCO thickness
An increased thickness of the IOH film can improve its electron mobility on CIGS samples as
shown in section 5.4.3. However, an increased film thickness results in an increase of parasitic
optical absorption and can decrease the short circuit current density in solar cells. We investigated
therefore the usability of IOH front contacts with thicknesses in the range of approx. 450 nm to
750 nm in CIGS modules with Zn(O,S) as buffer layer and highly resistive layer and compared
them to modules with AZO front contact. The CIGS absorber layer was deposited by a sequential
process at AVANCIS. After IOH deposition the modules were annealed in vacuum at
≈
180
°
C
for 1 h. As a reference one non-annealed module with 550 nm IOH was fabricated. Based
on the results presented in section 5.4.3 we assumed an improvement of the series resistance
with increased IOH thickness. However, CIGS samples with the present configuration and
roughness were not studied before and can therefore result in unexpected results, as for this
CIGS absorber a rather round-shaped grain structure with smooth surfaces is assumed. In fact,
for this configuration an IOH thickness of 450 nm was sufficient to achieve series resistances as
low as 1.1
Ω
cm
2
. Figure 6.17 presents the characteristic module parameters of the samples as
cell equivalents, before and after light soaking. In the following the results are discussed in more
detail.
After light soak the short circuit current density of the annealed IOH module is
≈
1.4 mA/cm
2
higher than the non-annealed reference module, indicating solid phase crystallization due to
the annealing. Within the annealed IOH modules no trend in
jsc
was observed. However, the
jsc
after light soaking improved by
≈
0.5 mA/cm
2
when annealed IOH was applied as front
contact instead of AZO. Modules with the annealed IOH showed average fill factors of
≈
67.7 %,
higher than the best FF of the AZO module with 64.8 %. The high FF results from the low
series resistance of
≈
1
Ω
cm
2
for the modules with annealed IOH, regardless of the IOH film
thickness. No dependence was found between the shunt resistance and the variation of the front
contact. However, a drop in
Voc
was observed in the modules which were fabricated with IOH
front contact, more pronounced in the annealed IOH modules after light soak. The efficiency of
the best AZO reference module after light soak was 14.5 %. The best IOH modules after light
soak (750 nm) had an efficiency of 14.3 %and before light soak (650 nm) 14.5 %.
The results show again that an IOH front contact can improve the short circuit current density in
CIGS modules compared to an AZO reference. No adverse effect in
jsc
was found with increased
IOH thickness due to parasitic optical absorption. We assume that this effect results from a shift
of the interference fringes to more favorable wavelengths, which counteracts absorption losses of
the reflection spectra due to changes of the film thickness. This might result in a configuration
most favorable for the module with 750 nm IOH, although the highest parasitic absorption and
thus the lowest
jsc
is expected. In the IOH modules the fill factor was not significantly limited
128
6.3 Application of strategies for improved TCO electron mobility in CIGS solar cells
0.66
0.64
0.62
0.60
0.58
V
OC
/ V
36
35
34
33
32
31
j
SC
/ mA cm
-2
70
60
50
40
FF / %
15
14
13
12
11
10
9
8
η / %
before LS
after LS
4000
3000
2000
1000
0
R
Sh,Cell
/
W
cm
2
AZO Ref 1
AZO Ref 2
IOH 550 nm as grown
IOH 450 nm anneald
IOH 550 nm anneald
IOH 650 nm annealed
IOH 750 nm annealed
10
8
6
4
2
0
R
S,Cell
/
W
cm
2
AZO Ref 1
AZO Ref 2
IOH 550 nm as grown
IOH 450 nm anneald
IOH 550 nm anneald
IOH 650 nm annealed
IOH 750 nm annealed
Figure 6.17: J-V parameter as cell equivalent of CIGS modules with different TCO configurations before
(black) and after (red) light soaking
by the series resistance, as it is the case in the AZO reference modules. However, no dependence
of the series resistance on the IOH film thickness was found, thus we assume that the TCO sheet
resistance was similar for the annealed IOH films, although a low sheet resistance is expected for
thicker films. This can be caused by the following: during deposition films are heated by plasma
irradiation, which can induce crystalline nuclei in the films. With continued film deposition, these
nuclei induce a crystalline phase in the newly deposited material, leading to an total increased
crystalline fraction and to strain in the films. This deteriorates the electrical properties, limiting
improvement of the sheet resistance at high film thicknesses [184].
The drop in
Voc
was observed for the annealed as well as the non-annealed IOH modules, thus it
can not be explained by deterioration of the absorber during annealing. We assume that the
Zn(O,S) buffer layer might react with -OH or -H in the plasma during deposition of the first IOH
layers. This could lead to changes in the Zn(O,S) composition and the formation of Zn(O,S,OH).
Changes in composition can lead to unfavorable band alignments and recombination currents at
the interface [185] lowering the open circuit voltage, similar to the observed results. However, no
Voc drop was observed by Steigert et al. [71] in CIGS cells with Zn(O,S)/IOH window.
129
Application of Indium Oxide based TCOs in CIGS solar cells
The main findings can be summarized as follows:
•
Thick In
2
O
3
:H did not reduce the short circuit current density in CIGS modules, probably
due to shifts of reflection fringes.
•
For this configuration a low series resistance and reasonable fill factor was achieved with
annealed In2O3:H, also at film thicknesses of 450 nm.
•Similar module efficiencies could be achieved with In2O3:H and ZnO:Al front contact.
6.4 Conclusion
The findings obtained in this chapter are summarized in the following Chart diagram (Figure
6.18).
In this section the applicability of indium oxide based TCOs as front contact in CIGS cells and
modules was studied. For the successful implementation as a front contact several requirements
have to be fulfilled. Two basic requirements are (i) a suitable band line-up between the TCO
(here annealed IOH) and the highly-resistive layer (here i-ZnO) and (ii) the thermal stability
of the cell due to annealing of the TCO films. In this study the band line-up of annealed,
crystallized IOH and i-ZnO was measured. The conduction band offset was calculated to be
∆ECBM
= (-0.234
±
0.2) eV. The large error results from the large uncertainty of the assumed
In
2
O
3
band gap. However, numerical simulations revealed no adverse effect on the solar cell
parameter for such conduction band offset. Thus we conclude that the band alignment of these
two materials is suitable for photovoltaic applications. Annealing of IOH layers, which were
deposited on CdS buffered CIGS samples resulted in degradation of the solar cell parameter.
We assumed that this was due to a poor CIGS/CdS interface quality. To improve the thermal
stability of the cells, we suggested two different strategies beside the improvement of the interface:
(i) The deposition parameter of the indium oxide based TCOs has to be modified in such a
way, that low crystallization temperatures can be realized. Annealing of such TCO layers and
corresponding CdS buffered CIGS solar cells at 150
°
C revealed no adverse effect. In fact, the
solar cell properties improved. (ii) Substitution of the CdS buffer to thermally more stable buffers
(e.g. Zn(O,S)) results in an increased temperature process window during annealing. In fact, no
degradation of Zn(O,S) buffered solar cells could be observed after annealing at 180 °C.
As the mentioned requirements were satisfied, the IOH films were applied as front contacts in
CIGS modules. Indeed a gain of the
jsc
(up to 2.8 mA/cm
2
) was observed for modules with IOH
front contact in comparison to a module with AZO front contact. However, the fill factor was
limited due to a high series resistance. It was shown, that the increase of the series resistance
130
6.4 Conclusion
results from an increased TCO sheet resistance, caused by void formation in the layer and partly
also by pronounced crystalline growth. To improve the series resistance in modules therefore the
electrical properties, i.e., the electron mobility of the indium oxide based TCOs has to improve.
Thus the strategies for an improved TCO electorn mobility on CIGS samples presented in section
5.4 were evaluated for ther applicability: (i) The deposition of spin coated Ga
2
O
3
sol gel layers
as highly resistive layers resulted in a barrier with increased Ga
2
O
3
thickness. However, modules
with a thin layer showed reduced series resistances and improved fill factors and even higher
efficiencies than a reference module with standard i-ZnO/AZO window. This demonstrated the
potential of the application of spin coated amorphous layers as template for IOH deposition.
(ii) Etching of the CIGS absorber led beside of the improved electron mobility of the IOH films
also to an improved
Voc
by up to 40 mV compared to an non-etched reference. The etched solar
cells showed an overall improved efficiency by
≈
0.8 %. Additionally the series resistance of the
cell was found to decrease while the fill factor increased. Thus, also this strategy is suitable for
the application in CIGS solar cells. (iii) No beneficial effect was observed in modules when thick
IOH films were deposited as front contact (450 nm to 750 nm). For the used CIGS absorber type
(rather smooth/rounder shaped surface) already a thickness of 450 nm was sufficient to achieve
series resistances of
≈
1
Ω
cm
2
. The fill factor was also reasonable high. However, for the applied
layer configurations a reduced
Voc
was observed. In total, similar high module efficiencies could
be achieved with IOH front contacts.
We can conclude, that IOH films have a high potential as front contact in modules, as gain in
the short circuit density with reasonable fill factors was demonstrated. However, specific CIGS
layer configuration and surface topography is required to for full beneficial effects.
131
Application of Indium Oxide based TCOs in CIGS solar cells
Basic Requirements
Application of Indium Oxide Based TCOs as Front
Contact in CIGS Devices
Suitable Band Alignment
IOH/ZnO interface
suitable; = - 0.234 ± 0.2 eV∆ECBM
Thermal Stability of Cells During
Annealing
CdS buffered cells with i-ZnO as
highly resistive layer
Cell degradation probably due to
bad interface
Modification of the IOH deposition
parameter to reduce crystallization
temperature
No degradation of CdS buffered
cells during annealing at 150°C
Substitution of buffer layer
material
No degradation of Zn(O,S)
buffered cells during annealing at
180°C
Front Contact in Modules
Improved ; no degradation ofjSC
other parameters
Limitation of the due to highFF
IOH sheet resistances on CIGS
samples
Improvement of electrical
properties of IOH when grown on
CIGS samples (see section 5.4)
Spin coated GaO sub-layer
X
- suitable when very thin
- improved module efficiencies
possible
- thick layer cause cell degradation
due to barrer, thermal impact
Improved by 2.8 mA/cm whenjSC
2
exchanging AZO with as grown
IOH on i-ZnO/CdS
IOH front contact on several
buffer/highly resistive layers
Smoothening of CIGS surface by
etching
effective, even improved VOC
possible
Increase of IOH thickness
- for applied configuration 450 nm
was suitable
- no improvement with thicker
films
- similar efficiencies with AZO and
IOH front contact
Improved jSC in CIGS modules with IOH front contact was demonstraded,
but required for full beneficial effectspecific CIGS configuration
Issue Investigation Result
Issue Investigation Result
Figure 6.18: Main findings concerning the applicability of indium oxide based TCOs (i.e., In
2
O
3
:H,
In2O3:H,W) as front contact in CIGS solar cells and modules
132
CHAPTER 7
Conclusions and Outlook
The topic of this thesis were new high-mobility transparent conductive oxides (TCOs) for the
application as front contact in Cu(In,Ga)(S,Se)
2
(CIGS) devices. For this hydrogen doped (and
tungsten co-doped) indium oxide (In
2
O
3
:H, In
2
O
3
:H,W) thin films were investigated. The aim
was to take advantage of the combined high transparency and low resistivity of these materials
and to implement them as the front contact in CIGS solar cells and modules. The target was to
improve the current collection and efficiency in CIGS modules and thus to reduce the relative
losses with regard to CIGS solar cells. Moreover a successful implementation in modules requires
a homogeneous large-scale process, which was developed in this thesis.
We successfully transferred the deposition process of In
2
O
3
:H to an in-line pulsed DC magnetron
sputtering tool for thin film deposition on 30 x 30 cm
2
substrates. We achieved a homogeneous
deposition with thickness variations of 2.1 %. The effect of the deposition parameter water
vapor and oxygen partial pressure on the film properties was investigated. We observed, that
compared to an RF sputtering process, the process window, depending on the pulse parameter
during pulsed DC sputtering, might be more narrow for the water vapor supply when aiming for
amorphous In
2
O
3
:H thin films. For a duty cycle of 96 %a pronounced crystalline growth was
observed at
pH2O
= 0.2x10
−3
Pa. Increased oxygen supply correlated well with reduction of the
charge carrier density of the films, as oxygen vacancies are known to act as donors.
Furthermore the influences of the annealing atmospheres vacuum and air (220
°
C 30 min) on the
In
2
O
3
:H film properties were studied. We observed that both thermal treatments led to solid
phase crystallization of the amorphous phase. Presumably due to diffusion of atmospheric species
during annealing in air the charge carrier densities of those films was lower compared to films
annealed in vacuum, resulting in an increased resistivity, a lower free carrier absorption in the near
infrared region and a reduced Burstein-Moss shift of the optical band gap. A pronounced porous
structure evolved after annealing of films deposited in high oxygen partial pressures, similar to
133
Conclusions and Outlook
films deposited at high water vapor pressures. This resulted in an accelerated degradation of
the electrical properties, i.e., the electron mobility, of the crystallized films in damp heat. To
improve the charge carrier density and thus the conductivity of films after the annealing in air,
we developed a bi-layered In
2
O
3
:H film with a thin (about 30 nm) partly crystalline cap layer,
compared to the mainly amorphous bulk (about 160 nm). The cap-layer presumably counteracted
diffusion of atmospheric species into the film during annealing in air and thus improved the
charge carrier density of the films without significantly affecting the electron mobility or optical
absorption of the films.
Implementation of the indium oxide based TCOs in CIGS devices requires their deposition on
different sub-layers, which can affect the growth and properties of the TCOs. The poly-crystalline
structure of i-ZnO was observed to promote crystalline growth of In
2
O
3
:H films, as revealed
by X-ray diffraction measurements. This effect was observed on planar as well as on textured
substrates, i.e., rough CIGS. The crystalline fraction of as-grown In
2
O
3
:H films increased with
decreased ZnO grain size. Furthermore the preferred orientation of the films, as indicated by the
texture coefficient, changed from (222) to (400) with increased crystallinity. The results indicate
formation of crystalline nuclei especially at ZnO grain boundaries, presumably preferred in (400)
orientation. In contrast, In
2
O
3
:H, deposited within the same deposition run on amorphous
Zn(O,S) layers exhibit a mainly amorphous growth. The increased crystallinity after growth
induced strain in the films and led to decreased charge carrier densities and mobilities, before, as
well as after annealing. To suppress the crystalline growth of In
2
O
3
:H on poly-crystalline ZnO,
the hydrogen supply during the growth of the first layers at the ZnO interface must be increased
to facilitate the incorporation of hydrogen into the indium oxide film. However, this effect results
in narrowing of the possible process window.
Residual water in sub-layers of In
2
O
3
:H affects the electrical properties of the TCO after annealing,
presumably due to enhanced diffusion into the indium oxide layer and thus interference with the
crystallization process. This effect was demonstrated with spin coated In
x
O
y
and Ga
x
O
y
sol-gel
layers. Furthermore small crystallites within the spin-coated In
x
O
y
films promote crystalline
growth of sputtered In
x
O
x
:H films, which were deposited on top, leading to poor electrical
properties already after deposition. However, spin-coated Ga
x
O
y
films, which were pre-annealed
at 200
°
C provide a suitable template for the growth of amorphous In
x
O
x
:H films, which exhibit
high electron mobility after annealing.
Furthermore the specific roughness of CIGS affects the growth and electrical properties of the
indium oxide based TCOs In
2
O
3
:H and In
2
O
3
:H,W. Mainly amorphous In
2
O
3
:H films, grown on
Zn(O,S)-coated CIGS samples, showed a reduced mobility with increased roughness. In
2
O
3
:H
films that were grown on i-ZnO-coated Zn(O,S)/CIGS samples were partly crystalline and showed
in addition to the decreased mobility also a reduced charge carrier density. The loss in electron
134
Conclusions and Outlook
mobility was caused by the formation of voids inside the TCO film. As revealed by STEM and
SEM measurements, the location of most of such voids could be correlated with CIGS grain
boundaries. Spiky grain edges and sharp regressions induced void formation. The loss of the
electron mobility could be correlated empirically with an increase of the RMS roughness by an
linear regression. When compared to substrates with significantly changed topography, the loss
of the electron mobility could be described more accurately by the ratio of the median local slope
and the estimated grain size, as determined by the AFM measurement.
The slope of the linear regression was approximately equal for mainly amorphous grown and
pronounced crystalline grown In
2
O
3
:H films, indicating that the increased crystalline fraction
does not influence these observed losses or the formation of voids significantly. Solid phase
crystallization did not lead to sufficient improvement of the electron mobility. In fact, the deficit
in electron mobility was found to increase compared to reference films deposited on planar glass
substrates. Thus we can conclude, that to achieve high-mobility indium oxide films on CIGS
samples, the electron mobility has to be improved already in the as-grown state. No influence
of the deposition parameters water vapor pressure and oxygen partial pressure as well as of
the deposition rate on the limitation of the electron mobility by voids was observed. This was
demonstrated for In
2
O
3
:H,W films, deposited by reactive plasma deposition and for In
2
O
3
:H
films deposited by reactive plasma deposition and sputtering. However, the electron mobility of
tungsten doped indium oxide films was found to be less affected when the films were grown on
rough CIGS samples.
The degraded electrical properties of the TCOs on CIGS samples led to an increased sheet
resistance and thus to an increased series resistance in CIGS modules, limiting the fill factor and
efficiency.
Three strategies to minimize the observed losses in electron mobility in indium oxide films grown
on CIGS samples were developed and applied to CIGS devices:
•
Deposition of spin coated Ga
x
O
y
as the highly resistive layer smoothed the CIGS sample
surface by covering spiky and sharp structures. This consequently improved the electron
mobility to values equal to reference films deposited on planar glass substrates, although the
CIGS samples showed still a RMS roughness of 50 nm. We demonstated that by applying a
thin Ga
x
O
y
layer as the highly resistive layer in CIGS modules in combination with In
2
O
3
:H
front contact an improved
jsc
by more than 2 mA/cm
2
and overall improved efficiency
by more than 1 %can be achieved compared to a reference module with i-ZnO/ZnO:Al
window. A potential barrier, induced by the Ga
x
O
y
layer was reduced significantly by light
soaking.
135
Conclusions and Outlook
•
Etching of the CIGS surface with an acid bromine solution preferentially removes spiky
structures. Consequently the electron mobility of In
2
O
3
:H deposited on the etched CIGS
samples improved compared to films deposited on a non-etched reference. Corresponding
solar cells with etched CIGS absorbers showed an improved
Voc
and overall efficiency. Thus,
etching of CIGS absorbers is an effective process for the improvement of the elctron mobility
of indium oxide TCOs on CIGS.
•
At higher film thicknesses the grains may coalesce, thus leading to a less porous film.
The electron mobility was observed to increase with higher thickness of the TCOs. The
effect was more pronounced for films deposited by reactive plasma deposition than for the
sputtered films. However, on rather smooth CIGS samples In
2
O
3
:H films with 450 nm
thickness were sufficient to achieve a reasonable series resistance and fill factor in the
respective CIGS module. Furthermore the
jsc
of modules with In
2
O
3
:H front contact was
higher than in the reference modules with ZnO:Al front contact.
Additionally the band alignment of crystallized In
2
O
3
:H and ZnO was evaluated by XPS/UPS
measurements. We determined a valence band offset of
∆EV BM
= (-0.78
±
0.06) eV and a
conduction band offset of
∆ECBM
= (-0.23
±
0.21) eV. Thus the conduction band of In
2
O
3
:H is
energetically below the conduction band of ZnO. The large error resulted from the uncertainty
of the assumed band gap of indium oxide. According to numerical calculations, such an offset
has no negative influence on the solar cell properties, i.e. the fill factor.
The thermal treatment, which is required to initiate solid phase crystallization in the amorphous
phase of indium oxide films can affect the properties of a corresponding solar cell. In particular
CdS buffered solar cells were found to degrade due to annealing at temperatures above approx.
160
°
C. With increased temperature the
Voc
decreased due to reduction of the doping concentration
of the CIGS absorber and the corresponding broadening of the depletion zone. However, we
demonstrated that reduction of the required crystallization temperature to 150
°
C did not lead to
degradation of the CIGS solar cells with In
2
O
3
:H front contact. In fact, the solar cell parameters
improved.
Substitution of the CdS buffer to a thermally more stable buffer, i.e., Zn(O,S) led to more
stable solar cells with In
2
O
3
:H front contact, no degradation of the solar cell parameter was
observed after annealing at 180 °C. In fact, the efficiency improved. Moreover, an increased jsc
was demonstrated already prior to the annealing compared to a reference cell with ZnO:Al front
contact.
The results show that by implementing high mobility indium oxide films as front contact in CIGS
solar cells or modules, improved short circuit current densities and similar or even improved
efficiencies compared to the corresponding reference samples can be achieved. To realize a
136
Conclusions and Outlook
reasonable series resistance in CIGS modules, formation of voids in the amorphous grown TCOs
needs to be avoided, e.g. by specific CIGS layer configurations.
Outlook
Topics of future reasearch might be the development of additional stratagies for improved electron
mobilities in indium oxide based TCOs grown on rough CIGS samples. Additionally sol-gel layers
with improved conductivity might be beneficial for the application as highly resistive or buffer
layers in CIGS solar cells in order to avoid the formation of a barrier while smoothing the CIGS
sample surface.
Another topic might be detailed investigations of the stability of the pulsed DC magnetron
sputtering process. Moreover the influence of the pulse parameters on the process stability and
film properties needs to be analyzed.
137
APPENDIX A
Supplementary Information
A.1 General description of standard characterization methods
In this section the standard methods, which were used to characterize the samples are described
briefly.
Material characterization
Spectral Ellipsometry (SE)
With spectral ellipsometry the change of the optical polarization state of a electromagnetic wave
can be described after linearly polarized light is reflected from a thin film sample. The change
is determined by the ellipsometric angles
Ψ
and
∆
, which are related to the complex Fresnel
reflection coefficients according to
ρ=rP
rs
=tanΨei∆(A.1)
where
rp
and
rS
are the Fresnel-reflection coefficients parallel and perpendicular to the plane of
incidence.
tanΨ
represents the amplitude ratio between the parallel and vertical component,
∆
described the change in phase difference. By fitting the obtained data to a model, representing
the sample, many film properties, such as the film thickness, optical constants and surface
roughness can be determined.
138
1.1 General description of standard characterization methods
4 point probe
The simplest way to measure the sheet resistance of a thin film is by 4 point measurements. 4
linearly aligned needles in equidistant distance are used. It is required that the distance between
the needles is much smaller than the sample size, which is assumed to be infinite for simplicity.
Current is applied through the two outer needles, the voltage is measured by the two inner
needles. The sheet resistance is calculated by [19]
RSq =π
ln(2)
U
I(A.2)
Hall effect measurements
The film resistivity, charge carrier density and mobility can be extracted by measurements in the
van der Pauw geometry. Four contact needles are set on the outer corner of the sample surface,
the contacts must be small compared to the sample size. It is required that all sides of the sample
are equal. For resistivity measurements current (
I
) is applied at two contacts while voltage is
measured via the two other contacts. This procedure is repeated in different sequences. From the
followed Hall effect measurements the charge carrier density and mobility can be assessed. The
basic principle is that a permanent magnetic field (
B
) is applied perpendicular to the sample
surface. Due to the Lorenz force moving charge carriers are forced transversely to the direction
of movement, causing Hall voltage
VH
. This procedure is repeated with altered polarity of the
magnetic field. As a result the sheet carrier density (ns) can be calculated with
ns=IxBy
eVH
(A.3)
where e is the carrier charge. With knowledge of the film thickness (d) the charge carrier density
can be assessed:
n=ns
d(A.4)
The Hall mobility (µH) can be calculated with
µH=1
ensRSq
(A.5)
A detailed description of the measurement procedure can be found elsewhere [20,186]. To obtain
accurate results, the film must be homogeneous and without cracks. In multi-layered films all
layers contribute to the measurements [187].
139
Supplementary Information
Glow Discharge Optical Emission Spectroscopy (GDOES)
Glow Discharge Optical Emission Spectroscopy (GDOES) is a spectroscopic method for qualitative
and quantitative analysis of the composition of materials. The cathode (sample) is set in an
noble-gas environment and is bombarded by positive noble gas ions and atoms generated by
an electrical discharge. Material of the sample is sputtered, partly ionized and excited to emit
photons corresponding to characteristic spectral lines [95].
Transmission Electron Microscopy (TEM)
Transmission Electron Microscopy (
TEM
) provides information on the structural properties
on the nano and atomic scale. For such measurements uniformly thin samples are required to
allow transmission of electrons. Lamellas of the specimen are typically prepared by conventional
methods, such as mechanical dimpling and FIB. Thinning of the sample can induce damage or
change the surface conditions.
Scanning Electron Microscopy (SEM)
A scanning electron microscope (SEM) was used to image the sample with high magnifications.
A focused electron beam scans the sample surface in high vacuum. To enable focusing of the
beam, the sample surface must be conductive. The electrons are accelerated through an electric
field towards the sample surface, where electrons are back-scattered or secondary electrons are
released due to elastic and inelastic scattering. The electrons are imaged by detectors, therefore
a high magnification of the sample surface can be realized. Images were taken in top view or
cross section layout.
Electron Backscatter Diffraction (EBSD)
Electron backscatter diffraction is a special form of electron microscopy. When an electron
beam impinges a crystal located at the samples surface (which is tilted), backscattered electrons
incident on lattice planes and are diffracted. The diffraction patterns (Kikuchi or EBSD pattern)
provide information about the crystal orientation.
Photoelectron Spectroscopy
XPS
is a widely used surface analysis method based on the photoelectric effect and provides
information of the surface composition. When X-rays with a defined energy
hν
(>1 keV) impinge
the sample surface, core electrons are exited and released from the atom. The kinetic energy
140
1.1 General description of standard characterization methods
Ekin
of the electrons is measured by an electron detector. The electron binding energy
EB
can
be calculated as
EB=Ehν −Ekin −Φspec (A.6)
with
Φspec
as the work function of the spectrometer. The electron binding energy depends on
the specific element, electron orbital and chemical environment of the atom.
UPS
is used to evaluate the valence band region. In contrast to
XPS
, ionizing radiation of a few
eV is used to induce the photoelectric effect. Typically ultraviolet photons are produced by a gas
discharge lamp filled with helium, resulting in energies of 21.2 eV (He I) (used in this study) and
40.8 eV (He II). More details concerning the methods can be found elsewhere [95,121,188].
Atomic Force Microscopy
Atomic force microscopy was used to evaluate the topography of the samples. In general either
the height of the tip or the distance of tip/surface is held constant. The distance between the
tip and the surface determines the predominate force and results by the measurement type. In
contact mode the total van der Waals force becomes positive (repulsive), as the atoms are in
contact. An advantage of this mode is the good resolution, but however, the surface of the
sample easily gets damaged due to the contact with the tip. In true non contact mode the total
van der Waals force is negative (attractive). The cantilever vibrates in free space in the vicinity
of the resonant frequency. A change of the tip/sample distance results in a shift of the resonant
frequency and change of the topographic information. The advantage of this method prevention
of tip damage, but typically results in weak tip-surface interaction. A compromise between these
tow operation modes is the tapping or intermittent mode (Dynamic Force Microscope (
DFM
)).
Similar to the non-contact mode the tip is vibrating in the free space, but repeatedly taps the
surface, resulting in contact of the tip and surface, similar to the contact mode. This yields good
resolution and low damage of the tip or surface [18,189].
141
Supplementary Information
A.2 Additional information concerning investigations of IOH
layers on ZnO
100
80
60
40
20
0
Crystalline fraction X
C
of as-deposited In
2
O
3
:H films / %
5045403530
<D>
AFM
/ nm
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
Figure A.1: Dependence of the crystalline fraction of as grown In
2
O
3
:H on the lateral grain size, adapted
from Erfurt et al. (Supporting Information) [100]
142
1.2 Additional information concerning investigations of IOH layers on ZnO
Figure A.2: X-ray diffraction patterns of annealed In
2
O
3
:H films grown on glass, Zn(O,S) and ZnO
layers; Reference patterns of In
2
O
3
and ZnO were taken from PDF 00-006-0416 and PDF 01-070-8070,
respectively; patterns were shifted for improved differentiation, adapted from Erfurt et al. (Supporting
Information) [100]
0.25
0.20
0.15
0.10
0.05
0.00
Texture coefficient T
c(hkl)
/ arb. units
5045403530
<D>
AFM
/ nm
ZnO40
ZnO130
ZnO130T
ZnO130p
ZnO200
(222)
(400)
(440)
Figure A.3: Dependence of the texture coefficient of as grown In
2
O
3
:H on the lateral grain size, adapted
from Erfurt et al. (Supporting Information) [100]
143
Supplementary Information
Figure A.4: TEM images of the interface of ZnO and
In2O3:H
, adapted from Erfurt et al. (Supporting
Information) [100]
Figure A.5: TEM images of ZnO/In
2
O
3
:H with additional measurement spots in the ZnO film with
measured inter-planar distances and the corresponding crystalline assignment, colored boxes represent the
area taken for FFT analysis, adapted from Erfurt et al. (Supporting Information) [100]
144
1.3 Additional information concerning investigations of the low TCO electron mobility on CIGS
samples
A.3 Additional information concerning investigations of the low
TCO electron mobility on CIGS samples
Table A.1: Sample assignment; the thickness of the Zn(O,S), ZnO and IOH layers is 60 nm, 130 nm and
330 nm, respectively; note that samples 4-3998-15-0 and 4-4041-12-4 were prepared without a Zn(O,S)
layer
Substrate CIGS process time ID .../Zn(O,S)/i-ZnO ID .../Zn(O,S) TCO RMS
Mo/CIGS 107 min 4-3995-1-2 4-3995-1-4 IOH 118 nm
Mo/CIGS 71 min 4-3995-11-2 4-3995-11-4 IOH 89 nm
Mo/CIGS 46 min 4-3995-6-2 4-3995-6-4 IOH 27 nm
glass/CIGS 107 min 4-3998-2-2 4-3998-2-4 IOH 65 nm
glass/CIGS 71 min 4-3998-6-2 4-3998-6-4 IOH 55 nm
glass/CIGS 46 min 4-3998-4-2 4-3998-4-4 IOH 40 nm
glass - 4-3998-15-0 4-3998-13-0 IOH 0.4 nm
glass - 5-15-2-83 5-15-2-81 IOH 170 nm
Mo/CIGS 107 min 4-3996-18-4 - AZO 113 nm
glass/CIGS 107 min 4-3998-11-4 - AZO 59 nm
glass - 4-4041-12-4 - AZO 0.4 nm
As PDT the HZB-CIGS absorber were treated with NaF after deposition, the AIST-CIGS
absorber with NaF/KF. In case of the PDT carried out at HZB two different deposition duration
of 4 min and 8 min were done. GDOES measurements ware carried out of AIST samples with
and without a PDT and of HZB samples with 8 min and without PDT, each for a layer stack
with and without an intrinsic ZnO layer. As no standard reference sample was measured, the
intensities are only relative. The total relative intensities of the HZB samples without ZnO layers
are overall higher, as these were measured with different Ar pressure than the other samples.
No correlation between the relative sodium amount in the TCO layer and the electron mobility
could be observed.
145
Supplementary Information
120
100
80
60
40
20
RMS roughness / nm
410400390380370360350
AFM grain size / nm
Figure A.6: Correlation of the CIGS sample RMS roughness and the determined AFM grain size using
the Watershed method
1.2x10
-3
1.00.80.60.40.20.0
MLS/Grain Size / nm
-1
{ µ
e
} = - 23572 { MLS/GS } + 93.2
{ µ
e
} = - 35870 { MLS/GS } + 91.5
R
2
= 0.971
R
2
= 0.951
0.50.40.30.20.10.0
Median Local Slope (MLS) / arb. units
{ µ
e
} = - 56.2 { MLS } + 89.8
{ µ
e
} = - 93.8 { MLS } + 89.4
R
2
= 0.964
R
2
= 0.967
100
90
80
70
60
50
40
µ
e
/ cm
2
V
-1
s
-1
200150100
50
0
RMS roughness / nm
{ µ
e
} = - 0.56 { RMS } + 88.7
{ µ
e
} = - 0.33 { RMS } + 89.4
R
2
= 0.973
R
2
= 0.979
5x10
20
4
3
2
1
0
n
e
/ cm
-3
annealed on...
... glass/CIGS
IOH/i-ZnO/Zn(O,S)
IOH/Zn(O,S)
... glass
IOH/i-ZnO/Zn(O,S)
IOH/Zn(O,S)
Figure A.7: Comparison of the impact of the RMS roughness, median local slope and median local slope
per grain/structure size of the substrate/sub-layers on the charge carrier density and electron mobility of
sputtered IOH films deposited on ZnO/Zn(O,S)- and Zn(O,S)-coated samples after annealing in vacuum
at 180
°
C for 1 h; for the linear fits the films on the textured glass sample were only taken into account
for the fit on the dependence on the median local slope per grain size
146
1.3 Additional information concerning investigations of the low TCO electron mobility on CIGS
samples
50
45
40
35
30
25
20
µ
e
/ cm
2
V
-1
s
-1
PDT
In
2
O
3
:H,W
w/
w/o
i-ZnO
HZB CIGS / HZB CdS
HZB CIGS / AIST CdS
AIST CIGS / AIST CdS
HZB CIGS
AIST CIGS
SLG
without
(a) In2O3:H,W
50
45
40
35
30
25
20
µ
e
/ cm
2
V
-1
s
-1
PDT
In
2
O
3
:H
w/
w/o
i-ZnO
HZB CIGS / HZB CdS
AIST CIGS / AIST CdS
AIST CIGS
HZB CIGS
SLG
without
(b) In2O3:H
Figure A.8: Electron mobility of (a) In
2
O
3
:H,W and (b) In
2
O
3
:H thin films deposited on SLG and CIGS
samples in dependence of the relatively PDT amount used for the CIGS absorber preparation
2.0
1.5
1.0
0.5
0.0
Relative Intensity / arb. units
500400300200100
0
Time / s
HZB-CIGS with i-ZnO
Zn
In
Mo
Na
Cd
with PDT: thin, dotted line
(a)
4
3
2
1
0
Relative Intensity / arb. units
300250200150100
50
0
Time / s
HZB-CIGS without i-ZnO
Zn
In
Mo
Na
Cd
with PDT: thin, dotted line
(b)
Figure A.9: GDOES measurements of (a) IOH/ZnO/CdS/CIGS/Mo and (b) IOH/CdS/CIGS/Mo,
respectively with and without a NaF PDT; the CIGS absorber and PDT was deposited at HZB
147
Supplementary Information
2.0
1.5
1.0
0.5
0.0
Relative Intensity / arb. units
500400300200100
0
Time / s
AIST-CIGS with i-ZnO
Zn
In
Mo
Na
Cd
with PDT: thin, dotted line
(a)
2.0
1.5
1.0
0.5
0.0
Relative Intensity / arb. units
500400300200100
0
Time / s
AIST-CIGS without i-ZnO
Zn
In
Mo
Na
Cd
with PDT: thin, dotted line
(b)
Figure A.10: GDOES measurements of (a) IOH/ZnO/CdS/CIGS/Mo and (b) IOH/CdS/CIGS/Mo,
respectively with and without a NaF/KF PDT; the CIGS absorber and PDT was deposited at AIST
Figure A.11: Elemental composition the sample sputtered-In
2
O
3
:H/i-ZnO/CdS/CIGS; indium was not
detected within the cracks
148
1.3 Additional information concerning investigations of the low TCO electron mobility on CIGS
samples
CIGS
ZnO/
CdS
In O :H
2 3
Figure A.12: Overview of as grown sputtered-In
2
O
3
:H grown on i-ZnO/CdS/CIGS, measured by STEM;
Voids are clearly visible in the In2O3:H layer
149
Supplementary Information
CIGS
ZnO/CdS
In O :H
2 3
Pt/C
Figure A.13: Additional STEM measurements of as grown sputtered In
2
O
3
:H grown on i-ZnO/CdS/CIGS;
Voids are clearly visible in the In2O3:H layer
150
Symbols and Acronyms
pO2Oxygen partial pressure
IOH Hydrogen doped indium oxide
NIR Near infrared
TCO Transparent conductive oxide
ALD Atomic Layer Deposition
RPD Reactive Plasma Deposition
DC direct current
RF radio frequency
i-ZnO intrinsic zinc oxide
GI-XRD grazing incidence X-ray diffraction
BB-XRD Bragg Brentano X-ray diffraction
FFT Fast Fourier Transformation
DFM Dynamic Force Microscope
TEM Transmission Electron Microscopy
STEM Scanning Transmission Electron Microscopy
a-In2−2xGa2xO3amorphous indium gallium oxide
FTIR Fourier-transform infrared spectroscopy
eGB extended grain boundaries
RMS root mean square
151
Symbols and Acronyms
MLS median local slope
RGA residual gas analyzer
HZB Helmholtz-Zentrum Berlin für Materialien und Energie
AIST National Institute of Advanced Industrial Science and Technology
PDT post deposition treatment
IGZO Indium Gallium Zinc Oxide
CIGS Cu(In,Ga)(S,Se)2
In2O3:H,W hydrogen and tungsten co-doped indium oxide
In2O3:H hydrogen doped indium oxide
CBO conduction band offset
VBO valence band offset
UPS ultraviolet photoelectron spectroscopy
XPS X-ray photoelectron spectroscopy
VBM valence band maximum
CBM conduction band minimum
EQE external quantum efficiency
IQE internal quantum efficiency
SRH Shockley–Read–Hall
CBD chemical bath deposition
HR highly resistive
AFM atomic force microscopy
FIB focused ion beam
TCO transparent conductive oxide
152
Bibliography
[1]
International Renewable Energy Agency (Irena). Global energy transformation: A roadmap
to 2050. Policy report [2018]. Ww.irena.org/publications.
[2]
WWW.CIGS-PV.NET. White Paper for CIGS Thin-Film Solar Cell Technology. Tech.
rep. [2015].
[3]
S. Frontier. Solar Frontier Achieves World Record Thin-Film Solar Cell Efficiency of 22.9%
[2017].
[4]
V. Bermudez and A. Perez-Rodriguez. Understanding the cell-to-module efficiency gap
in Cu(In,Ga)(S,Se)2 photovoltaics scale-up. Nature Energy, 3(6):466–475 [2018]. ISSN
2058-7546. doi: 10.1038/s41560-018-0177-1.
[5]
T. Koida, Y. Ueno and H. Shibata. In
2
O
3
-Based Transparent Conducting Oxide Films
with High Electron Mobility Fabricated at Low Process Temperatures. physica status solidi
(a), p. 1700506 [2018]. ISSN 18626300. doi: 10.1002/pssa.201700506.
[6]
A. E. Delahoy and S. Y. Guo. Transparent and semitransparent conducting film deposition by
reactive-environment, hollow cathode sputtering. Journal of Vacuum Science & Technology
A: Vacuum, Surfaces, and Films, 23 (4):1215–1220 [2005]. ISSN 0734-2101, 1520-8559. doi:
10.1116/1.1894423.
[7]
F. Meng, J. Shi, Z. Liu et al. High mobility transparent conductive W-doped In2o3 thin
films prepared at low substrate temperature and its application to solar cells. Solar Energy
Materials and Solar Cells, 122:70–74 [2014]. ISSN 09270248. doi: 10.1016/j.solmat.2013.
11.030.
[8]
C. Warmsingh, Y. Yoshida, D. W. Readey et al. High-mobility transparent conducting
Mo-doped In2o3 thin films by pulsed laser deposition. Journal of Applied Physics, 95(7):3831–
3833 [2004]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.1646468.
[9]
Y. Meng. A new transparent conductive thin film In2o3:Mo. Thin Solid Films, 394(1-
2):218–222 [2001]. ISSN 00406090. doi: 10.1016/S0040-6090(01)01142-7.
153
Bibliography
[10]
M. F. A. M. van Hest, M. S. Dabney, J. D. Perkins et al. Titanium-doped indium oxide: A
high-mobility transparent conductor. Applied Physics Letters, 87(3):032111 [2005]. ISSN
0003-6951, 1077-3118. doi: 10.1063/1.1995957.
[11]
E. Fortunato, D. Ginley, H. Hosono et al. Transparent Conducting Oxides for Photovoltaics.
MRS Bulletin, 32(03):242–247 [2007]. ISSN 0883-7694, 1938-1425. doi: 10.1557/mrs2007.29.
[12]
S. C. Dixon, D. O. Scanlon, C. J. Carmalt et al. n-Type doped transparent conducting
binary oxides: an overview. Journal of Materials Chemistry C, 4(29):6946–6961 [2016].
ISSN 2050-7526, 2050-7534. doi: 10.1039/C6TC01881E.
[13]
J. K. Wassei and R. B. Kaner. Graphene, a promising transparent conductor. Materials
Today, 13(3):52–59 [2010]. ISSN 13697021. doi: 10.1016/S1369-7021(10)70034-1.
[14]
N. F. Mott and L. Friedman. Metal-insulator transitions in VO
2
, Ti
2
O
3
and Ti
2- x
V
x
O
3
. Philosophical Magazine, 30(2):389–402 [1974]. ISSN 0031-8086. doi:
10.1080/14786439808206565.
[15]
P. P. Edwards and M. J. Sienko. Universality aspects of the metal-nonmetal transition
in condensed media. Physical Review B, 17(6):2575–2581 [1978]. ISSN 0163-1829. doi:
10.1103/PhysRevB.17.2575.
[16]
I. Hamberg and C. G. Granqvist. Evaporated Sn-doped In
2
O
3
films: Basic opti-
cal properties and applications to energy-efficient windows. Journal of Applied Physics,
60(11):R123–R160 [1986]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.337534.
[17]
P. Drude. Zur Elektronentheorie der Metalle. Annalen der Physik, 306(3):566–613 [1900].
ISSN 00033804, 15213889. doi: 10.1002/andp.19003060312.
[18]
S. Neubert. Innovative front contact systems for a-Si:H/
µ
c-Si:H solar cells based on
thermally treated zinc oxide. Dissertation, Technical University of Berlin, Berlin [2015].
[19]
D. S. Ginley, H. Hosono and D. C. Paine. Handbook of transparent conductors. Springer,
New York [2010]. ISBN 978-1-4419-1637-2 978-1-4419-1638-9. OCLC: ocn495781319.
[20]
A. Luque and S. Hegedus. Handbook of Photovoltaic Science and Engineering. John Wiley
& Sons, Ltd., second edn. [2011]. ISBN 978-0-470-72169-8.
[21]
Y. Zhang and M. H. White. A quantum mechanical mobility model for scaled NMOS tran-
sistors with ultra-thin high-K dielectrics and metal gate electrodes. Solid-State Electronics,
52(11):1810–1814 [2008]. ISSN 00381101. doi: 10.1016/j.sse.2008.08.003.
[22]
D. B. Buchholz, Q. Ma, D. Alducin et al. The Structure and Properties of Amorphous
Indium Oxide. Chemistry of Materials, 26 (18):5401–5411 [2014]. doi: 10.1021/cm502689x.
154
Bibliography
[23]
H. Brooks. Theory of the Electrical Properties of Germanium and Silicon. In Advances in
Electronics and Electron Physics, vol. 7, pp. 85–182. Elsevier [1955]. ISBN 978-0-12-014507-2.
doi: 10.1016/S0065-2539(08)60957-9.
[24]
R. Dingle. XCIV. Scattering of electrons and holes by charged donors and acceptors
in semiconductors. The London, Edinburgh, and Dublin Philosophical Magazine and
Journal of Science, 46(379):831–840 [1955]. ISSN 1941-5982, 1941-5990. doi: 10.1080/
14786440808561235.
[25]
C. Erginsoy. Neutral Impurity Scattering in Semiconductors. Physical Review, 79(6):1013–
1014 [1950]. ISSN 0031-899X. doi: 10.1103/PhysRev.79.1013.
[26]
K. Seeger. Semiconductor Physics: an introduction. Advanced Texts in Physics. Springer
Berlin, Berlin, 9. ed edn. [2010]. ISBN 978-3-662-09855-4 978-3-642-06023-6. OCLC:
846198919.
[27]
J. W. Orton. The story of semiconductors. Oxford University Press, Oxford [2004]. ISBN
978-0-19-853083-1. OCLC: ocm56694906.
[28]
R. L. Petritz. Theory of Photoconductivity in Semiconductor Films. Physical Review,
104(6):1508–1516 [1956]. ISSN 0031-899X. doi: 10.1103/PhysRev.104.1508.
[29]
J. Y. W. Seto. The electrical properties of polycrystalline silicon films. Journal of Applied
Physics, 46(12):5247–5254 [1975]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.321593.
[30]
G. Baccarani, B. Riccò and G. Spadini. Transport properties of polycrystalline silicon films.
Journal of Applied Physics, 49(11):5565–5570 [1978]. ISSN 0021-8979, 1089-7550. doi:
10.1063/1.324477.
[31]
I. Hamberg, C. G. Granqvist, K. F. Berggren et al. Band-gap widening in heavily Sn-
doped In 2 O 3. Physical Review B, 30(6):3240–3249 [1984]. ISSN 0163-1829. doi:
10.1103/PhysRevB.30.3240.
[32]
A. Pflug, V. Sittinger, F. Ruske et al. Optical characterization of aluminum-doped zinc
oxide films by advanced dispersion theories. Thin Solid Films, 455-456:201–206 [2004].
ISSN 00406090. doi: 10.1016/j.tsf.2004.01.006.
[33]
NREL. Reference Solar Spectral Irradiance: Air Mass 1.5.
Https://rredc.nrel.gov/solar/spectra/am1.5/; Accessed 12.08.2018.
[34]
R. Menner, D. Hariskos, V. Linss et al. Low-cost ZnO:Al transparent contact by reac-
tive rotatable magnetron sputtering for Cu(In,Ga)Se2 solar modules. Thin Solid Films,
519(21):7541–7544 [2011]. ISSN 00406090. doi: 10.1016/j.tsf.2011.01.392.
155
Bibliography
[35]
S. Calnan and A. Tiwari. High mobility transparent conducting oxides for thin film solar cells.
Thin Solid Films, 518(7):1839–1849 [2010]. ISSN 00406090. doi: 10.1016/j.tsf.2009.09.044.
[36]
C. Guillén and J. Herrero. Optical, electrical and structural characteristics of Al:ZnO thin
films with various thicknesses deposited by DC sputtering at room temperature and annealed
in air or vacuum. Vacuum, 84(7):924–929 [2010]. ISSN 0042207X. doi: 10.1016/j.vacuum.
2009.12.015.
[37]
R. Cebulla, R. Wendt and K. Ellmer. Al-doped zinc oxide films deposited by simultaneous
rf and dc excitation of a magnetron plasma: Relationships between plasma parameters and
structural and electrical film properties. Journal of Applied Physics, 83(2):1087–1095 [1998].
ISSN 0021-8979, 1089-7550. doi: 10.1063/1.366798.
[38]
K. Tominaga, N. Umezu, I. Mori et al. Transparent conductive ZnO film preparation by
alternating sputtering of ZnO:Al and Zn or Al targets. Thin Solid Films, 334(1-2):35–39
[1998]. ISSN 00406090. doi: 10.1016/S0040-6090(98)01112-2.
[39]
S. Körner, M. Hartig, R. Muydinov et al. Serial cosputtering for aluminum doping
manipulated zinc oxide as front contact for Cu(In,Ga)Se
2
solar cells. Japanese Journal of
Applied Physics, 57(8S3):08RC18 [2018]. ISSN 0021-4922, 1347-4065. doi: 10.7567/JJAP.
57.08RC18.
[40]
Y. Igasaki and H. Saito. The effects of deposition rate on the structural and electrical
properties of ZnO:Al films deposited on (1120) oriented sapphire substrates. Journal of
Applied Physics, 70(7):3613–3619 [1991]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.
349258.
[41]
K. Ellmer. Resistivity of polycrystalline zinc oxide films: current status and physical
limit. Journal of Physics D: Applied Physics, 34(21):3097–3108 [2001]. ISSN 0022-3727,
1361-6463. doi: 10.1088/0022-3727/34/21/301.
[42]
O. Tuna, Y. Selamet, G. Aygun et al. High quality ITO thin films grown by dc and RF
sputtering without oxygen. Journal of Physics D: Applied Physics, 43(5):055402 [2010].
ISSN 0022-3727, 1361-6463. doi: 10.1088/0022-3727/43/5/055402.
[43]
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(1):110–114 [2000]. ISSN 09215107. doi: 10.1016/S0921-5107(00)00477-3.
[44]
K. Ellmer and R. Mientus. Carrier transport in polycrystalline ITO and ZnO:Al II: The
influence of grain barriers and boundaries. Thin Solid Films, 516(17):5829–5835 [2008].
ISSN 00406090. doi: 10.1016/j.tsf.2007.10.082.
156
Bibliography
[45]
F. Kurdesau, G. Khripunov, A. da Cunha et al. Comparative study of ITO layers deposited
by DC and RF magnetron sputtering at room temperature. Journal of Non-Crystalline
Solids, 352(9-20):1466–1470 [2006]. ISSN 00223093. doi: 10.1016/j.jnoncrysol.2005.11.088.
[46]
T. Koida, H. Fujiwara and M. Kondo. Hydrogen-doped In2o3 as High-mobility Transparent
Conductive Oxide. Japanese Journal of Applied Physics, 46(No. 28):L685–L687 [2007].
ISSN 0021-4922. doi: 10.1143/JJAP.46.L685.
[47]
J. Keller, L. Stolt, M. Edoff et al. Atomic layer deposition of In
2
O
3
transparent conductive
oxide layers for application in Cu(In,Ga)Se
2
solar cells with different buffer layers: Atomic
layer deposition of In
2
O
3
transparent conductive oxide layers. physica status solidi (a),
213(6):1541–1552 [2016]. ISSN 18626300. doi: 10.1002/pssa.201532883.
[48]
H. Scherg-Kurmes, S. Körner, S. Ring et al. High mobility In2o3:H as contact layer for
a-Si:H/c-Si heterojunction and
µ
c-Si:H thin film solar cells. Thin Solid Films, 594:316–322
[2015]. ISSN 00406090. doi: 10.1016/j.tsf.2015.03.022.
[49]
M. Morales-Masis, E. Rucavado, R. Monnard et al. Highly Conductive and Broadband
Transparent Zr-Doped In
2
O
3
as Front Electrode for Solar Cells. IEEE Journal of
Photovoltaics, 8(5):1202–1207 [2018]. ISSN 2156-3381, 2156-3403. doi: 10.1109/JPHOTOV.
2018.2851306.
[50]
E. Kobayashi, Y. Watabe and T. Yamamoto. High-mobility transparent conductive thin
films of cerium-doped hydrogenated indium oxide. Applied Physics Express, 8(1):015505
[2015]. ISSN 1882-0778, 1882-0786. doi: 10.7567/APEX.8.015505.
[51]
S. Z. Karazhanov, P. Ravindran, P. Vajeeston et al. Phase stability, electronic structure,
and optical properties of indium oxide polytypes. Physical Review B, 76(7) [2007]. ISSN
1098-0121, 1550-235X. doi: 10.1103/PhysRevB.76.075129.
[52]
R. S. Roth. Classification of perovskite and other ABO3-type compounds. J. Res. Nat. Bur.
Stand, 58(2):75–88 [1957].
[53]
M. Marezio. Refinement of the crystal structure of In
2
O
3
at two wavelengths. Acta Crys-
tallographica, 20(6):723–728 [1966]. ISSN 0365-110X. doi: 10.1107/S0365110X66001749.
[54]
R. L. Weiher and R. P. Ley. Optical Properties of Indium Oxide. Journal of Applied
Physics, 37(1):299–302 [1966]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.1707830.
[55]
K. Irmscher, M. Naumann, M. Pietsch et al. On the nature and temperature dependence
of the fundamental band gap of In
2
O
3
: Nature and temperature dependence of the
fundamental band gap of In
2
O
3
. physica status solidi (a), 211(1):54–58 [2014]. ISSN
18626300. doi: 10.1002/pssa.201330184.
157
Bibliography
[56]
V. Scherer, C. Janowitz, A. Krapf et al. Transport and angular resolved photoemission
measurements of the electronic properties of In
2
O
3
bulk single crystals. Applied Physics
Letters, 100(21):212108 [2012]. ISSN 0003-6951, 1077-3118. doi: 10.1063/1.4719665.
[57]
C. Janowitz, V. Scherer, M. Mohamed et al. Experimental electronic structure of In
2
O
3
and Ga
2
O
3
. New Journal of Physics, 13(8):085014 [2011]. ISSN 1367-2630. doi:
10.1088/1367-2630/13/8/085014.
[58]
P. D. C. King, T. D. Veal, F. Fuchs et al. Band gap, electronic structure, and surface
electron accumulation of cubic and rhombohedral In 2 O 3. Physical Review B, 79(20)
[2009]. ISSN 1098-0121, 1550-235X. doi: 10.1103/PhysRevB.79.205211.
[59]
A. Walsh, J. L. F. Da Silva, S.-H. Wei et al. Nature of the Band Gap of In 2 O 3 Revealed
by First-Principles Calculations and X-Ray Spectroscopy. Physical Review Letters, 100(16)
[2008]. ISSN 0031-9007, 1079-7114. doi: 10.1103/PhysRevLett.100.167402.
[60]
Y. Shigesato, I. Yasui, Y. Hayashi et al. Effects of water partial pressure on the activated
electron beam evaporation process to deposit tin-doped indium-oxide films. Journal of
Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 13(2):268–275 [1995].
ISSN 0734-2101, 1520-8559. doi: 10.1116/1.579409.
[61]
T. Koida, M. Kondo, K. Tsutsumi et al. Hydrogen-doped In2o3 transparent conducting
oxide films prepared by solid-phase crystallization method. Journal of Applied Physics,
107(3):033514 [2010]. doi: 10.1063/1.3284960.
[62]
S. Limpijumnong, P. Reunchan, A. Janotti et al. Hydrogen doping in indium oxide: An
ab initio study. Physical Review B, 80(19) [2009]. ISSN 1098-0121, 1550-235X. doi:
10.1103/PhysRevB.80.193202.
[63]
P. D. C. King, R. L. Lichti, Y. G. Celebi et al. Shallow donor state of hydrogen in In 2 O
3 and SnO 2 : Implications for conductivity in transparent conducting oxides. Physical
Review B, 80(8) [2009]. ISSN 1098-0121, 1550-235X. doi: 10.1103/PhysRevB.80.081201.
[64]
B. Macco, H. C. M. Knoops and W. M. M. Kessels. Electron Scattering and Doping
Mechanisms in Solid-Phase-Crystallized In2o3:H Prepared by Atomic Layer Deposition. ACS
Applied Materials & Interfaces, 7(30):16723–16729 [2015]. doi: 10.1021/acsami.5b04420.
[65]
T. Koida, H. Fujiwara and M. Kondo. Structural and Electrical Properties of Hydrogen-
Doped In2o3 Films Fabricated by Solid-Phase Crystallization. Journal of Non-Crystalline
Solids, 354(19-25):2805–2808 [2008]. ISSN 00223093. doi: 10.1016/j.jnoncrysol.2007.09.076.
[66]
H. Wardenga, M. Frischbier, M. Morales-Masis et al. In Situ Hall Effect Monitoring of
Vacuum Annealing of In2o3:H Thin Films. Materials, 8(2):561–574 [2015]. ISSN 1996-1944.
doi: 10.3390/ma8020561.
158
Bibliography
[67]
B. Macco, Y. Wu, D. Vanhemel et al. High mobility In
2
O
3
:H transparent conductive
oxides prepared by atomic layer deposition and solid phase crystallization. physica status
solidi (RRL) - Rapid Research Letters, 8(12):987–990 [2014]. ISSN 18626254. doi: 10.
1002/pssr.201409426.
[68]
J. Keller, A. Aijaz, F. Gustavsson et al. Direct comparison of atomic layer deposition and
sputtering of In 2 O 3 :H used as transparent conductive oxide layer in CuIn 1-x Ga x Se 2
thin film solar cells. Solar Energy Materials and Solar Cells, 157:757–764 [2016]. ISSN
09270248. doi: 10.1016/j.solmat.2016.07.012.
[69]
L. Barraud, Z. Holman, N. Badel et al. 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]. ISSN 09270248. doi: 10.1016/j.solmat.2013.03.024.
[70]
M. Boccard, N. Rodkey and Z. C. Holman. Properties of hydrogenated indium oxide prepared
by reactive sputtering with hydrogen gas. In 2016 IEEE 43rd Photovoltaic Specialists
Conference (PVSC), pp. 2868–2870 [2016]. doi: 10.1109PVSC.2016.7750178.
[71]
A. Steigert, I. Lauermann, T. Niesen et al. Sputtered Zn(O,S)/In2o3:H window layers for
enhanced blue response of chalcopyrite solar cells. physica status solidi (RRL) - Rapid
Research Letters, 9(11):627–630 [2015]. ISSN 18626254. doi: 10.1002/pssr.201510318.
[72]
S. Li, Z. Shi, Z. Tang et al. Study on the hydrogen doped indium oxide for silicon
heterojunction solar cell application. Journal of Alloys and Compounds, 705:198–204 [2017].
ISSN 09258388. doi: 10.1016/j.jallcom.2017.02.133.
[73]
J. Yu, J. Bian, W. Duan et al. Tungsten doped indium oxide film: Ready for bifacial copper
metallization of silicon heterojunction solar cell. Solar Energy Materials and Solar Cells,
144:359–363 [2016]. ISSN 09270248. doi: 10.1016/j.solmat.2015.09.033.
[74]
F. Engelhardt, L. Bornemann, M. Köntges et al. Cu(In,Ga)Se2 solar cells with a ZnSe
buffer layer: interface characterization by quantum efficiency measurements. Progress in
Photovoltaics: Research and Applications, 7(6):423–436 [1999]. ISSN 1062-7995, 1099-159X.
doi: 10.1002/(SICI)1099-159X(199911/12)7:6<423::AID-PIP281>3.0.CO;2-S.
[75]
U. Rau and H. Schock. Cu(In,Ga)Se2 thin-film solar cells. In Solar Cells, pp. 303–349.
Elsevier [2005]. ISBN 978-1-85617-457-2. doi: 10.1016/B978-185617457-2/50013-5.
[76]
R. Scheer and H. W. Schock. Chalcogenide photovoltaics: physics, technologies, and thin
film devices. Wiley-VCH, Weinheim, Germany [2011]. ISBN 978-3-527-31459-1. OCLC:
ocn664325819.
159
Bibliography
[77]
D. Shvydka and V. Karpov. Modeling of non-uniformity losses in integrated large-area
solar cell modules. In Conference Record of the Thirty-first IEEE Photovoltaic Specialists
Conference, 2005., pp. 359–362. IEEE, Lake buena Vista, FL, USA [2005]. ISBN 978-0-
7803-8707-2. doi: 10.1109/PVSC.2005.1488143.
[78]
P. Grabitz, U. Rau and J. Werner. A multi-diode model for spatially inhomogeneous solar
cells. Thin Solid Films, 487(1-2):14–18 [2005]. ISSN 00406090. doi: 10.1016/j.tsf.2005.01.
027.
[79]
M. A. Green. Accuracy of analytical expressions for solar cell fill factors. Solar Cells,
7(3):337–340 [1982]. ISSN 03796787. doi: 10.1016/0379-6787(82)90057-6.
[80]
K. Iwata, T. Sakemi, A. Yamada et al. Improvement of ZnO TCO film growth for
photovoltaic devices by reactive plasma deposition (RPD). Thin Solid Films, 480-481:199–
203 [2005]. ISSN 00406090. doi: 10.1016/j.tsf.2004.11.072.
[81]
M. D. Heinemann. CIGSe Superstrate Solar Cells: Growth and Characterization of CIGSe
Thin Films on Transparent Conductive Oxides. Dissertation, Technical University of Berlin
[2015].
[82]
P. Jackson, R. Wuerz, D. Hariskos et al. Effects of heavy alkali elements in Cu(In,Ga)Se
2
solar cells with efficiencies up to 22.6%. physica status solidi (RRL) - Rapid Research
Letters, 10(8):583–586 [2016]. ISSN 18626254. doi: 10.1002/pssr.201600199.
[83]
N. Nicoara, T. Lepetit, L. Arzel et al. Effect of the KF post-deposition treatment on grain
boundary properties in Cu(In, Ga)Se2 thin films. Scientific Reports, 7(1) [2017]. ISSN
2045-2322. doi: 10.1038/srep41361.
[84]
A. Chirilă, P. Reinhard, F. Pianezzi et al. Potassium-induced surface modification of
Cu(In,Ga)Se2 thin films for high-efficiency solar cells. Nature Materials, 12(12):1107–1111
[2013]. ISSN 1476-1122, 1476-4660. doi: 10.1038/nmat3789.
[85]
J.-P. Bäcker. In situ investigation of the rapid thermal reaction of Cu-In-Ga precursors to
Cu(In,Ga)Se2 thin-film solar cell absorbers. Dissertation, Freie Universität Berlin [2018].
[86]
S. S. Schmidt, C. Wolf, H. Rodriguez-Alvarez et al. Adjusting the Ga grading during fast
atmospheric processing of Cu(In,Ga)Se
2
solar cell absorber layers using elemental selenium
vapor: Adjusting the Ga grading in fast processing of CIGSe. Progress in Photovoltaics:
Research and Applications, 25(5):341–357 [2017]. ISSN 10627995. doi: 10.1002/pip.2865.
[87]
T. Dalibor, P. Eraerds, M. Grave et al. Advanced PVD buffers on the road to GW-scale
CIGSSe production. pp. 1–5. IEEE [2017]. ISBN 978-1-5090-5605-7. doi: 10.1109/PVSC.
2017.8366300.
160
Bibliography
[88]
C. A. Kaufmann. Chemical Bath Deposition of Thin Semiconductor Films for Use as
Buffer Layers in CuInS2 Thin Film Solar Cells. Dissertation, University of Oxford, Oxford
[2002].
[89]
S. Merdes, F. Ziem, T. Lavrenko et al. Above 16% efficient sequentially grown
Cu(In,Ga)(Se,S)
2
-based solar cells with atomic layer deposited Zn(O,S) buffers: CIGSSe-
based solar cells with atomic layer deposited Zn(O,S) buffers. Progress in Photovoltaics:
Research and Applications, 23(11):1493–1500 [2015]. ISSN 10627995. doi: 10.1002/pip.2579.
[90]
T. Koida, J. Nishinaga, Y. Ueno et al. Impact of front contact layers on performance
of Cu(In,Ga)Se
2
solar cells in relaxed and metastable states. Progress in Photovoltaics:
Research and Applications [2018]. ISSN 10627995. doi: 10.1002/pip.3017.
[91]
T. Koida, Y. Ueno, J. Nishinaga et al. Cu(In,Ga)Se
2
Solar Cells with Amorphous In
2
O
3
-Based Front Contact Layers. ACS Applied Materials & Interfaces [2017]. ISSN 1944-8244,
1944-8252. doi: 10.1021/acsami.7b07092.
[92]
N. Zhou, M.-G. Kim, S. Loser et al. Amorphous oxide alloys as interfacial layers with
broadly tunable electronic structures for organic photovoltaic cells. Proceedings of the
National Academy of Sciences, 112 (26):7897–7902 [2015]. ISSN 0027-8424, 1091-6490. doi:
10.1073/pnas.1508578112.
[93]
M. Marinkovic. Contact resistance effects in thin film solar cells and thin film transistors.
Dissertation, Jacobs University, Bremen [2013].
[94]
B. D. Viezbicke, S. Patel, B. E. Davis et al. Evaluation of the Tauc method for optical
absorption edge determination: ZnO thin films as a model system: Tauc method for optical
absorption edge determination. physica status solidi (b), 252(8):1700–1710 [2015]. ISSN
03701972. doi: 10.1002/pssb.201552007.
[95]
U. Rau, D. Abou-Ras and T. Kirchartz. Advanced characterization techniques for thin
film solar cells. Wiley-VCH, Weinheim, Germany [2011]. ISBN 978-3-527-41003-3. OCLC:
ocn676728907.
[96]
Electron Backscatter Diffraction (EBSD). Https://www.helmholtz-
berlin.de/forschung/oe/ee/si-pv/analytik/rem/edx1_en.html; Accessed 09.08.2018.
[97]
M. Birkholz, P. F. Fewster and C. Genzel. Thin film analysis by X-ray scattering. Wiley-
VCH, Weinheim [2006]. ISBN 978-3-527-31052-4.
[98]
K. Frost, D. Kaminski, G. Kirwan et al. Crystallinity and structure of starch using wide
angle X-ray scattering. Carbohydrate Polymers, 78 (3):543–548 [2009]. ISSN 01448617. doi:
10.1016/j.carbpol.2009.05.018.
161
Bibliography
[99]
R. A. Young, editor. The Rietveld method. No. 5 in International Union of Crystallography
monographs on crystallography. International Union of Crystallograhy ; Oxford University
Press, [Chester, England] : Oxford ; New York [1993]. ISBN 978-0-19-855577-3.
[100]
D. Erfurt, M. D. Heinemann, S. S. Schmidt et al. Influence of zno-based sub-layers on the
growth of hydrogen doped indium oxide. ACS Appl. Energy Mater., 1:5490 – 5499 [2018].
doi: 10.1021/acsaem.8b01039.
[101]
R. Ajimsha, A. K. Das, V. K. Sahu et al. Valance band offset of TiO 2 /CuGaO 2
hetero-structure measured by x-ray photoelectron spectroscopy. Solar Energy Materials and
Solar Cells, 140:446–449 [2015]. ISSN 09270248. doi: 10.1016/j.solmat.2015.04.035.
[102] D. Nečas and P. Klapetek. Gwyddion [2016].
[103]
Bruker. OPUS. Https://www.bruker.com/de/products/infrared-near-infrared-and-raman-
spectroscopy/opus-spectroscopy-software.html; Accessed 10.08.2018.
[104] WaveMetrics. IGOR Pro [2014].
[105]
D. Erfurt, M. D. Heinemann, S. Körner et al. Improved electrical properties of pulsed
dc magnetron sputtered hydrogen doped indium oxide after annealing in air. Materials
Science in Semiconductor Processing, 89:170 – 175 [2019]. ISSN 1369-8001. doi: https:
//doi.org/10.1016/j.mssp.2018.09.012.
[106]
T. Koida, H. Shibata, M. Kondo et al. Correlation between oxygen stoichiometry, structure,
and opto-electrical properties in amorphous In
2
O
3
:H films. Journal of Applied Physics,
111(6):063721 [2012]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.3696978.
[107]
G. C. E. Jost, A. N. Hamri, F. Köhler et al. Reliability aspects of hydrogen-doped indium
oxide. physica status solidi (a), 213(7):1751–1759 [2016]. ISSN 1862-6319. doi: 10.1002/
pssa.201532806.
[108]
J. B. Varley, H. Peelaers, A. Janotti et al. Hydrogenated cation vacancies in semiconducting
oxides. Journal of Physics: Condensed Matter, 23(33):334212 [2011]. ISSN 0953-8984,
1361-648X. doi: 10.1088/0953-8984/23/33/334212.
[109]
T. Koida. Amorphous and crystalline In2o3-based transparent conducting films for
photovoltaics. physica status solidi (a), 214(2):1600464 [2017]. ISSN 1862-6319. doi:
10.1002/pssa.201600464.
[110]
H. Scherg-Kurmes, S. Seeger, S. Körner et al. Optimization of the post-deposition annealing
process of high-mobility In2o3:H for photovoltaic applications. Thin Solid Films, 599:78–83
[2016]. ISSN 00406090. doi: 10.1016/j.tsf.2015.12.054.
162
Bibliography
[111]
P. Prathap, G. G. Devi, Y. Subbaiah et al. Growth and characterization of indium oxide
films. Current Applied Physics, 8(2):120–127 [2008]. ISSN 15671739. doi: 10.1016/j.cap.
2007.06.001.
[112]
E. Burstein. Anomalous Optical Absorption Limit in InSb. Physical Review, 93(3):632–633
[1954]. ISSN 0031-899X. doi: 10.1103/PhysRev.93.632.
[113]
T. S. Moss. The Interpretation of the Properties of Indium Antimonide. Proceedings of
the Physical Society. Section B, 67(10):775–782 [1954]. ISSN 0370-1301. doi: 10.1088/
0370-1301/67/10/306.
[114]
T. Tohsophon, A. Dabirian, S. De Wolf et al. Environmental stability of high-mobility
indium-oxide based transparent electrodes. APL Materials, 3(11):116105 [2015]. ISSN
2166-532X. doi: 10.1063/1.4935125.
[115]
D. Greiner, N. Papathanasiou, A. Pflug et al. Influence of damp heat on the optical and
electrical properties of Al-doped zinc oxide. Thin Solid Films, 517(7):2291–2294 [2009].
ISSN 00406090. doi: 10.1016/j.tsf.2008.10.107.
[116]
D. Greiner, S. Gledhill, C. Köble et al. Damp heat stability of Al-doped zinc oxide films on
smooth and rough substrates. Thin Solid Films, 520(4):1285–1290 [2011]. ISSN 00406090.
doi: 10.1016/j.tsf.2011.04.190.
[117]
U. Rau and M. Schmidt. Electronic properties of ZnO/CdS/Cu(In,Ga)Se2 solar cells —
aspects of heterojunction formation. Thin Solid Films, 387(1-2):141–146 [2001]. ISSN
00406090. doi: 10.1016/S0040-6090(00)01737-5.
[118]
U. Rau, P. O. Grabitz and J. H. Werner. Resistive limitations to spatially inhomogeneous
electronic losses in solar cells. Applied Physics Letters, 85(24):6010–6012 [2004]. ISSN
0003-6951, 1077-3118. doi: 10.1063/1.1835536.
[119]
R. Klenk, A. Steigert, T. Rissom et al. Junction formation by Zn(O,S) sputtering yields
CIGSe-based cells with efficiencies exceeding 18%: Junction formation by Zn(O,S) sputtering.
Progress in Photovoltaics: Research and Applications, 22 (2):161–165 [2014]. ISSN 10627995.
doi: 10.1002/pip.2445.
[120]
T. Koida, Y. Kamikawa-Shimizu, Akimasa Yamada et al. Cu(In,Ga)Se
2
Solar Cells
With Amorphous Oxide Semiconducting Buffer Layers. IEEE Journal of Photovoltaics,
5(3):956–961 [2015]. ISSN 2156-3381, 2156-3403. doi: 10.1109/JPHOTOV.2015.2396356.
[121]
K. Ellmer, A. Klein and B. Rech, editors. Transparent conductive zinc oxide: basics and
applications in thin film solar cells. No. 104 in Springer series in materials science. Springer,
Berlin [2008]. ISBN 978-3-540-73611-0. OCLC: ocn181423059.
163
Bibliography
[122]
V. Gupta and A. Mansingh. Influence of postdeposition annealing on the structural and
optical properties of sputtered zinc oxide film. Journal of Applied Physics, 80(2):1063–1073
[1996]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.362842.
[123]
E. Bachari, G. Baud, S. Ben Amor et al. Structural and optical properties of sputtered
ZnO films. Thin Solid Films, 348(1-2):165–172 [1999]. ISSN 00406090. doi: 10.1016/
S0040-6090(99)00060-7.
[124] Y. E. Lee, J. B. Lee, Y. J. Kim et al. Microstructural evolution and preferred orientation
change of radio-frequency-magnetron sputtered ZnO thin films. Journal of Vacuum Science
& Technology A: Vacuum, Surfaces, and Films, 14(3):1943–1948 [1996]. ISSN 0734-2101,
1520-8559. doi: 10.1116/1.580365.
[125]
W. Water and S.-Y. Chu. Physical and structural properties of ZnO sputtered films. Materials
Letters, 55(1-2):67–72 [2002]. ISSN 0167577X. doi: 10.1016/S0167-577X(01)00621-8.
[126]
S. Nasser, H. Afify, S. El-Hakim et al. Structural and physical properties of sprayed
copper–zinc oxide films. Thin Solid Films, 315(1-2):327–335 [1998]. ISSN 00406090. doi:
10.1016/S0040-6090(97)00757-8.
[127]
F. van de Pol, F. Blom and T. Popma. R.f. planar magnetron sputtered ZnO films I:
Structural properties. Thin Solid Films, 204(2):349–364 [1991]. ISSN 00406090. doi:
10.1016/0040-6090(91)90074-8.
[128]
O. Kluth, G. Schöpe, J. Hüpkes et al. Modified Thornton model for magnetron sputtered
zinc oxide: film structure and etching behaviour. Thin Solid Films, 442(1-2):80–85 [2003].
ISSN 00406090. doi: 10.1016/S0040-6090(03)00949-0.
[129]
S.-N. Bai and T.-Y. Tseng. Electrical and optical properties of ZnO:Al thin films grown by
magnetron sputtering. Journal of Materials Science: Materials in Electronics, 20(3):253–256
[2009]. ISSN 0957-4522, 1573-482X. doi: 10.1007/s10854-008-9712-3.
[130]
P. Cannard and R. Tilley. New intergrowth phases in the ZnO-In2o3 system. Journal of Solid
State Chemistry, 73(2):418–426 [1988]. ISSN 00224596. doi: 10.1016/0022-4596(88)90127-2.
[131]
C. H. Yi, I. Yasui and Y. Shigesato. Oriented Tin-Doped Indium Oxide Films on <001>
Preferred Oriented Polycrystalline ZnO Films. Japanese Journal of Applied Physics,
34(Part 1, No. 3):1638–1642 [1995]. ISSN 00214922. doi: 10.1143/JJAP.34.1638.
[132]
Y.-C. Liang, C.-C. Liu, C.-C. Kuo et al. Structural and opto-electronic properties of
transparent conducting (222)-textured Zr-doped In2o3/ZnO bilayer films. Journal of Crystal
Growth, 310(16):3741–3745 [2008]. ISSN 00220248. doi: 10.1016/j.jcrysgro.2008.05.053.
164
Bibliography
[133]
D. Shin, J. H. Shin, H. Y. Lee et al. Characteristics of IZO/AZO and AZO/IZO Bi-layer
transparent conducting thin films prepared by using PLD. Journal of the Korean Physical
Society, 60(10):1662–1665 [2012]. ISSN 0374-4884, 1976-8524. doi: 10.3938/jkps.60.1662.
[134]
X. Sun, L. Wang and H. Kwok. Improved ITO thin films with a thin ZnO buffer layer
by sputtering. Thin Solid Films, 360(1-2):75–81 [2000]. ISSN 00406090. doi: 10.1016/
S0040-6090(99)01077-9.
[135]
P. Thilakan, C. Minarini, S. Loreti et al. Investigations on the crystallisation properties
of RF magnetron sputtered indium tin oxide thin films. Thin Solid Films, 388(1-2):34–40
[2001]. ISSN 00406090. doi: 10.1016/S0040-6090(01)00820-3.
[136]
H. Nakazawa, Y. Ito, E. Matsumoto et al. The electronic properties of amorphous and
crystallized In2o3 films. Journal of Applied Physics, 100(9):093706 [2006]. ISSN 0021-8979,
1089-7550. doi: 10.1063/1.2358829.
[137]
Y. S. Jung and S. S. Lee. Development of indium tin oxide film texture during DC magnetron
sputtering deposition. Journal of Crystal Growth, 259(4):343–351 [2003]. ISSN 00220248.
doi: 10.1016/j.jcrysgro.2003.07.006.
[138]
W. Yin, K. Smithe, P. Weiser et al. Hydrogen centers and the conductivity of I n 2 O
3 single crystals. Physical Review B, 91(7) [2015]. ISSN 1098-0121, 1550-235X. doi:
10.1103/PhysRevB.91.075208.
[139]
Y. Wu, B. Macco, D. Vanhemel et al. Atomic Layer Deposition of In
2
O
3
:H from InCp
and H
2
O/O
2
: Microstructure and Isotope Labeling Studies. ACS Applied Materials &
Interfaces, 9(1):592–601 [2017]. ISSN 1944-8244, 1944-8252. doi: 10.1021/acsami.6b13560.
[140]
M. D. Heinemann, M. F. A. M. van Hest, M. Contreras et al. Amorphous oxides as electron
transport layers in Cu(In,Ga)Se
2
superstrate devices: Amorphous oxides in Cu(In,Ga)Se
2
superstrate devices. physica status solidi (a), 214 (5):1600870 [2017]. ISSN 18626300. doi:
10.1002/pssa.201600870.
[141]
S. Merdes, R. Sáez-Araoz, A. Ennaoui et al. Recombination mechanisms in highly efficient
thin film Zn(S,O)/Cu(In,Ga)S2 based solar cells. Applied Physics Letters, 95(21):213502
[2009]. ISSN 0003-6951, 1077-3118. doi: 10.1063/1.3266829.
[142]
P. Innocenzi. Infrared spectroscopy of sol–gel derived silica-based films: a spectra-
microstructure overview. Journal of Non-Crystalline Solids, 316(2-3):309–319 [2003]. ISSN
00223093. doi: 10.1016/S0022-3093(02)01637-X.
[143]
C. M. Ghimbeu, M. Lumbreras, J. Schoonman et al. Electrosprayed Metal Oxide Semicon-
ductor Films for Sensitive and Selective Detection of Hydrogen Sulfide. Sensors, 9(11):9122–
9132 [2009]. ISSN 1424-8220. doi: 10.3390/s91109122.
165
Bibliography
[144]
P. Singh and D. Kaur. Room temperature growth of nanocrystalline anatase TiO2 thin
films by dc magnetron sputtering. Physica B: Condensed Matter, 405(5):1258–1266 [2010].
ISSN 09214526. doi: 10.1016/j.physb.2009.11.061.
[145]
R. Henríquez, E. Muñoz, E. A. Dalchiele et al. Electrodeposition of In
2
O
3
thin films from
a dimethylsulfoxide based electrolytic solution. physica status solidi (a), 210(2):297–305
[2013]. ISSN 18626300. doi: 10.1002/pssa.201228534.
[146]
M. Bouttemy, P. Tran-Van, I. Gerard et al. Thinning of CIGS solar cells: Part I: Chemical
processing in acidic bromine solutions. Thin Solid Films, 519(21):7207–7211 [2011]. ISSN
00406090. doi: 10.1016/j.tsf.2010.12.219.
[147]
Z. Jehl, F. Erfurth, N. Naghavi et al. Thinning of CIGS solar cells: Part II: Cell
characterizations. Thin Solid Films, 519(21):7212–7215 [2011]. ISSN 00406090. doi:
10.1016/j.tsf.2010.12.224.
[148]
T. Mise and T. Nakada. Microstructural properties of (In,Ga)2se3 precursor layers for
efficient CIGS thin-film solar cells. Solar Energy Materials and Solar Cells, 93(6-7):1000–
1003 [2009]. ISSN 09270248. doi: 10.1016/j.solmat.2008.11.028.
[149]
V. Deprédurand, T. Bertram, M. Thévenin et al. Alternative Etching for Improved Cu-rich
CuInSe2 Solar Cells. MRS Proceedings, 1771:163–168 [2015]. ISSN 1946-4274. doi:
10.1557/opl.2015.447.
[150]
T. Jäger, Y. E. Romanyuk, S. Nishiwaki et al. Hydrogenated indium oxide window layers
for high-efficiency Cu(In,Ga)Se 2solar cells. Journal of Applied Physics, 117(20):205301
[2015]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.4921445.
[151]
H. Kitami, M. Miyashita, T. Sakemi et al. Quantitative analysis of ionization rates of
depositing particles in reactive plasma deposition using mass-energy analyzer and Langmuir
probe. Japanese Journal of Applied Physics, 54(1S):01AB05 [2015]. ISSN 0021-4922,
1347-4065. doi: 10.7567/JJAP.54.01AB05.
[152]
W. Yang, Z. Liu, D.-L. Peng et al. Room-temperature deposition of transparent conduct-
ing Al-doped ZnO films by RF magnetron sputtering method. Applied Surface Science,
255(11):5669–5673 [2009]. ISSN 01694332. doi: 10.1016/j.apsusc.2008.12.021.
[153]
W. Yang, Z. Wu, Z. Liu et al. Room temperature deposition of Al-doped ZnO films on
quartz substrates by radio-frequency magnetron sputtering and effects of thermal annealing.
Thin Solid Films, 519(1):31–36 [2010]. ISSN 00406090. doi: 10.1016/j.tsf.2010.07.048.
[154]
K. H. Kim, K. C. Park and D. Y. Ma. Structural, electrical and optical properties
of aluminum doped zinc oxide films prepared by radio frequency magnetron sputtering.
166
Bibliography
Journal of Applied Physics, 81(12):7764–7772 [1997]. ISSN 0021-8979, 1089-7550. doi:
10.1063/1.365556.
[155]
T. Koida, T. Kaneko and H. Shibata. Carrier Compensation Induced by Thermal Annealing
in Al-Doped ZnO Films. Materials, 10(2):141 [2017]. ISSN 1996-1944. doi: 10.3390/
ma10020141.
[156]
W.-J. Jeong and G.-C. Park. Electrical and optical properties of ZnO thin film as a function
of deposition parameters. Solar Energy Materials and Solar Cells, 65(1-4):37–45 [2001].
ISSN 09270248. doi: 10.1016/S0927-0248(00)00075-1.
[157]
A. Grimm, D. Kieven, R. Klenk et al. Junction formation in chalcopyrite solar cells by
sputtered wide gap compound semiconductors. Thin Solid Films, 520(4):1330–1333 [2011].
ISSN 00406090. doi: 10.1016/j.tsf.2011.04.150.
[158]
J. Lindahl, J. Keller, O. Donzel-Gargand et al. Deposition temperature induced conduction
band changes in zinc tin oxide buffer layers for Cu(In,Ga)Se 2 solar cells. Solar Energy
Materials and Solar Cells, 144 :684–690 [2016]. ISSN 09270248. doi: 10.1016/j.solmat.2015.
09.048.
[159]
T. C. Kaspar, T. Droubay and J. E. Jaffe. ZnO/Sn:In
2
O
3
and ZnO/CdTe band offsets
for extremely thin absorber photovoltaics. Applied Physics Letters, 99(26):263504 [2011].
ISSN 0003-6951, 1077-3118. doi: 10.1063/1.3672218.
[160] M. Burgelman. SCAPS [2015].
[161]
P. D. C. King, T. D. Veal, D. J. Payne et al. Surface Electron Accumulation and the Charge
Neutrality Level in In 2 O 3. Physical Review Letters, 101(11) [2008]. ISSN 0031-9007,
1079-7114. doi: 10.1103/PhysRevLett.101.116808.
[162]
F. Rüggeberg and A. Klein. The In2o3/CdTe interface: A possible contact for thin film
solar cells? Applied Physics A, 82(2):281–285 [2006]. ISSN 0947-8396, 1432-0630. doi:
10.1007/s00339-005-3329-7.
[163]
M. Ruckh, D. Schmid and H. W. Schock. Photoemission studies of the ZnO/CdS interface.
Journal of Applied Physics, 76(10):5945–5948 [1994]. ISSN 0021-8979, 1089-7550. doi:
10.1063/1.358417.
[164]
S.-K. Hong, T. Hanada, H. Makino et al. Band alignment at a ZnO/GaN (0001) heteroint-
erface. Applied Physics Letters, 78(21):3349–3351 [2001]. ISSN 0003-6951, 1077-3118. doi:
10.1063/1.1372339.
167
Bibliography
[165]
F. Säuberlich, J. Fritsche, R. Hunger et al. Properties of sputtered ZnO films and its
interfaces with CdS. Thin Solid Films, 431-432:378–381 [2003]. ISSN 00406090. doi:
10.1016/S0040-6090(03)00251-7.
[166]
E. A. Kraut, R. W. Grant, J. R. Waldrop et al. Precise Determination of the Valence-Band
Edge in X-Ray Photoemission Spectra: Application to Measurement of Semiconductor
Interface Potentials. Physical Review Letters, 44(24):1620–1623 [1980]. ISSN 0031-9007.
doi: 10.1103/PhysRevLett.44.1620.
[167]
M. Perego, G. Seguini and M. Fanciulli. XPS and IPE analysis of HfO2 band alignment
with high-mobility semiconductors. Materials Science in Semiconductor Processing, 11(5-
6):221–225 [2008]. ISSN 13698001. doi: 10.1016/j.mssp.2008.10.008.
[168]
C. J. Dong, W. X. Yu, M. Xu et al. Valence band offset of Cu
2
O/In
2
O
3
heterojunction
determined by X-ray photoelectron spectroscopy. Journal of Applied Physics, 110 (7):073712
[2011]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.3641637.
[169]
X. K. Ning, Z. J. Wang, Y. N. Chen et al. Valence-band offset and forward-backward
charge transfer in manganite/NiO and manganite/LaNiO
3
heterostructures. Nanoscale,
7(48):20635–20641 [2015]. ISSN 2040-3364, 2040-3372. doi: 10.1039/C5NR06026E.
[170]
J. Si, S. Jin, H. Zhang et al. Experimental determination of valence band offset at
PbTe/CdTe(111) heterojunction interface by x-ray photoelectron spectroscopy. Applied
Physics Letters, 93(20):202101 [2008]. ISSN 0003-6951, 1077-3118. doi: 10.1063/1.3028028.
[171]
V. Srikant and D. R. Clarke. On the optical band gap of zinc oxide. Journal of Applied
Physics, 83(10):5447–5451 [1998]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.367375.
[172]
T. Kamiya and M. Kawasaki. ZnO-Based Semiconductors as Building Blocks for Active
Devices. MRS Bulletin, 33(11):1061–1066 [2008]. ISSN 0883-7694, 1938-1425. doi: 10.
1557/mrs2008.226.
[173]
H. Song, G. Zheng, A. Yang et al. The growth of ZnO on buffer layers and the valence
band offset determined by X-ray photoemission spectroscopy. Solid State Communications,
150(41-42):1991–1994 [2010]. ISSN 00381098. doi: 10.1016/j.ssc.2010.08.022.
[174]
T. Jäger, Y. E. Romanyuk, B. Bissig et al. Improved open-circuit voltage in Cu(In,Ga)Se2
solar cells with high work function transparent electrodes. Journal of Applied Physics,
117(22):225303 [2015]. ISSN 0021-8979, 1089-7550. doi: 10.1063/1.4922351.
[175]
W. Witte, R. Carron, D. Hariskos et al. IZO or IOH Window Layers Combined with Zn(O,S)
and CdS Buffers for Cu(In,Ga)Se
2
Solar Cells. physica status solidi (a), 214(12):1700688
[2017]. ISSN 18626300. doi: 10.1002/pssa.201700688.
168
Bibliography
[176]
J.-H. Wi, W.-J. Lee, D.-H. Cho et al. Characteristics of temperature and wavelength
dependence of CuInSe
2
thin-film solar cell with sputtered Zn(O,S) and CdS buffer layers:
Temperature and wavelength dependence of CuInSe
2
thin-film solar cell. physica status
solidi (a), 211(9):2172–2176 [2014]. ISSN 18626300. doi: 10.1002/pssa.201431232.
[177]
S.-Y. Park, E.-W. Lee, S.-H. Lee et al. Investigation of ZnO/CdS/CuInxGa1-xSe2 interface
reaction by using hot-stage TEM. Current Applied Physics, 10(3):S399–S401 [2010]. ISSN
15671739. doi: 10.1016/j.cap.2010.02.043.
[178]
S. Kijima and T. Nakada. High-Temperature Degradation Mechanism of Cu(In,Ga)Se
2
-Based Thin Film Solar Cells. Applied Physics Express, 1:075002 [2008]. ISSN 1882-0778,
1882-0786. doi: 10.1143/APEX.1.075002.
[179]
S. Merdes, V. Malinen, F. Ziem et al. Zn(O,S) buffer prepared by atomic layer deposition
for sequentially grown Cu(In,Ga)(Se,S)2 solar cells and modules. Solar Energy Materials
and Solar Cells, 126:120–124 [2014]. ISSN 09270248. doi: 10.1016/j.solmat.2014.03.044.
[180]
H. Hoppe, M. Seeland and B. Muhsin. Optimal geometric design of monolithic thin-film
solar modules: Architecture of polymer solar cells. Solar Energy Materials and Solar Cells,
97:119–126 [2012]. ISSN 09270248. doi: 10.1016/j.solmat.2011.09.037.
[181]
N. F. Cooray, K. Kushiya, A. Fujimaki et al. Large area ZnO films optimized for graded
band-gap Cu(InGa)Se2-based thin-film mini-modules. Solar Energy Materials and Solar
Cells, 49(1-4):291–297 [1997]. ISSN 09270248. doi: 10.1016/S0927-0248(97)00055-X.
[182]
M. D. Heinemann, J. Berry, G. Teeter et al. Oxygen deficiency and Sn doping of amorphous
Ga
2
O
3
. Applied Physics Letters, 108 (2):022107 [2016]. ISSN 0003-6951, 1077-3118. doi:
10.1063/1.4938473.
[183]
M. Ruckh, D. Schmid, M. Kaiser et al. Influence of substrates on the electrical properties
of Cu(In,Ga)Se2 thin films. Solar Energy Materials and Solar Cells, 41-42:335–343 [1996].
ISSN 09270248. doi: 10.1016/0927-0248(95)00105-0.
[184]
A. Steigert, I. Lauermann, R. Muydinov et al. Crystallization of sputtered In2o3 and
In2o3:H [2016].
[185]
D. Kieven, A. Grimm, I. Lauermann et al. Band alignment at sputtered ZnSx O1-
x/Cu(In,Ga)(Se,S)2 heterojunctions. physica status solidi (RRL) - Rapid Research Letters,
6(7):294–296 [2012]. ISSN 18626254. doi: 10.1002/pssr.201206195.
[186]
National Institute of Standards and Technology (NIST). Resistivity and Hall Mea-
surements. Https://www.nist.gov/pml/engineering-physics-division/popular-links/hall-
effect/resistivity-and-hall-measurements; Accessed 09.08.2018.
169
Bibliography
[187]
B. Arnaudov, T. Paskova, S. Evtimova et al. Multilayer model for Hall effect data analysis
of semiconductor structures with step-changed conductivity. Physical Review B, 67(4)
[2003]. ISSN 0163-1829, 1095-3795. doi: 10.1103/PhysRevB.67.045314.
[188]
R. W. Cahn, P. Haasen and E. J. Kramer. Materials science and technology: a comprehensive
treatment. VCH, Weinheim ; New York [1991]. ISBN 978-3-527-26813-9.
[189]
Park Systems. Basic Contact AFM & Dynamic Force Microscope (DFM). Tech. rep.
[2018]. Https://www.parksystems.com/images/spmmodes/standard/2_Basic-Contact-
AFM-and-Dynamic-Force-Microscopy-(DFM).pdf; Accessed 10.08.2018.
170
List of Publications
First-authorship
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Reiner Klenk, Rutger
Schlatmann, "Improved electrical properties of pulsed DC magnetron sputtered hydrogen
doped indium oxide after annealing in air", Materials Science in Semiconductor Processing
89 (2019) 170–175; doi: 10.1016/j.mssp.2018.09.012
•Darja Erfurt
, Marc D. Heinemann, Sebastian S. Schmidt, Stefan Körner, Bernd Szyszka,
Reiner Klenk, Rutger Schlatmann, "Substrate influence on the growth of hydrogen doped in-
dium oxide", ACS Appl. Energy Mater. 2018, 1, 5490-5499; doi: 10.1021/acsaem.8b01039
Co-authorship
•
Houda Ennaceri,
Darja Erfurt
, Lan Wang, Tristan Köhler, Abdelhafed Taleb, Asmae
Khaldoun, Abdallah El Kenz, Abdelilah Benyoussef, Ahmed Ennaoui, "Deposition of
multifunctional TiO
2
and ZnO top-protective coatings for CSP application", Surface &
Coatings Technology 298 (2016) 103–113; DOI: 10.1016/j.surfcoat.2016.04.048
•
Houda Ennaceri, Lan Wang,
Darja Erfurt
, Wiebke Riedel, Gauri Mangalgiri, Asmae
Khaldoun, Abdallah El Kenz, Abdelilah Benyoussef, Ahmed Ennaoui, "Water-resistant
surfaces using zinc oxide structured nanorod arrays with switchable wetting property",
Surface &Coatings Technology 299 (2016) 169–176; DOI: 10.1016/j.surfcoat.2016.04.056
•
Sri Hari Bharath Vinoth Kumar, Ruslan Muydinov, Tat‘yana Kol’tsova,
Darja Erfurt
,
Alexander Steigert, Oleg Tolochko, and Bernd Szyszka, "Graphene assisted effective hole-
extraction on In
2
O
3
:H/CH
3
NH
3
PbI
3
interface: Studied by modulated surface spectroscopy",
Appl. Phys. Lett. 112, 011604 (2018); DOI: 10.1063/1.5017579
•
Stefan Körner, Manuel Hartig, Ruslan Muydinov,
Darja Erfurt
, Reiner Klenk, Bernd
Szyszka, and Rutger Schlatmann, "Serial cosputtering for aluminum doping manipulated
171
List of Publications
zinc oxide as front contact for Cu(In,Ga)Se
2
solar cells", Japanese Journal of Applied
Physics 57, 08RC18 (2018); DOI: 10.7567/JJAP.57.08RC18
•
Ruslan Muydinov, Alexander Steigert, Markus Wollgarten, Pawel Michalowski, Ulrike
Bloeck, Andreas Pflug,
Darja Erfurt
, Reiner Klenk, Stefan Körner, Iver Lauermann, Bernd
Szyszka, "Crystallisation phenomena of In
2
O
3
:H films", Materials 2019, 12(2), 266; DOI:
10.3390/ma12020266
Conference Contributions
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Reiner Klenk, Christian
A. Kaufmann, Rutger Schlatmann, " Influence of the Zn(O,S) buffer onto the In
2
O
3
:H
front contact in CIGS solar cells" Poster presentation, IW-CIGSTech 7 Munic, Germany,
23.06.2016 - 23.06.2016
•
Marc D. Heinemann, Klaus Ellmer, Moritz Kölbach, Dieter Greiner,
Darja Erfurt
, Reiner
Klenk, Rutger Schlatmann, Christian A. Kaufmann, "Amorphous InGaO
x
, a versatile
electron transport layer for solar cells", Oral presentation, 20th International Conference
on Ternary and Multinary Compounds (“ICTMC-20”) Halle (Saale), Germany, 05.09.2016
- 09.09.2016
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Christian A. Kaufmann,
Rutger Schlatmann, "Integration of IOH as a front contact layer for CIGS cells and
modules", Oral presentation, 6th International Symposium on Transparent Conductive
Materials Platanias-Chania Crete, Greece, 09.10.2016 - 13.10.2016
•Darja Erfurt
, "Wasserstoff-dotiertes In
2
O
3
als TCO mit hoher Mobilität", Oral presentation,
6. EFDS-Workshop "Transparente leitfähige Materialien (TCO/TCM): Festkörperphysikalis-
che Grundlagen und Technologien" Erfurt, Germany, 22.11.2016 - 23.11.2016
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Reiner Klenk, Rutger
Schlatmann, "In
2
O
3
:H with high mobility prepared by DC sputtering and annealing in air",
Oral presentation, EMRS Spring 2017 Strasbourg, France, 22.05.2017 - 26.05.2017
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Reiner Klenk, Rutger
Schlatmann, "Hydrogen doped Indium Oxide as a high mobility TCO prepared by DC
sputtering and annealing in air", Poster presentation, IW-CIGSTech8 Stuttgart, Germany,
30.05.2017 - 30.05.2017
•
Stefan Körner, Ruslan Muydinov,
Darja Erfurt
, Manuel Hartig, Bernd Szyszka, Reiner
Klenk, "Doping manipulated AZO as front TCO by using serial co-sputtering for CIGS
solar cells", Oral presentation, PVSEC27 Otsu, Japan, 12.11.2017 - 17.11.2017
172
List of Publications
•Darja Erfurt
, Marc D. Heinemann, Stefan Körner, Bernd Szyszka, Reiner Klenk, Rutger
Schlatmann, "Challenges of an Hydrogen Doped Indium Oxide Window Layer in CIGS
Modules", Oral presentation, PVSEC-27, 27th International Photovoltaic Science and
Engineering Conference Otsu, Japan, 12.11.2017 - 17.11.2017
•Darja Erfurt
, Marc D. Heinemann, Reiner Klenk, Takashi Koida, Jiro Nishinaga, Yukiko
Kamikawa, Hajime Shibata, Hasan A. Yetkin, Bernd Szyszka, Rutger Schlatmann, "Strate-
gies to improve the electron mobility of IOH on CIGS samples", Poster presentation,
IW-CIGSTech 9 Stuttgart, Germany, 18.06.2018 - 18.06.2018
•
Stefan Körner,
Darja Erfurt
, Reiner Klenk, Michael Kirsch, Bernd Szyszka, Rutger Schlat-
mann, "Thickness variation of zinc magnesium oxide as buffer layer in CIGS solar cells",
Poster presentation, IW-CIGSTech 9 Stuttgart, Germany, 18.06.2018 - 18.06.2018
•
Bernd Szyszka, Ruslan Muydinov, Stefan Körner, Manuel Hartig, Marlene Härtel,
Darja
Erfurt
, Reiner Klenk, Christian A. Kaufmann, Bernd Stannowski, Rutger Schlatmann,
Alexander Steigert, Michael Siemers, Stephan Ulrich, Volker Sittinger, Andreas Pflug, "High
mobility TCOs for photovoltaics", Oral presentation to be published at 7th International
Symposium on Transparent Conductive Materials Crete, Greece, 14.10.2018 - 19.10.2018
173
Acknowledgement
First I would like to thank Prof. Dr. Bernd Szyszka for supervising this thesis. In addition I
want to thank Prof. Dr. Rutger Schlatmann for the opportunity to do my PhD at the PVcomB.
I am also very grateful to Prof. Dr. Günter Bräuer for reviewing this thesis. Dr. Jürgen Bruns
is acknowledged for being the chairperson of the defence.
My sincere thanks to Dr. Marc Heinemann and Dr. Reiner Klenk for the continuous support,
your feedback and the great supervision. Also Christian Kaufmann is acknowledged for his
supervision, in particular at the beginning of my PhD. Especially I want to thank Prof. Dr.
Rutger Schlatmann and Dr. Takashi Koida for the change to join the AIST for four months. Dr.
Takashi Koida is further acknowledged for supervising me at AIST and introducing me into the
Japanese culture. I had a wonderful time and learned a lot! Furthermore I want to acknowledge
Dr. Hajime Shibata, Dr. Jiro Nishinaga and Dr. Yukiko Kamikawa and all former colleagues at
AIST for their support during my stay.
Katja Mayer-Stillrich and Manuel Hartig are acknowledged for the support on the LOS1, Michael
Kirsch for the deposition of ZnO and Zn(O,S) films. Carola Klimm and Iris Dorbandt are
acknowledged for support with EBSD and SEM measurements, respectively. I also want to thank
Ulrike Bloek, Dr. Sebastian Schmidt and Dr. Chen Li for sample preparation and providing
TEM measurements. Also I want to thank Alexander Steigert for the support and help with
XPS/UPS measurements.
Many thanks to the whole PVcomB and especially the CIGS-team for maintain the systems
and labs, for providing all kind of samples and the general support. Among other things I
acknowledge Bianka Bunn for the preparation of glass substrates, Christian Wolf, Ralf Habrecht,
Marc Heinemann, Jakob Lauche, Tim Münchenberg and Tobias Bertram for the preparation
of CIGS absorber. Iris Dorbandt and Tim Münchenberg are acknowledged for CdS deposition.
I acknowledge all students at PVcomB for the support to characterize the solar cells. Sonja
Cinque and Guillermo Farias are acknowledged for the module structuring. Dr. Reiner Klenk,
Alexandros Cruz and Matthias Müller are acknowledged for proof reading the manuscript of this
thesis.
174
Acknowledgement
Furthermore I want the acknowledge all members of the PVcomB for the great working atmosphere
and distinct cake culture. Special thanks to my PhD colleagues, i.e. Stefan Körner, Tim Kodalle,
Alexandros Cruz, Jan-Peter Bäcker, Natalie Preissler, Hasan Yetkin and more for the great time
in the "Doktorandenbüro". In addition I would like to thank Daniel Meza for beeing such a nice
PhD spokespersons and for organizing the "Stammtisch".
I acknowledge the German Federal Ministry for Economic Affairs and Energy for the financial
support and for founding the TCO4CIGS project (contract number 0325762). Also I would like
to thank all the partners of the project for the continuous exchange of knowledge.
In particular I want to thank my friends and especially my parents for the unlimited support
during the last years.
175
