Synthesis and Characterization of Novel Composite
Photoelectrodes based on Chalcopyrite and Silicon for
the Visible Light-driven Hydrogen and Oxygen
Evolution
vorgelegt von
Dipl.-Ing.
Anahita Azarpira
geb. in Tehran
von der Fakultät III- Prozesswissenschaften- Institut für Energietechnik
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften
-Dr.-Ing.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. John Banhart
Gutachter: Prof. Dr. Frank Behrendt
Gutachter: Assoc. Prof. Dr. Thomas Schedel-Niedrig
Tag der wissenschaftlichen aussprache: 9. May 2016
Berlin 2016
Eidesstaatliche Erklärung
Hiermit versichere ich an Eides statt, die vorliegende Arbeit selbstständig und eigen-
händig angefertigt zu haben.
Anahita Azarpira
09.05.2016
Acknowledgement
First of all I would like to express my deepest gratitude to my advisors Prof. Dr. Thomas
Schedel-Niedrig and Dr. Michael Lublow for their patience, motivation, enthusiasm and
continuous support throughout my thesis work. Their guidance and training in research
using various methods, their support in analysis, interpretation and discussions of the
results, presented in this thesis, is highly appreciated.
I would like to give a heartfelt, special thanks to Dr. Michael Lublow for his detailed
advice in electrochemistry and surface science. His creativity and perseverance makes
him an outstanding mentor and helped me solving even most intricate problems.
I am very grateful to Prof. Dr. Frank Behrendt who made it possible to complete
this work in an inspiring atmosphere. It has been an honor to be his Ph.D. student.
Many valuable results were achieved and new ideas were developed during discussions in
seminars under his supervision.
I would also like to thank Prof. Dr. Roel van de Krol for giving me the opportunity
to realize this thesis in his department at the Helmholtz-Zentrum Berlin and Ms. Mirjam
Guerra for assistance with organizational issues.
I would like to thank Dr. Peter Bogdano for his kind assistance for DEMS and IPCE
measurements and Dr. Sebastian Fiechter for giving me detailed introduction to XRD
measurements and data analysis. The assistance by Dr. Aafke Bronneberg for deposition
of
TiO2
by means of ALD and by Christian Höhn and Alex Steigert for XPS measure-
ments is highly appreciated. I appreciate very much, too, the support by Dr. Daniel
Abou-Ras and Ulrike Bloeck for SEM and TEM measurements.
I am very thankful for the collaboration with all my colleagues at the Helmholtz-Zentrum
Berlin who, each of them in his or her respective eld, provided assistance and advice.
Particularly the support given by Katarzyna Olech and Dianna Stellmach in DEMS mea-
surements is highly appreciated.
I would also like to thank Prof. Dr. Christian Herbert-Fischer to let me work in his
ILGAR laboratory. I want to thank present and former members of the ILGAR team for
their support and kindness.
I am very thankful to Dr. Christian Kaufmann for providing me with chalcopyrite
substrates and supporting my project with his expertise in analysis of semiconductor
properties. I am deeply thankful to Prof. Dr. Matthias Driess and Johannes Pfrommer
from Technical University Berlin for providing me with ZnO:Co catalyst powders.
Above all, I am extraordinarily grateful to my parents, Kian and Mohammad for be-
lieving in me and for being the mentors in my life. I deeply appreciate their eorts in
raising me always in a thoughtful and warm-hearted way and making me what I am
today. I deeply hope to meet always their expectations to full extent, now and in the
future.
Kurzfassung
In dieser Arbeit werden eziente und stabile Mehrschicht-Halbleiter-Elektroden besch-
rieben, die für die beiden Halbzellenreaktionen der lichtgestützten Wasserspaltung, d.h.
die Entwicklung von Wassersto (HER) bzw. Sauersto (OER), entwickelt wurden.
Für die eziente und stabile Entwicklung von Wassersto in sauren Elektrolyten wur-
den neuartige Photokathoden entwickelt, die hochqualitative Dünnschicht-Absorber aus
Cu(In,Ga)Se2
mit
TiO2
-Schichten verbinden, welche zusätzlich mit Pt dotiert wurden.
Variierte Konzentrationen von Pt wurden systematisch untersucht, um eine simultane
Optimierung zu erzielen bzgl. i) der Leitfähigkeit der
Pt −TiO2
- Schichten, ii) der elek-
trokatalytischen Aktivität und iii) der Lichteinkopplung in den Chalkopyrit Absorber. Es
wird nachgewiesen, dass die graduelle Erhöhung der Pt-Konzentration durch ein Ezienz-
und Stabilitäts-Maximum der resultierenden Photokathode läuft (bei etwa 5 vol. % der
Pt-haltigen Präparationslösung). Bei diesem Maximum wird eine optimierte Lichteinkop-
plung in den Chalkopyrit-Absorber realisiert, die Photostromdichten von 15
mAcm−2
am thermodynamischen Potential der Wasserstoentwicklung ermöglicht. Photokatho-
den, die diesem Parameter entsprechend hergestellt werden, arbeiten stabil über mehr
als 24 Stunden ohne Zeichen von Degradation.
Für die entsprechende Präparation von Halbleiter-basierten Photoanoden wurde das Ver-
fahren der elektrophoretischen Abscheidung angewendet. Hiermit wurden zwei Wasserox-
idationskatalysatoren, ZnO:Co und
RuO2
, auf Halbleitersubstraten immobilisiert. Für
die Sauerstoentwicklung in alkalischen Lösungen wurde eine detaillierte Analyse durchge-
führt, um optimale Parameter für die Abscheidung von ZnO:Co Katalysatoren auf u-
oridierte Zinnoxidschichten aus verschiedenen organischen Lösungen zu bestimmen. Die
nachfolgende Auswertung der elektrochemischen Aktivität zeigte eine deutliche Abhängig-
keit von der verwendeten organischen Lösung mit höchster Aktivität nach Abscheidung
aus Acetonitril und geringster Aktivität nach Abscheidung aus Ethanol. Die detaillierte
Analyse der jeweiligen Schichten mit Hilfe verschiedener Analyseverfahren zeigte, dass
die Veränderung der elektrochemischen Aktivität durch eine entsprechende Veränderung
der Grösse der aktiven Oberäche verursacht wird. Es konnte weiterhin gezeigt werden,
dass die weniger aktiven aber deutlich lichtdurchlässigeren ZnO:Co-Phasen, die durch
Abscheidung aus Ethanol gewonnen wurden, erfolgreich in einer kombinierten Anord-
nung mit einer kostensparenden Dreifach-Solarzelle verwendet werden können. Durch
diese Anordnung wurde eine Ezienz von 5% in der Umwandlung von Lichtenergie in
chemische Energie, d.h. Wassersto, erzielt. Um schliesslich eine Silicium-basierte Pho-
toanode herzustellen, die für die Sauerstoentwicklung in sauren Elektrolyten geeignet
ist, wurde eine neuartige Herangehensweise entwickelt: die Anwendbarkeit des ezien-
testen Wasseroxidationskatalysators,
RuO2
, zur Polymerisation von Alkoholen wurde hi-
erzu ausgenutzt. Dadurch konnte eine stabile organische Schutzschicht hergestellt wer-
den, welche zum ersten Mal den Langzeitbetrieb eines Silicium-
RuO2
Schichtsystems als
OER-Photoanode ermöglichte. Die organische Schutzschicht wurde hierbei durch eine
Iod-gestützte elektrochemisch-reduzierende Polymerisation realisiert, die zeitgleich mit
der elektrophoretischen Abscheidung von
RuO2
stattfand. Eine Analyse der möglichen
Reaktionswege legt nahe, dass die
RuO2
-induzierte Katalyse auf
E2
-Eliminationsschritten
beruht, welche für eine
sp3−sp2
Umformung der Kohlenstobindungen innerhalb des
Filmes sorgen. Für die beiden photoelektrochemischen Anwendungsformen, die photo-
voltaische und die photoelektrokatalytische, wurden Photostromdichten von 20
mAcm−2
bzw. 15
mAcm−2
sowie Stabilitäten von 8 bzw. 24 Stunden erzielt. Die organische
Schutzschicht ermöglicht schliesslich Silicium-Photo-spannungen von 500 mV, was auf
eine ausserordentlich hohe Grenzächenqualität hindeutet.
Abstract
In the presented work, ecient and stable multi-junction semiconductor electrodes
are introduced for the two half-cell reactions of photo-assisted splitting of water, the hy-
drogen evolution reaction (HER) and the oxygen evolution reaction (OER).
For ecient and stable HER in acidic electrolytes, novel composite photocathodes were
developed which functionalize device-grade
Cu(In,Ga)Se2
thin-lm absorbers in con-
junction with electrocatalytic Pt-implemented
TiO2
layers. Varying Pt-concentrations
were systematically investigated in order to optimize simultaneously (i) the conductivity
of the
Pt −TiO2
lms, (ii) the electrocatalytic activity, and (iii) light-guidance toward
the chalcopyrite absorber. It is shown that the gradual increase of the Pt-concentration
passes through an eciency- and stability-maximum of the device (at about 5 vol.% Pt
of the precursor solution). At this maximum, optimized light-incoupling into the chal-
copyrite light-absorber was achieved and 15
mAcm−2
at the thermodynamic potential
for
H2
-evolution (0 V vs. RHE) were realized. Devices, fabricated according to this op-
timized parameter, operated over more than 24 hours with no sign of degradation.
For the corresponding preparation of semiconductor-based photoanodes, electrophoretic
deposition was used for formation of two dierent water oxidation catalysts, ZnO:Co and
RuO2
on semiconductor supports. Firstly, for OER in alkaline solutions, an extensive
analysis was carried out in order to determine optimized parameters for electrophoretic
deposition of pre-synthesized ZnO:Co catalysts from varied organic solvents on uorinated
tin oxide. Evaluation of the electrochemical activity proved a clear solvent-dependence
with highest activity upon deposition from acetonitrile and lowest activity upon deposi-
tion from ethanol. Detailed analysis of the respective lms by various methods showed
that the change in electrochemical activity is caused by a corresponding variation in
the size of the active surface area. It is furthermore shown that less active but highly
transparent ZnO:Co phases, prepared from ethanol-containing suspensions, can be suc-
cessfully employed in a stacking conguration with low-cost triple-junction solar cells.
Thereby, solar-to-hydrogen eciencies of up to 5.0% were achieved. Secondly, for de-
vising a silicon-based photoanode, applicable to OER in acidic media, a novel approach
was developed: the capacity of the most ecient water oxidation catalyst in acidic elec-
trolytes,
RuO2
, was exploited towards alcohol polymerization. Thereby, a stable organic
protection layer could be formed which allows for the rst time long-term operation
of silicon-
RuO2
junction as OER-photoanode. The interfacial layers are generated via
iodine-mediated electro- reductive polymerization of alcohols, simultaneously forming
during electrophoretic transport of
RuO2
. Reaction chemistry analyses suggest that the
RuO2
-induced catalysis introduces E2-elimination reactions which result in a carbon
sp3
-
sp2
transformation within the lm. For the two modes of photoelectrochemical operation,
the photovoltaic and the photoelectrocatalytic mode, 20
mAcm−2
and 15
mAcm−2
pho-
tocurrent densities, respectively, were obtained with operational stability for 8 and 24 hrs.
The interfacial organic-protection layer enables Si photovoltages of 500mV, demonstrating
an extraordinary electronic interface quality.
Contents
1 Introduction 1
2 Concept of photoelectrochemical water splitting 7
2.1 Principle of photoelectrochemical cells . . . . . . . . . . . . . . . . . . . 7
2.2 Semiconductor photoelectrode material . . . . . . . . . . . . . . . . . . . 11
2.3 Description of electrolytes used for splitting of water . . . . . . . . . . . 12
2.4 Semiconductor (SC)/ electrolyte junctions: . . . . . . . . . . . . . . . . . 13
3 Introduction to photoelectrode architectures based on Si and chalcopy-
rite supports 17
3.1 Introduction to the used material combinations for HER . . . . . . . . . 17
3.1.1
Cu(In,Ga)Se2
chalcopyrite for HER . . . . . . . . . . . . . . . . 17
3.1.2 Pt-doped
TiO2
as electrocatalytic protection layer on CIGSe supports 18
3.2 Introduction to the used materials combinations for OER . . . . . . . . . 19
3.2.1 n-Silicon as an absorber for OER . . . . . . . . . . . . . . . . . . 19
3.2.2 ZnO:Co as catalyst for OER in alkaline electrolytes . . . . . . . . 20
3.2.3
RuO2
as catalyst for OER in acidic electrolytes . . . . . . . . . . 20
4 Experimental section 23
4.1 Samplepreparation.............................. 23
4.1.1 Ion-Layer-Gas-Reaction (ILGAR) . . . . . . . . . . . . . . . . . . 23
4.1.2 Electrophoretic deposition (EPD) . . . . . . . . . . . . . . . . . . 24
4.2 Analyticalmethods ............................. 27
4.2.1 Electrochemical analysis . . . . . . . . . . . . . . . . . . . . . . . 27
4.2.1.1 Current-Voltage (CV) behavior . . . . . . . . . . . . . . 27
4.2.1.2 Incidence photon to charge carrier conversion eciency
(IPCE) ........................... 28
4.2.1.3 Mott-Schottky measurements . . . . . . . . . . . . . . . 29
4.2.1.4 Dierential electrochemical mass spectroscopy (DEMS) . 30
4.2.2 Spectroscopic analysis . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2.2.1 Grazing incidence X-ray diraction (XRD) . . . . . . . . 31
4.2.2.2 Surface photoelectron spectroscopy(PES) . . . . . . . . . 32
4.2.3 Morphological-chemical analysis . . . . . . . . . . . . . . . . . . . 34
4.2.3.1 Scanning Electron Microscopy coupled with Energy Dis-
persive X-ray (SEM/EDX) Analysis . . . . . . . . . . . 34
4.2.3.2 Transmission Electron Microscopy (TEM) . . . . . . . . 35
4.2.4 Opticalanalysis............................ 36
4.2.4.1 Ultraviolet-visible spectroscopy . . . . . . . . . . . . . . 36
4.2.4.2 Fourier Transform Infrared Spectroscopy (FTIR) . . . . 37
4.2.4.3 Surface photovoltage spectroscopy (SPV) . . . . . . . . 37
5 Results and discussions 39
5.1 Characterization of doped
TiO2
/
Cu(In,Ga)Se2
photocathodes . . . . . . 39
5.1.1
Cu(In,Ga)Se2
as an absorber in heterojunction photocathode . . 40
5.1.2 Eect of the
TiO2
deposition temperature on the device performance 42
5.1.3 The Pt-doped
TiO2
layer: variation of the doping density . . . . . 44
5.1.3.1 The twofold functionality of Pt as dopant and catalyst . 46
5.1.3.2 Correlation of optical behavior with doping density . . . 50
5.1.3.3 Dependence of the energy of the tail states on the doping
density ........................... 52
5.1.3.4 IPCE - DEMS measurements . . . . . . . . . . . . . . . 54
5.1.3.5 pH dependent performance: CV and stability measurements 55
5.1.4 Interface engineering by thin interfacial
TiO2
deposited by Atomic
Layer Deposition (ALD) . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 EPD-prepared semiconductor electrodes for electrocatalytic and photo-
assisted evolution of oxygen . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2.1 ZnO:Co on FTO as electrocatalyst in alkaline electrolytes . . . . . 62
5.2.1.1 Eect of the electrophoretic mobility on the quality of the
EPDprocess ........................ 62
5.2.1.2 Eect of the organic solvent on the catalytic activity of
ZnO:Co electrodes . . . . . . . . . . . . . . . . . . . . . 66
5.2.1.3 Optical analysis of ZnO:Co on FTO by UV-VIS and SPV 69
5.2.1.4 Morphological analysis by SEM/ EDX/ TEM . . . . . . 74
5.2.2 Characterization of
RuO2
on FTO as electrocatalyst in acidic elec-
trolytes ................................ 79
5.2.3
RuO2
/Si photoanode for acidic electrolytes . . . . . . . . . . . . 80
5.2.3.1
RuO2
/Si: formation principles . . . . . . . . . . . . . . . 82
5.2.3.2 Characterization of the
RuO2
/Si photoanode . . . . . . 90
5.2.3.3 pH dependent performance of the
RuO2
/n-Si photoanodes 94
5.2.3.4 Photoelectrochemical solar cell application . . . . . . . . 94
6 Summary and outlook 97
7 Appendices 99
7.1 References................................... 99
7.2 Listofgures................................. 107
7.3 Listofpublications.............................. 113
7.4 Patents .................................... 115
7.5 Conference contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.6 CurriculumVitae............................... 117
1 Introduction
Clean production of hydrogen fuel from sunlight and water is the most desirable approach
for establishing a green economy for future generations [1]. Hydrogen (
H2
) represents a
very attractive, emission-free alternative fuel, whose energy density per mass unit (143
kJ/g) is approximately 100 times higher than for batteries and approximately three times
higher than for standard liquid fuels [2]. Therefore,
H2
is assumed to be a very promising
candidate for substitution of fossil energy resources. Most of the conventional techniques
for
H2
production require a source of carbon. One of the sustainable and environmental
friendly methods for
H2
production is solar water splitting. Water is an abundant source
of hydrogen, and approximately
∼3.5 ×1013 L
of water is needed to store the energy the
world uses in one year (
4.7 ×1020 J
) in the form of hydrogen. This corresponds to 0.01%
of the annual rainfall, or 0.000002% of the amount of water in the world's oceans [3]. The
overall reaction for water splitting is written as:
2H2O←→ 2H2+O2∆G = +237 kJ/moleH2
(1.1)
This reaction with positive Gibbs energy, G, is an endothermic reaction, which means
that external energy has to be provided to the system in order to realize the splitting of
water.
Figure 1.1: solar to hydrogen pathways.
Several dierent conversion routes can be followed in order to utilize sun light for
splitting of water. Figure 1.1 summarizes these pathways. The rst three routes are
based on solar thermal systems. The solar thermochemical route STC)[4] is based on
photon-to-thermal energy conversion followed by a thermal-to-chemical conversion step.
The concentrating solar thermal route (CST) [5] is a three steps process involving photon-
to-heat, heat-to-electricity and electricity-to-chemical conversion steps. The third route
is ultra-high temperature thermolysis [6], a one-step process, being performed at elevated
temperatures of 2200
◦
C-3000
◦
C or at 900
◦
C -1200
◦
C with the assistance of Pt/Ru cata-
lysts [7]. The fourth route is a two-step process that combines photovoltaics (PV) with
electrolysis. The conversion of light to electricity is realized by PV while the conversion
of electrical to chemical energy is carried out in electrolysers. Although all these path-
ways provide hydrogen as alternative fuel, most of them require multi-step processing
and eciency losses occur at each step. Moreover, ultra-high temperature treatments as
in thermolysis represent further energetic disadvantages. Therefore, from an economic
1
point of view, the application of photocatalytic and photoelectrocatalytic (PEC) conver-
sion processes using state-of-the-art semiconductors is an appealing alternative approach
to produce hydrogen from water and light [8]. For this purpose, semiconductor-based
architectures (as used in high-ecient photovoltaics) are employed in order to eciently
separate photoinduced charge carriers (electrons and holes) and to drive subsequently the
water-splitting reaction [9]. This photoelectrochemical (PEC) route is a one-step process
and is usually performed at room temperature and under ambient pressure conditions.
In 1972 A. Fujishima and K. Honda reported for the rst time direct water splitting
in a photoelectrochemical cell using a crystalline
TiO2
(rutile) anode and a Pt cathode
under ultraviolet (UV) irradiation and at an external bias of 840 mV [10-11]. Since then,
considerable eorts have been devoted to accomplish further advancements in this eld.
In 1981 a PEC device based on single crystal p-InP, covered by a thin Ru layer and with
solar-to-hydrogen (STH) eciency of 12% was reported by A. Heller [12]. A comparable
eciency of 12.1% was reported by Lewerenz et al. [13] in 2010 for p-InP covered by Rh
nanoparticles. In 2000, Turner et al. [14-15] and Licht et al. [16] reported 12% and 18%
STH based on III-V
(p−GaInP2/GaAs)
and Si/III-V monolithic devices, respectively.
Very recently, May et al. [17] reported a very ecient monolithic device based on multi
junction GaInAs/GaInP with Rh and
RuO2
cocatlaysts reaching an eciency of 14%.
However, all these high-eciency architectures are based on rare and therefore expensive
materials which make a scale-up process to large-area electrodes dicult. Since 2010,
signicant eorts have been done to replace noble-metals with earth-abundant materials
and to substitute very expensive III-V semiconductors by other appropriate supports. In
2011, Nocera et al. [18] reported a monolithic device based on triple-junction amorphous
silicon (a-Si) and Co catalysts for OER and NiMoZn catalysts for HER, realizing an STH
of 2.5%. In 2013, van de Krol et al. reported 4.7% in STH employing photoactive
BiVO4
[19] coupled with double-junction a-Si and enhanced by Co catalyst, prepared according
to the route developed by Nocera (so-called Co-pi catalysts). Other electrocatalytic metal
oxides such as
WO3
[20] and
Fe2O3
[21] are in the spotlight of ongoing research, too, due
to their promising photoactivity.
In this context, one of the promising candidates as photocathode for HER within a PEC
water splitting device is copper chalcopyrite.
Cu(In,Ga)Se2
has a band gap of 1.2 eV
with suitable band position with respect to
H2
evolution, i.e. the conduction band mini-
mum is located above the theoretical potential for proton reduction in acidic electrolytes
(about 4.6 eV). The material can be furthermore fabricated by a relatively simple and
low-cost preparation process and proves high chemical stability [22-26]. The advantage of
chalcopyrite-based materials, besides having the right potential for hydrogen evolution, is
their ability to absorb light in the entire visible range, resulting in high photocurrent den-
sities with values of up to 38
mA/cm2
for world record
Cu(In,Ga)Se2
-based solar cells
with a photovoltaic conversion eciency beyond 20% [27]. The benet and challenge of a
chalcopyrite-based photocathode is the transfer of the unique photovoltaic performance
to an ecient photocathode for competitive and cheap solar hydrogen production. Ac-
cording to very recent investigations, chalcopyrite thin lms can be successfully employed
as composite photocathode;in combination with organic thin lms of polymeric carbon
nitride as surface modication, they can be used for reduction of protons by light-induced
photoelectrons [28-29]. Up to now, only a few studies have been reported in the litera-
ture concerning the capability of inorganic composites based on chalcopyrite thin lms
as photocathode for PEC water splitting devices [30-32]. In alkaline electrolytes, pho-
tocurrent densities of up to
8mAcm−2
at the thermodynamic potential for
H2
-evolution
2
have been reported recently [30]. However, the conversion of incident photons to excess
minority charge carriers remained low. Very recently, it has been shown that the stability
of
ZnO/CdS/Cu(InxGa1−x)Se2
device-grade solar cells in water and under illumination
imposes a severe problem due to photodegradation of the chalcopyrite under reducing
conditions [32]. Hence, the development of a composite (or heterojunction) device based
on p-type
Cu(In,Ga)Se2
thin lms and an appropriate protection layer withstanding
the photocorrosive conditions represents a considerable challenge for this novel type of
composite. In the rst part of this work, this challenge is addressed by systematic devel-
opment of a
TiO2
protection layer with incorporated Pt as simultaneous
H2
-catalyst and
electronic dopant. In complemental eorts, extensive research has been carried out in or-
der to also develop corresponding photoanodes for optimization of the counter-reaction,
i.e. the oxygen evolution reaction. Due to the slower kinetics of the OER, this half-cell
reaction is recognized as the bottleneck for ecient overall splitting of water [33-38]. The
OER reaction requires, in general, higher overpotentials with respect to the thermody-
namic potential (+1.23V vs. RHE at pH0) in comparison to the corresponding hydrogen
evolution reaction. These overpotentials lead to partial loss of energy and reduce the
overall eciency of the water splitting reaction [39-41]. In order to enhance this reac-
tion, combination of semiconductors with co-catalysts is therefore inevitable. In 2015
Jaramillo et al. [42] performed a comprehensive study on overpotential and stability
of dierent electrocatalysts employed in HER and OER. Using electrocatalysts which
are not responsive to illumination (e.g.
ZnO :Co,NiO2,IrO2,RuO2
), the development
of ecient photoelectrode architectures is facilitated: these materials permit individual
testing of the catalytic activity in the dark, i.e. independent on light-absorption behavior
and photon-to-charge-carrier transformation. Subsequently, those electrocatalysts, that
combine most eectively electrochemical activity and benecial optical properties, can
be immobilized on photo-active supports in order to test the synergetic properties of the
thereby fabricated junctions. This modular approach is usually termed in the literature
as photo-assisted conversion process. Already many eorts have been done by Trasatti
and Lodi [43] to realize ecient evolution of oxygen in acidic electrolytes employing
IrO2
and
RuO2
. Although these catalysts are very ecient for the OER, the use of expen-
sive iridium and ruthenium makes large-scale application cost-ineective. Recent work
on metastable cobalt-oxidic materials by Shao Horn. et al. showed the great potential
in employing intrinsically metastable materials (e.g.
LiCoO2
) for the in-situ formation
of amorphous, cobalt-oxidic water oxidation catalysts (WOC) [44-46]. Pfrommer [47]
and coworkers have shown that employing Co-substituted ZnO (ZnO:Co) as a precat-
alyst leads to higher hole-conductivity in the in-situ formed WOC nanocomposite due
to the formation of core-shell structures with the polar ZnO:Co precatalyst integrated
into the lm [48]. It should be noted that pre-synthesized catalyst powders require a
suitable immobilization method in order to functionalize the catalysts on appropriate
semiconductors. In the case of ZnO:Co, electrophoretic deposition (EPD) from powder
suspension in organic solvent has been successfully employed to fabricate electrodes with
stable activity at pH7 and pH12. EPD represents in general a cost-eective technology
for realization of colloidal coatings in many elds of application [49-50]. Powder materials
are suspended in a chemically inert solvent and are transported under the inuence of
a (direct-current, DC) electric eld towards the supporting electrode [49-53]. The key
parameter in EPD is the velocity of the particles which determines the deposition rate
and is dependent on the physical properties of the solvent. In the case of ZnO:Co, the
inuence of varied organic solvent on the properties of the resulting electrodes has not
3
yet been suciently explored. Likewise, optimized preparation conditions for application
of ZnO:Co lms also in photo-assisted evolution of oxygen have not been identied. Par-
ticularly the latter research subject can help transferring ecient electrocatalysts to the
important eld of photoelectrolytic splitting of water.
For all semiconductor-based (photo-)electrodes, stability under operational (cathodic or
anodic) conditions is of utmost importance. In recent approaches, nano-scaled water ox-
idation catalysts have been deposited on metallic overlayers or have been employed as
compact oxide lms in order to avoid oxidation of the substrate upon photoelectrochem-
ical splitting of water [54-57]. Both schemes come with their individual disadvantages:
metallic protection layers have to be devised suciently thick in order to compensate
structural aws upon unwanted self-oxidation. Increased thickness, however, limits the
light reaching the semiconductor substrate. Oxide lms of suciently pin-hole free qual-
ity, on the other hand, may introduce non-negligible serial resistance to the electrode and
require mild and complex preparation conditions in order not to compromise the substrate
[58]. For both materials, metals and oxides, junction formation with the semiconductor is
critical since the achievable photovoltage is easily reduced by partial Fermi-level pinning
at the interfacial region [59]. A third material class, organic protection layers, nally,
is considered a promising alternative route for electrode protection in the future. Or-
ganic layers can oer nearly ideal electronic surface passivation but have been employed
so far only in
H2O
-containing electrolytes below the water oxidation potential due to
their increased susceptibility to self-oxidation [60-61]. In this work,
Cu(In,Ga)Se2
and
n-Si are chosen as absorber materials for application as photocathodes and photoanodes,
respectively. In the rst part, the development of transparent conductive oxide (TCO)
lms, composed of Pt-doped
TiO2
, is described as simultaneously electrocatalytic and
protective surface lm on
Cu(In,Ga)Se2
absorbers. High STH eciencies result from
optimized light-guidance into the chalcopyrite support. The superior electrocatalytic
performance is established by incorporation of nanoscaled Pt-clusters into the
TiO2
over-
layer. The layers were deposited by chemical vapor deposition of undoped and Pt-doped
TiO2
thin lms on
Cu(In,Ga)Se2
substrates. The synthesis is characterized by a molec-
ular gas phase deposition technique, ILGAR (Ion Layer Gas Reaction)[62-63], at elevated
temperatures of up to 400
◦
C. Thereby, phase pure anatase Pt-doped and undoped
TiO2
thin lms were obtained. The resulting
TiO2
/
Cu(In,Ga)Se2
photocathodes showed high
stability in acidic to neutral electrolytes.
In the second part of the project immobilization of presynthesized catalysts on FTO
and n-Si by EPD is reported for preparation of electro/(photo-)anodes, stable in alkaline
and acidic electrolytes. Nano-particulate ZnO:Co catalysts show high stability in alkaline
electrolytes while
RuO2
catalysts are well-known OER catalyst in acidic electrolytes. The
role of the organic solvent upon electrophoretic deposition of ZnO:Co and
RuO2
is inves-
tigated and structural, electrochemical as well as optical properties are correlated to the
applied preparation procedure. In the case of ZnO:Co, it will be shown that the organic
solvent considerably inuences the size of the active surface area of the deposited lms.
It is thereby possible to identify acetonitrile as organic solvent which provides higher
activity than the reported acetone solution [47]. Secondly, and most important, ethanol
will be identied as promising solvent for deposition onto photo-active semiconductor
supports: although the observed onset potentials of ethanol-prepared samples are lowest,
optical transparency is highest. Consequently, it can be shown that light-incoupling into
a triple-junction solar cell is improved which helps overcompensating the lower activity
of the electrocatalytic layer. The resulting PV-electrocatalyst-device on the laboratory
4
scale realizes 5.0% STH eciency, demonstrating thereby the potential of low-cost and
highly ecient solar-driven splitting of water. Strong light absorption by acetonitrile- and
acetone-prepared samples, on the other hand, limits the corresponding eciency to about
1%. In the case of
RuO2
, a novel synthesis scheme for preparation of polymeric protection
layers is reported. Suspensions of
RuO2
in alcohol/iodine are shown to allow for complex
reactions such as dehydration of the alcohol [64], formation of terminal aldehyde groups
and
sp3−sp2
transformations under electrochemical conditions. A single-step fabrication
route creates thereby an organic interface on silicon that integrates near-ideal electronic
and electrochemical passivation in an ultrathin polymeric layer of 3-4 nm thickness and
allows for extended operation of Si/Polymer/
RuO2
in acidic electrolytes for more than
one day.
5
2 Concept of photoelectrochemical water splitting
2.1 Principle of photoelectrochemical cells
PEC cells utilize light energy i.e. photons to perform electrochemical reactions (here,
splitting of water into
H2
and
O2
). The PEC cell usually consists of an anode and
a cathode, immersed into an electrolyte and connected to each other via an external
wire. A power supply is added to the circuitry in order to drive, if necessary, the reac-
tion in question with additional bias potential. In so-called monolithic congurations,
both electrodes are integrated into a single device. In half-cell investigations for photo-
electrochemical water splitting, typically either the anode or the cathode consists of a
photoactive semiconductor while the other electrode is a Pt counter-electrode. Therefore
semiconductors are considered to be the main component of a PEC cell. Fundamental
process steps in a PEC cell under operational conditions are summarized below:
I) Absorption of incident light by the semiconductor and generation of charge carriers
(electron-hole pairs). This process is based on the internal photo-eect requiring
photons with energies higher than the band gap of the semiconductor.
II) Separation and transportation of generated charge carriers. Electron -hole pairs are
separated due to the presence of an electric eld inside of the semiconductor (space
charge region, SRG) and transported under an external bias in opposite directions.
III) Extraction of generated charge carriers and production of
H2
at the cathode and
O2
at the anode. Electrons and holes accumulate at the respective solid-liquid
junctions of cathode and anode and induce corresponding electrochemical reactions
in the surface-near region.
All these processes as separation, transportation and extraction are due to the rec-
tifying nature of the semiconductor/electrolyte junction and, if required, an additional
external bias, all of which this described in detail in section 2.4. For an optimized, near-
ideal case, suitable semiconductors have to be identied which not only absorb the highest
portion of the solar spectrum (visible light) but also have appropriate positions of the
respective minima of the conduction and valence band,
CB
and
VB
, i.e. straddling the
thermodynamic potentials of
H2
and
O2
evolution, respectively. If no kinetic limitation
induces additional overpotentials, no external driving potential is required in this case.
In reality, the systems based on single light absorber components have low eciency in
optical absorption as, for instance,
TiO2
. Although
TiO2
oers suitable band position
for both
H2
and
O2
evolution the large band gap (
∼
3.2 eV) restricts light absorption to
UV, reducing thereby the usable amount of the solar spectrum to 4% only [65]. Figure
2.1 shows the relative positions of the conduction and valance band minima of several
semiconductors. According to this scheme, not all of them are suitable for simultaneous
HER and OER but only one of the half-reactions.
7
Figure 2.1: Position of conduction and valence band minima,
CB
and
VB
, of semiconductors with respect to the thermo-
dynamic potentials for
H2
and
O2
evolution, respectively [69].
For those materials which surpass with their minima both redox potentials, the band
gap is too large in order to absorb more than UV light. To overcome this low eciency,
the use of tandem PEC absorber was proposed. Such architecture contains two semi-
conductors, one suitable for
H2
evolution and one suitable for
O2
evolution. Recently,
for two-junction PEC photoelectrodes, prototypes based on III-V semiconductors such as
InGaP/GaAs [15], AlGaAs/Si [66] and InGaP/InGaAs [67] were reported, reaching STH
eciencies between 10 and 20% [68].
The overall water splitting reaction is divided in two half-cell reactions and depends on
the pH value of the electrolyte (alkaline or acidic)[3]. For an alkaline electrolyte (pH14),
the reduction and oxidation reactions with respect to the Normal Hydrogen Electrode,
NHE, are:
4H2O+4e−←→ 2H2+4OH−Eo
red = −0.828 V vs.NHE
(2.1)
4OH−+4h+←→ 2H2O+O2Eo
ox = −0.401 V vs.NHE
(2.2)
For an acidic environment, the oxidation and reduction reactions are:
4H++4e−←→ 2H2Eo
red = +0.000 V vs.NHE
(2.3)
2H2O+4h+←→ 4H++O2Eo
ox = −1.229 V vs.NHE
(2.4)
This dierence in anodic and cathodic reactions for acidic and alkaline electrolytes is
due to the dierent concentrations of
H+
and
OH−
ions in the respective media. For neu-
tral electrolytes both pathways can be expected to occur simultaneously [70]. Depending
on the conduction type of the semiconductor the motion of photogenerated electrons and
holes is dierent. For n-type semiconductors, the photogenerated electrons are directed
toward the back contact (proton reduction) and photogenerated holes are directed to-
ward the semiconductor/electrolyte interface (oxygen evolution). The opposite scheme
applies for p-type semiconductors. A PEC device may be composed of one semicon-
ductor as photoelectrode and a metal as counter-electrode or may be composed of two
semiconductors as photoanode and photocathode, respectively. Figure 2.2 shows the cor-
responding energy diagrams of dierent potential PEC congurations. The conguration
can consist of one photoelectrode for oxygen (a) or hydrogen (b) evolution and a metal
8
as counter electrode for the counter half-cell reaction. For this scheme, an optimal size of
the photoanode's band gap was estimated to be ca. 2.0 eV. The maximum energy con-
version eciency for a single absorber is approximately 23% as calculated by Shockley
and Queisser [71]. A further possible conguration consists of two photoelectrodes (c);
one photoelectrode is acting as photocathode for evolution of hydrogen and the other as
photoanode for evolution of oxygen [18,72-76]. In this case, the band gap of the respective
semiconductors can be smaller than the required overall minimum potential of 1.23 V.
Both semiconductors combine their light-absorption capabilities in a so-called Z-scheme
to exceed with their respective contributions the value of 1.23 V. In all these three cong-
urations, a complex and concerted activity of two electrodes, cathode and anode, under
illumination is required [77-78]. An alternative conguration of a water splitting PEC
cell is the back-to-back wireless design (d) in order to minimize the costs associated with
fabricating monolithic devices [79-80].
Figure 2.2: Dierent congurations of PEC cell.
Heterojunction devices prove increasing eciency in solar-to-
H2
conversion but ne-
cessitate consideration of a number of serious boundary conditions [80-83]. The use of
relatively small band gap materials (i.e., 1.1-1.7 eV) is desired in order to eciently ab-
sorb light in the range of the terrestrial solar spectrum. Simultaneously, the respective
positions of conduction band minimum and valence band maximum have to be adapted
to the standard potentials of
H2
and
O2
evolution to provide the necessary overall pho-
tovoltage (>1.23 V). Beside a suitable band gap, they should particularly prove stability
in aqueous electrolytes, small ohmic losses for charge carrier transport, and the water
oxidation/reduction kinetics should proceed faster than surface-mediated charge-carrier
recombination. Photocorrosion or photopassivation under cathodic conditions represents
a potential risk for some semiconductors [84], but almost all semiconductors are suscep-
tible to rapid oxidation under anodic conditions [85,64]. Therefore, surface protection of
the supporting semiconductor has to be realized such that long-term operation in aque-
ous electrolytes is guaranteed. In order to facilitate the oxidation and reaction kinetics,
and minimize the recombination of photogenerated charge carriers, suitable co-catalysts
are deposited on the semiconductor surface to provide a high density of active sites for
HER and OER [86]. Furthermore, the electrode design has to ensure that photons are
eciently converted to charge carriers and that charge transport is not limited by junc-
tion barriers or ohmic losses. In order to assess the eciency of a (stable) PEC system
quantitatively, the solar to hydrogen (STH) conversion eciency is used:
STH = ((|Jsc(mAcm−2|×1.23(V)×ηF)
P(mWcm−2))AM1.5
(2.5)
9
Here,
Jsc
denotes the short-circuit photocurrent density, measured at the thermodynamic
potential for
H2
evolution,
ηf
the Faradaic eciency, i.e. the percentage of charge ow
consumed exclusively for HER, and P the incident illumination power density. All these
parameters have to be measured under standard solar illumination conditions (AM1.5G)
[87]. If the photocurrent in Eq. 2.5 is not determined at the thermodynamic potential,
the Applied Bias Photon-to Current Eciency (ABPE) is used instead:
ABPE = ((|Jphoto(mAcm−2|×(1.23(V) − V))
Ptotal(mWcm−2))AM1.5
(2.6)
Here, V is the additional voltage that is applied to the cell and
Jphoto
denotes the
photocurrent measured at this voltage.
High eciency photoelectrochemical water splitting devices require the integration
of electrocatalysts (ECs) with light absorbing semiconductors (SCs) [64, 86-98]. Elec-
trocatalysts at the surface of the SCs are minimizing the kinetic losses associated with
oxygen- or hydrogen evolution [88]. Although the amount of the catalyst determines the
active sites for reaction, this amount should be well chosen in order to prevent unwanted
blockage of the incoming light in its passage toward the absorber surface. Figure 2.3
shows the activity and stability of several electrocatalysts for hydrogen evolution (red)
and oxygen evolution (green) for alkaline and acidic electrolytes [42].
Figure 2.3: Activity of Various catalysts for HER and OER in acidic and alkaline electrolytes.
As shown in the image, the lowest overpotential for HER catalysts in acidic electrolyte
(1M
H2SO4
) are Pt [99], NiMo [100] and NiMoCo [101]. For alkaline electrolytes, NiMo
proves the highest activity [102]. For OER, most of the non-noble metals show instability
under oxidative conditions in acidic electrolytes. Only Ir [103] and Ru [103-104] catalyst
are stable in acids. Under alkaline conditions, most catalysts operate with comparable
activity for OER. It has been postulated that oxide-based OER catalysts operate via a
common mechanism that includes the formation of a
OH∗
surface hydroxide intermediate
which is oxidized to a surface hydroperoxy
OOH∗
intermediate [105-108]. Although many
of these catalysts are comparable in activity, NiMoFe-(b) was observed to be the most
stable catalyst for OER in alkaline electrolytes [109].
The key point in the performance of these hetero-structure devices is the type of the
junction formed between SCs and ECs and also between SCs and the electrolyte. These
junctions are described in detail in section 2.4. First, dierent types of the semiconductor-
electrolyte junctions are explained and subsequently the eect of SCs-ECs junctions with
respect to the performance of the resulting photoelectrodes is described. In absence of an
electrocatalyst, the semiconductor forms a liquid junction with the aqueous electrolyte.
10
The dierence between the isolated SC Fermi level
Ef,n
and the solution redox potential,
EO2/OH−
, determines the band bending under equilibrium conditions and thus the upper
limit of the attainable photovoltage.
When an EC is added on the SC surface, the EC layer modies not only the rate of electron
transfer to and from the solution but also the equilibrium and non-equilibrium interface
energetics. Some ECs, such as Pt or dense crystalline
RuO2
and
IrO2
, are metallic or
metallic-like, respectively, and thus the SC/EC interface is expected to behave according
to the well-developed theory of SC/metal Schottky contacts [88].
2.2 Semiconductor photoelectrode material
Solid materials depend on their electron distribution and are classied as metals, semi-
conductors or insulators. This classication is shown schematically in Figure 2.4. The
valence band represents the highest energy range of occupied electron states and the con-
duction band represents correspondingly the lowest range of vacant electronic states. The
dierence in energy between the minima of
CB
and
VB
is called band gap (
Eg
) and deter-
mines the properties of the material. The band gap can be determined by measurement
of the absorption coecient vs. wavelength [110] according to:
α=(hν −Eg)m
hν
(2.7)
Here, m depends on the nature of the optical transition: m = 1/2 is applied for a
direct band gap, and m = 2 for an indirect gap. Furthermore, h denotes the planck
constant and
ν
is the frequency of light. By extrapolation of a plot of (
αhν)m−1
vs.
hν
to zero absorption, the band gap is determined [3].
As shown in Figure 2.4, there exists an overlap of
CB
and
VB
in the case of metals.
If both bands are separated by a suciently large value (>4eV), on the other hand, the
material behaves like an insulator. When the dierence is smaller, the material is called
a semiconductor and the electrons can be excited by visible light into the conduction
band. The vacant position of an excited electron in the valance band is called hole and
can move through space by transfer of an electron to the vacancy. Beside position of
the conduction and valence band of a semiconductor, the Fermi level is a further key
parameter. It is dened as the energy level at which the probability of occupation by an
electron is 1/2 and shows strict dependence on the doping concentration of the material
[110].
Figure 2.4: Comparison of
VB
and
CB
positions of a metal, a semiconductor and an insulator[3].
11
Depending on the nature of the impurities in a semiconductor which give rise to ad-
ditional electronic states, the semiconductors are classied in two groups: (I) in intrinsic
semiconductors, crystal defects generate additional states within the band gap close to
CB
and
VB
, respectively. At a rate, increasing with temperature, these states can be lled
either with electrons to generate excess holes in the valence band or by holes to generate
excess conduction electrons. Consequently, the Fermi level position is lowered or raised,
respectively, to indicate the nature of the conductivity (p-type or n-type, respectively). In
absence of these states, the Fermi level assumes a mid-gap position
EF=1/2(EV+EC)
[111]. (II) For extrinsic semiconductors, atoms of another material have to be added
to induce corresponding states within the band gap. The resulting conductivity of the
material is given by[3]:
σ=neµe+peµh
, where n and p denote the respective carrier
concentration (
cm−3
), while e is the charge of an electron = 1.602
×10−19
C and
µe(µh)
the electron (hole) mobility
(cm2/(Vs))
. Semiconductor doping can be realized by dier-
ent methods like epitaxy, diusion or ion implantation. Depending on whether the added
impurities are electron donors or acceptors, the semiconductor's Fermi level is shifted to-
ward the conduction band or valance band, respectively. Figure 2.5 schematically shows
the energy levels of n- and p-type semiconductors.
Figure 2.5: Schematic diagram of the energy levels of an n-type semiconductor (a) and a p-type semiconductor (b).
The concentration of electrons in the
CB
and holes in the
VB
, under equilibrium is
given by:
n=Nce−(EC−EF)/kT with Nc=2(2πm∗
eKT
h2)3/2
(2.8)
p=Nve−(EF−Ev)/kT with Nv=2(2πm∗
hKT
h2)3/2
(2.9)
If the defect state lies within
∼2kBT
near
CB
or
VB
, almost all states are ionized at
room temperature. Those dopants are called shallow dopants. If the electronic defect
state is located at a larger distance, then the dopants are called deep-level donors or
acceptors and may act as recombination centers for light-induced electron-hole pairs [111].
2.3 Description of electrolytes used for splitting of water
An electrolyte provides cations and anions in order to conduct electricity by the motion
of ions. The electrolyte may be subject to reversible processes as an iodide/triiode con-
taining electrolytes where charges are consumed to oxidize
I−
(anion) to
I−
3
(cation) and
vice versa, or to irreversible processes as in water electrolysis. Due to the low conduc-
tivity of neutral water (pH7), the equilibrium of
H+
and
OH−
concentration is shifted
to more acidic or alkaline conditions. This change is realized by additions of salts and
12
thereby ion dissociation. The addition of these salts, however, introduces further chem-
ical species to the electrolyte which then are in competition with the desired reactions
of hydrogen and oxygen evolution: an anion from the electrolyte is in competition with
hydroxide ions (
OH−
) to give up an electron. If the electrolyte anion has a lower standard
potential than hydroxide, it will be oxidized rather than the hydroxide and no oxygen
gas will be produced. The corresponding competitive conditions hold true for cations
with a higher standard potential than a hydrogen ion. Consequently, cations with lower
electrode potentials than
H+
like
Li+
,
K+
,
Ba+2
,
Na+
have to be used and anions with
higher electrode potentials than
OH−
like
SO−
4
. Common electrolytes are therefore sul-
furic acid, potassium hydroxide, sodium sulfate, and potassium phosphate. The redox
potential of water electrolysis shows pH dependence and can be calculated by:[112]
E=E0−pH ×0.059 −Eref
(2.10)
Here,
E0
is equal to
1.23V
at 25
◦C
and
Eref
is the potential of the used reference electrode.
2.4 Semiconductor (SC)/ electrolyte junctions:
Depending on the position of the Fermi energy of the semiconductor and the chemical
potential of the solution (redox potential), the direction of the band bending in the semi-
conductor varies. Figure 2.6 shows the dierent energy levels for p-type (a) and n-type
(b) semiconductors and the redox potential position of the electrolyte (c) on the common
vacuum scale. In this picture,
χ
and
φ
are the semiconductor electron anity and work
function, respectively. When a semiconductor is immersed in the redox electrolyte, de-
pending on the position of the Fermi level of the semiconductor and the redox potential
of the electrolyte, the resulting band banding in the semiconductor interface varies.
Figure 2.6: energy levels in (a) p-type semiconductor, (b) n-type semiconductor and (c) redox electrolyte [113].
Figure 2.7 shows the three scenarios for semiconductor-electrolyte junctions in the
dark under equilibrium for n-type (a, c) and p-type (d, f) SCs. The rst scenario holds
when in the n-type or p-type semiconductor, the Fermi level and
Eredox
of the electrolyte
is at the same energy (a/d); therefore, after contact formation, no net charge transfer
occurs and this situation is named as at band potential (
EFB
). The determination of
the at band potential and charge carrier concentration (
ND
), can be performed via
Mott-Schottky measurements as explained in section 4.2.1.3. The second scenario holds
when the Fermi level for an n-type semiconductor is higher than
Eredox
of the electrolyte
(b). After contact formation the electrons ow from the semiconductor to the electrolyte.
During this process, positive charges are left behind in the semiconductor. As a conse-
quence, a space charge region (depletion layer) is formed within the semiconductor which
results in an upward band bending. This argument applies to p-type semiconductor when
13
the Fermi level is lower than the redox potential. In this case, holes ow from the semi-
conductor to the electrolyte. Therefore the semiconductor is depleted of majority charge
carriers and a downward band bending in the semiconductor occurs.
Figure 2.7: Eect of the energy levels of semiconductor and electrolyte on the band edges in the interior of an n-type (a,c)
and a p-type (d,f) semiconductor upon contact formation [114].
The third scenario has to be considered when the Fermi level in the n-type semi-
conductor is lower than the redox potential of the electrolyte or for p-type SCs if the
Fermi level is higher than the redox potential of the electrolyte. In this case, electrons
and holes from the electrolyte ow to the n-type and p-type SCs, respectively. There-
fore, an excess of majority charge carriers accumulates in the space charge region. As a
consequence, a downward and upward band bending in the n-type and p-type SCs are
observed, respectively. The charge transfer behavior of the semiconductor (as mentioned
above) depends on the type of its junction. If there is accumulation of majority charge
carriers at the surface, the semiconductor behaves like a metal. But if the semiconductor
is in depletion, then only few charge carriers are available. Then, the reaction rate is
very slow, unless the electrode is exposed to radiation with photon energies higher than
Eg
. Under illumination photo-induced charge carriers are generated and electrons are
excited to the conduction band. For n-type SCs, exhibiting an upward band bending,
electrons move toward the back contact and holes move toward the solid-liquid interface.
There, electrons from the redox couple can be extracted and the semiconductor acts as
photoanode.
14
Figure 2.8: Ideal current-voltage behavior of an n-type semiconductor (a) and p-type semiconductor (b) under illumination
(solid blue curve) and in the dark (dashed green curve).
For p-type semiconductors with downward band bending, electrons move toward the
solid-liquid interface and react with reduction species in the solution whereas holes move
toward the back contact. In this situation, the p-type SC acts as photocathode. Figure 2.8
shows the ideal behavior of n-type (a) and p-type (b) semiconductors under illumination
(solid blue curve) and in the dark (dashed green curve). In these graphs, all the three
situations for the space charge region are depicted. In depletion, no current is owing
in the dark due to the band bending and the low amount of available charges. Under
illumination and after the at band region is passed the current increases and reaches
nally saturation values which reect the limit given by the illumination intensity. In
accumulation and in the dark, the behavior of the semiconductors is similar to metals.
In reality, the onset of currents is limited by low reaction kinetics and not as fast as
shown in Figure 2.8. Therefore, the increase in current density is considerably slower. To
improve sluggish charge transfer kinetics, co-catalysts are coupled to the semiconductor.
Figure 2.9 shows the performance of n-type (a) and p-type (b) semiconductors with co-
catalysts (red dashed curve) and without (solid blue curve). As shown in this gure, the
use of co-catalysts shifts the onset potential to the left for anodic conditions and to the
right for cathodic conditions. In other words, the respective overpotential is decreased.
Depending on the transparency of the catalysts and the amount of coverage of the surface,
the saturation current density varies. When catalysts absorb some portion of the incident
light, the saturation current density is reduced.
15
Figure 2.9: behavior of (a) n-type semiconductor, (b) p-type semiconductor under illumination (solid blue curve) and with
co-catalysts (dashed red curve).
Based on previous studies, the aim of this work is to develop and to optimize novel
photoelectrodes for HER and OER using earth abundant semiconductors. In chap-
ter 3, materials used in this project are introduced. In chapter 4, experimental tech-
niques are described in detail. Chapter 5 presents results and detailed discussions.
This chapter is divided in two main sub-sections. Section 5.1 shows the performance
of
Pt :TiO2/Cu(In,Ga)Se2
photocathodes. The results are shown to prove the twofold
functionality of Pt as catalyst and dopant in the
TiO2
layer. Section 5.2, focuses on the
immobilization of pre-synthesized catalysts on conductive substrates and demonstrates
that EPD is a promising deposition method for nanoparticulate catalysts at room tem-
perature. Two pre-synthesized catalysts, ZnO:Co for alkaline electrolytes and
RuO2
for
acidic electrolytes were investigated for preparation.
16
3 Introduction to photoelectrode architectures based
on Si and chalcopyrite supports
3.1 Introduction to the used material combinations for HER
3.1.1
Cu(In,Ga)Se2
chalcopyrite for HER
Solar cells based on CIGSe thin lms have yielded the highest conversion eciency among
all thin lm technologies [115]. Their recent record eciencies, realized by alkali metal
post-deposition treatment (PDT), reached 21.7% [116]. They are designed in a substrate
conguration and the deposition temperature is observed to have great impact on their
eciency. In contrast to silicon, CIGSe is a direct semiconductor material which means
that the thickness needed for sucient light absorption is in the range of micrometers
and does not require additional light trapping concepts like silicon.
The CIGSe material used in this work was prepared by Ch. Kaufmann et al. [117],
Mo layers were sputtered on glass substrates to form a non-blocking contact with the
deposited CIGSe lm. The CIGSe deposition process is characterized by a three-stage
Physical Vapor Deposition (PVD) process [118-121]. In the rst stage, two dierent pre-
cursors of In-Se and Ga-Se are evaporated and deposited at 330
◦
C, thereby forming a
homogeneous layer in the form of In-Ga-Se. The second and third stages are performed
using a sequence of dierent temperatures ranging from 330
◦
C to 525
◦
C. In the second
stage, Cu-Se precursors are deposited until the ratios of [Cu]/ ([In]+[G])-Se) reach 1.15.
During the third stage the precursor of (In,Ga)-Se is deposited until the layer becomes
Cu-poor and the ratio of [Cu]/([In]+[G])-Se) reaches around 0.84-0.88. The CIGSe layer
is p-type and has a band gap of 1.12 eV and a thickness of 3 nm. The CIGSe solar
cell is completed when a CdS buer layer (
Eg=2.4
eV) and a ZnO window layer are
deposited by chemical bath deposition and Radio Frequency sputtering (RF-sputtering),
respectively. ZnO has a band gap of 3.3 eV. It is therefore transparent for visible light
and due to its high doping level it provides high electron conductivity.
âFigure 3.1(a) shows the current-voltage behavior of a CIGSe solar cell in the dark
and under illumination. Its saturation current density and open circuit voltage reaches 37
mAcm−2
and +0.6V under illumination, respectively. Figure 3.1(b) shows the external
quantum eciency of the CIGSe solar cell of up to 90%. Figure 3.1(c) shows a SEM
cross-section and Figure 3.1(d) shows the band alignment of the CIGSe solar cell under
zero-bias voltage. During the growth of CIGSe on Mo, a thin interface layer of
MoSe2
is formed. This layer acts as a quasi-ohmic contact [115]. By deposition of n-type layers
of CdS and ZnO, band bending occurs in the junction according to n/p-heterojunctions.
Furthermore, two regions are formed inside of the CIGSe absorber: one is the space charge
region with depletion of majority charge carriers and the second one is a quasi-neutral
region (QNR). Despite the suitable band alignment of CIGSe solar cells with the ZnO
window layer with respect to potential usage as photocathode for hydrogen evolution,
both of the window layer (ZnO) and the buer layer (CdS) suer from serious photocor-
rosion in long-term photocatalytic reactions [122]. Therefore, a new stable TCO is to be
developed for replacement of the ZnO:CdS layer. As one of the most promising stable
oxides,
TiO2
was chosen and its functionalization on the CIGSe absorber for evolution of
hydrogen is described in the next section [123].
17
Figure 3.1: (a) Light and dark C-V curves for an ideal solar cell. (b) External quantum eciency of CIGSe-based solar
cell devices with layers with dierent Cu/(In+Ga) ratios. (c) SEM cross section of a CIGSe solar cell. (d) Schematic band
diagram of a CIGS solar cell under zero-bias voltage [115].
3.1.2 Pt-doped
TiO2
as electrocatalytic protection layer on CIGSe supports
TiO2
is a versatile material that is widely used for coatings in industrial applications
such as plastic, paper, inks, bers, food and pharmaceutical production and processing.
Furthermore,
TiO2
is abundantly available, one of the most stable oxides in aqueous
electrolytes, non-corrosive, environmentally friendly and cost-eective. In 1972 the pho-
tolysis of water on
TiO2
was discovered by Honda-Fujishima [124].
TiO2
has a large band
gap of 3.05 eV for the rutile and 3.2 eV for the anatase crystal structure. Its energy levels
make it suitable for water splitting i.e. the conduction band minimum of
TiO2
is more
negative than the reduction energy level of water (
EH+/H2=0
V), while the valance band
maximum is more positive than the oxidation energy level of water (
EO2/H2O= +1.23
V) [125]. Therefore it has been considered for decades as candidate for photocatalytic
water splitting. A schematic of the
TiO2
band alignment for both rutile and anatase is
shown in Figure 3.2.
Despite the many advantageous of
TiO2
, there are several issues to be addressed in order
to maximize the potential of
TiO2
. First of all, its large bang gap enables it to harvesting
just 4% of the solar spectrum in the range of UV light whereas the visible light is around
50% of the solar spectrum. A second issue is the fast recombination of
CB
electrons
and
VB
holes to release energy in the form of unproductive heat or photons [126]. A
third issue is the high probability of facilitating the backward reaction (recombination
of hydrogen and oxygen into water) after decomposition of water into its elements. Due
to poor absorption of
TiO2
and the rapid recombination of photogenerated
CB
electrons
and
VB
holes, its maximum theoretical STH eciency is limited to 2.2% [127].
18
Figure 3.2:
TiO2
band alignment [126].
In order to solve the problems mentioned above, several approaches have been pro-
posed:[128] elemental doping (using transition metals: Cu, Co, Ni, Cr, Mn, Au, Ag,
Pt)[129-131], hydrogen treatment of
TiO2
[132-133], non-metal doping of
TiO2
(N, S, C,
B) [134-137], composites of
TiO2
with semiconductors having lower band energies (CdS,
Cu(In,Ga)Se2
), sensitization of
TiO2
with dyes [138-139],
TiO2
doping with an up-
conversion luminescence agent [140-141] and adding electron donors or sacricial reagents
to react with photo-generated
VB
holes and thereby to enhance electron-hole separation.
In this work, two dierent methods were combined to achieve high ecient photocathodes
for hydrogen evolution.
TiO2
prepared by the ILGAR process (section 4.1.1) was doped
with platinum as co-catalyst for hydrogen evolution, in order to minimize the recombi-
nation rate, and deposited on
Cu(In,Ga)Se2
. The role of the co-catalyst is to lower the
electrochemical overpotential associated with the multi-electron water reduction reaction
and to provide an interface with enhanced electron-hole separation. Therefore, the
TiO2
layer acts as catalytically active TCO layer for the CIGSe absorber. The electrochemical
analysis is shown in section 5.1.
3.2 Introduction to the used materials combinations for OER
3.2.1 n-Silicon as an absorber for OER
Silicon-based electrodes are earth abundant and low in production costs. They absorb
a wide range of the solar spectrum and can directly convert solar energy into chemical
energy. Therefore, they have attracted great attention for preparation of articial pho-
tosynthetic systems. Silicon atoms are arranged at the corners of a tetrahedron with a
center silicon atom [110]; its lattice constant is 5.43 Å. Its electronic conguration is char-
acterized by lled K and L shells with extra four valance electrons in the
3s2, 3p2
state.
Silicon has an indirect band gap of 1.12 eV. The bulk properties of silicon depend on the
crystal growth technique and also on further addition of electronically active impurities
[142]. However, using silicon as photoanode is demanding. First because of the position of
its valance band maximum (5.17 eV) which is smaller than the oxygen evolution potential
at pH0 (5.83 eV). Therefore transferring holes from the valance band to water molecules
requires an external bias potential. Due to this problem, silicon has been used generally
as secondary layer in tandem cell congurations in connection with other semiconductors
with suitable band alignment. The second problem of silicon-based photoanodes is its
poor electrochemical stability under anodic conditions. As soon as the electrolyte reaches
19
the silicon surface, a passivating
SiO2
lm is formed. The formation of pinhole-free free
passivation or protection layers is therefore inevitable. The third problem is the lack of
its catalytic functionality. Since water oxidation requires a high activation potential, it
is necessary to increase the activity of silicon by deposition of catalytic layers.
So far many candidates like metals such as Ni, Pt, Ru and Ir [143-144] and metal oxides
such as NiO,
RuO2
, MnO,
TiO2
, FeNiO and NiCrO [145-146] were explored as protection
layers for silicon. These metal and metal oxides mostly have higher work functions than
silicon that can form Schottky junctions upon contact with silicon and introduce a built-in
electric eld at the interface which promotes separation and transport of photogenerated
carriers [145]. Apart from these metal and metal oxides, other materials also can be
used as protection layer for n-silicon photoelectrodes such as metal silicide [147-148],
Co-Pi [149-150] and conductive polymers [151-152]. In this work, n-type silicon is used
as the primary light absorption material and photo-carrier generation layer. ZnO:Co
and
RuO2
pre-synthesized catalysts were used as catalytic material at pH14 and pH0.3,
respectively. It will be shown that side-reactions, occurring concomitantly during EPD,
result in simultaneous formation of an interfacial carbon-rich lms suciently stable to
prevent photocorrosion of the silicon substrates.
3.2.2 ZnO:Co as catalyst for OER in alkaline electrolytes
Compunds of earth abundant Co (e.g.
Co3O4
) have received considerable attention in
both electrochemical and photochemical applications for catalytic water oxidation. An
amorphous cobalt phosphate based material (Co-Pi) was discovered by Nocera et al.
which catalyzes water oxidation at pH7 with low overpotentials. The Co-Pi catalysts is
formed during electro-deposition from solutions containing
Co+2
and 0.1 M potassium
phosphate buered at pH7 [34]. Its catalytic performance was associated with oxygen
vacancies at the surface and the presence of Co oxo/hydroxyl species, whose oxidation
numbers vary between 2, 3 and 4. Therefore, minimizing the size of the particles enhances
the surface activity of the catalyst. Recently, Nocera et al. showed that
LiCoO2
and
LiCoPO4
upon certain anodization conditions transformed into amorphous catalytically
active cobalt oxide materials similar to electrodeposition of amorphous cobalt-oxide (Co-
Pi) [33,46].
In this context Pfrommer et al. showed that Co-substituted ZnO structures (ZnO:Co)
is an interesting electrocatalyst for oxygen evolution. The substitution leads to higher
p-type conductivity and higher hole density. ZnO:Co nanoparticles were synthesized
by solvolysis of a heterobimetallic single-source precursor (SSP) mixture composed of
tetranuclear cubanodic clusters of the di-2-pyridylmethandiolate-ligand [153-154]. The
solvolytic reaction commenced in benzylamine at low temperature (180
◦
C) for 10 minutes
[17]. The green powder of ZnO:Co has an average crystal size of 17.2 nm, 42.2% cobalt
content and 47.3
m2g−1
surface area and a band gap of 2.4 eV.
3.2.3
RuO2
as catalyst for OER in acidic electrolytes
Ruthenium (IV) oxide has a rutile crystal structure. It is highly stable at low pH values
and has great capacity to store charges when used in aqueous solutions. It shows a very
low overpotential for oxygen evolution in acidic electrolytes. It can have two dierent
crystal surface orientations (110) and (100). According to the literature, the OER activity
of
RuO2
(100) is higher than the activity of (110) surfaces. This observation is accounted
for by the higher density of active metal sites on the (100) surface in comparison to the
20
(110) orientation.
The key parameter that indicates the behavior of
RuO2
is its chemical surface state.
Highly hydrated ruthenium dioxide (
H2O > 24%
) is not a good oxygen evolution catalyst
and under anodic conditions degradation and transformation to
RuO4
is observed. By
annealing highly hydrated ruthenium dioxide at 144
◦
C for 5 hours, a partially dehydrated
form of
RuO2
.
xH2O
can be formed. This state is highly catalytically active and stable
under anodic conditions. This catalytic activity, however, decreases upon increase of
the annealing temperature due to full dehydration of the particles. Thereby, the specic
surface area becomes smaller and the degree of crystallinity is increased. At 900
◦
C
annealing temperature an anhydrous black powder will be formed. This anhydrous form
of
RuO2
is well-known to be corrosion-resistant but due to its very low specic area
it has a low catalytic activity for
O2
generation [155]. Here, in this work, nominally
anhydrous
RuO2
is deposited on n-type Si to simultaneously protect it from degradation
under anodic conditions and to provide an ecient
O2
catalyst for water oxidation.
21
4 Experimental section
4.1 Sample preparation
4.1.1 Ion-Layer-Gas-Reaction (ILGAR)
The ion layer gas reaction (ILGAR) process [156] is patented by the Helmholtz-Zentrum
Berlin für Materialien und Energie. It is a cyclic process involving decomposition of an
aerosol on a heated substrate followed by sulfurization to form a highly pure and con-
formal sulde layer such as
In2S3,Sb2S3,Bi2S3
and
CuInS2
. The ILGAR method was
rst developed as dip-ILGAR [157] and later improved by an improved spray ILGAR
technique [63] which allows deposition of suldes on dierent substrates including planar
and structured surfaces. The deposition rate is limited by the adhesion of the liquid lm.
To better control the deposition rate and for application in in-line processes the spray
version of the ILGAR technique proved to be extremely valuable.
The spray ILGAR technique also can be modied for deposition of metal oxides in a
way similar to spray pyrolysis. By this technique a precursor solution is nebulized with
an ultrasonic nebulizer and directed towards a hot substrate by a nitrogen (
N2
) ow as
carrier gas. On the heated substrate the solution dries, decomposes and forms a metal
oxide mixture in intimate contact with the substrate. Because of the simplicity of this
method which does not require vacuum equipment it has been employed for deposition of
thin lms of metal oxides, spinel oxides, chalcogenides and superconducting compounds
[158]. In this project, this technique was used for deposition of crystalline
TiO2
layers
on
Cu(In,Ga)Se2
(CIGSe) substrates, using the setup shown in Figure 4.1. The pre-
cursor solution was 80 mL of titanium solution 0.01 M prepared by dissolving titanium
diisopropoxide (
Ti(OCH(CH3)2)4
) from ALDRICH in ethanol (99.99%) and mixing in
an ultrasonic bath for 10 min. The precursor solution was nebulized by the ultrasonic
generator into a ne mist; the mist was then transported sequentially through the glass
tube to the heated substrate (CIGSe) by the
N2
gas ow and decomposed and oxidized
there.
Dierent spray times and temperatures were tested in order to nd the optimum deposi-
tion parameters with respect to the resulting photoelectrochemical activity and stability
upon evolution of hydrogen. Those optimal conditions were 40 min of spraying at 400
◦
C.
The thickness of the layer deposited at these conditions is around 300 nm. According
to analysis by x-ray diraction (XRD), surface-photovoltage measurements (SPV) and
x-ray photoelectron spectroscopy (XPS), this layer is crystalline and contains high con-
centrations of carbon resulting from the organic solvent. In order to furthermore optimize
the performance of the photocathodes, the
TiO2
layers were doped with Pt as hydrogen
evolution catalyst. The 0.001 M Pt precursor was prepared by dissolving platinium-
hexachloride from Sigma-Aldrich in ethanol. Three doping levels (2, 5 and 10 vol.% of
the precursor solution) were tested to determine the optimum concentration level (results
will be shown in 5.1.3).
23
Figure 4.1: Schematic of the Ion-Layer-Gas (ILGAR) set-up for deposition of
TiO2
.
4.1.2 Electrophoretic deposition (EPD)
Electrophoretic deposition (EPD) is a deposition technique for nanoparticulate lms [49-
50,159], which is known since the work by Ruess et al. in 1808 by observing an electric eld
induced movement of clay particles in water. In this process powder material is dispersed
in an organic solvent. Upon electrochemical equilibrium conditions surface charges form
around the powder particles (lyoshphere formation) giving them a corresponding net
momentum in the presence of an electrostatic eld. There are several mechanisms causing
charge formation on the particles [160] like (a) selective adsorption of ions onto the solid
particle from the liquid (b) dissociation of ions from the solid phase into the liquid (c)
adsorption or orientation of dipolar molecules at the particle surface and (d) electron
transfer between the solid and the liquid phase due to dierences in their work functions.
This charge induction yields a net electrical charge on one side of the interfacial region
and a charge of opposite sign on the other side, which is known as electrical double layer.
The charged particles in the suspension, under the inuence of a DC electric eld, migrate
in the bulk of the suspension towards the conductive electrodes. This migration step
depends on the bulk properties of the suspension like conductivity, viscosity, particle
concentration and the surface charge density of the particles (so-called zeta potential of
the particles). Depending on which electrode the deposition occurs, two types of EPD
are dened. I) Cathodic electrophoretic deposition: when the particles are positively
charged and deposited on the cathode and II) anodic electrophoretic deposition: when
the deposition of negatively charged particles occurs on the positive electrode (anode).
The last step of the process is the coagulation and deposition of the particles on the
electrode. The schematic of the EPD process is illustrated in Figure 4.2.
24
Figure 4.2: Schematic of the EPD process.
The quality of the EPD process depends on the amount of the particles reaching the
electrode at a certain amount of time which is subject to the quality of the suspension.
The parameters aecting the quality of the suspension are divided in two main groups, the
ones related to the chemistry of the suspension such as particle size, conductivity, viscosity
of the suspension and dielectric constant of the liquid and those related to the physical
parameters such as the deposition time and the applied voltage. All these parameters
are correlated in the well-known and widely used theory of electrophoresis as developed
by Hamaker, [49-50, 159-160] Ishihara et al. [161] and Chen and Liu [162]. The Hamaker
law states a direct relation between the deposited mass and the electrophoretic mobility
of the solvent. The electrophoretic mobility (
µe
) of the particles in the suspension under
the inuence of a uniform electric eld, determines the EPD rate and the velocity of the
particles in suspension, as given by Eq. (4.1):
w(t) = Zt2
t1
µeEACsdt
(4.1)
Here w(kg) is the deposited yield,
µe
(
m2V−1s−1
) is the electrophoretic mobility,
E(V
m−1
) is the eld strength, A(
m2
) is the electrode surface area and
Cs
(kg/
m3
) is
the concentration in suspension. Ishihara et al. [161] and Chen and Liu [162] correlated
the electrophoretic mobility (
µe
) to the zeta potential of the particles in suspension. The
zeta potential indicates the amount of the charge on the surface of the particles suspended
in the solvent and plays a role in the stabilization of the suspension by determining the
intensity of repulsive interaction between the particles. The stability of the suspension
is governed by the sum of attractive and repulsive forces between the particles, mainly
of electrostatic and van der Waals nature. In order to prevent particle agglomeration,
a high electrostatic repulsion due to high particle charge is required determining (i)the
direction and migration velocity of the particles during EPD and (ii) the density of the
deposit. The Zeta potential [163] is calculated according to Eq. (4.2):
ζ=3µeη/(20r)
(4.2)
Here,
0
is the permittivity of free space (8.854
e−12
(C/Vm)),
r
and
η
are, respectively,
the permittivity and viscosity of the suspension medium,
ζ
(mV) is the zeta potential of
25
particles in suspension and
µe
(
m2V−1s−1
) is the electrophoretic mobility of the particles
in the solvent. The viscosity of the suspension is calculated by Einstein's formula [164]
for the eective viscosity of a dilute suspension of rigid spherical particles in an ambient
uid, given by Eq. (4.3):
ˆ
η≈η(1+2.5Φ)
(4.3)
Here,
η
and
Φ
are the viscosity of the ambient uid and volume fraction of spheres in
the suspension respectively.
In this project, the EPD technique was used for preparation of (photo-)anodes. For
alkaline electrolytes, ZnO:Co was chosen as catalyst for oxygen evolution. The pre-
synthesized ZnO:Co catalyst (6 mg) was dispersed in 15 mL of an organic solvent and
3 mg iodine was added to the mixture as an oxidizing agent and stirred for 5 min in an
ultrasonic bath. The electrophoretic deposition was carried out in a 10 mL beaker. Two
uorine-doped tin-oxide (FTO) samples (Solaronix, Suisse, sheet resistance (7
Ωcm−2
))
were pre-cleaned by acetone and used as working and counter electrodes, respectively.
The EPD set-up is schematically shown in Figure 4.3. After lling the beaker with
the suspension, the electrophoretic deposition was carried out by applying -10 V to the
support electrode using a BioLogic potentiostat, model VSP-300.
In order to investigate the eect of the organic solvents on the quality of the EPD
process, three dierent organic solvents, ethanol, acetone and acetonitrile were used (re-
sults are shown in section 5.2). The chemical properties of these solvents are listed in
Table 4.1.
Figure 4.3: Schematic illustration of the electrophoretic deposition (EPD) set-up.
Silicon samples (n-type Si(111) and Si(100), doping density
ND∼6×1015
, ABC
Company, Germany) were pre-cleaned by ethanol and water and chemically etched in
HF(50%) or
NH4F
(40%) depending on the respective surface orientation: (I) Si(111)
was etched in consecutive steps by
NH4F
(100 s and 10 min, respectively) and nally
dipped in HF (50%). (II) Si(100) was etched in a 3:1 solution of HF(50%) and ethanol
for 30 s and 10 s with intermediate rinsing with water and drying with
N2
.
Solvent Chemica formula Density Dielectric Viscosity Dipole
constant moment
Ethanol
CH3−CH2−OH
0.789 g/ml 24.55 1.2cP 1.7
Acetone
CH3−C(= O) − CH3
0.786 g/ml 21 0.32cP 2.7
Acetonitrile
CH3−C≡N
0.786 g/ml 37.5 0.37cP 3.44
Toluene
C6H5−CH3
0.867 g/ml 2.38 0.59cP 0.36
Table 4.1:
chemical and physical properties of the used solvents.
26
4.2 Analytical methods
4.2.1 Electrochemical analysis
Electrochemistry studies the chemical reactions taking place at the interface of an elec-
trode and an ionic conductor called electrolyte. The charge ow, comprising ionic con-
duction in the electrolyte and electron ow in the outer circuitry is then monitored for
analysis.
4.2.1.1 Current-Voltage (CV) behavior
Cyclic voltammetry (CV) is a technique for studying the electrochemical activity of the
catalysts. In this method, the voltage is repeatedly scanned between two potential val-
ues and applied between the sample under investigation (the working electrode) and the
counter-electrode. The absolute energetic position of the working electrode is measured
against a calibrated reference electrode. This potential (
E1
) is ramped linearly with time
toward the second predetermined limit (
E2=E1+∆E
). Simultaneously the current den-
sity is measured between the working electrode and the counter electrode and plotted
vs. the applied potential. The forward scan will end when the potential reaches the
nal value (
E2
). Then, the direction of the scanning steps is reversed and the backward
scan continues toward the initial value (
E1
) to complete the cycle. To investigate the
photoactivity of the electrode these measurements were performed under illumination.
Figure 4.4: Typical cyclic voltammogram where
ipc
and
ipa
show cathodic and anodic peak currents, respectively, for a
reversible reaction [165].
Typical CV measurements at the anodic region are shown in Figure 4.4. The recorded
CV curves reveal several features indicative for the reaction mechanisms involved. The
rst part shows the capacitive current or non-faradaic current. This current is owing
through an electrochemical cell and is observed upon charging or discharging of the elec-
trical double layer and does not involve in any chemical reaction. The second part is the
faradaic current indicating charge transfer of electrons across the electrode-electrolyte
interface. Depending on the direction of the voltage scanning, electrons are either gained
or released by the electrodes. By scanning the voltage of the electrode to more negative
values, the energy of the electrons in the electrode increases, providing thereby the elec-
trons with sucient energy to occupy vacant states on the species in the electrolyte. In
this case reduction takes place like for hydrogen evolution. On the other hand, by driving
the voltage more positive, the electrons in the electrode loose energy, thereby electrons
27
from species in the electrolyte are transferred to the electrode. In this case oxidation
takes place like in oxygen evolution. The faradaic process will be continued till all the
material at the surface of the electrode is oxidized; the resulting peak is called anodic
peak. After reaching the nal potential the scan direction is reversed and reduction of the
substrate continues until
Epc
is reached where the reduction is nished. If the reaction
at the surface is reversible, the integral ratio of cathodic current and anodic current is
equal to 1.
Measurements described here were done in a three electrode conguration. The ma-
terial to be tested was positioned as working electrode; two steel clips attached to the
surface of the sample were applied as front contacts. The surface area of the working
electrode, exposed to the electrolyte, was 0.5
cm2
. A platinum wire was used as counter
electrode and the reference electrode was a KCl-saturated Ag/AgCl electrode. The ref-
erence electrode circuit uses a large series resistance (
>1011Ω
) such that, practically
speaking, no current exchange takes place. Therefore, it enables determination of the
absolute value of the working electrode potential in a reliable way. All these three elec-
trodes were connected through a potentiostat (Biologic VSP-300). A schematic of the
used electrochemical cell is shown in Figure 4.5.
Figure 4.5: Schematic of the photoelectrochemical cell.
4.2.1.2 Incidence photon to charge carrier conversion eciency (IPCE)
Incidence photon to to charge carrier conversion eciency (IPCE) is a measure to charac-
terize the eciency of a photoelectrode to convert incident light into photoelectrochem-
ical reactions. It is equal to the external quantum eciency and is the ratio between
the amount of collected carriers and the number of incident photons on the sample. The
schematic of the setup is shown in Figure 4.6. The principle of IPCE measurements is
based on illumination of the sample by monochromatic light and recording of the result-
ing photocurrent density. The setup consists of a 300W Xenon lamp from LOT-oriel
GmbH as light source with a continuous spectrum from 200 nm to 700 nm. A Prince-
ton Instruments (Acton, MA) SP2150i monochromator with entrance and exit slits of 1
mm was used to select light with dened narrow wavelength width. After passing the
monochromator, the light beam was chopped by a shutter (LS6T2-NL, Vincent Asso-
ciates, Rochester, NY) resulting in a periodic sequence of illumination and darkness in
2s time intervals. Thereby, the net photocurrent could be determined by subtraction of
currents observed under illumination and in the dark.
28
Figure 4.6: IPCE set up [3].
After passage of the light through two long-pass lters, used to remove higher orders
of diracted light, it illuminated the electrode in the PEC cell. A platinum wire was
used as counter electrode. The electrodes in the PEC cell were connected to a PRS237A
potentiostat (Ameteck Inc., Berwyn, PA) in order to apply a bias potential and to measure
the current densities. The data were recorded by a LabView program, automatically
decrementing the wavelength from 900 nm to 300 nm at a scan rate of 30 nm/min. The
IPCE is calculated by Eq. 4.4 as below:
IPCE(λ) = hc
(qλ)
jphoto(λ)
P(λ)
(4.4)
Here, c and
λ
are the speed and the wavelength of light respectively,q is the elementary
charge,
jphoto
and P are the photocurrent density and the light power intensity at a given
wavelength.
IPCE is also useful for studying the degradation behavior of devices. If the measurements
show a decrease in IPCE during repeated cycles, deterioration of the photoconversion
properties of the active material have to be assumed. A changing shape of the IPCE
curve, on the other hand, may point to morphological alterations in the absorbing layer.
4.2.1.3 Mott-Schottky measurements
Mott-Schottky measurements were performed to study the semiconductor -electrolyte
interface and to determine the at band potential
EFB
and the charge carrier concentra-
tion
ND
of the photoelectrodes. The measurements were carried out in the dark using
a Zahner Messystem electrochemical workstation, IM6, combining a potentiostat and a
frequency response analyzer. A potential range from 1.5 V to -0.5 V was applied on the
sample with oscillation amplitude of 10 mV with frequencies of 50 kHz, 13 kHz, 8 kHz
and 3kHz, respectively. The space charge capacitance (
Csc
) was measured as a function
of the applied potential (E). Results were plotted as 1/(
C2
sc
) vs. E, allowing the extrap-
olation of
ND
from the slope and
EFB
from the intercept with the horizontal axis. This
approach follows from the Mott-Schottky equation [166] (Eq. 4.5):
1
C2
sc
=2
q0NDA2(E−EFB −kT/q)
(4.5)
29
Here,
0
is the permittivity of free space and
is the relative dielectric constant of the
material of the photoanode. A is the cross sectional area exposed to the electrolyte,
ND
is
the charge carrier concentration, E is the applied potential,
EFB
is the at band potential
and k is the Boltzmann constant.
4.2.1.4 Dierential electrochemical mass spectroscopy (DEMS)
Dierential electron mass spectroscopy (DEMS) [167-168] was used to detect the pro-
duced gases during cyclic voltammetry. The inlet system between the electrochemical
cell and the dierentially pumped vacuum system of the mass spectrometer (Balzers;
QMI 420, QME 125, QMA 125 with 901 o axis SEM) consists of a porous hydrophobic
membrane, covered with a 100 nm thick Au-layer. This layer serves as contact area to
the working electrode. Oxygen or hydrogen, which is formed in an electrochemical ex-
periment at the working electrode, diuses to some extent into the mass spectrometer
where it is monitored simultaneously with the electrochemical data. A schematic of the
cell is shown in Figure 4.7. All experiments were carried out in
N2
purged atmosphere
at scan rates of 2
mVs−1
. As reference, an Ag/AgCl electrode (+290 mV NHE) and as
a counter electrode a Pt-wire were used. For alkaline electrolytes KOH 1N pH 14 and
NaOH 0.1M pH13, and for acidic electrolyte
H2SO4
0.5M pH0.3 and for buer solution
kPi ph7 were used.
Figure 4.7: Schematic of the Dierential Electrochemical Mass Spectroscopy (DEMS) setup.
30
4.2.2 Spectroscopic analysis
4.2.2.1 Grazing incidence X-ray diraction (XRD)
X-ray diraction (XRD), carried out in (conventional) reection mode, is a method used
for phase identication of crystalline materials and atomic spacing. With this method,
X-rays penetrate the entire sample's volume. Therefore, the sensitivity of the method is
highest with respect to the distance of atomic planes parallel to the surface. The most
important feature is that the incident angle corresponds to half of the scattering angle.
A corresponding mode is the grazing incidence technique which is applied in order
to enhance the sensitivity for scattering contributions from the lm and to suppress
contributions from the substrate. In this mode the X-rays are impinging on the surface
at an angle
αi
and the scattering occurs only in the horizontal plane. The x-ray photons
are then collected at an angle
αf=αi
. At low incidence angles the absorption mainly
limits the penetration depth of the radiation and therefore the gained information can
mainly be attributed to the surface. An X-ray diractometer consists of three basic
elements: an X-ray tube, a sample holder, and an X-ray detector. A schematic of the
setup is shown in Figure 4.8.
Figure 4.8: (a) X-ray diraction in reection mode. (b) Grazing incidence X-ray diraction (GIXRD) mode.
X- rays are produced by bombarding a metal target (Cu, Mo) with a beam of electrons
emitted from a heated lament (often tungsten). The incident beam will ionize electrons
from the K-shell (1s) of the target atom and x-rays are emitted as the resulting vacancies
are lled by electrons dropping down from the L (2p) or M (3p) levels. This gives rise
to
Kα
and
Kb
lines. Filtering, by foils a crystal monochromator, is required to produce
monochromatic X-rays needed for diraction. These X-rays are collimated and directed
onto the sample. As the sample and detector are rotated, the intensity of the reected
X-rays is recorded. When the geometry of the incident X-rays, impinging on the sample,
satises the Bragg Equation constructive interference occurs and a peak in intensity is
observed. A detector records and processes this X-ray signal and converts the signal to
a count rate which is then transferred to a device such as a printer or computer monitor.
In this project, the crystallinity of
TiO2
, ZnO:Co and
RuO2
layers was measured by
means of grazing incidence X-ray diraction (XRD) in Bragg-Brentano geometry, using
a Bruker AXS (D8 Advance) diractometer. In order to achieve high resolution and
an excellent signal to noise ratio the X-ray beam was directed via a Göbel mirror. For
detecting the diracted radiation, a SOL-X energy dispersive detector was employed. The
measured diractograms were analyzed with the software EVA3. The radiation source
used was Cu-K alpha with a wavelength of 1.5406 Å. The size of the crystal (t) in a sample
is estimated from the width of the measured diraction peak using the Debye-Scherrer
31
correlation [169]:
t=0.9λ/(β.cosθ)
(4.6)
Here, t is the crystal size,
λ
is the wavelength of the incident radiation,
β
is the width
of the diraction peak which is determined at full width of half maximum of the signal
peak.
4.2.2.2 Surface photoelectron spectroscopy(PES)
Photoelectron spectroscopy (PES) is a non-destructive surface-sensitive technique for
analyzing the surface chemistry of a material by measuring the kinetic energy of photo-
electrons emitted from the sample due to the photoelectric eect. The photoelectric eect
describes the emission of electrons from the surface when the surface is hit by electro-
magnetic radiation. Depending on the source of the ionization energy, PES is divided in
to two categories (I) Ultraviolet Photoelectron Spectroscopy (UPS) which uses noble gas
discharge lamps like He as the source of radiation with excitation energies lower than 41
eV and (II) X-ray photoelectron Spectroscopy (XPS) which uses X-ray excitation energies
in the range of 1000-1500 eV. UP excitation is used to assess the valence band, whereas
XPS can characterize deep core-levels. The comparison of UPS and XPS is shown in
Figure 4.9.
Figure 4.9: comparison between UPS and XPS[170].
All measurements were performed under high vacuum conditions at
3.4×10−10
mbar.
For XPS measurements, a monochromatized Al X-ray source was used. The X-rays
excite photoelectrons in the sample mostly in the rst 10 nm of the surface layer which
restricts the so-called information depth to a very thin surface lm. If the binding energy
of the excited electrons is lower than the excitation energy (here less than 1486.74 eV)
the electrons will be emitted as photoelectrons and detected in a channeltron analyzing
detector. The entire excitation process can be conceived to occur in three steps:
I) The atom will absorb the photon and its energy is transferred to an electron.
II) The excited electron will travel to the surface of the sample. During this motion
the electron can lose some of its kinetic energy by scattering processes.
III) The excited electron will be released from the surface into the vacuum. An electron
analyzer measures the kinetic energy and the number of emitted photoelectrons.
32
In the used set-up, a PHOIBOS analyzer 100 served as detector with 5 channels and
diameter of 100 mm. An XPS spectrum was obtained as plot of number of detected
electrons versus the binding energy, calculated according to Eq. 4.7 [171].
EK=hν −EB−Φspec
(4.7)
Here,
EK
is the kinetic energy of the photoelectrons measured by the system,
EB
is
the binding energy of the emitted electrons,
hν
is the energy of the X-ray photons,
ν
is
the frequency of the incident light and h is Planck's constant (
h=6.626 ×10−34 Js
). The
aluminum X-ray source provided an excitation energy of 1486.7 eV.
Φspec
is the work
function of the XPS system, a constant value that depends on the material properties
of the spectrometer, which has to be determined prior to sample analysis (4.287 eV).
Typically, the energy-dissipative curve of an XPS survey can show various features in the
spectrum to be associated to dierent emission processes [172]:
I) Sharp peaks due to photoelectrons created within the rst few atomic layers (elas-
tically scattered).
II) Spin-orbit splitting (if incomplete lled shells contain unpaired electrons).
III) A broad background structure due to electrons from deeper regions of the solid
which are ineslastically scattered (reduced EK).
IV) Satellite structures, which arise when a core electron is removed by photoionization.
In this case, a sudden change of the eective charge occurs due to the loss of shielding
electrons. This kind of peaks has two forms:
•
Shake-up satellites: the emitted electron interacts with a valence electron and
excites it (shakes it up) to a higher energy level. As a consequence the energy
of the core electron is reduced and a satellite structure appears a few eV below
(EK scale) the core level position.
•
Shake-o satellites: the valence electron is ejected from the atom completely
(to the continuum). This eect is visible as broadening of the core level peak
or as contribution to the background signal.
V) Plasmons which are created by collective excitations of valence band electrons:
•
Extrinsic Plasmon: excited as the photoelectron propagates through the solid
after the photoelectric process.
•
Intrinsic Plasmon: screening response of the solid to the sudden creation of
the core hole in one of its atoms.
VI) Auger peaks produced by X-rays (transitions from L to K shell: O KLL or C KLL).
These peaks are well dened in terms of the binding energy of the electronic states of
the atoms. Depending on the number of the protons in the atom, these binding energies
can be specied for each atom. Figure 4.10 shows the relation between electron binding
energies and atomic number (Z).
33
Figure 4.10: Relation between electron binding energies and atomic number (Z)[173].
Assignment of the core-level lines to specic oxidation states was done by using pub-
lished data (J. Chastain and R. C. King Jr. (Eds.), Handbook of X-Ray Photoelectron
Spectroscopy, Physical Electronics, Inc., Minnesota, USA (1995) and the XPS data base
of the National Institute of Standards and Technology, NIST, USA). By comparison of
these binding energies with the references, the elemental composition, the chemical state
and the electronic state of the elements within a material can be characterized. The ma-
chine used in this project for surface chemical analysis was an XPS-ESCA system (Specs,
Germany).
4.2.3 Morphological-chemical analysis
4.2.3.1 Scanning Electron Microscopy coupled with Energy Dispersive X-ray
(SEM/EDX) Analysis
Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray (EDX) are non-
destructive methods for obtaining high resolution morphological and chemical analysis
of samples. In these measurements the surface of a sample is bombarded with electrons
emitted from a heated tungsten lament or from a eld emission cathode. The energy of
the incident electrons depend on the evaluation objectives and lies between 100 eV and
30 keV. The depth of the penetrating of the beam ranges from 10 to several 100 nm and
is related to the spatial resolution of the data. Figure 4.11 shows the interaction of the
incident electron beam with a sample. The interaction of this high energy beam with
the sample results in a variety of signals: (I) secondary electrons (II) backscattered elec-
trons (BSE), (III) visible light, (IV) X-ray photons and (V) Auger electrons. Secondary
electrons and backscattered electrons are commonly used for microscopic imaging of the
samples. The analysis of secondary electrons serves for analysis of the sample's surface
morphology while backscattered electrons allow for achieving high contrasts in the com-
position of multi-phase samples. The intensity of backscattered electrons is correlated to
the atomic number of the element within the sampling volume.
34
Figure 4.11: Interaction of the electron beam with a sample in SEM [174].
Depending on the lling form of the vacancies created by the secondary electrons, two
types of signals can be detected: Auger electrons and X-ray photons. In Auger emission,
the remaining hole belongs to inner shells e.g. to the K shell, and is lled by an electron
from an outer shell (L1). The released energy is transferred subsequently to another elec-
tron (here: L3) which is then ejected as Auger electron. X-ray photons are emitted when
the vacancies, created by secondary electrons, are lled with electrons from higher orbital
levels and the released energy is emitted as a photon. X-rays provide information about
the elemental composition of the sample and they are detected by an energy dispersive
detector that displays the signal as a spectrum of intensity versus X-ray energy.
Here, the morphology of the semiconductor was investigated by a scanning electron mi-
croscope (SEM), LEO Gemini 1530, from Zeiss, Germany, and a Thermo Fisher EDX
system. The EDX software, used for elemental analysis, was Noran System Six (NSS).
The thicknesses of the samples were determined by using a Veeco-Dektak 8 stylus pro-
lometer [175] which measures the vertical displacement of a diamond tip electromechan-
ically. The gantry moves the stylus with adjustable speed and force across the surface of
the sample. A stylus with a radius of 12.5
µ
m was used.
4.2.3.2 Transmission Electron Microscopy (TEM)
Transmission electron microscopy is a microscopy technique in which the sample is char-
acterized by a beam of electrons. Since in TEM electron beams are used as probe signal,
much higher resolution is achieved in comparison to light microscopes due to the much
shorter wavelength associated to electrons. The TEM set-up has three main parts: (I)
the electron gun to produce the electron beam and a condenser to focus the beam on the
object, (II) a series of lenses, a movable specimen stage, and (III) an image recording
system that consists of a uorescent screen for viewing and focusing the image and a
digital camera. To prepare a cross-sectional sample, the specimen is cut into two small
slices and glued on each other with epoxy in a face-to-face conguration. The area size
of the slices is about 0.5
mm2
each. The sample is subsequently thinned to about 4-6
microns and polished by ion milling. TEM measurements were carried out under vacuum
conditions. The electron beam hit the specimen and passed through it. Depending on
the state of the material (amorphous, crystalline phase, lattice constant etc.) some of
the electrons are scattered and are not detected. Behind the sample at the bottom of
35
the microscope, electrons, not aected by scattering, hit a uorescent screen. As result,
a gray image with varying contrast is obtained in dependence on the amount of scattered
electrons. In this project, the microscope C5, 12 from Phillips equipped with a LaB6
cathode (lanthanum hexaboride) was operated at an accelerating voltage of 120 kV. The
material under investigation present in thin layers was embedded in epoxy resin and cut
with an Ultramikroton by Reichert & Young using a diamond knife (diatoms) at an angle
of 45
◦
.
4.2.4 Optical analysis
4.2.4.1 Ultraviolet-visible spectroscopy
Ultraviolet-visible spectroscopy is an optical method which measures the transmittance
or reectance of light after interaction with a sample as a function of the wavelength.
It contains information about the optical band gap, the type of the band gap (direct or
indirect) and the absorption coecient of the material. The principle of UV-VIS spec-
troscopy is based on irradiation of a sample with light of the visible and ultraviolet spec-
trum. Molecules containing bonding electrons (
π
) or non-bonding electrons (n-electrons)
will absorb the energy of the incident light and these electrons will be excited to higher
anti-bonding molecular orbitals. As a result, the intensity of the impinging light attenu-
ates along the penetration path [110]. Optical measurements of thin lms were performed
using a Perkin Elmer Lambda 950 UV/VIS spectrometer equipped with an integrating
sphere, the used light spectrum was in the range of 200 nm to 800 nm, and transmit-
tance or reectance spectrums were measured. The absorption (A) was calculated by the
relationship:
A+T+R=1
(4.8)
Here, T refers to the transmission and R to the reection coecient, respectively.
In order to determine, how far light of a particular wavelength can penetrate into a
material before it is absorbed, the absorption coecient was calculated using the Beer-
Lambert Law [172]:
I(t) = I0exp(−αt)
(4.9)
Here,
α
is the absorption coecient and I(t) is the intensity of the transmitted light after
passing through the sample with thickness t.
I0
is the incident intensity. I(t) and
I0
are
equal to T and (1-R) respectively, therefore
T= (1−R)exp(−αt)
(4.10)
The absorption coecient depends on the material and also on the wavelength of the light.
A low absorption coecient indicates that light is only poorly absorbed. If the material
is thin enough, it will appear almost transparent to that wavelength. The intersection
of the absorption coecient with the wavelength-axis is equal to the band gap of the
material. Near the band gap of the material, the absorption coecient is given by Eq.
(4.11)[110]:
α∝(hν −Eg)m
(4.11)
Here, h
ν
is the photon energy and m is a constant that determines whether the
material has a direct band gap or indirect band gap. For direct band gap materials,
m is equal to 1/2 or 3/2. For indirect band gap materials, excitation of electrons is
accompanied by absorption or release of a phonon in order to conserve the momentum,
and m is equal to 2 or 3.
36
4.2.4.2 Fourier Transform Infrared Spectroscopy (FTIR)
Fourier Transform Infrared Spectroscopy (FTIR) is a nondestructive optical method for
material analysis. The method applies illumination of the sample with infrared (IR) light.
Some of the frequencies will be absorbed due to vibrational excitation of the bonds of the
atoms. These frequencies are detected as characteristic absorption features. The strength
of these peaks corresponds to the amount of the respective species in the material. Since
each dierent material is composed of unique atomic bonds, none of them produce the
same FTIR ngerprint. The FTIR spectra of solutions and deposited lms were measured
using a Bruker TENSOR 27 FTIR Spectrometer with OPUS software.
4.2.4.3 Surface photovoltage spectroscopy (SPV)
Surface Photo Voltage spectroscopy (SPV) [176] is a contactless and highly sensitive
method for measuring light induced charge separation in conductive layers. With this
technique photo-induced charge separation at the surface is detected in the presence of
an external voltage. Thereby, information about the absorption edge and distribution of
surface states is obtained and separation and recombination mechanisms can be charac-
terized [177-178].
An SPV signal arises whenever there is an asymmetry in the separated charge and there-
fore a change in the surface work function. This change can occur when photogenerated
charge carriers are separated in space or when non-uniform generation and recombination
of charge carriers is present. Depending on the amount of charge (Q) separated in space
and the distance (d) between the centers of the separated charges, the resulting SPV
signal is given by Eq. 4.12 [179]:
SPV =Qd
×0
(4.12)
Here, the constant
is the relative dielectric constant of the layer and
0=8.85×10−14
F/cm. The SPV signals is measured in two congurations: (a) in the Kelvin probe
mode and (b) in the xed capacitor mode (Figure 4.12). In the Kelvin probe mode,
the contact potential dierence (CPD) is measured between the sample and a vibrating
metal electrode in the dark and under steady light illumination. The metal electrode
and the sample form a capacitor whose capacitance changes due to the vibration of the
electrode. A voltage is applied to counter the alternating current generated by the varying
capacitance. Under illumination, the capacitance changes which leads to a change in the
applied voltage. The change in the applied voltage is equivalent to the change in the CPD.
The dierence (
∆CPD
) between the CPD in the dark and under illumination represents
the SPV signal.
In the xed capacitor mode, which was used in this project, measurements were done at
room temperature under low vacuum of
10−2
mbar. A thick (10-30
µ
m) mica sheet was
placed between the sample and the electrode (cylindrical
SnO2:F
). The sample, mica
sheet and an FTO electrode were pressed together by using a cardanic spring to form a
parallel plate capacitor. The capacitance of this arrangement was of the order of 100 pF.
Then the sample was illuminated by a halogen lamp. The incident light was chopped at
a frequency of 8 Hz. When the sample is illuminated, the change in the surface potential
results in a change in capacitance.
37
Figure 4.12: (a) Kelvin probe mode and (b) Fixed capacitor mode.
As SPV signal, the change in capacitance is measured across the measurement re-
sistance (10 G
Ω
) using a lock-in amplier (EG & G, 7260 DSP). The lock-in amplier
measured the signal in-phase (X) and by a 90
◦
phase shifted signal (Y) with respect to
the chopped light. The in-phase (X) signal contains information about the direction of
charge separation: a positive or negative X signal implies that photo-induced electrons
are separated towards the internal or outer surface, respectively. The 90
◦
phase shifted
signal (Y) characterizes the phase shift between the SPV signal and the chopped light,
the amplitude (R) and phase angle (
ϕ
). The SPV amplitude R and phase angle
ϕ
are
determined from the signals X and Y as shown below:
SPVamp =R=√X2+Y2
(4.13)
ϕ=tan−1(Y/X)
(4.14)
The X and Y signals can be positive or negative, the sign of the signals depends on
the direction of charge separation and how fast the charge separation is. A positive X
signal and negative Y signal indicate that charge separation of electrons proceeds towards
the internal surface, and the magnitudes of the signals depends on how fast this charge
separation occurs. Fast charge separation leads to a high positive X signal and a low
negative Y signal and slow charge separation leads to a low positive X signal and a
high negative Y signal. If the X signal is negative and the Y signal gets positive, this
case indicates charge separation towards the external surface. In this case, if the charge
separation is fast, a high negative X signal and low positive Y signal is measured but
upon slow charge separation, a low negative X signal and high positive Y signal will be
measured. The R signal shows the maximum amplitude and the onset of the SPV signal
which is a signature of the mobility band gap. It includes contributions from charge
separation from states below the band gap. It should be noted that this band gap is
related to the density of electronic defect states and has to be distinguished from the
optical band gap. By dividing the normalized R signal with respect to the photon ux,
a t of the band edge by an exponential function results in the energy of the tail states:
Rnormalized
φphotonflux
=Aexp(hν
Etail
)
(4.15)
The energy of tail states indicates the distribution of defect states within the band gap
of the material. The phase angle of the SPV signal gives information about the direction
of charge separation. Values of 0
◦
and 180
◦
for
ϕ
or close to them mean a fast spectral
response and values close to
±
90
◦
implies a slow time response [180].
38
5 Results and discussions
Solar-driven splitting of water into hydrogen and oxygen requires a complex and con-
certed activity of two electrodes, cathode and anode, under illumination [77-78]. For this
purpose, state of the art semiconductors are presently employed with increasing success
but necessitate consideration of a number of serious boundary conditions [80-82]. The
electrode design has to ensure that photons are eciently converted to charge carriers
and that charge transport is not limited by junction barriers or ohmic losses. Most im-
portant, surface protection (of the semiconductor) has to be realized such that long-term
operation in aqueous electrolytes is guaranteed. In this work, commercial chalcopyrite
and silicon absorbers are used as fundamental photoelectrodes. This chapter contains the
experimental results obtained in this study and is divided in two main sections. The rst
section includes the preparation and characterization methods of the novel photocathode
based on chalcopyrite as an absorber and
TiO2
as conductive protection layer for hydro-
gen evolution.
The second section introduces electrophoretic deposition as an eective technique to
deposit presynthesized catalysts on conductive substrates for preparation of (photo-
)electrodes. Two case studies based on this method were performed with pre-synthesized
ZnO:Co and
RuO2
in order to prepare electro/(photo-)anodes for alkaline and acidic
electrolytes, respectively. Silicon (n-type Si(100), doping density
ND∼6×1015
, ABC
Company, Germany) serves as an absorber for photoanode preparation. In the subsec-
tions, the electrodes were characterized by:
I) Optical measurements to determine the transmission, reection and absorption be-
havior of the thin lms (UV-VIS), and the molecular ngerprint of the sample
(FTIR).
II) Spectroscopic analysis to establish the phase (GIXRD), elemental composition and
chemical state of the elements, relative composition of the constituents in the surface
region and the valence band structure and oxidation states (XPS).
III) Morphological-chemical analysis to identify the homogeneity and elemental compo-
sition of the lms (SEM/EDX).
IV) PEC performance under at
100mWcm−2
illumination intensity. This characteriza-
tion also includes mass spectroscopy (DEMS), and incident photon to charge carrier
conversion eciency (IPCE).
5.1 Characterization of doped
TiO2
/
Cu(In,Ga)Se2
photocathodes
As discussed in chapter 2, for developing an ecient photocathode, the use of relatively
small band gap materials (i.e., 1.1-1.7 eV) is desired in order to eciently absorb light in
the range of the terrestrial solar spectrum.
Simultaneously, the respective positions of conduction band minimum and valence band
maximum have to be adapted to the standard potentials of
H2
evolution. One of the
suitable groups of materials in this respect are polycrystalline chalcopyrite lms. Chal-
copyrite thin lm absorbers such as
CuInS2,Cu(In,Ga)Se2
, and
CuGaSe2
, are very
attractive candidates for solar-driven evolution of hydrogen because they have suitable
39
band gap energies (
≈
1.0-1.7 eV) [24-26] and can, in principle, produce a signicant por-
tion of the required photovoltage. The advantage of chalcopyrite-based materials, besides
having the right potential for hydrogen evolution, is their ability to absorb light in the
entire visible range, resulting in high photocurrent densities, with values of up to 38
mAcm−2
for world record
Cu(In,Ga)Se2
based solar cells and photovoltaic conversion
eciency beyond 20% [27].
The stability of ZnO/CdS/
Cu(In,Ga)Se2
device-grade solar cells in water under illumi-
nation turns out to be a severe problem [32]. The ZnO layer (TCO layer of the solar
cell) is well-known to be unstable in aqueous environments [181]. Hence, the aim of this
project is the development of a composite (or heterojunction) device based on p-type
Cu(In,Ga)Se2
device-grade chalcopyrite and an appropriate n-type transparent conduc-
tive oxide (TCO), stable under photo-corrosive conditions.
One of the most promising TCO candidates is
TiO2
. The deposition method for prepara-
tion of
TiO2
lms is the spray ILGAR technique described in detail in section 4.1.1. The
rst part of this chapter reports the optimum deposition temperature of the
TiO2
. In
the second part, the performance of the chalcopyrite solar cell with and without
TiO2
is
compared. In the third part, Pt is introduced, rstly, as a catalyst to enhance the HER
rate and, secondly, as a dopant to optimize the conductivity of
TiO2
.
5.1.1
Cu(In,Ga)Se2
as an absorber in heterojunction photocathode
In order to identify the optimum multi junction photocathode based on a chalcopyrite
solar cell, three architectures were tested. In a rst conguration,
TiO2
was deposited
directly on the TCO layer (ZnO) of the solar cell. In a second conguration, the ZnO layer
was removed and
TiO2
was deposited on the buer layer (CdS). In a third conguration,
both ZnO and the buer layer were removed, and the
TiO2
lm was deposited directly on
the CIGSe absorber. The current-voltage behavior of these three dierent multi junctions
in
H2SO4
0.5 M is depicted in Figure 5.1. Dark current densities are throughout shown
in blue color while corresponding photocurrent densities are shown in red color. The
solid curves show the performance of the multi junction photocathode with
TiO2
top
layer deposited by ILGAR. Dashed curves show the performance of the same composite
without
TiO2
layer at the top. Figure 5.1(a) compares the current-voltage behavior of
the chalcopyrite solar cell in
H2SO4
0.5 M in the dark (blue dashed curve) and under
illumination (red dashed curve). In this gure, it is shown that the chalcopyrite solar cell
degraded immediately after getting in contact with the electrolyte. This results clearly
proves that ZnO is not stable in aqueous media. In order to protect the ZnO from
corrosion and benet from the solar cell photovoltage, a
TiO2
layer was deposited on the
TCO of the solar cell (by ILGAR process at 400
◦
C for 40 min). As shown in Figure
5.1(a), in this case the activity of the photocathode in the dark (blue solid line) and
under illumination (red solid line) is the same, i.e. a high dark current and no photo
activity is observed. This observation points to an instability of the ZnO lm during
the ILGAR process. Therefore, the entire junction was compromised already during the
ILGAR preparation process.
40
Figure 5.1: Current-voltage behavior (a)
TiO2
:ZnO/CdS/CIGSe (solid line) and ZnO/CdS /CIGSe (dashed line), (b)
TiO2
:CdS/CIGSe (solid line) and CdS/CIGSe and (c)
TiO2
:CIGSe (solid line), bare CIGSe (dashed line).
According to the second conguration, the ZnO layer of the solar cell was removed
in order to use the remaining
CdS/Cu(In,Ga)Se2
junction as support (and absorber).
Electrochemical results are shown in Figure 5.1(b). The activity of this heterojunction
in the dark (dashed blue curve) and under illumination (dashed red curve) was similar
to the solar cell, i.e. fast degradation was observed and the samples did not show any
photoactivity. After deposition of
TiO2
on CdS (buer layer) the performance of the
multi-junction photocathode improved. The dark current were limited to lower values of
3
mAcm−2
at 1.1 V (blue solid curve). Under illumination the photoelectrode reached
a saturation current density of 25
mAcm−2
at 1.1 V (red solid curve). Despite this con-
siderable improvement, this heterostructure is still characterized by a high overpotential
with respect to the thermodynamic potential of 500 mV. One of the reasons, causing this
increased overpotential, is considered to be interdiusion of CdS at high temperature with
the
Cu(In,Ga)Se2
absorber during ILGAR deposition of
TiO2
. Since the presence of the
CdS buer layer proved to be deleterious, it was removed and just
Cu(In,Ga)Se2
was
used as an absorber. The current-voltage behavior of
Cu(In,Ga)Se2
with and without
protection by
TiO2
(
TiO2
/
Cu(In,Ga)Se2
) is shown in Figure 5.1(c). It can be seen that
bare
Cu(In,Ga)Se2
still shows disadvantages as photocathode in acidic media (high dark
current, no photoactivity and high over potential) but after deposition of
TiO2
at the
top, the sample proves to be suciently protected from corrosion. The current-voltage
behavior shows no dark current, and the saturation photocurrent density reaches high
values of 25
mAcm−2
while the overpotential is considerably reduced down to 200 mV.
The reason why still a relatively high overpotential is observed is attributed to the low
conductivity of
TiO2
. In the next section, it will be shown that the conductivity of the
TiO2
layer can be successfully improved by Pt doping.
41
5.1.2 Eect of the
TiO2
deposition temperature on the device performance
The deposition temperature of
TiO2
is a crucial parameter for optimization of the per-
formance of the photocathode. To nd out the optimum deposition temperature (
Td
),
the temperature control of the sample holder at the ILGAR set-up was adjusted to a
sequence of dierent temperatures. Then the titanium precursor 0.01 M was sprayed
(nitrogen ow: 30 LPM) on an FTO substrate and heated to a predened temperature.
Grazing-incidence XRD (GI-XRD) results, obtained after deposition, are shown in Fig-
ure 5.2. The black, red, green and blue curves correspond to 250
◦
C, 350
◦
C, 400
◦
C and
450
◦
C deposition temperature, respectively. Because of dierent instrumental require-
ments, GI-XRD measurements were carried out for
TiO2
layers deposited on FTO, while
SPV and PEC measurements were performed for
TiO2
layers deposited on
Cu(In,Ga)Se2
substrates.
Figure 5.2 shows that, layers deposited at 250
◦
C and 350
◦
C are amorphous. All the
observed peaks correspond to the FTO substrates indicated by dashed lines. By increas-
ing the temperature of the sample holder to 400
◦
C (green curve), the amorphous
TiO2
layer becomes crystalline. The peak positions are in accordance with the anatase phase.
For deposition at 400
◦
C, the FTO peaks still are visible at the angels 25.59
◦
, 33.66
◦
,
51.76
◦
, 61.76
◦
and 65.73
◦
. By increasing the temperature to 450
◦
C, the crystallinity of
the layer increases and most of the FTO peaks are not visible anymore. This observa-
tion is attributed to the larger
TiO2
thickness caused by the deposition at an increased
temperature.
Figure 5.2: Grazing incidence X-ray diraction (GIXRD) at 0.3
◦
angle of incidence of ILGAR
TiO2
deposited on FTO at
dierent deposition temperatures: 250
◦
C (black curve), 350
◦
C (red curve), 400
◦
C (green curve) and 450
◦
C (blue curve).
42
Figure 5.3 shows the eect of the deposition temperature on the SPV-determined mo-
bility edge and on the formation of defect states near the band gap of the
TiO2
/
Cu(In,Ga)Se2
composite. The left image in Figure 5.3 (a) shows the normalized SPV amplitude ver-
sus photon energy for bare
Cu(In,Ga)Se2
(purple curve), and for
TiO2
/
Cu(In,Ga)Se2
deposited at 250
◦
C (black curve), 350
◦
C (red curve), 400
◦
C (green curve) and 450
◦
C
(blue curve). SPV results prove that bare CIGSe has two absorption edges, one with an
onset between 0.8-0.99 eV and another one at 1.6 eV. With heating of the
Cu(In,Ga)Se2
substrate to increasing temperatures, this second absorption edge becomes less and less
visible.
Figure 5.3: Surface photovoltage measurements: (a) Normalized SPV amplitude and (b) normalized SPV divided by photon
ux for
TiO2:CIGSe
samples prepared at dierent deposition temperatures: bare CIGSe (purple curve), 250
◦
C (black
curve), 350
◦
C (red curve), 400
◦
C (green curve) and 450
◦
C ( blue curve).
The strong variation of the onset energies for the lower band edge (between 0.8 and
about 1 eV) is attributed to the increasing incorporation of carbon and nitrogen into the
heterojunction and diusion of copper atoms. It is well known that disordering processes
in materials can lead to defect states in the band gap close to the band edge and result
in an exponential drop in the SPV signal towards lower photon energies [182]. The tail
state energy,
Et
, is a characteristic quantity of the degree of disorder in a semiconductor
and describes the defect states below the band gap which participate in charge separation
[180]. In order to obtain the energy parameter of the exponential tail states (
Et
), the
normalized SPV amplitude is divided by the photon ux and tted by an exponential
function in the region below the band gap. Figure 5.3(b) shows
Et
values for bare CIGSe
(330 meV) and for
TiO2
/
Cu(In,Ga)Se2
at 250
◦
C, 350
◦
C, 400
◦
C and 450
◦
C (340, 450,
750 and 460 meV, respectively).
PEC measurements were carried out for samples prepared at dierent
Td
to determine the
dependence of the performance of the composite photocathodes on
Td
(Figure 5.4). On
the left hand side (Figure 5.4(a)) the current-voltage behavior of
TiO2
/
Cu(In,Ga)Se2
for
Td
: 250
◦
C (black curve) and 350
◦
C (red curve) is shown and on the right hand side
43
(Figure 5.4(b)) the corresponding current-voltage behavior of
TiO2
/
Cu(In,Ga)Se2
for
Td
: 400
◦
C (green curve) and 350
◦
C (blue curve) is depicted. The dashed lines show
the activity of the composite photocathode in the dark and the solid lines show the
performance under illumination. For samples with an amorphous
TiO2
layer at the top
(prepared at 250
◦
C and 350
◦
C), high dark currents, low photoactivity and high onset
potentials were observed. Moreover, sample degradation proceeds fast.
Figure 5.4: Eect of deposition temperature on the current-voltage behavior of the
TiO2/Cu(In,Ga)Se2
photocathode:
TiO2
sprayed at (a) 250
◦
C (blue curve), 350
◦
C (red curve) and (b) 400
◦
C (green curve), 450
◦
C (orange curve). Electrolyte:
0.5 M
H2SO4
, pH0.3. Photoelectrodes were measured under near AM1.5 (100
mW/cm2
) conditions, with Pt as counter
electrode and Ag/AgCl as reference electrode.
For samples with crystalline
TiO2
layers, it is observed that the dark current density
is limited. For a deposition temperature of
Td
= 400
◦
C the resulting sample shows the
lowest overpotential of about 200 mV. At 600 mV a saturation photocurrent density of 24
mAcm−2
is reached. A further increase of the deposition temperature to 450
◦
C, however,
results in a decrease of the photoactivity of the junctions, indicating thereby an optimum
deposition temperature near 400
◦
C.
A potential physical property, to be related to the observed decrease in performance, is
the decreased conductivity of the
TiO2
layer when the deposition temperature exceeds
400
◦
C. This conclusion will be discussed in connection with so-called four-probe resistivity
measurements to be discussed below (see Figure 5.7).
5.1.3 The Pt-doped
TiO2
layer: variation of the doping density
Although
TiO2
is a stable oxide in aqueous electrolytes it has the drawback of poor
conductivity. As mentioned in section 3.1.2, one of the approaches to improve the con-
ductivity in the
TiO2
layer is elemental doping, using transition metals [183]. Here, Pt
was chosen as dopant metal due to its high activity for HER and also due to its stability
in aqueous solutions. In order to identify the optimum value for the doping density, var-
ious concentrations of the platinum molecular precursor in the ILGAR process, from 2
44
vol.% up to 10 vol.% were tested. The resulting photocurrent densities for the Pt-doped
TiO2:Cu(In,Ga)Se2
samples as a function of the applied potential is shown in Fig-
ure 5.5. Solid curves indicate the results obtained under illumination of 100
mWcm−2
(AM1.5), dashed curves show the activity of the samples in the dark. The photocathode
was illuminated from the front side through the electrolyte (
H2SO4
0.5 M pH0.3).
The solid black curve shows the performance of the undoped
TiO2
on CIGSe with an on-
set potential for hydrogen evolution of -200 mV (corresponding to an overpotential of 200
mV). Results for doped samples prove that, in general, by addition of the Pt-precursor to
the solution, the onset potential shifts from negative values towards more positive values,
i.e. the overpotential is reduced. This shift of the onset potential can be attributed to
both Pt-doping, which increases the conductivity, and Pt nanoparticles (NPs) inclusions,
which enhances the electrocatalytic activity of the
TiO2
layer.
Figure 5.5: (a) Current-voltage behavior of undoped
TiO2:Cu(In,Ga)Se2
(black curve) and Pt-implemented
TiO2:
Cu(In,Ga)Se2
devices. The best current-voltage behavior shows an anodically shifted onset potential at +230 mV (green
curve); Electrolyte: 0.5 M
H2SO4
, pH0.3. Photoelectrodes were measured under near AM1.5 (100 mW/
cm2
) conditions.
The red, green and blue solid curves are showing the performance of samples with
2, 5 and 10% Pt doping density, respectively. The vertical dashed line serves as a mark
for
ERHE ≈
0 V and emphasizes the potential gain in hydrogen evolution achieved by
the
Pt :TiO2/Cu(In,Ga)Se2
composite photoelectrodes in comparison to the undoped
device. By increasing the Pt density from 2 to 5 vol.%, the performance of the photo-
cathode is improved and improves from 30
mAcm−2
saturation photocurrent density to
37
mAcm−2
with an onset potential of +230 mV. Further increase of the doping density
to 10 vol.%, however, decreases the performance dramatically: the saturation photocur-
rent density decreases to 25
mAcm−2
and the onset potential shifts to +100 mV. An
optimum photoelectrocatalytic device performance is therefore achieved for 5 vol.% Pt:
a saturation photocurrent density of 37
mAcm−2
corresponds, in fact, to values obtained
45
to reported best chalcopyrite solar cells [184]. Furthermore, 15
mAcm−2
photocurrent
density obtained at the redox the potential already exceed industrial minimum require-
ments of 10
mAcm−2
by 50%. For potentials more negative than about -0.4 V RHE ,
the device performance of the photocathode is only limited by the excess minority charge
carriers (electrons) generated by the external white light source (
≈
38
mAcm−2
; simulat-
ing AM1.5 conditions). Several measurements including UV-VIS, SPV, SEM/TEM and
four probe measurements were performed to clarify the role of the Pt for the eciency of
the photoelectrode. The results are reported in the following sections.
5.1.3.1 The twofold functionality of Pt as dopant and catalyst
In this section, by comparative analysis of TEM, XPS (The SEM, TEM and XPS mea-
surements were done by Dr. M. Lublow, Dr. A. Fischer from Uni-Freiburg and A. Steigert
from HZB, respectively)and XRD results, the twofold functionality of Pt as dopant and
catalyst is demonstrated [183]. The Pt doping concentration in the
TiO2
thin lms was
modied by applying various concentrations of the platinum molecular precursor in the
ILGAR process, from 2 vol.% up to 10 vol.%. Corresponding TEM images for 2%, 5%
and 10% Pt doping are shown in Figure 5.6 (a), (b) and (c), respectively. For low plat-
inum precursor concentrations of 2 vol.%, no Pt-cluster formation is detected by TEM,
i.e. Pt-atoms are most likely incorporated into the
TiO2
host-lattice either on
Ti+4
cation
sites or at interstitial sites. For 5 vol.%, aggregation of Pt nanoparticles (NPs) is observed
in the
TiO2
layer (Figure 5.6 (b)). For higher concentrations of 10 vol.%, the Pt NPs
increase further in size and also aggregate directly at the
TiO2
-
Cu(In,Ga)Se2
interface.
Figure 5.6 also shows that the morphology of the
TiO2
layer changes by increasing the
Pt concentration from a ake-like morphology (2% doping) to big crystals (10% doping).
The conductivity of the
TiO2
layer doped by Pt, was determined by four probe mea-
surements. The results are shown in Figure 5.7. The blue line shows the resistivity of
the Pt doped
TiO2
layer after deposition on glass. The green line shows the resistivity
of the layer after deposition on
Cu(In,Ga)Se2
. The red line shows the thickness of the
corresponding layers (Pt:
TiO2
/glass) measured by a Dektak step prolometer and shows
the reducing trend for thickness upon increase of the doping level. The thickest layer
shows a thickness of 350 nm for undoped
TiO2
. The thickness of Pt-doped
TiO2
layers,
prepared from solutions containing 2%, 5% and 10%, is 180 nm, 115 nm and 50 nm,
respectively. These results are in good agreement with UV-VIS measurements (section
5.1.3.2). According to the conductivity results in Figure 5.7, the conductivity increases
with increasing Pt-concentration from 2% to 5% due to incorporation of Pt-atoms in the
layer and decreases upon further increase of the concentration to 10% due to formation
of Pt-clusters (TEM image 5.6(c)).
46
Figure 5.6: Surface morphology (SEM, left) and Pt-distribution (TEM, right) of
TiO2
thin lms on CIGSe supports
upon increasing Pt concentrations. Characterization by SEM clearly shows an increase of the anatase crystallite size for
incremented Pt concentrations. For 5 and 10 vol.%, agglomeration of Pt to nano-clusters is visible in the corresponding
TEM images [183].
47
Figure 5.7: Resistivity and thickness of undoped and Pt-doped
TiO2
.
GI-XRD and XPS spectroscopy results are discussed in the following to identify the
oxidation state of Pt inside the
TiO2
layer. The grown
TiO2
thin lms on the CIGSe are
characterized by measurements carried out at a grazing X-ray angle of incidence of 0.3
◦
.
Measurements were done for three samples as shown in Figure 5.8: the black curve shows
the XRD result obtained for undoped
TiO2
on CIGSe, the red and green curves show
the corresponding results for 2 and 5 vol.% Pt doping. The results suggest that
TiO2
deposited at 400
◦
C is characterized by an anatase phase (Figure 5.8). The CIGSe pattern
is not visible for the undoped
TiO2
layer since the layer thickness is larger in comparison
to doped
TiO2
, as proven by Dektak step prolometry. The increasing Pt(200) signal at
46.5
◦
for the doped samples indicates the incremented incorporation of metallic Pt in the
TiO2
layer.
Figure 5.8: (a) Grazing incidence X-ray diraction (GIXRD) at 0.3
◦
angle of incidence of undoped ILGAR
TiO2
paste
(black curve), 2 vol.% Pt doped
TiO2
(red curve) and 5 vol.% Pt doped
TiO2
(green curve). All lms were prepared on
CIGSe substrates.
48
X-ray photoelecron spectroscopy (XPS) of the Pt 4f core-level signal was carried
out for a Pt:
TiO2
lm prepared from solutions with 5 vol.% Pt-precursor concentration
(Figure 5.9). According to XPS analysis, the actual Pt concentration is only about 0.1
atom%. Due to the high surface sensitivity of the method, this result is valid for an about
10 nm thin surface region. The value of 0.1 atom%, on the other hand, is close to the
detection limit of this technique. The Pt
4f7/2
and Pt
4f5/2
signals show a pronounced
asymmetric broadening most likely due to the contributions of Pt in an oxidation state
of +2 and +4, denoted in the gures as Pt(II) and Pt(IV), respectively. The signal
Pt(0) at a binding energy near 71 eV is caused by metallic platinum, i.e., Pt-NPs.
According to recent results reported in the literature,
Pt4+
and
Pt2+
ions can occupy
either
Ti4+
sites or interstitial sites in rutile or in the anatase lattice [185]. As a result
of similar ionic radii of
Pt2+
(
Rion
= 0.80 Å) and
Pt4+
(
Rion
= 0.63 Å) ions in the
sixfold coordination geometry of anatase,
TiO2
(
Ti4+
(
Rion
= 0.61 Å) [25] either
Pt2+
or
Pt4+
ions (with electron congurations of
Pt2+
[Xe]
4f14 5d8
and
Pt4+
[Xe]
4f14 5d6
)
can partially replace
Ti4+
ions leading to an extra n-type doping of the
TiO2
anatase in
the bulk. The relative contributions of Pt(0), Pt(II), and Pt(IV) in Figure 5.9 indicate
predominantly a Pt-dopant oxidation state of +2 rather than +4. Three contributions
to the envelope curve were identied and labeled as Pt(0), i.e. metallic Pt, Pt(II) and
Pt(IV). The oxidation states (II) and (IV) are attributed to PtO,
Pt(OH)2
and
PtO2
,
respectively. The low signal-to-noise ratio impedes accurate analysis with respect to both
the background signal and the identied components.
Figure 5.9: X-ray photoelectron spectroscopy of the Pt4 core level for the 5 vol. % Pt-
TiO2
lm [183].
Figure 5.10: X-ray photoelectron spectroscopy of the Ti 2p (a), O 1s (b) and N 1s (c) signals measured for 5% Pt-doped
TiO2
on CIGSe [183].
49
Figure (5.10(a)) shows the XPS analysis of the Ti 2p signal for the sample with 5% Pt
doping. For analysis of the data, only one doublet peak was applied. The small variation
of the residuum between 459 eV and 462 eV suggests the presence of a further component.
However, no convergence was achieved assuming a second Ti 2p related doublet structure.
XPS analysis of the O 1s signal measured for 5% Pt-doped
TiO2
on CIGSe is shown in
Figure (5.10(b)). Two sub-structures were identied to be attributed to oxygen in
TiO2
(near 530.7 eV) and hydroxyl groups (near 532.5 eV). XPS analysis of the N 1s signal
measured for 5% Pt-doped
TiO2
on CIGSe is shown in Figure (5.10(c)). The core-level
signal
I1
is assigned to substitutional nitrogen at an oxygen site, the signal
I2
may either
be related to interstitial N or
NHx
.
Figure 5.11: XPS analysis of the C 1s signal measured for 5% Pt-doped
TiO2
on CIGSe. (a) before electrochemistry, (b)
after electrochemistry, (c) comparison of the C 1s signals in (a) and (b)[183].
Figure 5.11 shows the respective components for
sp3
C-C bonds, C-OH and C-OOH.
Due to the broadening of the
sp3
peaks, it is not possible to exclude
sp2
carbon. These
components are detected before as well as after photoelectrochemical evolution of hydro-
gen. The much lower intensity of the signal after operation, however, suggests that carbon
is mainly conned to the surface, i.e. carbon species are adsorbed from the environment
rather than incorporated in the oxide lm.
5.1.3.2 Correlation of optical behavior with doping density
The optical properties of the doped
TiO2
layers, deposited on glass, were investigated
by UV-VIS measurements. Figure 5.12 shows the transmission (a) and reectance spec-
trum (b) and the absorption coecient (c) which was calculated according to the Beer-
Lambert Law [110]. In these gures, the black curve shows the spectrum for undoped
TiO2
and the red, green and blue curves show the spectra for the samples with 2%, 5%
and 10 vol.% doping density. The transmission and reectance spectra reveal that the
50
optical properties of the
TiO2
lm are sensitively inuenced by gas phase proportion
of the Pt precursor and demonstrate the nonlinear change in light absorption upon in-
crementing Pt-concentrations. The measurements show that the transparency of doped
samples is higher than of undoped
TiO2
(black curve) and is increased by increment-
ing Pt-concentrations from 2 to 5 vol.%. The transparency decreases, however, for the
photocathode device with the highest Pt concentration in the
TiO2
layer (10 vol.% Pt
precursor).
Comparison of the reectance behavior demonstrates that the reection of the doped
layers is higher than for undoped
TiO2
[183]. The reection decreases with incrementing
Pt-concentrations. The sample with 2 vol.% Pt doping has the highest reection. The
lowest reection is observed for the 10 vol.% Pt doping concentration. This change in the
optical behavior is also visible to the naked eye: a color change is observed upon increase
of the doping level which is attributed to decreasing lm thickness and formation of
Pt-clusters. Samples prepared from solutions with 2 vol.% Pt-precursor concentrations
appear yellowish while samples with 10 vol.% appear more brownish.
Figure 5.12: UV-VIS transmission (a) and reection (b) of undoped and Pt-doped
TiO2
/CIGSe(c) Determination of the
optical band gap of undoped and Pt-doped
TiO2
/CIGSe.
The increased density and average size of the Pt-NPs limits the magnitude of light
that reaches the chalcopyrite absorber (Figure5.5). Furthermore, increased electron-hole
recombination at the
TiO2
/Pt-NP interface can lower the achievable saturation pho-
tocurrent density. Consequently, saturation photocurrent densities are limited to about
25
mAcm−2
(blue curve; Figure 5.5). The photocathode device with the lowest Pt con-
centration in the
TiO2
layer (red curve; Figure 5.5), in turn, proves higher saturation
photocurrent densities. But they remain still below those of the most ecient photocath-
ode device with
TiO2
lms grown in 5 vol.% Pt-precursor gas phase concentration.
The absorption coecient,
α
, of the samples was determined according to 1=
α
+ R + T
where R and T represent reectance and transmission of the samples, respectively. The
calculated spectra,
(α−hν)2
, were subsequently extrapolated by straight lines (Figure
5.12(c)). Here,
hν
represents the photon energy. According to the results indicated by
the inset, an optical band gap of about 3.6 eV was determined for all samples. It is
noteworthy that this band gap is typically detected for
TiO2
prepared by spray pyrolysis.
51
5.1.3.3 Dependence of the energy of the tail states on the doping density
SPV measurements were used for investigation of the charge separation mechanism in
the
TiO2
layers. In this chapter, results for the mobility edge of dierent
TiO2
layers
and charge separation mechanisms across the
TiO2
/FTO junction are presented. Charge
separation across the interface was measured by modulated SPV (explained in detail in
4.2.4.3) in the xed parallel plate capacitor conguration and as a function of the depo-
sition method. Three dierent samples are compared in Figure 5.13; commercial
TiO2
paste (blue curve) prepared by screen printing, undoped
TiO2
prepared by ILGAR (green
curve) and Pt doped
TiO2
also prepared by the ILGAR process (red curve) after deposi-
tion on uorinated tin oxide (FTO). Figure 5.13(a) shows the normalized SPV amplitude
of these three samples. According to the results, the onset of the SPV absorption (
Eabs
)
(mobility edge) for the commercial
TiO2
is 3.2 eV, i.e. it is close to the optical band
gap of
TiO2
. This value is shifted to 1.05 eV in the case of
TiO2
samples prepared by
the ILGAR process. This position of the absorption edge is due to embedding of carbon
and nitrogen atoms inside the
TiO2
structure during the deposition process. According
to Figure 5.13(a), Pt-doping, in fact, gives rise to an additional defect band centered at
3.2 eV and pointing thereby to an increased density of donor-like states near the conduc-
tion band. These defect bands contribute to an increase in absorption and higher charge
separation [183].
Figure 5.13(b-d) shows the X and Y signals for commercial
TiO2
paste (b), undoped
TiO2
(c) and Pt:
TiO2
(d) deposited on FTO. For all samples X signals were positive
and Y signals were negative. This means that photo-generated electrons were separated
preferentially towards the internal interface. The amplitude of both X and Y signals for
commercial
TiO2
paste is higher than for ILGAR-grown
TiO2
. Comparison of Figure
5.13(c) and (d) shows that Pt-doping resulted in a decrease of the amplitude of the X
and Y signals. For all samples a common peak arises at 3.2 eV corresponding to the band
gap of
TiO2
. The absorption peak at 1.05 eV for samples prepared by ILGAR was also
detected in the X and Y signals. New signals around 2.5 ev and 2.7 eV were observed in
the defect region below the band gap of
TiO2
for Pt doped
TiO2
(Figure 5.13(d)) which
are not detectable for the undoped
TiO2
sample.
52
Figure 5.13: Surface photovoltage measurements (SPV): (a) Normalized SPV amplitude, in-phase (blue curves) and phase
shifted by 90
◦
signals (red curves) for commercial
TiO2
paste (b), undoped
TiO2
(c) and 5%Pt:doped
TiO2
(d) prepared
by the ILGAR process.
53
5.1.3.4 IPCE - DEMS measurements
Quantum eciencies characterize the performance of a photocathode in photon-to-charge
carrier conversion eciency. Figure 5.14 presents the corresponding IPCE data, measured
at a constant potential of
−0.4
V RHE, i.e. in the saturation photocurrent density region.
Four dierent doping concentrations (0, 2, 5, and 10 vol.% Pt-precursors, respectively)
are compared. Measurements were performed over a wide range of wavelengths (300-800
nm). The black curve shows the IPCE of undoped
TiO2
deposited on CIGSe. The IPCE
data are below 8% over the visible light range. The red, green and blue curves belong
to 2, 5, 10 vol.% Pt doping density, respectively. The IPCE results are in good agree-
ment with the corresponding PEC performance of the photocathodes: the eciency of
the sample increases by incrementing of the Pt density from 2 to 5% and decreases by
further doping to 10 vol.%. This nding implies better charge separation with Pt up to
5 vol.% and is also in good agreement with conductivity measurements (Figure 5.7).
Figure 5.14: Incident-photon-to-charge-carrier conversion eciency (IPCE) data of four dierent Pt-doped
TiO2−
Cu(In,Ga)Se2
heterojunctions. (a) green squares: 5 vol.% Pt; (b) red squares: 2 vol.% Pt; (c) blue squares: 10
vol.% Pt compared with the undoped
TiO2−Cu(In,Ga)Se2
heterojunction photocathode (d) black squares. The IPCE
data have been measured at -0.4 V vs. RHE. Electrolyte: 0.5 M
H2SO4
, pH0.3 [183].
The IPCE data of the most ecient photocathode device reveals an essentially high
eciency of
≈
80% over the complete visible light range from 400 to 800 nm (green
squares). The observed cut-o energies can be related to the optical band gap of ILGAR-
TiO2
(near 345 nm corresponding to 3.6 eV, respectively,
Cu(In,Ga)Se2
(
≈
1000 nm
corresponding to
≈
1.2 eV) [117,186].
In order to prove the evolution of hydrogen gas and to exclude formation of
H2Se
, HER
of the most ecient (5%)Pt-doped
TiO2
/
Cu(In,Ga)Se2
photocathode was followed by
dierential (photo-) electrochemical mass spectroscopy (DEMS) as shown in Figure 5.15.
The onset for hydrogen evolution (blue solid line: mass signal of
H2
related to the left
y-axis) of the photocathode occurs at a positive, anodic potential of about 110 mV versus
the RHE reference electrode.
54
Figure 5.15: Hydrogen evolution behavior of the most ecient Pt-implemented
TiO2
-
Cu(In,Ga)Se2
photocathode
under illumination (red curve) and in the dark (blue curve) detected by dierential photoelectrochemical mass spectroscopy
(DEMS). The inset shows the hydrogen evolution with respect to CV curves under illumination. Electrolyte: 0.5 M
H2SO4
,
pH0.3.
This onset potential is lower than the corresponding one shown in Figure 5.5 and is
caused by the employed gas membrane of the DEMS setup. This membrane is in close
contact to the sample, limiting thereby mass transport to the sample and causing an
additional overpotential. The inset shows the hydrogen evolution mass signal (red curve)
with respect to the CV curve under illumination (black curve). The slight increase of
the
H2
-mass signal in the dark is attributed to incomplete sealing of free parts of the
molybdenum back contact to which electrical contacts were connected. In our custom-
made setup, not all evolved
H2
gas can be directed to the spectrometer therefor the
results of Figure 5.15 are qualitatively only and it is not possible to evaluate the amount
of generated
H2
.
5.1.3.5 pH dependent performance: CV and stability measurements
The photoelectrochemical behavior of the most ecient photocathode (5 vol% Pt-precursor
solution) is shown in Figure 5.16 for aqueous electrolytes of pH 0.3 (green curves), pH7
(red curves) and pH14 (blue curves). Solid lines show the original data and dashed
lines are IR corrected data, i.e. data corrected for uncompensated solution resistance,
R, as determined by impedance spectroscopy. The corrected potentials are calculated by
subtraction of the potential drop across the solution, IR, from the applied potential U
according to: (
Unew =Umeasured −IR
).
As shown in this Figure, at the redox potential, current densities of 14, 13 and 3
mAcm−2
were measured for pH values of 0.3, 14 and 7, respectively. The current-voltage
behavior of the sample in pH7 shows that the saturation photocurrent density is obtained
at signicantly higher potentials which has to be attributed to the low conductivity of
the electrolyte at pH7. In comparison to the most ecient devices for photoassisted
evolution of hydrogen published so far, planar Si-Pt and InP-Rh heterojunctions [187-188]
with eciencies of 9.6% and 14.5%, respectively, the
TiO2
-
Cu(In,Ga)Se2
system shows
55
higher saturation photocurrent densities but operates less active, i.e. the samples show
lower photovoltages and, consequently, a negatively shifted onset. However, our CIGSe
substrates are characterized by uniform doping while both the Si and InP absorbers were
fabricated as homojunctions with a highly doped surface-near region, improving thereby
considerably the achievable photovoltage.
Figure 5.16: Current-voltage behavior of 5 vol.% Pt doped
TiO2
-
Cu(In,Ga)Se2
composite photocathode devices in
dierent electrolytes:
H2SO4
0.5M pH 0.3 (green curve), KOH 1N pH 14 (blue curve) and KPi buer phosphate solution
pH7 (red curve). All measurements were carried out in a three-electrode mode vs. Ag/AgCl reference electrode.
To examine the long-term stability of the new composite devices, the cathodic pho-
tocurrent density has been measured at the thermodynamic reduction potential of hydro-
gen evolution,
E0
= 0 V RHE, in
H2SO4
electrolyte pH0.3 (Figure 5.17). No degradation
was observed within 25 h for the 5 vol.%
Pt :TiO2/Cu(In,Ga)Se2
composite photocath-
odes. A cathodic photocurrent density of
≈
9
mAcm−2
was monitored over 25 h. The
external light source was turned on under potential and turned o after 25h in order to
prove the maintained photoactivity of the device and to exclude electrochemical degrada-
tion currents which can be typically observed also in the dark. In order to highlight this
important stability, the corresponding turn-over-number (TON) of active surface sites of
the photoelectrocatalyst surface was calculated. A total amount of charge of about 810 C
cm−2
[
≈
9
mAcm−2×25 h×3600 s
] is supplied by the photoelectrocatalytic sample. By
assuming that each surface atom provides one active site (
≈1015
atoms
cm−2
), a high
TON of
≈2.5 ×106
is calculated for the produced solar hydrogen molecules per active
site which corresponds to a Turn-over frequency (TOF) of about 690 hydrogen molecules
per second and per active site [183].
56
Figure 5.17: Photocurrent density time behavior over 25 (20) hours of the most ecient Pt-doped
TiO2/Cu(In,Ga)Se2
heterojunction photocathode in electrolytes of pH0.3 (pH7). Photoelectrodes were measured under near AM1.5 (100
mW/cm2
) conditions at the hydrogen reduction potential 0 V vs. RHE. The spikes correspond to light on/light-o in
order to assess the dark current density in a short time of several minutes.
In Figure 5.17(b), stability investigations for the sample in solutions of pH7 and
pH0.3 are shown for 20 (25) h. The calculated STH is 2%. The long-term stability in
neutral solutions with overpotential-free evolution of hydrogen proves the advantageous
concept with promising implications for future industrial applications. After 24 hours the
electrolyte was renewed and the sample was cleaned by rinsing in ultra-pure water. Sub-
sequently the photocurrent density reached again values near 8
mAcm−2
. This nding
suggests that active surface sites might have been contaminated during extended opera-
tion by the buer solution. The stability at pH14 is limited about 2h.
5.1.4 Interface engineering by thin interfacial
TiO2
deposited by Atomic
Layer Deposition (ALD)
SEM analysis of very low ecient samples revealed that Pt nanoparticles were forming
at the interface and surface of the
TiO2/Cu(In,Ga)Se2
junction. Figure 5.18 shows some
of these SEM images. In (a) two big clusters formed on the surface of
Cu(In,Ga)Se2
and
TiO2
. EDX mapping (b) proved that these particles are composed of Pt. Also in
(c) and (d), the cross section and a top view of one of these samples is depicted and
shows that a layer, composed of Pt nanoparticles, was formed at the interface of
TiO2
and
Cu(In,Ga)Se2
.
57
Figure 5.18: (a) and (b) SEM / EDX image of a low ecient sample, with formation of Pt clusters on the surface of a
Cu(In,Ga)Se2
, (c) SEM image of cross section of the sample, (d) the top view SEM image of the sample shows formation
of an interface layer with Pt nanoparticles.
The formation of a suitable metal-semiconductor junction is critical for the perfor-
mance of the electrodes. As described in section (2.4), depending on the work function
of the semiconductor and the metal, two types of barriers can be observed: the Schottky
barrier and the ohmic barrier. The assumed band bending for a p-type semiconductor is
proposed and shown in Figure 5.19 (a/b). According to UPS measurements, the electron
anity and work function of
Cu(In,Ga)Se2
are 3.8 eV and 4.82 eV, respectively. The
work function of Pt depends on the size of the nanoparticle and is between 5.3 and 5.7
eV. If Pt comes in direct contact with
Cu(In,Ga)Se2
formation of an ohmic contact
is assumed (Figure 5.18(b)). In this case,
Cu(In,Ga)Se2
is characterized by at-band
conditions or slightly upward band bending which prohibits separation of photogenerated
carriers. Therefore no light-induced charge transfer occurs and the performance of the
sample is low.
Figure 5.19: Band alignment of a p-semiconductor with a metal for dierent work function relations (a)
Φm< Φp
, (b)
Φm> Φp
and (c) Proposed band alignment of the 5%Pt-doped
TiO2−Cu(In,Ga)Se2
[183].
58
In order to derive an energetic scheme for the
TiO2
/
Cu(In,Ga)Se2
junction, work
function,
Φ
, energetic distance,
∆
Fermi-VBM, between valence band maximum (VBM)
and Fermi energy,
EFermi
, in
Pt :TiO2
have been measured by means of UV-light excited
photoelectron spectroscopy (UPS). The derived energetic position of the
TiO2
conduction
band minimum (CBM) of
∼
3.4(2) eV, i.e. the electron anity,
χ
, is negative with respect
to the thermodynamic potential for proton reduction. A splitting of the quasi-Fermi levels
of the minority charge carriers,
En,p
F
, occurs at the heterojunction under illumination and
results in a net photovoltage. The proposed band alignment of the most ecient Pt-doped
TiO2−Cu(In,Ga)Se2
composite photocathode is depicted in Figure 5.19(c) together with
the electrochemical position of both the hydrogen reduction potential
µH+/H2
e
, and the
water oxidation potential,
µH2O/O2
e
, on the right hand side [117,185-186].
In order to minimize the recombination rate at the interface of chalcopyrite the
(
Cu(In,Ga)Se2
) and
TiO2
, and also to avoid the formation of semiconductor-metal (Pt)
junctions, a thin layer (10 nm) of amorphous
TiO2
was deposited by atomic layer deposi-
tion (ALD) on the
Cu(In,Ga)Se2
surface. Subsequently, Pt doped
TiO2
was deposited
by ILGAR. Figure 5.20 shows the comparison of the respective current-voltage behavior
of the samples without Pt doping (black curve), 2%, 5% and 10% Pt:
TiO2
/
Cu(In,Ga)Se2
(in red, green and blue curves) and the sample with an ALD
TiO2
interlayer and 5%Pt:
TiO2
top layer (brown curve).
As shown in this graph an anodically shifted onset potential at +400 mV is observed
for the sample with ALD
TiO2
interlayer. The slope of the graph indicates, however, an
increase in resistivity of the sample. Although the onset potential improved, the pho-
tocurrent density at the redox potential shows no improvement. Saturation photocurrent
densities reach, as expected, the same value independently on the presence of the inter-
layer.
Figure 5.20: Comparison of the current-voltage behavior of the undoped
TiO2−Cu(In,Ga)Se2
(black curve),
Pt :
TiO2−Cu(In,Ga)Se2
compsoite photocathode with
Pt :TiO2−TiO2(ALD) − Cu(In,Ga)Se2
sample. The best
current-voltage behavior is observed for the sample with
TiO2
(by ALD) interface layer and shows an anodically shifted
onset potential at +400 mV (brown curve); Electrolyte: 0.5 M
H2SO4
, pH0.3. Photoelectrodes were measured under near
AM1.5 (100 mW/
cm2
) conditions.
59
5.2 EPD-prepared semiconductor electrodes for electrocatalytic
and photo-assisted evolution of oxygen
Electrochemical evolution of oxygen requires immobilization of suitable electrocatalysts
on semiconductor supports. Two types of electrocatalysts were investigated, (i) Co-
substituted ZnO, ZnO:Co, and (ii)
RuO2
in order to realize the OER in alkaline as well
as in acidic electrolytes. Although FTO is a common semiconductor support for evo-
lution of oxygen, the stabilization of this support in high-concentrated electrolytes is
demanding. It will be shown that the electrophoretic deposition of the electrocatalysts
resulted concomitantly in formation of a carbon-rich protection layer which stabilized the
FTO support upon long-time operation. This nding made it possible to choose nally
the even more oxidation-sensitive silicon as support for photo-assisted evolution of oxy-
gen (my work focuses here exclusively on the Si/
RuO2
junction while the corresponding
Si/ZnO:Co system was investigated by colleagues of mine). It is noteworthy that the
application of silicon in an inorganic-organic conguration schemes represents thereby a
major progress. Although doped-silicon is commercially available and its band gap makes
it a suitable candidate for photoanode preparation, the susceptibility of this semiconduc-
tor to photo-degradation and passivation in aqueous electrolytes, especially under anodic
conditions, makes ecient surface protection schemes inevitable to guarantee long-term
stable operation. The challenge for proper surface protection is to realize optimum elec-
tronic and optical properties in order to benet most from the technologically advanced
semiconductor substrate. In recent approaches, metallic overlayers or compact oxide lms
have been employed with increasing success and oxidation of the semiconductor upon pho-
toelectrochemical splitting of water could be avoided for extended periods of time [78, 80,
144, 147, 189-191]. Both protection schemes require individual considerations: metallic
protection layers have to be devised suciently thick in order to compensate structural
aws upon unwanted self-oxidation. Increased thickness, however, attenuates the light
reaching the absorber. Suciently pin-hole free oxide lms, on the other hand, can intro-
duce non-negligible serial resistance to the electrode and, also require mild and complex
preparation conditions in order not to compromise the substrate. For both materials,
metals and oxides, junction formation with the semiconductor is critical since the achiev-
able photovoltage is easily reduced by partial Fermi-level pinning at the interfacial region
[192]. A third material class, organic overlayers is considered a promising alternative
route for semiconductor protection: surface terminal groups obtained by alkylation pro-
cedures oer nearly ideal electronic passivation and ensure maximum in-built potentials
and therefore highest photovoltages [193-195]. Moreover, junction formation with single
layers of graphene have been suggested which can help overcoming intricate preparation
steps associated with chemical surface functionalization [77]. So far, improved stability
could be shown only in aqueous electrolytes where the redox reactions occur at potentials
considerably smaller than the water oxidation potential. Therefore, an organic protec-
tion scheme for state-of-the-art semiconductors, applicable to solar water splitting, still
remains a major challenge.
In the following rst part the eect of the organic solvents is investigated upon de-
position of ZnO:Co on uorinated tin oxide supports (FTO). The results section will
be nally completed by analysis of the corresponding deposition process of
RuO2
onto
silicon.
61
5.2.1 ZnO:Co on FTO as electrocatalyst in alkaline electrolytes
Three organic solvents (ethanol, acetone and acetonitrile) with varied electrophoretic
mobility have been employed for deposition of nanocrystalline ZnO:Co particles onto
uorinated tin oxide supports using the electrophoretic deposition technique. The depo-
sition method is described in section 4.1.2. For fundamental investigations, the catalysts
were deposited on FTO. The results is expanded for deposition of presenthesized
RuO2
on n-Si (section 5.2.3) in order to prepare a photoanode working in acidic electrolyte.
5.2.1.1 Eect of the electrophoretic mobility on the quality of the EPD pro-
cess
In Figure 5.21(a), the eect of the organic solvent on the lm thickness (left axis,solid
lines) and the density of the layers (right axis, dashed lines) is summarized for var-
ied deposition times. The red, blue and green colors are used for samples prepared in
ethanol, acetone and acetonitrile, respectively. Thicknesses of the layers were measured
by DEKTAK prolometry and the density of the layers was calculated by measuring the
deposited mass. In total, nine samples were prepared by suspending ZnO:Co electrocat-
alysts in ethanol, acetone and acetonitrile. Electrophoretic deposition was carried out
for 4, 7 and 10 min each. As shown in Figure 5.21(a), the resulting layer thicknesses
are lowest for ethanol and highest for acetonitrile for each deposition time. The inverse
dependence is observed for the calculated density. In general, the deposition rate for
each solvent is observed to be constant, i.e. the layer thickness increases linearly with
time. The lowest deposition rate occurs in ethanol solution and the highest in acetonitrile
solution [196]. The visual appearance of the samples after 10 min is depicted as inset in
Figure 5.21(b). An obvious color variation is observed from dark green after deposition
from acetonitrile to light green after deposition from ethanol. As shown in Figure 5.21(a),
as dashed lines, the variation of the density of the layers proves an increasing porosity
with deposition time. This nding is particularly important for electrocatalytic processes
where an increased active surface area is crucial for an enhanced electrochemical activity.
This observed trend implies that under the applied deposition potential smallest parti-
cles reach the substrate rst, forming thereby a compact layer, while bigger particles are
subsequently deposited with a large number of voids in between. Figure 5.21(a) shows
that the densities of the layers prepared in ethanol (red dashed line) are higher than
the corresponding densities of those samples prepared in acetone (blue dashed line) and
acetonitrile (green dashed line). In the following, the measured layer properties are dis-
cussed in terms of EPD theory, i.e. electrophoretic mobility and EPD deposition rate,
respectively. By measuring the deposited mass and using Eq.(1) (Hamaker law),[176] the
electrophoretic mobility of ZnO:Co in dierent solvents can be calculated by equation
(4.1).
62
Figure 5.21: (a) Eect of the organic solvent and the deposition time on the thickness and the density of the layers. (b)
Role of the organic solvent on the electrophoretic mobility of ZnO:Co particles in the suspension [196].
As described in section 4.1.2, electrophoretic mobility inuences the speed of the
particles in the media. The results of
µe
for ZnO:Co particles are shown in Figure
5.21(b) and Table 5.1. Higher
µe
values for samples prepared in acetonitrile solvent
indicates that the ZnO:Co catalysts are moving with higher velocities in this solution
in comparison to acetone and ethanol solvents. The lowest velocity of the particles is
observed for the ethanol solvent. This velocity depends on the interaction of the particles
with the applied eld.
When the catalysts are suspended in the solvent, surface charges are forming around
them. These charges are responsible for stabilization and movements of the particles
under an applied electric eld present in the suspension. Four types of forces can aect
the behavior of the particles.
1. Electromagnetic force (
Ef
) which corresponds to the interaction of the surface
charges with the electric eld and is equal to the product of the number of charges
and the electric eld strength (
Ef
). The magnitude of the electric eld was kept
constant (
10 Vcm−1
) during the EPD process.
2. Drag force (
Fd
) : this force exerts a force on the particle in opposite direction of
its motion. The magnitude of the drag force depends on the velocity of the particle
and the properties of the uid (density) and the size of the particles.
3. Gravitation force (
Fg
): this force points downwards and depends on the mass of
the particles.
63
4. Buoyancy force (
Fb
): this force points upwards and is caused by the uid, reducing
thereby the weight of an immersed particle. It is equal to the product of the uid's
viscosity (
ρf
), g (9.8
ms−2
) and the volume of the particles.
The schematic of these forces is shown in Figure 5.22. If the electromagnetic force is
larger than the sum of the three other forces, the particle moves toward the anode and
is deposited there. Otherwise the particle settles at the bottom of the beaker in the case
of a buoyancy force lower than the gravitational force.
Figure 5.22: Diagram of forces inuencing the movement of the dispersed particle.
Since the magnitude of the electric eld was kept constant (
10 Vcm−1
) during the EPD
process, the magnitude of the electromagnetic force depends on the amount of charges
induced at the surface of the particle built upon electrochemical equilibration with the
solvent. Dierent parameters like polarity and viscosity of the suspension also inuence
the amount of induced charges. All these parameters are summarized as zeta potential
parameter (
ζ
) in the Smoluchowski correlation, given by Eq.(4.2) described in section
4.1.2.
The electrophoretic mobility and the zeta potential of ZnO:Co particles in dierent sol-
vents were accordingly calculated by Eq. 4.1 and Eq. 4.2 and are given in Figure 5.23.
As shown in this gure, both the electrophoretic mobility and the zeta potential de-
creased slightly during increasing deposition times for all the solvents [196]. This eect
is accounted for by the progressive consumption of iodine which is required to induce a
positive charge at the surface of the particles. The decreasing iodine concentration then
results in a lower zeta potential and, consequently, smaller deposition rates.
64
Figure 5.23: (a), (b) Eect of the organic solvent as a function of deposition time on the electrophoretic mobility and the
zeta potential of the ZnO:Co particles [196].
As shown in Figure 5.23, there is a big dierence between the zeta potential of ZnO:Co
particles dissolved in ethanol and those dissolved in acetone and acetonitrile. This eect
can be understood by the dierent activity of the solvents towards iodine and nally the
suspended particles. Acetonitrile and acetone are polar aprotic solvents; the nitrile group
in acetonitrile and the carbonyl group in acetone are proton acceptors. Polar aprotic
solvents cannot participate in hydrogen bonding because of lack of O-H or N-H groups.
Therefore, iodine in the solution is free and induces a positive charge at the ZnO:Co
particles. In the case of ethanol, the polar-protic solvent participates in hydrogen bond-
ing which is a powerful intermolecular force. Hydrogen bonds are more easily formed
between hydrogen of the hydroxyl group and the nucleophile iodine. Therefore the con-
centration of iodine in the ethanol solution decreases in comparison to acetonitrile and
acetone. As a consequence, the amount of positive surface charge at the surface of the
ZnO:Co particles is lowered and the horizontal movement of the particles towards the
cathode is more impeded by vertical forces, i.e. gravitation. As a further consequence,
formation of hydroiodic acid is facilitated in ethanol. The generation of acid can inu-
ence the deposition of the layers in two ways: rstly, the ZnO:Co particles are partially
etched, i.e. they decrease in size; secondly, already deposited particles can be released
from the surface if the etching process aects the interface between the FTO support
and the particles. Both eects therefore contribute to an overall limited layer thickness.
Moreover, the apparent lm porosity is reduced since small particles can be immobilized
in a denser layer structure than larger particles. It should be noted that also for ace-
tone formation of hydroiodic acid can be expected: due to the keton-enol tautomerism
of acetone, a small fraction of -OH bonds are present in solution. It is assumed that
the kinetics of this transformation is shifted towards the enol species in the presence of
iodine. Therefore, lm properties of acetone-prepared samples, i.e. the density of the
65
lms, as well as electrophoretic parameters, i.e. the EPD deposition rate, are expected to
reveal the impact of minute concentrations of hydroiodic acid. In fact, the data of Figure
5.21 demonstrate that layer thickness and porosity as well as electrophoretic mobility
are all slightly smaller than observed for acetonitrile. Only the calculated zeta potential
in Figure 5.23(b) appears increased which has to be attributed to the larger dielectric
constant of acetonitrile. The results are summerized in table 5.1.
ZnO:Co in ZnO:Co in ZnO:Co in
Ethanol Acetone Acetonitrile
µ
(
cm2V−1S−1)×10−8
4 min EPD 1.46 18.3 25.5
7 min EPD 1.46 16.7 23
10 min EPD 1.17 16.1 22.5
ζ
potential(mv)
4 min EPD 12.1 47.9 42.7
7 min EPD 12.1 43.7 38.4
10 min EPD 9.7 42.2 37.1
Deposited Mass(mg)
4 min EPD 0.04 0.5 0.7
7 min EPD 0.07 0.8 1.1
10 min EPD 0.08 1.1 1.5
Thickness(nm)
4 min EPD 49 654.58 1102.74
7 min EPD 125.29 1496.94 2318.36
10 min EPD 165.34 2330.87 3585.11
Table 5.1:
Electrophoretic deposition data
5.2.1.2 Eect of the organic solvent on the catalytic activity of ZnO:Co elec-
trodes
The electrochemical behavior of nanocrystalline ZnO:Co particles deposited onto uo-
rinated tin oxide supports (FTO) samples was investigated by cyclic voltammetry in
1N KOH, pH14. Under these conditions, the redox potential for oxygen evolution is at
E0
= + 0.2 V versus Ag/AgCl. Figure 5.24(a) shows the current-voltage behavior in a
three-electrode arrangement for the samples prepared in acetone (blue curve), acetonitrile
(green curve) and ethanol (red curve). Corresponding two-electrode measurements are
shown in Figure 5.24(b). Evaluation of the electrochemical activity for evolution of oxygen
proves a clear solvent-dependence with highest activity upon deposition from acetonitrile
and lowest activity upon deposition from ethanol. Pronounced oxidation and reduction
peaks around 0.4 V Ag/AgCl are visible for samples prepared in acetone and acetonitrile.
XPS results, As already reported earlier[47], a transition of Co(II) to Co(III) occurs upon
OER. For ethanol-prepared samples this reversible oxidation-reduction behavior is much
less visible. These observations are clearly related to the dierence in porosity of the re-
spective samples and therefore the dierence in active surface area. Correspondingly, the
current onset potentials are visibly improved for acetonitrile and acetone in comparison
to ethanol. Galvanostatic determination of the potential corresponding to 1
mAcm−2
therefore proves the enhanced activity over ethanol-prepared samples (see inset in Figure
5.24(a)): the respective potentials are 362, 300 and 290 mV, conrming thereby the slight
improvement by usage of acetonitrile-based suspensions.
66
Figure 5.24: (a) Current voltage behavior of ZnO:Co prepared in dierent organic solvents ethanol (red line), acetone
(blue line) and acetonitrile (green line) and deposited on FTO substrates. (b) Galvanostatic measurements of the layers
corresponding to a current density of
1mAcm−2
. Electrolyte: KOH 1N.
Using dierential electrochemical mass spectroscopy it is nally proven that com-
mencing evolution of oxygen is detected beyond potentials measured above (Figure 5.25).
The right curve shows the
O2
mass signal for a sample prepared in acetonitrile (Figure
5.25(a)) and the left curve shows the
O2
mass signal for a sample prepared in ethanol
(Figure 5.25(b)). Since the performance of the samples prepared in acetone and acetoni-
trile are very similar, corresponding mass signals are only shown for the acetonitrile and
ethanol solvent. Putting the observation of thicker and more porous lms, prepared in
acetonitrile and acetone (Figure 5.25a), into relation with higher OER activity (Figure
5.24) and lower overpotentials (inset in Figure 5.24), it is concluded that the size of the
active surface area is the most crucial distinguishing lm property. This property, in turn,
is directly related to the dierence in electrophoretic mobility of the ZnO:Co particles in
the respective organic solvent. It should be noted that due to the specics of the DEMS
setup diusion-related additional overpotentials are present and have to be related to the
small spacing between the sample and the used membrane.
67
Figure 5.25: Dierential electrochemical mass spectroscopy (DEMS) measurements for
O2
evolution on ZnO:Co deposited
on FTO, dissolved in acetonitrile (a) and ethanol (b). Electrolyte: KOH 1N.
Stability of the respective ZnO:Co lms in 1N KOH, pH14, was measured over a
period of 24 hours and is shown in Figure 5.26. Potentials were adjusted such that
current densities are close to 4
mAcm−2
in the beginning. For clarity, the curves in
Figure 5.26(a) are presented after applying a positive small shift. For all samples a
current decrease by about 10% is observed. Cross-check analysis of the behavior of bare
FTO in 1N KOH, however, proved pronounced instability of the support (Figure 5.26(a)).
In order to achieve more insight into the long-term behavior, an ethanol-prepared sample
was additionally tested over 72 hours (Figure 5.26(b)). After that time, the electrolyte
was renewed as indicated by an arrow. It can be seen that the renewed electrolyte permits
partial compensation of the current decrease. Initial high values, however, are not reached
anymore. This nding can be attributed either to a loss in activity of the catalytic
material or to structural aws within the lm which lead to progressive degradation of
the FTO support. These results point to an ecient protection of the FTO support by
the ethanol-prepared lm which is, according to the analysis above, characterized by the
highest structural density of the investigated lms.
68
Figure 5.26: Stability assessment of ZnO:Co, prepared from ethanol-containing suspensions, at pH14. For comparison, the
rapid degradation of a bare FTO substrate is indicated.
The nature of these protecive properties for the support have been recently inves-
tigated by M. Lublow who suggested the formation of a carbonnaceous oxyhydroxide,
concomitantly forming at the support's surface during EPD (see planned publication
No. 6 in section 7.3). As precursor material for the formation of the oxyhydroxide,
dissolution products of ZnO:Co in the presence of hydroiodic acid appear likely. It can
be therefore assumed that the prepared ZnO:Co/FTO electrodes enclose an interfacial
region of carbon-containing composite material with amorphous zinc and cobalt oxyhy-
droxides rather than to be composed of pure ZnO:Co nanocrystals. This nding of a
dense carbonaceous oxide layer with protective properties was inspirational for direct de-
position of water oxidation catalysts onto oxidation-sensitive semicundoctor supports to
be discussed further below with the example of
RuO2/Si
. The investigation of the role
of carbon (both
sp3
and
sp2
hybridized) as addition to transition oxide metal layers is
continued.
5.2.1.3 Optical analysis of ZnO:Co on FTO by UV-VIS and SPV
Optical measurements include transmission spectroscopy and normalized SPV ampli-
tudes of ZnO:Co lms prepared in dierent solvents for 7 min deposition time each. The
results are shown in Figure 5.27(a-b), respectively. The red, green and blue curves show
the behavior of the samples prepared in ethanol, acetonitrile and acetone, respectively.
The samples were illuminated from the front side, i.e. light passed through the ZnO:Co
lm rst and only photons, not absorbed by ZnO:Co reached the FTO support. Opti-
cal transmittance spectra for ZnO:Co lms prepared in dierent solvents for 7 min are
shown in Figure 5.27(a). While acetonitrile-and acetone-prepared samples show strong
69
attenuation of the incident light, highest transmittance is achieved with samples prepared
from ethanol solutions. Transmittance values near 70% therefore suggest this preparation
method as suitable for photoactive supports to be discussed further below. The optical
band gap of ZnO:Co, obtained by extrapolation of the low-energy feature to the x-axis is
about 2 eV. The second absorption edge near 3.3 eV corresponds to the optical band gap
of the FTO support. In Figure 5.27(a), the SPV response indicates that photo-induced
excess charge carriers are separated in space [197-198]. The SPV amplitude for all the
samples exhibited two response peaks (as in corresponding UV-VIS spectra), the curves
increased sharply at a photon energy of about 1.7-1.8 eV indicating the onset of tail
states (defect levels) within the band gap of the material (see extrapolated straight lines
in Figure 5.27(b)). The second structure, to be attributed to FTO, is not shown in Figure
5.27(b) for clarity. It should be noted that the onset energy of electronic states extending
into the band gap is always lower than the corresponding optical band gap.
Figure 5.27: optical characterization of ZnO:Co deposited on FTO substrates (a) Transmission spectra as a function of
wavelength.
X-ray diraction patterns of ZnO:Co powder (before deposition) and ZnO:Co/FTO
junctions before and after electrochemical measurements (EC) are shown in Figure 5.28.
GIXRD measurements were carried out at 0.2
◦
angle of incidence. The results reveal a
polycrystalline, highly oriented lm directly after electrophoretic deposition. The dom-
inant peaks are assigned to 101, 100 and 002 reections. Using the Scherrer equation,
average particle sizes of 24, 17 and 21 nm were calculated for the corresponding peaks,
respectively. After EC measurements there is a visible change in the X-ray reection
spectra (Figure 5.28 (b)). The intensity of the peaks appears reduced and crystal sizes of
18, 12 and 16 nm are obtained. In correspondence to earlier ndings by Pfrommer et al.
[47], this change has to be attributed to a continuous transformation of the crystalline
into an amorphous phase, i.e. the nanocrystalline powder acts as a core-precatalyst whose
surface area continuously transforms to the nal electrocatalytic material upon evolution
70
of oxygen.
Figure 5.28: GIXRD diagram of ZnO:Co prepared in ethanol (red line), acetone (blue line) and acetonitrile (green line)
(a) after electrophoretic deposition and (b) after electrochemical measurements.
Correspondingly, elemental analysis by EDX proves considerable dierences in the
Zn:Co ratio before and after OER. The results are shown in Figure 5.29, the colored lines
show the EDX pattern for as-prepared samples and the black line indicates the sample
after electrochemistry. For all samples, a decreased Zn:Co ratio is observed pointing
to preferential dissolution of zinc oxide. Before electrochemical testing, the respective
Zn:Co ratios for acetone-, acetonitrile- and ethanol-prepared samples were: 2.5:1, 2.5:1,
and 2.6:1, respectively. After EC, ratios of 0.4:1, 0.5:1 and 0.1:1 were calculated. The
variation of these latter numbers can be understood by the dierence of the respective
(Zn-poor) surface to (Zn-rich) volume ratio after electrochemistry. According to Figure
5.21, ethanol prepared samples show the lowest lm thickness and porosity, i.e. only
small nanoparticles are deposited with a smaller surface-to-volume ratio than acetonitrile-
and acetone-prepared lms. Consequently, the Zn-poor surface after OER region has a
stronger impact on the overall Zn:Co ratio.
71
Figure 5.29: EDX spectra of ZnO:Co deposited on FTO in acetonitrile (a), ethanol (b) and acetone (c) before and after
electrochemistry measurements.
Figure 5.30: XPS analysis of Co 2p signal (a) before electrochemistry for samples prepared in acetone (blue curve),
acetonitrile (green curve) and ethanol (red). (b) XPS results for the Co 2p signal for acetone prepared sample before
and after electrochemistry (c) and (d) comparative deconvolution of the Co 2p and Zn 2p signals for samples prepared in
acetone and ethanol, respectively.[196].
72
In Figure 5.30, XPS analysis of ZnO:Co is shown after deposition on FTO from
acetone-, acetonitrile- and ethanol-prepared suspensions. The as-prepared samples (Fig-
ure 5.30 a) showed high resistivity of the lms. As a consequence, photoelectron emission
resulted in electrostatic charging, and the spectra were apparently shifted to higher bind-
ing energies. The respective shifts (about 3 eV for acetonitrile and acetone and about
10 eV for acetone) were corrected using the C 1s core-level signal of adventitious car-
bon as reference energy. For all samples, pronounced shake-up signals were detected at
binding energies of about 787 and 803 eV. The presence of these features is attributed
to a Co(II) oxidation state [199]. After evolution of oxygen in 1N KOH, acetonitrile-
and ethanol-prepared samples did not show any charging while binding energies of the
acetone-prepared sample still were shifted by about 3 eV (Figure 5.30 b). Detailed quan-
titative analysis was therefore carried out only for the Co 2p and Zn 2p signals, measured
for acetonitrile- and ethanol-prepared lms (Figures 5.30 c and d). Shake-up satellite fea-
tures are much less pronounced, indicative for the transition of the cobalt oxidation state
from Co(II) to Co(III). Deconvolution of the Co 2p signal resulted in three substructures.
The main peak lines at about 779.9 and 781.2 eV are attributed to
Co3O4
and
Co(OH)2
,
respectively. A third structure at about 782.5 eV results from imperfect subtraction of the
2p signal from the measured curve, i.e. this feature indicates the onset of the following
satellite structure. The corresponding Zn 2p signals exhibit two sub-structures at about
1021 and 1021.5 eV. The main struture at 1021.5 eV is attributed to ZnO. The smaller
feature at lower binding energy cannot be unambiguously identied. In order to deter-
mine the Co:Zn ratio of the lms within the escape depth of photoelectrons, i.e. within
about 1 nm [200], the integral areas,
ICo
and
IZn
, beneath the curves were calculated
and normalized by the corresponding photoionization cross sections of the 2p core-level
signals for an excitation energy of 1486.3 eV (
σCo
= 0.26,
σZn
=0.39). The ratio can then
be expressed as
σ−1
Co ×ICo/σ−1
Zn ×IZn
. For the acetonitrile-prepared sample, a ratio of
3:1 was calculated in approximate agreement of corresponding EDX analysis which re-
sulted in a (more volume-sensitive) value of 2:1. The deviation in the ratio, obtained by
the two methods, can be explained by the fact that XPS probes more the active surface
area, i.e. the sites where Zn is leaching into the electrolyte upon evolution of oxygen.
For the ethanol-prepared sample, the corresponding XPS-determined ratio is 1.7:1. This
value is smaller than for the acetonitrile-prepared sample (3:1) and much smaller than the
one obtained by EDX analysis (10:1). Two eects become thereby apparent: rstly, the
ethanol-prepared sample is less active, in correspondence with measured C-V curves, and
leaching of Zn proceeds at a slower rate; secondly, the volume of the lm suered from
loss of Zn already upon preparation. Probably, the HI-containing suspension dissolved
considerable amounts of the ZnO:Co particles, and the molecular dissolution products
were deposited as an amorphous Zn-poor lm. The amorphous morphology was already
suggested by TEM analysis. Remnant particles, nally, were deposited on the top of
this amorphous layer due to slower deposition kinetics in comparison to molecular mate-
rial. Thereby, XPS detects the stoichiometric ratio of these particles at the surface while
EDX probes more the volume of the lm which appears to strongly deviate from the pre-
synthesized ZnO:Co particles. These ndings help clarifying the lower performance of
the ethanol-prepared sample: while the surface area appears to consist of the same active
phase as in the case of acetonitrile-prepared samples, the amorphous sub-layer structure
may introduce either structural impediments, i.e. the lm is to smooth to introduce an
enlarged surface area, or an additional barrier for charge transfer, caused by the amor-
phous lm, lowers the activity. (XPS measurments were done by Ch. Höhn at HZB and
73
XPS analysis was carried out under supervision of Dr. M. Lublow)
5.2.1.4 Morphological analysis by SEM/ EDX/ TEM
SEM and TEM images for ZnO:Co are shown in Figures 5.31-33 deposited from the
three dierent solvents, acetonitrile, acetone and ethanol, respectively. Left-row images
depict the materials structure before electrochemical operation. Right-row images show
the corresponding analyses after electrochemistry. On the larger scale of the SEM images,
the least changes are visible with samples deposited from ethanol (Figure 5.33 (a/b)). Due
to the higher porosity and therefore higher active surface area, those changes are more
pronounced for samples prepared from acetonitrile and acetone (Figure 5.31 and 5.32,
respectively). Closer inspection by TEM analysis of small lamellae (and selected area
electron diraction images) reveals the commencing transition to an amorphous phase
by exposure to the electrolyte under anodic potential (compare the respective images
c through f). This transition is almost completed with ethanol-prepared samples while
with acetonitrile and acetone prepared samples, distinguishable diraction patterns can
still be observed. (TEM measurements were done by C. Göble Pfrommer at TU-Berlin)
74
Figure 5.31: Microscopy characterization of ZnO:Co morphologies after EPD using acetonitrile as solvent. (a) and (b) SEM
surface view before and after electrochemistry, respectively. (c) and (d) corresponding TEM images of ZnO:Co lamellae
before and after electrochemistry, respectively. (e) and (f) Magnied TEM images of ZnO:Co lamellae before and after
electrochemistry, respectively [196].
75
Figure 5.32: Microscopy characterization of ZnO:Co morphologies after EPD using acetone as solvent. (a) and (b) SEM
surface view before and after electrochemistry, respectively. (c) and (d) corresponding TEM images of ZnO:Co lamellae
before and after electrochemistry, respectively. (e) and (f) Magnied TEM images of ZnO:Co lamellae before and after
electrochemistry, respectively [196].
76
Figure 5.33: Microscopy characterization of ZnO:Co morphologies after EPD using ethanol as solvent. (a) and (b) SEM
surface view before and after electrochemistry, respectively. (c) and (d) corresponding TEM images of ZnO:Co lamellae
before and after electrochemistry, respectively. (e) and (f) Magnied TEM images of ZnO:Co lamellae before and after
electrochemistry, respectively [196].
77
It is nally shown that the enhanced optical transparency of ethanol-prepared samples
can be successfully employed for solar-cell controlled direct water splitting. The schematic
of the setup with a triple-junction solar cell (amorphous silicon) is depicted in Figure 5.34.
The solar cell characteristic under illumination of 100
mWcm−2
is shown in black.
The electrocatalytic layer is xed to the photoactive glass front-side of the triple-junction
silicon solar cell. Short-circuit photocurrent densities of about 4.2
mAcm−2
results in an
eciency of 5% for direct photo-assisted water splitting. Corresponding photocurrent-
voltage curves, measured with Pt-counter electrode, are presented in Figure 5.35. Due to
the only small light attenuation by ZnO:Co (prepared in ethanol), about 70% of the short
circuit photocurrent can be exploited for light-driven evolution of oxygen. The eciency
thereby amounts to:
η=1.23V×J0
A×100 mWcm−2
(5.1)
Here,
J0
denotes the short-circuit photocurrent density.
Corresponding measurements with acetonitrile- and acetone-prepared samples are also
shown with green and blue curves in Figure 5.35(b). They exhibit much lower eciencies
due to large light absorption in the thick electrocatalytic layers. It is worth noting that
the modular architecture presented here allows for simple exchange of the ZnO:Co/FTO
top-layer once its activity has fallen below a minimum threshold value. By considering
the use of low-cost electrocatalysts, this approach appears to be an attractive alternative
to approaches where the catalysts are permanently xed to the photoactive supports.
Figure 5.34: Schematic setup of the combined supports, an amorphous/microcrystalline silicon solar cell (bottom) and
a ZnO:Co/FTO heterojunction (top). In the top-view, the O-ring is depicted in decreased size in order to stress the
photoactive area beneath.
78
Figure 5.35: (a) Photocurrent-voltage behavior of the silicon triple-junction solar cell (black curve) and the stacking
conguration with a top ZnO:Co/FTO layer, prepared from ethanol (red curve). (b) Corresponding stacking congurations
with ZnO:Co/FTO layers prepared from acetone (green curve) and acetonitrile (black curve). For comparison the solar
cell behavior without electrocatalysts is shown in red [196].
5.2.2 Characterization of
RuO2
on FTO as electrocatalyst in acidic elec-
trolytes
The harsh acidic and oxidizing conditions at the anode render most catalysts inactive or
unstable, except for oxides of Ir or Ru. It turns out that
RuO2
is more active than
IrO2
[42], but it is somewhat unstable towards degradation towards
RuO4
[155]. According
to the theoretical volcano model developed by Rossmeisl, Norskov and coworkers[201],
RuO2
is the most active pure metal oxide catalyst for OER, because it exhibits optimal
binding behavior to reaction intermediates [42]. Consequently, in this section presynthe-
sized
RuO2
catalyst was used instead of ZnO:Co on FTO and n-Si substrates in order
to prepare (photo-)anodes for acidic electrolytes. EPD deposition parameters were kept
constant as in the case of ZnO:Co particles. Firstly, the eect of the chosen organic
solvent on the electrode performance was investigated using ethanol, acetone and ace-
tonitrile and FTO substrates. Figure 5.36(a) shows the performance of the
RuO2
/FTO
electrode tested in
H2SO4
0.5M in a three electrode conguration and with Ag/AgCl
reference electrode. The green, blue and red curves show the current-voltage behavior
of the samples prepared in acetonitrile, acetone and ethanol, respectively. As shown in
this graph, the electrocatalytic activity of the samples prepared in acetonitrile and ace-
tone show pronounced oxidation-reduction peaks in comparison to the sample prepared
in ethanol. Again, as for ZnO:Co, acetonitrile-prepared electrodes (on FTO) prove lowest
overpotentials (about 320 mV for 10
mAcm−2
). The trend of the C-V curves for
RuO2
for the other organic solvents is similar to the behavior observed for ZnO:Co catalysts.
Figure 5.36(b) shows the eect of the organic solvent on the density and thickness of
the deposited layers. Similar to ZnO:Co, the sample prepared in ethanol has the thinnest
layer (17 nm) and the sample prepared in acetonitrile has the highest thickness (500 nm).
The sample prepared in acetone has a thickness of 226 nm. The calculated densities
demonstrate that, as expected, the deposited layer in ethanol has the highest density
while the acetonitrile-prepared sample has the lowest density. Corresponding conclusions
with respect to the porosity hold true as in the case of ZnO:Co.
79
Figure 5.36: (a) Current-voltage behavior of ethanol (red curve), acetone (blue curve) and acetonitrile (green curve)
prepared samples in
H2SO4
0.5 M. (b) Eect of the organic solvent on the thickness and the density of the layers. (c)
Role of the organic solvent on the electrophoretic mobility of
RuO2
particles in the suspension.
Electrophoretic mobilities and zeta potentials of
RuO2
were calculated according to
Eq. (4.1) and Eq. (4.2). The results (Figure 5.36(c)) are in good agreement with the
trends of the thickness and density of the layers. According to these results,
RuO2
par-
ticles have highest electrophoretic mobility and zeta potential in the acetonitrile solvent
that means the particles gain more charges in the solution. Therefore, under an external
electric eld the particles have higher speed (higher electrophoretic mobility) and reach
sooner the electrode (FTO). Particles in ethanol solvent has the least amount of charges
(lowest zeta potential), therefore interaction with the external electric eld is low.
5.2.3
RuO2
/Si photoanode for acidic electrolytes
As shown in Figure 5.36(a), the
RuO2
/FTO electrodes have approximately 200-300 mV
overpotential. To overcome this overpotential the
RuO2
catalyst were deposited on n-Si
to form a hetero-junction photoanode. In this preparation two critical points should be
considered. Firstly,
RuO2
is a black powder and absorbs the entire incident light. If it
completely covers the absorber (n-Si), the eciency of the photoanode decreases dramati-
cally. Therefore the aim is to prepare a thin layer, which is less light absorbing. Secondly,
Si is well known to be unstable under anodic conditions. Therefore, the deposited layer
not only should be catalytically active for OER but also should be pinhole-free to protect
the Si substrate from corrosion under anodic condition. It will be shown that the EPD
deposition of
RuO2
results in an polymeric interface that integrates near-ideal electronic
and electrochemical passivation in an ultrathin layer of 3-4 nm overall thickness (Figure
80
5.38).
In order to prepare photoanodes from
RuO2
electrocatalyst, presynthesized
RuO2
was
dissolved in three organic solvents (ethanol, acetone and acetonitrile) and deposited on
n-Si by EPD under cathodic voltage of -10 V for 20 min. Subsequently, the photoanodes
were tested in
H2SO4
0.5M under illumination (AM1.5). The current-voltage behavior of
the photoanodes is shown on the left-hand side of Figure 5.37(a-c). The electrocatalytic
activity was assessed after deposition on FTO (dashed curves) and n-Si (solid curves)
substrates. Red, green and blue curves show the performance of the sample prepared in
ethanol, acetonitrile and acetone, respectively. After deposition on silicon supports, the
performance of the sample prepared in ethanol (red solid curve) was completely dierent
from samples prepared in acetone and acetonitrile. The sample prepared in ethanol under
illumination reaches a saturation current density of 12
mAcm−2
at the redox potential.
But under illumination the photoelectrodes prepared in acetonitrile and acetone show
only a sluggish current increase and are susceptible to degradation within minutes. Since
all the materials and procedures were kept constant and only the organic solvent was
varied, it is assumed that the major dierence of these samples is related to dierent
interfacial regions formed during EPD.
Figure 5.37: (a-c) Current-voltage behavior of ethanol (red curve), acetone (blue curve) and acetonitrile (green curve)
prepared samples. Dashed lines refer to FTO substrates, solid lines to the corresponding behavior for silicon substrates.
(d-f) corresponding XPS analysis of the carbon C 1s signal [202].
81
Figure 5.38: (a) TEM image of Si/Carbon/
RuO2
, (b) SEM image of a particle-free region (resolution 100 nm) and (c)
EDX mapping of a Si/
RuO2
edge [202].
In order to identify the chemical and electronic composition of the respective inter-
faces, XPS analysis was carried out (XPS measurements were done by Ch. Hohn at HZB
and analyses were done under supervision of Dr. M. Lublow). Results are shown on the
right-hand side of Figure 5.37(d-f). Deconvolution of the XPS Ru 2p/ C 1s signal for
ethanol-prepared samples reveals that the interface formed between H- terminated silicon
and the
RuO2
particles shows mainly
sp3
and
sp2
-carbon [202]. It shows furthermore
contributions of C-OH and C=O bonds. In contrast, acetone and acetonitrile-prepared
samples do not show any
sp2
-carbon.
Figure 5.38(a) shows a TEM cross section image of an ethanol-prepared sample. A weak
but visible contrast between the thin (
∼
3-4 nm) very smooth top-surface layer and the
carbon-containing glue (used for preparation for cross-sectional TEM analysis) points to
formation of a homogeneous lm between the silicon support and the
RuO2
particles.
Figure 5.38 (b) shows an SEM image recorded at a tilt angle of the detector with respect
to the surface normal by 30
◦
. It can be seen that some parts of the surface are covered by
RuO2
particles. A cross-sectional sample was prepared and analyzed for three hours by
EDX surface mapping. Upon cutting and breaking of the sample, particles near the edge
came o by tensile stress forces exerted on the support. Thereby, a particle-free region
was exposed. There, and also between the particles, the microscopic probe reaches the
silicon surface and its interface. A clear contrast in both, the carbon and the Si signal
was obtained suggesting a homogeneous organic interfacial region.
5.2.3.1
RuO2
/Si: formation principles
In order to gain insight into the actual structure of this interfacial region, several model
experiments were carried out. The results, to be discussed in the following, suggest the
presence of a Si/
SiO2
junction with an adjacent organic lm encompassing two distin-
guishable sub-layers with dierent electronic and electrochemical properties. According
to the XPS results for sample prepared in ethanol (Figure 5.37 (d)), a scheme of this
entire structure is presented in Figure 5.39. It depicts in (a) the entire device structure
82
comprising the n-Si(100) substrate, an interfacial silicon dioxide layer and the organic
junction containing carbon-
sp3
and carbon-
sp2
sub-layers. Grey spheres indicate
RuO2
nanoparticles at the top as prepared by electrophoretic deposition. In (b), the respective
interfaces and their functions within the device structure are shown. The device and the
interfaces were individually analyzed by the indicated electrochemical reactions: evolu-
tion of oxygen in acidic electrolytes (
H2SO4
0.5 M) and conversion to electrical energy
using an
I−/I−
3
redox couple (section 5.2.3.4). These individual interfaces result from
the combined electrochemical and electrophoretic reactions under cathodic conditions as
identied by the following step-wise analysis [202].
Figure 5.39: Device scheme and interfaces of Si/
SiO2
/Carbon(
sp3
)/Carbon (
sp2
)/
RuO2
.
Step I: Role of iodine
Firstly, an H-terminated n-Si electrode was exposed to pure ethanol for 15 min at -10
V. Secondly, another H-terminated n-Si sample was exposed to a mixture of ethanol and
iodine under the same electrochemical conditions. The photocurrent-voltage behavior of
these samples is shown in Figure 5.40 (a) and (b). The results show that while cathodic
polarization of silicon in ethanol yields a low open circuit photovoltage (140 mV) and
no surface protection in 0.5 M
H2SO4
(fast degradation), the corresponding treatment
in ethanol-iodine results in a high photovoltage of 550 mV and initial surface protection
during anodic cyclovoltammetry. It is thereby demonstrated that the combination of
ethanol with iodine introduces high-quality electronic surface passivation to the silicon
surface, i.e. the surface is defect-free and the surface recombination velocity, the most
crucial parameter for photoelectrode applications, is extremely low. XPS analysis, ob-
tained for the sample prepared in the mixture of ethanol and iodine, is shown in Figure
5.41 for C 1s (a), Si 2p (b) and I 3d (c). Deconvolution of the XPS C 1s signal (Figure
5.41(a)) shows three binding energies at 287.4 eV, 286.8 eV and 285.55 eV. Please note
that carbon Auger analysis was added as inset to serve as comparative reference sig-
nal for corresponding results obtained for samples with deposited
RuO2
. Deconvolution
of the C 1s signal in Figure 5.41(a) suggests the presence of
sp3
-carbon with C-O and
83
C-OH bonds. The Si 2p signal in Figure 5.41(b) shows that only minute amounts of
SiO2
is formed during preparation. Distinctive amounts of I 3d, however, are observed.
Iodine appears in three dierent bonding states. The largest contribution near 621 eV
(3d 5/2) has to be attributed to molecular iodine. Two further levels at 619.5 and near
620 eV are detected to be assigned to reaction of iodine with the organic solvent. It
is therefore assumed that coupling of ethyl groups to the silicon surface occurs during
the electrochemical treatment at -10 V and that this surface conditioning stabilizes the
sample initially under anodic conditions in
H2SO4
at pH0.3. This
sp3
-carbon layer forms
the organic interface directly adjacent to the Si/
SiO2
interface depicted in the scheme of
Figure 5.39(b).
Figure 5.40: Photocurrent-voltage behavior of Si(100) in 0.5 M
H2SO4
after electrochemical surface conditioning in ethanol
and ethanol-iodine, respectively.
Figure 5.41: XPS C 1s, Si 2p and I 3d signals of Si(100) after electrochemical surface conditioning in ethanol-iodine [202].
Figure 5.42 shows UPS analysis of the cut-o region of secondary electrons of H-
terminated (a) and ethanol-iodine-prepared silicon (b), respectively. After etching in HF
(50%), the hydrogen-terminated surface shows two work-function features in the cut-
o region: the higher value of 4.15 eV is indicative for the position of the Fermi-level
of n-doped Si. The low-energy shoulder is assigned to partial surface coverage of the
sample causing a surface dipole with reduced work function energy. After conditioning
in ethanol-iodine, a reverse dipole is observed by ultra-violet photoelectron spectroscopy,
pointing to partial coverage of the surface with a dierent terminal group, decreasing
thereby the silicon work function of 4.3 eV to about 4.1 eV.
84
Figure 5.42: UPS analysis of the cut-o region of secondary electrons of (a) H-terminated silicon and (b) ethanol-iodine-
prepared silicon.
The bonding to ethyl groups to the silicon surface under the applied electrochemi-
cal conditions is conceived as follows: remnant water in the ethanol solvent, introduces
slightly acidic conditions by hydroiodic acid formed due to disproportionation of iodine
into iodide, hypoiodite and protons. Under cathodic polarization (-10 V versus counter
electrode), ethanol molecules are considered to undergo substitution reactions to form
ethyl iodide:
CH3−CH2−OH +HI −→ CH3−CH2−I+H2O
, activating thereby the
molecules for reaction with the H-terminated silicon surface.
Step II: Role of
RuO2
In order to identify the role of
RuO2
upon formation of the organic interface, a correspond-
ing XPS analysis was carried out after EPD-deposition of
RuO2
onto silicon. As an XPS
reference,
RuO2
powder was physically adsorbed from a water solution, i.e. the powder
was neither subject to electrochemical processing nor in contact with an organic solvent.
The results are shown in Figure 5.43(a/b). It should be noted that XPS carbon-analysis
of the surface area in the presence of
RuO2
requires particularly careful deconvolution
of the binding energy range between 270 and 290 eV: in this range, the respective sig-
nals of the Ru 3d and the C 1s level are superimposed on each other. The individual
components are therefore separately depicted in the gure. Furthermore, the complex
surface chemistry of ruthenium oxide gives rise to satellite structures, whose origin is still
debated. In order to facilitate disentanglement of the respective contributions of carbon
and ruthenium, also Auger C KVV and Ru MNN emission signals were measured and
presented as insets in the corresponding gures. These signals appear at distinguishable
energies and allow assessing the relative strength of carbon and ruthenium contributions,
respectively. Analysis of the C 1s core-level signal demonstrates that after electrophoretic
deposition, carbon appears in
sp2
-hybridization (at 284.05 eV in Figure 5.43 (b)). No
corresponding signal is observed for
RuO2
physically adsorbed on silicon (Figure 5.43(a)).
Additionally, carboxyl (C-OH) and carbonyl groups (C=O) are present as detected by the
corresponding signals at 286.85 and 288.0 eV, respectively. The surface of the sample was
covered by about 50% with
RuO2
nanoparticles. UPS measurements in Figure 5.43 show
that the measured work function increases after electrophoretic deposition from about 5
eV (corresponding to the work function of
RuO2
) to 5.7 eV.
85
Figure 5.43: XPS analysis of the Ru 3d and C 1s core-level signals of (a)
RuO2
powder after physical adsorption from
aqueous particle suspensions onto silicon and (b)
RuO2
deposited for 30 min onto silicon from ethanol-iodine containing
suspensions. The C 1s deconvoluted signals and UPS analysis of the cut-o region of secondary electrons of each case are
shown below [202].
Figure 5.44 shows the XPS analysis of the Ru 3d and C 1s core-level signals of Si/
RuO2
after extended deposition for 60 min onto silicon from ethanol-iodine containing suspen-
sions. With longer deposition times, the silicon surface is increasingly covered by
RuO2
(the coverage increases to about 80%). This increase can also be deduced from the re-
spective signal intensities of the carbon and ruthenium Auger emission lines shown as
insets. Despite lower photoelectron emission strength of the C 1s electrons, carbon is
again clearly detected in
sp2
-hybridization at 284.2 eV. The respective ratios of C-H,
C-OH and C=O have slightly changed in comparison to Figure 5.43(b). Furthermore, a
low contribution of metallic ruthenium (about 0.1%) suggests the involvement of the cat-
alyst in reductive electrochemical reactions which lead to the observed
sp2
-rich organic
interface. It is this
sp2
-containing interface that forms the second sub-layer within the
organic lm depicted in Figure 5.39(b).
86
Figure 5.44: XPS analysis of the Ru 3d and C 1s core-level signals and UPS analysis of
RuO2
deposited for 60 min onto
silicon from ethanol-iodine containing suspensions [202].
Figure 5.45: FTIR analysis of the used solutions, ethanol (red), ethanol/iodine after electrochemistry (blue) and
ethanol/iodine/
RuO2
after electrochemistry (green).
As independent further analysis method, FTIR spectra were recorded for the three dif-
ferent solutions after deposition or surface conditioning, respectively. The spectra are
shown in Figure 5.45 for ethanol (red curve), mixture of ethanol/iodine (blue curve)
and mixture of ethanol/iodine/
RuO2
(green curve). Surface conditioning of silicon in
ethanol-iodine containing solutions results in only small changes in comparison with pure
ethanol [202]. The OH-vibration signal near 3400
cm−1
appears slightly broadened and
may give an indication for a dierent OH-containing species in solution, presumably
H2O
produced by ethanol dehydration. After deposition from suspensions of
RuO2
powder in
ethanol-iodine, FTIR proves the presence of C=C double bonds (signal near 1600
cm−1
),
87
CO2
(signal near 2400
cm−1
) and a considerably broadened OH-vibration, indicative for
a strong increase of water concentration. The C=O bonds are thought to result from a
dierent reaction mechanism that may involve intermediate formation of acetaldehyde.
In Figure 5.46 analysis of the I 3d signal is presented for samples prepared in ethanol for
30 min (a), for 60 min (b), acetone (c) and acetonitrile (d). A very low I 3d signal is
detected after deposition from ethanol-iodine solutions. In the case of acetone and ace-
tonitrile, the large signals near 620 and 619 eV indicate high reactivity of iodine towards
the organic solvent molecules.
Figure 5.46: XPS Iodine I 3d analysis after deposition from ethanol- (a/b), acetone- (c) and acetonitrile- (d) iodine
suspensions of
RuO2
.
Figure 5.47 shows the XRD spectrum of the sample prepared in a
RuO2
/ethanol/iodine
mixture. The blue curve shows the XRD spectrum of an as-prepared sample and the red
curve shows the corresponding spectrum after annealing of the sample at 580
◦
C in air.
In the upper image of Figure 5.47, lines to be attributed to the organic interface are
labeled as 1 and 2, respectively. A pronounced peak near 21
◦
is accompanied by a weaker
structure around 44
◦
. The reection pattern at 22
◦
lies between those measured for
polyethylene and doped polyacetylene, i.e. the two potential building blocks of the poly-
meric material [203-204]. After transformation to
CO2
by annealing in air, these lines
are not visible anymore (lower images 5.47). In contrast to (undoped) polyacetylene,
however, the d-spacing is larger and the main XRD line is shifted by about 2
◦
towards
lower angles.
88
Figure 5.47: XRD analysis of Si/
RuO2
, prepared from ethanol-iodine containing suspensions before (blue) and after
annealing (red).
Based on these observations, the mechanism that leads to the formation of the pro-
tection layer is assumed to follow two main reaction routes[202]:
1. in the presence of
RuO2
, dehydration of ethanol proceeds at an elevated rate ac-
cording to FTIR analysis. Coupling of ethyl iodide ions results in a (
CH2)n
network
with substitutions of hydrogen for iodine (
−H2C−HI−CH2−
). The process follows
likely a
RuO2
-surface-mediated pathway as recently reported for ethyl iodides on
Au-surfaces where ethyl iodide bonds to surface oxygen by release of iodine from
C-I bonds [203].
2. Under the applied reducing conditions, formation of an ethoxide is supported, i.e.
ethanol in the presence of
RuO2
and under cathodic conditions forms the anion
CH3−CH2−O−
. It should be noted that this reaction is the cathodic coun-
terpart to alkoxide formation by anodic dissolution of metallic species. Ethox-
ide acts as a strong base which suppresses, by steric hindrance, substitution reac-
tions and favors E2 elimination (dehydrohalogenation) to produce carbon double
bonds within the (
CH2)n
network at sites where hydrogen is substituted by io-
dine:
I−+ −CH2−CHI−−→ −HC =CH − +HI +I−
. Consequently, and in
agreement with XPS analysis, iodine is removed from the network by this process
(Figure 5.46). Finally, formation of acetaldehyde introduces C=O bonds which
connect to the layer as terminal groups. These bonds, and remnant C-OH bonds,
are stabilizing the carbon atoms against electrophilic attack in the acidic solution.
In fact, a dehydrohalogenation process of the alcohol appears essential for carbon-
sp2
formation: corresponding experiments with acetone and acetonitrile did not
show any signicant formation of carbon double bonds (Figure 5.37 C 1s signals)
89
while iodine is detected in high concentrations (Figure 5.46(c/d)). Accordingly,
the resulting photoelectrochemical performance as water splitting photoelectrode
remained insucient in these cases (Figure 5.37(b/c)). In order to account for the
assumed p-type conductivity of the organic interface, terminal oxygen-containing
groups, withdrawing electrons from carbon, may be identied as possible cause.
5.2.3.2 Characterization of the
RuO2
/Si photoanode
The photoelectrochemical performance for light-induced splitting of water at pH0.3 (0.5
M
H2SO4
) is depicted in Figure 5.48 (a). The red curve is related to a surface coverage of
RuO2
by about 80% when prepared from
RuO2
/ethanol/iodine solution. The high open-
circuit photovoltage (400 mV) results in a pronounced shift of the photocurrent onset with
respect to the thermodynamic potential for water splitting (1.23 V RHE). Photocurrent
densities near 10
mAcm−2
are reached at this potential, i.e. without overpotential. Auger
analysis (shown as inset in gure 5.55(a)) shows predominantly ruthenium MNN emission
(near 1210 eV), conrming the high coverage with the catalyst. The carbon signal is low
due to high coverage of
RuO2
on the surface.
Figure 5.48: Operation of the Si/
SiO2
/
RuO2
electrode for photo-assisted water splitting. (a) The photocurrent-voltage
behavior is shown for two dierent distributions of
RuO2
particles. (b) Constant potential measurements in 0.5M
H2SO4
under illumination (AM1.5)[202].
90
In order to increase the light absorption by silicon substrate, a non-polar solvent,
toluene, was added to the solution. The surface morphology thereby changes such that
instead of a homogeneously distribution of
RuO2
particles,
RuO2
clusters are formed
(SEM picture in Figure 5.49). Thereby, larger areas of the silicon substrate are accessible
to illumination by the used white-light source.
These results are also conrmed by XPS analysis of the Ru 3d and C 1s signals of
the sample (Figure 5.50). XPS analysis also conrmed the lower coverage of the surface
by
RuO2
particles. Deconvolution of the XPS Ru 3d, C 1s core-level signals and UPS
measurements of Si/
RuO2
after deposition for 30 min onto silicon from ethanol-toluene-
iodine containing suspensions is shown in Figure 5.50. Comparison of the C 1s signals
measured for samples prepared from ethanol-iodine (Figure 5.43(b)) and from ethanol-
toluene-iodine show that the carbon
sp2:sp3
ratio has decreased. This eect is attributed
to a decrease in dehydrogenation reactions caused by the more scattered distribution
of the catalytic
RuO2
particles (SEM image in Figure 5.49) [202]. The UPS cut-o
region (Figure 5.50)shows similar features as observed after preparation in ethanol-iodine
(without toluene) but the measured work function has decreased to 4.9 eV.
This improved light-management with
RuO2
cluster formation, leads consequently to
a higher open-circuit photovoltage, increased photocurrent densities at the water split-
ting potential and nally to higher saturation photocurrent densities (blue curve in Figure
5.48(a)). It should be noted that the open-circuit photovoltage of 500 mV is close to corre-
sponding value after surface alkylation from ethanol/iodine solutions as described above.
Auger analysis (blue curve in the inset) conrms a larger carbon emission at 1225 eV,
conrming higher exposure of the carbon-covered interface to the electrolyte. SEM sur-
face analysis (images in Figure 5.49) conrms that addition of toluene results in particle
aggregation with an overall more scattered distribution. Thereby, light reaches the silicon
interface more easily. Corresponding IPCE measurements conrm the increased genera-
tion of electron-hole pairs under illumination of the sample prepared in the mixture of
ethanol/toluene/idonine. Figure 5.49 shows the comparative surface and IPCE analysis
of Si/
RuO2
after deposition from ethanol-iodine containing suspensions with (red), with-
out toluene addition (blue) and of bare Si (black). The analysis shows that the sample
with lower coverage reached 40% eciency and with higher coverage only 20%. For com-
parison, the immediate degradation of bare silicon is depicted as black curve. In Figure
5.48(b), nally, the photocurrent behavior during 24 (25) hours of operation at a constant
potential of 1.1 V (RHE) is depicted for ethanol and ethanol-toluene prepared samples.
Both preparation routes result in stable operation during this period of time and maintain
high photocurrent densities. The samples thereby greatly surpass reported stabilities of
organic protection schemes employing chemical surface alkylation or graphene layers.[77]
91
Figure 5.49: Comparative surface and IPCE analysis of Si/
RuO2
after deposition from ethanol-iodine containing sus-
pensions with (red) and without toluene addition (blue). SEM surface analysis (upper images) conrms that addition of
toluene results in particle aggregation with an overall more scattered distribution.
Figure 5.50: XPS analysis of the Ru 3d and C 1s core-level signals of Si/
RuO2
after deposition for 30 min onto silicon
from ethanol-toluene-iodine containing suspensions [202].
It should be noted, however, that further decrease of the
RuO2
surface coverage
increases the risk for photodegradation. This observation suggests that carbon-
sp2
for-
mation requires a sucient spatial distribution of the catalyst and a suciently high
catalytic surface area. Otherwise diusion of reaction intermediates toward the silicon
surface is limited, inhibiting thereby pinhole free growth of the protection layer.
The scheme for the
RuO2
-catalyzed formation of an organic interface is depicted in
gure 5.51. Red circular areas indicate the spatial distribution of reaction intermediates
produced in the presence of the
RuO2
catalyst. Aggregations of particles consequently
result in wider radii of the circular areas. It is assumed that, for successful protection of
the silicon surface, these areas have to overlap in order to achieve a nearly pin-hole free
92
surface layer. The use of ethanol-toluene mixtures allows more scattered distribution of
particle clusters due to the wider range of diusion of reaction intermediates.
Figure 5.51: Scheme for the
RuO2
-catalyzed formation of an organic interface.
Figure 5.52 shows dierential electrochemical mass spectroscopy measurements (DEMS)
for
RuO2
/n-Si under illumination (AM1.5). The blue and black curves show the C-V
behavior and the
O2
mass signals respectively. It is proven that oxygen evolution is
commencing before the redox potential (at 1053 mV vs. RHE). The dierence in the
onset potential in these measurements and gure 5.48 is due to the potential drop across
the membrane in the DEMS setup during the measurement. This membrane is in close
contact to the sample, limiting thereby mass transport to the sample and causing an
additional overpotential.
Figure 5.52: Dierential electrochemical mass spectroscopy (DEMS) measurements for
O2
evolution on
RuO2
/ n-Si
photoanode, electrolyte
H2SO4
0.5 M.
93
5.2.3.3 pH dependent performance of the
RuO2
/n-Si photoanodes
The performance of the
RuO2
/n-Si photoanode was tested in dierent electrolytes, KOH
pH14 (blue curve), Kpi pH7 (green curve) and compared with the current-voltage be-
havior of the photoanode in acidic electrolyte (
H2SO4
pH0.3 red curve). Figure 5.53 (a)
shows the current-voltage behavior of the photoanode under illumination (AM1.5) and
in a three-electrode conguration with Ag/AgCl reference electrode. In KOH the current
density reaches 7
mAcm−2
at the redox potential and in the KPi electrolyte the current
reaches 1.7
mAcm−2
. At pH7, due to the low conductivity of the electrolytes, the resis-
tivity is higher. Therefore, the current increases very slowly. Figure 5.53(b) shows the
corresponding stability assessment at a constant potential for the three pH values. The
photoanode proves stability from neutral to acidic electrolytes. It has a high stability
(25 h) in acidic electrolytes (red curve) and moderate stability for around 5 h at pH7
(green curve). The stability of the sample at pH14 (blue curve) is shown for two cases:
freshly prepared (blue dashed line) and after operation at pH0.3 for several hours (blue
solid curve). As demonstrated by the respective curves, the rate of degradation depends
on the history of the sample: the freshly prepared sample degrades fast upon exposure
to the electrolyte at anodic potentials. A sample pre-conditioned (for 15 hours) in acidic
electrolytes shows short-term stability for about one hour. The greater susceptibility
to degradation in alkaline solutions is attributed to nucleophilic attack of carbon in the
organic layer.
Figure 5.53: (a) current-voltage behavior of
RuO2
/Si photoanode in
H2SO4
(red), Kpi pH7 (green) and KOH pH14
(blue). (b) Constant potential measurements of the photoanodes in the corresponding electrolytes.
5.2.3.4 Photoelectrochemical solar cell application
The entire device structure, as shown in Fig. 5.39, operates eciently for solar-driven
evolution of oxygen.
RuO2
particles withstand mechanical stress forces upon oxygen gas
evolution and cover also after 24 hours of operation the entire silicon surface area. In
contact with an
I−/I−
3
electrolyte, however, the
RuO2
surface layer is easily removed,
exposing thereby the top carbon surface to the solution. This eect provides a unique
opportunity to characterize the electrochemical properties of this lm individually. Fig-
ure 5.54 summarizes the behavior of the heterojunction in contact with the
I−/I−
3
redox
couple after removal of
RuO2
. The incident light excites the semiconductor electrode and
photogenerated electrons and holes are separated in the space charge region. Electrons
94
leave the silicon and travel to the Pt (counter electrode) to reduce
I−
3
and generated
holes travel to the surface of Si and oxidize
I−
to
I−
3
. Here, solar energy is converted to
electrical energy by the process of oxidation of
I−
to
I−
3
[206-207]. Data shown in Fig.
5.54(a) were measured in a two-electrode arrangement vs Pt-counter electrode, and are
indicated with respect to the measured redox potential (0.35 V vs. RHE).
Figure 5.54: Operation of the
Si/SiO2/RuO2
electrode as photoelectrochemical solar cell: In (a), two-electrode mea-
surements in a iodide/triiodide electrolyte are shown using a Pt-counter electrode. The sensitive electrolyte was purged by
nitrogen during both current-voltage cycling and long-time assessment of the short-circuit photocurrent (b).
95
Figure 5.55: SEM surface analysis of the Si photoelectrochemical solar cell.
Upon operation in the iodide/triiodide electrolyte, most of the
RuO2
particles are re-
moved from the surface. Only scattered small aggregations are still visible(Figure 5.55).
The stable and ecient operation of the cell, after removal of most of the
RuO2
parti-
cles, is attributed to the pronounced hole-conductivity of the Carbon(
sp3
)/ Carbon(
sp2
)
junction.
The photocurrent-voltage behavior indicates that also in this case high open-circuit
photovolategs of 500 mV are reached. Since our standard electrochemical cell was used,
a large volume of the (red colored)
I−/I−
3
electrolyte was required, thereby considerably
attenuating the incoming light. Consequently, short-circuit photocurrent densities were
limited to about 20
mAcm−2
. As Figure 5.54(b) shows even in this photovoltaic mode,
extended operation is realized without loss in performance (8 hours) and the results
demonstrate the good conduction properties for holes across the about 3-4 nm thick or-
ganic heterointerface. This good p-type conductivity is attributed to the presence of C=O
and C-OH terminal groups with their electron-withdrawing properties. Eects by iodine
incorporation, as observed for iodine-doped polyethylene, appear of minor importance
due to the very low signal strength of corresponding XPS I 3d signals. The developed
hybrid heterojunction allows therefore for ecient and stable operation in both, the pho-
tovoltaic and the photoelectrocatalytic (water splitting) solar energy conversion mode.
By combining interfacial surface alkylation with adjacent carbon-
sp2
protection layers,
the derived architecture competes successfully with recent inorganic photoelectrodes and
surpasses existing benchmarks of the photovoltage in solar-driven water splitting. The
high photovoltages of up to 500 mV are attributed to an inorganic-organic n-p hetero-
junction where the p-type polymeric protection layer induces a large in-built potential
within the silicon surface region.
96
6 Summary and outlook
1. In the rst part of this work, it has been shown for the rst time that the device-
grade thin lm solar cell concept based on p-type chalcopyrite thin lm absorbers
can be successfully transferred to photoelectrodes for the visible-light driven hy-
drogen evolution. In order to achieve ecient and stable photocathode devices,
new TCO thin lms based on photo-resistant Pt-doped
TiO2
were developed as
substitute of standard and chemically unstable ZnO window layers. Spray ILGAR
deposition was applied to deposit
TiO2
on
Cu(In,Ga)Se2
. For increase of the
conductivity and catalytic activity of the TCO layer, 5 vol.% Pt was added to the
TiO2
precursor solution. The conductivity of the layers was measured by four-
probe-technique and it could be shown that there is a linear correlation between Pt
concentration and conductivity of the layer up to 5% loading. Beyond this value,
the conductivity decreased due to agglomeration of Pt-nanoparticles. In acidic
electrolyte, the most ecient and long-term stable Pt:
TiO2
-
Cu(In,Ga)Se2
het-
erojunction allowed generation of solar hydrogen under visible light illumination by
1 equivalent sun (
∼100 mWcm−2
) with an onset potential of about +300 mV with
respect to the thermodynamic potential for
H2
-evolution (0 V RHE). Furthermore,
the stable photocathode devices proved ecient solar hydrogen evolution with an
IPCE of about 80 % in the complete visible light range. Photocurrent densities of
∼15mAcm−2
at the thermodynamic potential for
H2
-evolution were achieved. High
turn-over-numbers of
∼2.5×106
per active site were calculated. Further device im-
provements are in progress: (i) interface engineering of the Pt:
TiO2
-
Cu(In,Ga)Se2
heterojunction by introducing a thin layer of
TiO2
by ALD in order to avoid forma-
tion of deleterious metal-semiconductor junctions, (ii) doping of the
TiO2
thin lms
with earth-abundant elements such as V and Cr as substitutes of cost-extensive Pt,
and (iii) the utilization of chalcopyrite thin lms with higher band gap energies
such as
CuInS2
and
CuGaSe2
in order to increase the built-in driving potential
upon contact formation with
TiO2
. [183]
2. In the second part of this work, electrophoretic deposition was introduced as a
promising method for deposition of presynthesized catalysts on conductive sub-
strates. Fundamental research was performed to elucidate the eect of dierent
organic solvents, used upon EPD, on the device performance. A systematic study
was presented to correlate the deposition rates and lm structure properties of
ZnO:Co and
RuO2
thin lms with the electrophoretic mobility and the Zeta po-
tential in dierent organic solvents. The electrocatalytic activity of these particles
on FTO supports was shown to depend on the size of the active surface area of
the resulting lms. It could be shown that the use of acetonitrile results in a sur-
face morphology with highest active surface area. Consequently, the overpotentials
for evolution of oxygen (OER), observed at 1
mAcm−2
, could be reduced from
360 mV to 290 mV. Combining ZnO:Co electrocatalysts and low-cost photovoltaic
triple-junctions (a-Si), a proof-of-concept could be successfully demonstrated for an
ecient modular architecture for generation of solar hydrogen (up to 5% STH and
long-term stability for more than 25 hours). [196]
3.
RuO2
as a well-known catalyst for OER was tested instead of ZnO:Co in order to
prepare a stable photoanode for acidic electrolytes. An activity-solvent correlation
carried out for
RuO2
/FTO proved equivalent results as in the case of ZnO:Co/FTO.
97
Highest activity in acetonitrile was observed due to the higher surface area while
ethanol-prepared samples showed lowest activity, i.e. highest overpotentials. Using
n-type silicon as deposition supports, however, the role of the organic solvent greatly
diered from corresponding results obtained for ZnO:Co. Samples prepared in ace-
tonitrile and acetone proved to be unstable and showed no photoactivity. Samples
prepared in ethanol, in turn, proved unprecedented high photovoltages of 500-550
mV, high saturation photocurrent densities and long-term stability. TEM cross
section analysis revealed the formation of an ultra-thin polymeric layer of 3-4 nm
thickness which guaranteed the stable operation of the oxidation-sensitive silicon
support. XPS analysis, FTIR and Grazing-Incidence X-ray Diraction investiga-
tions were carried out in order to conceive the generation of polymeric protection
layers from alcohols in the presence of
RuO2
. A complex reaction scheme could
be thereby deduced, involving dehydration of the alcohol, formation of terminal
aldehyde groups and
sp3−sp2
transformations under electrochemical conditions.
The photoanode showed high stability in acidic electrolyte for more than 24 hours.
Addition of toluene to the precursor solution nally allowed for improvement of the
light-absorbing properties, and photocurrent densities at the redox potential could
be increased to 12
mAcm−2
. The prospect of future developments is appealing:
new reaction schemes, to be developed, may allow to more eciently synthesizing
this novel type of polymer. Further fundamental investigations are expected to pro-
vide schemes for doping and for stabilizing of the material also in alkaline media.
[202]
98
7 Appendices
7.1 References
[1] Statistical Review of World Energy 2013, http://www.bp.com / statisticalreview
[2] D. G. Nocera, ChemSusChem 2009, 2, 387-390.
[3] R. van de Krol, Photoelectrochemical measurements,in Photoelectrochemical Hydro-
gen Production (R. van de Krol and M. Grätzel), vol. 102 of Electronic Materials:
Science & Technology, ch. 3, pp. 69, 117, Springer US,2012.
[4] J. Petrasch, J. Klaisner, WIREs Energy and Environment 2012, 1, 347-361.
[5] I. Dincer, A. S. Joshi, Solar based hydrogen production systems Springer 2013.
[6] S. Z. Baykara, E. Bilgen, Hydrogen energy 1989, 14, 881-891.
[7] R. H. Crabtree, Energy Production and Storage: Inorganic Chemical Strategies for a
Warming World; 2010.
[8] K. Maeda, K. Domen, J. Phys. Chem. C 2007, 111(22), 7851.
[9] M. Bledowski,Visble light-driven photooxidation of water at hybride photoanodes,
2013.
[10] A. Fujishima and K. Honda, B Chem. Soc. Jpn. 1971, 44, 1148- 1150.
[11] A. Fujishima and K. Honda, Nature 1972, 238, 37.
[12] A. Heller, Acc. Chem. Res. 1981, 14, 154-162.
[13] H. J. Lewerenz, C. Heine, K. Skorupska, N. Szabo, T. Hannappel, T. Vo-Dinh, S. A.
Campbell, H. W. Klemm, A. G. Munoz, Energy Environ. Sci. 2010, 3, 748-760.
[14] O. Khaselev, A. Bansal and J. A. Turner, Int J Hydrogen Energ, 2001, 26, 127-132.
[15] O. Khaselev and J. A. Turner, Science, 1998, 280, 425-427.
[16] S. Licht, B. Wang, S. Mukerji, T. Soga, M. Umeno and H. Tributsch, J. Phys. Chem.
B, 2000, 104(38), 8920-8924.
[17] M. M. May, H. J. Lewerenz, D. Lackner, F. Dimroth, T. Hannappel, Nature Com-
munications 2015, 6, 8286.
[18] S. Y. Reece, J. A. Hamel, K. Sung, T. D. Jarvi1, A. J. Esswein, J. J. H. Pijpers, D.
G. Nocera, Science 2011, Vol. 334 no. 6056 pp. 645-648.
[19] F. F. Abdi, L. Han, A. H. M. Smets, M. Zeman, B. Dam, R. van de Krol, Nature
Communications 2013, 4, 2195.
[20] E. L. Miller, B. Marsen, D. Paluselli, R. Rocheleau, Electrochemical and Solid-State
Letters 2005, 8, A247-A249.
[21] K. Sivula, F. L. Formal, M. Grätzel, ChemSusChem 2011, 4, 432-449.
[22] H. Kaneko, T. Minegishi, K. Domen, Coatings 2015, 5, 293-311.
[23] R. Klenk, J. Klaer, R. Scheer, M. C. Lux-Steiner, I. Luck, N. Meyer, U. Ruhle, Thin
Solid Films 2005, 480, 509.
[24] J. Haarstrich, H. Metzner, M. Oertel, C. Ronning, T. Rissom, C. A. Kaufmann, T.
Unold, H. W. Schock, J. Windeln, W. Mannstadt, E. Rudigier-Voigt, Sol. Energy Mater.
Sol. Cells 2011, 95, 1028-1030.
[25] R. Caballero, C.A. Kaufamnn, V. Emova, T. Rissom, V. Homann, H. W. Schock,
Prog. in Photovolt: Res. Appl. 2013, 21, 30-46.
[26] M. Rusu, S. Wiesner, R. Wurz, S. Lehmann, S. Doka-Yamigno, A. Meeder, D. F.
Marron, M. Bar, V. Koteski, H. E. Mahnke, E. Arushanov, J. Beckmann, K. Hohn, W.
Fritsch, W. Bohne, P. Schubert-Bischo, M. Heuken, A. Jager-Waldau, A. Rumberg, Th.
Schedel-Niedrig, Sol. Energy Mater. Sol. Cells 2011, 95, 1555-1580.
99
[27] P. Jackson, D. Hariskos, E. Lotter, S. Paetel, R. Wuerz, R. Menner, W. Wischmann,
M. Powalla, Prog. in Photovolt: Res. Appl. 2011, 19, 894-897.
[28] F. Yang, M. Lublow, S. Orthmann, C. Merschjann, T. Tyborski, M. Rusu, S. Kubala,
A. Thomas, R. Arrigo, M. Hävecker, T. Schedel-Niedrig, ChemSusChem 2012, 5, 1227.
[29] F. Yang, V. Kuznietsov, M. Lublow, C. Merschjann, A. Steigert, J. Klaer, A. Thomas,
Th. Schedel-Niedrig, J. Mater. Chem. A 2013, 1, 6407.
[30] D. Yokoyama, T. Minegishi, K. Maeda, M. Katayama, J. Kubota, A. Yamada, M.
Konagai, K. Domen, Electrochem. Commun. 2010, 12, 6, 851-853.
[31] B. Neumann, P. Bogdano, H. Tributsch, J. Phys. Chem.C 2009, 113(49), 20980-
20989.
[32] T. J. Jacobsson, C Platzer-Björkman, M. Edo, T. Edvinsson, Int. J. Hyd. Ener.
2013, 38, 15027-15035.
[33] M. W. Kanan, D. G. Nocera, Science 2008, 321, 1072-1075.
[34] Y. Surendranath, M. Dinca, D. G. Nocera, J. Am. Chem. Soc. 2009, 131, 2615-2620.
[35] N. D. Morris, M. Suzuki, T. E. Mallouk, J. Phys. Chem. A 2004, 108, 9115-9119.
[36] P. G. Hoertz, Y. Kim, W. J. Youngblood,T. E. Mallouk, J. Phys. Chem. B 2007,
111, 6845-6856.
[37] W. J. Youngblood, S.- H. A. Lee, Y. Kobayashi, E. A. Hernandez-Pagan, P. G.
Hoertz, T. A. Moore, A. L. Moore, D. Gust, T. E. Mallouk, J. Am. Chem. Soc. 2009,
131, 926-927.
[38] B. O`Regan, M. Gräetzel, Nat. 1991, 353, 737-740.
[39] H. Dau, C. Limberg, T. Reier, M. Risch, S. Roggan, P. Strasser, ChemCatChem
2010, 2, 724-761.
[40] H. Dau, I. Zaharieva, Acc. Chem. Res. 2009, 42, 1861-1870.
[41] H. Inoue, T. Shimada, Y. Kou, Y. Nabetani, D. Masui, S. Takagi, H. Tachibana,
ChemSusChem 2011, 4, 173-179.
[42] C. C. L. McCrory, S. Jung, I. M. Ferrer, S. M. Chatman, J. C. Peters, T. F. Jaramillo,
J. Am. Chem. Soc. 2015, 137, 4347-4357.
[43] S. Trasatti, G. Lodi, "in electrodes of conductive metallic oxides" Elsevier 1981,
521-626.
[44] M. Risch, A. Grimaud, K. J. May, K. A. Stoerzinger, T. J. Chen, A. N. Mansour, Y.
Shao-Horn, J. Phy. Chem. C 2013, 117, 8628-8635.
[45] A. Grimaud, C. E. Carlton, M. Risch, W. T. Hong, K. J. May, Y. Shao-Horn, J. Phy.
Chem. C 2014, 117, 25926-25932.
[46] S. W. Lee, C. Carlton, M. Risch, Y. Surendranath, S. Chen, S. Furutsuki, A. Yamada,
D. G. Nocera, Y. Shao-Horn, J. Am. Chem. Soc. 2012, 134, 16959-16962.
[47] J. Pfrommer, M. Lublow, A. Azapira, C. Göbel, M. Lücke, A. Steigert, M. Pogrzeba,
P. W. Menezes, A. Fischer, T. Schedel-Niedrig, M. Driess, Angew. Chem. Int. 2014, 53,
5183-5187.
[48] D. M. Jang, I. H. Kwak, E. L. Kwon, C. S. Jung, H. S. Im, K. Park, J. Park, J. Phys.
Chem. C 2015, 119, 1921-1927.
[49] L. Besra, M. Liu, Materials Science 2007, 52, 1-61.
[50] M. Ammam, RSC Advances 2012, 2, 7633-7646.
[51] F. Begona, M. Rodrigo, J. Eur. Ceram. Soc. 2010, 30, 1069-1078.
[52] I. Corni, M. P. Ryan, A. R. Boccaccini, J. Eur. Ceram. Soc. 2008, 28, 1353-1367.
[53] O. Van der Biest, L. J. Vandeperre, Annu. Rev. Mater. Sci. 1999, 29, 327-352.
[54] J. Tsukada, H. Zanin, L. C. A. Barbosa, G. A. da Silva, H. J. Ceragioli, A. C.
100
Peterlevitz, R. F. Teo'lo, V. Baranauskasa, J. Electro chem. Soc. 2012, 159, D159-
D161.
[55] H. Wang, M. R. Shen, Z. Y. Ning, C. Ye, C. B. Cao, H. Y. Dang, H.S. Zhu, Appl.
Phys. Lett. 1996, 69, 1074-1077.
[56] A. H. C. Sirk, D. R. Sadoway, J. Electrochem. Soc. 2008, 155, E49-E55. [57] Y.
Namba, J. Vac. Sci. Technol. A 1992, 10, 3368-3370.
[58] B. Seger, S. D. Tilley, Th. Pedersen, P. C. K. Vesborg, O. Hansen, M. Grätzel, I.
Chorkendor, J. Mater. Chem. A, 2013,1, 15089-15094.
[59] J. E. Thorne, S. Li, C. Du, G. Qin, D. Wang, J. Phys. Chem. Lett. 2015, 6,
4083-4088.
[60] S. Avasthi, Y. Qi, G. K. Vertelov, J. Schwartz, A. Kahn, J. C. Sturm, Appl. Phys.
Lett. 2010, 96, 222109-222111.
[61] R. Har-Lavan, O. Yae, P. Joshi, R. Kazaz, H. Cohen, D. Cahen, AIP Advances
2012, 2, 012164 .
[62] N.A. Allsop, A. Schönmann, A. Belaidi, H.-J. Muer, B. Mertesacker, W. Bohne, E.
Strub, J. Röhrich, M. C. Lux-Steiner, Ch.-H. Fischer, Thin Solid Films 2006, 513, 52-56.
[63] Ch. H. Fischer, N.A. Allsop, S.E. Gledhill, T. Köhler, M. Krüger, R. Sa'ez-Araoz,
Y. Fu, R. Schwieger, J. Richter, P. Wohlfart, P. Bartsch, N. Lichtenberg, M. Ch. Lux-
Steiner; Sol. Energy Mater. Sol. Cells 2011, 95, 1518-1526.
[64] X. Liao, X. Zhang, K. Takai, T. Enoki, J. Appl. Phys. 2010, 107, 013709.
[65] M. Ni, M. K. H. Leung, D. Y. C. Leung, K. Sumathy, Renewable and Sustainable
Energy Reviews 2007, 11, 401-425.
[66] S. Licht, , B. Wang, S. Mukerji, T. Soga, M. Umeno, H. Tributsch, Int. J. Hydrogen
Energy, 2001, 26, 653-659.
[67] G. Peharz, F. Dimroth and U. Wittstadt, Int. J. Hydrogen Energy, 2007, 32, 3248-
3252.
[68] S. Hu, Ch. Xiang, S. Haussener, A. D. Bergercd, N. S. Lewis, Energy Environ. Sci.,
2013, 6, 2984-2993.
[69] R. van de Krol, Y. Liang, J. Schoonman, J. Mater. Chem., 2008, 18, 2311-2320.
[70] P. Hillebrand, Structure-function relationships of cobalt and manganese oxides as
catalysts for (photo-) electrochemical water splitting, TU-Berlin 2015.
[71] W. Shockley, H. J. Queisser, J. Appl. Phys. 1961, 32, 510-519.
[72] K. S. Joy, y. F. Joya, K. Ocakoglu, R. V. de Krol, Angew. Chem. Int. Ed., 2013,
52, 10426-10437.
[73] J. Luo, J. H. Im, M. T. Mayer, M. Schreier, M. K. Nazeeruddin, N. G. Park, S. D.
Tilley, H. J. Fan, M. Grätzel, Science, 2014, 345, 1593-1596.
[74] K. S. Joya, H. J. M. de Groot, ChemSusChem, 2014, 7, 73-76.
[75] R. Cao, W. Laia, P. Du, Energy Environ. Sci. 2012, 5, 8134-8157.
[76] M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. X. Mi, E. A. Santori,
N. S. Lewis, Chem. Rev. 2010, 110, 6446-6473.
[77] A. C. Nielander, M. J. Bierman, N. Petrone, N. C. Strandwitz, S. Ardo, F. Yang, J.
Hone, N. S. Lewis, J. Am. Chem. Soc. 2013, 135, 17246-17249.
[78] S. Hu, M. R. Shaner, J. A. Beardslee, M. Lichterman, B. S. Brunschwig, N. S. Lewis,
Science 2014, 344, 1005-1009.
[79] D. G. Nocera, Acc. Chem. Res. 2012, 45, 767-776.
[80] B. Mei, B. Seger, T. Pedersen, M. Malizia, O. Hansen, I. Chorkendor, P. C. K.
Vesborg, J. Phys. Chem. Lett. 2014 5, 1948-1952.
101
[81] A. G. Scheuermann, J. D. Prange, M. Gunji, C. E. D. Chidsey, P. C. McIntyre,
Energy Environ. Sci 2013, 6, 2487-2496.
[82] T. P. Luxton, M. J. Eick, K. G. Scheckel, J. Coll. Int. Sci. 2011, 359, 30-39.
[83] J. R. McKone, R. A. Potash, F. J. DiSalvoa , H. D. Abruna, Phys. Chem. Chem.
Phys. 2015, 17, 13984-13991.
[84] M. Nose, T. Kinumoto, H. S. Choo, K. Miyazaki, T. Abe, Z. Ogumi, FUEL CELLS,
2009 9, 284-290.
[85] M. Schriver, W. Regan, W. J. Gannett, A. M. Zaniewski, M. F. Crommie, A. Zettl,
ACS Nano 2013, 7, 5763-5768.
[86] J. Yang, D. Wang, H. Han, C. Li, Accounts of Chemical Research 2013, 1900-1909.
[87] H. Dotan, N. Mathews, T. Hisatomi, M. Grätzel, A. Rothschild; J. Phys. Chem.
Lett. 2014, 5, 3330-3334.
[88] F. Lin, S.W. Boettcher, Nature material 2014, 13, 81-86.
[89] C. C. L. McCrory, S. Jung, J. C. Peters, Th. F. Jaramillo, J. Am. Chem. Soc. 2013,
135, 16977-16987.
[90] M. Grätzel Acc. Chem. Res. 1981, 14, 376-384.
[91] U. Kölle, New J. Chem. 1992, 16, 157-169.
[92] A. J. Bard, M. A. Fox, Acc. Chem. Res. 1995, 28, 141-145.
[93] J. A. Turner, Science 2004, 305, 972-974.
[94] N. S. Lewis, D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A. 2006,103, 15729-15735.
[95] N. S. Lewis, Science 2007, 315, 798-801.
[96] G. W. Crabtree, M. S. Dresselhaus, MRS Bull. 2008, 33, 421-428.
[97] H. B. Gray, Nat. Chem. 2009, 1, 7-8.
[98] T. R. Cook, D. K. Dogutan, S. Y. Reece, Y. Surendranath, T. S. Teets, D. G. Nocera,
Chem. Rev. 2010, 110, 6474-6502.
[99] D. J. Walton, L. D. Burke, M. M. Murphy, Electrochim. Acta 1996, 41, 2747-2751.
[100] E. Navarro-Flores, Z. Chong, S. Omanovic, J. Mol. Catal. A: Chem. 2005, 226,
179-197.
[101] C. Fan, D. L. Piron, P. Paradis, Electrochim. Acta 1994, 39, 2715-2722.
[102] N. V. Krstajic', V. D. Jovic', L. Gajic'-Krstajic', B. M. Jovic', A. L. Antozzi, G. N.
Martelli, Int. J. Hydrogen Energy 2008, 33, 3676-3687.
[103] J. M. Spurgeon, J. M. Velazquez, M. T. McDowell, Phys. Chem. Chem. Phys.
2014, 16, 3623-3631.
[104] E. Tsuji, A. Imanishi, K.-i. Fukui, Y. Nakato, Electrochim. Acta 2011, 56, 2009-
2016.
[105] I. C. Man, H. Y. Su, F. Calle-Vallejo, H. A. Hansen, J. I. Martinez, N. G. Inoglu,
J. Kitchin, T. F. Jaramillo, J. K. Nørskov, J. Rossmeisl, J. ChemCatChem 2011, 3,
1159-1165.
[106] J. Rossmeisl, A. Logadottir, J. K. Nørskov, J. Chem. Phys. 2005, 319, 178-184.
[107] J. Rossmeisl, Z. W. Qu, H. Zhu, G. J. Kroes, J. K. Nørskov, J. Electroanal. Chem.
2007, 607, 83-89.
[108] M. T. M. Koper, J. Electroanal. Chem. 2011, 660, 254-260.
[109] M. Jayalakshmi, W. Y. Kim, K. D. Jung, O. S. Joo, Int. J. Electrochem. Sci. 2008,
3, 908-917.
[110] S. M. Sze, K. K. Ng, Physics of Semiconductor Devices, John Wiley & Sons Inc.
1969.
[111] K. Rajeshwar,Fundamentals of Semiconductor Electrochemistry and Photo-electrochemistry
Wiley-VCH 2007
102
[112] J. R. Mott, P. J. Munson, R. A. Kreuter, B. S. Chohan, D. G. Sykes, J. Chem.
Educ., 2014, 91 (7), pp 1028-1036.
[113] B.V. Van Zeghbroeck, Principles of Semiconductor Devices and Heterojunctions
2010.
[114] A. W. Bott,Electrochemistry of semiconductor 1998.
[115] A. Chiril'a, S. Buecheler, F. Pianezzi, P. Bloesch, Ch.Gretener, A. R. Uhl, C. Fella,
L. Kranz, J.Perrenoud, S. Seyrling, R. Verma, S. Nishiwaki, Y. E. Romanyuk, G. Bilger
, A. N. Tiwari, Nature Materials 2011, 10, 857-861.
[116] P. Jackson, D. Hariskos, R. Wuerz, O. Kiowski, A. Bauer, Th. M. Friedlmeier, M.
Powalla, Phys. Status Solidi RRL 2015, 9, 28-31.
[117] C. A. Kaufmann, R. Caballero, T. Unold, R. Hesse, R. Klenk, S. Schorr, M. Nichter-
witz, H. W. Schock, Solar Energy Materials & Solar Cells 2009, 93, 859-863.
[118] A. M. Gabor, J. R. Tuttle, M. H. Bode, A. Franz, A. L. Tennant, M. A. Contreras,
R. Nou, D. G. Jensen, A. M. Hermann, Sol. Energy Mater. Sol. Cells 1996, 41, 247-260.
[119] R.Nou,Y.Yan,J.Abu-Shama,K.Jones, M.Al-Jassim,B.Keyes, J. Alleman, R. Ra-
manathan, Proc.
29th
IEEE PVSC,NewOrleans, 2002, NREL/CP-520-31430.
[120] C. A. Kaufmann, A. Neisser, R. Klenk, R. Scheer, Thin Solid Films 2005, 480,
515-519.
[121] R. Hesse, J. R. K. Kamdoum, R. Caballero, Ch. A. Kaufmann, Ch. Stephan,
S. Schorr, D. Abou-Ras, Th. Unold, H. W. Schock, Phys. Status Solidi C 2009, 6,
1249-1252.
[122] X. Wang, G. Liu, G. Q. Lu, H. M. Cheng, J. of hydrogen energy 2010, 35, 8199-8205.
[123] A. A. Keller, H. Wang, D. Zhou, H. S. Lenihan, G. Cherr, B. J. Cardinale, R. Miller,
Z. Ji, Environ. Sci. Technol. 2010, 44 (6), pp 1962-1967.
[124] K. Hashimoto, H. Irie, A. Fujishima, Japanese Journal of Applied Physics 2005, 44,
8269-8285.
[125] D. O. Scanlon, Ch. W. Dunnill, J. Buckeridge, S. A. Shevlin, A. J. Logsdail, S. M.
Woodley, C. R. A. Catlow, M. J. Powell, R. G. Palgrave, I. P. Parkin, G. W. Watson,
Th. W. Keal, P. Sherwood, A. Walsh, A. A. Sokol, Nature Materials 2013, 12, 798-801.
[126] Ch. H. Liao, Ch. W. Huang, J. C. S. Wu, Catalysts 2012, 2, 490-516.
[127] A. B. Murphy, P. R. F. Barnes, L. K. Randeniya, I. C. Plumb, I. E. Grey, M. D.
Horne, J. A. Glasscock, Int. J. Hydrogen Energy 2006, 31, 1999-2017.
[128] A. Zaleska, Recent Patents on Engineering 2008, 2, 157-164.
[129] M. Anpo, Pure & Appl. Chem 2000, 72, 1787-1792.
[130] A. Fuerte, M. D. Hernàndez-Alonso, A. J. Maira, A. Martìnez-Arias, M. Fernàndez-
Garcìa, J. C. Conesaa , J. Soria, Chem. Commun. 2001, 24, 2718-2719.
[131] H . Yamashita, M. Harada, J. Misaka, J. Synchrotron Rad 2001, 8, 569-571.
[132] T. Ihara, M. Miyoshi, M. Ando, S. Sugihara, Y. Iriyama, J. Mater. Sci. 2001,
4201-4207.
[133] K. Takeuchi, I. Nakamura, O. Matsumoto, S. Sugihara, M. Ando, T. Ihara, Chem.
Lett. 2000, 29, 1354-1355.
[134] T. Ohno, T. Mitsui, M. Matsumura, Chem Lett 2003, 32, 364-365.
[135] Y. Liu, X. Chen, J. Li, C. Burda, J. Chemosphere 2005, 61, 11-18.
[136] J. C. Yu, L. Zhang, Z. Zheng, J. Zhao, J. Chem Mater 2003, 15, 2280 -2286.
[137] C. Sotelo-Vazquez, N. Noor, A. Kazas, R. Quesada-Cabrera, D. O. Scanlon, A.
Taylor, J. R. Durrant, I. P. Parkin, J. Chem. Mater. 2015, 27 (9), 3234-3242.
[138] T. Hirai, K. Suzuki, I. Komasawa, J. Colloid Interface Sci. 2001, 244, 262-265.
[139] D. Chatterjee, A. Mahata, Appl. Catal. B Environ. 2001, 33, 119-125.
103
[140] J. Wang, T. Ma, G. Zhang, Catal Commun 2007, 8, 607-611.
[141] W. Zhou, Y. Zheng, G. Wu, Appl. Surf. Sci. 2006, 253, 1387-1392.
[142] C. Kittel, Introduction to Solid State Physics, John Wiley & Sons,ISBN : 978-0-
471-41526-8
[143] YH. Lai, HS. Park , JZ. Zhang , PD. Matthews , DS. Wright , E. Reisner, Chemistry.
2015 Mar 2; 21(10): 3919-3923.
[144] Y. W. Chen, J. D. Prange, S. Dühnen, Y. Park, M. Gunji, Ch. E. D. Chidsey, P.
C. McIntyre, Nature Materials 2011, 10,539-544.
[145] Zh. Xia, X. Zhou , J. Li ,Y. Qu, Sci. Bull. (2015) 60(16):1395-1402.
[146] R. E. Rocheleau, E. L. Miller, A. Misra, Energy & Fuels 1998, 12, 3-10.
[147] F.R.F. Fan, R.G. Keil, A.J. Bard, J. Am. Chem. Soc. 1983, 105, 220-224.
[148] F. R. F. Fan, G. A. Hope, A. J. Bard, J. Electrochem. Soc. 1982, 129, 1647-1649.
[149] J. J. H. Pijpers, M. T. Winkler, Y. Surendranath, T. Buonassisic, D. G. Nocera,
Proc. Natl. Acad. Sci. USA 2011, 108, 10056-10061.
[150] E. R. Young, R. Costi, S. Paydavosi, Energy Environ. Sci. 2011, 4, 2058-2061.
[151] X. Li, W. Lu, W. Dong, Nanoscale 2013, 5, 5257-5261.
[152] W. Cui, Zh. Xia, Sh. Wu, F. Chen, Y. Li, B. Sun, ACS Appl Mater Interfaces
2015, ACS Appl. Mater. Interfacesdoi:10.1021/acsami. 5b01393.
[153] S. Polarz, A. V. Orlov, M. W. E. van den Berg, M. Driess, Angew. Chem. Int.
2005, 44, 7892-7896.
[154] M. L. Tong, S. L. Zheng, J. X. Shi, Y. X. Tong, H. K. Lee, X. M. Chen, Journal of
the Chemical Society, Dalton Transactions 2002, 1727-1734.
[155] J. C. Cruz, V. Baglio, S. Siracusano, V. Antonucci, A. S. Aricò, R. Ornelas, L.
Ortiz-Frade, Int. J. Electrochem. Sci. 2011, 6, 6607 - 6619.
[156] Deutsches Patent für ILGAR-Oxid-Schichten: Nr. 198 16 403.7 für ILGAR-Oxid-
Schichten
[157] H.-J. Muer, C.-H. Fischer, K. Diesner, M.C. Lux-Steiner, Solar Energy Materials
and Solar Cells, 67 (1-4), 121-127.
[158] P. S. Patil, Materials Chemistry and Physics 1999, 59, 185-198.
[159] J. J. Van Tassel, C. A. Randall, Key Engineering Materials 2006, 314 , 167-174.
[160] H. C. Hamaker, Trans Faraday Soc. 1940, 36, 279-87.
[161] T. Ishihara, K. Shimise, T. Kudo, H. Nishiguchi, T. Akbay, Y. Takita, J. Am.
Ceram. Soc. 2000, 83(8), 1921-1927.
[162] F. Chen, M. Liu, J Eur Ceram. Soc. 2001, 21, 127-134.
[163] M. von Smoluchowski, URN: urn:nbn:de:hebis:30-84830 .
[164] B. M. Haines, A. L. Mazzucato, arXiv.org 2011 arXiv:1104.1102.
[165] D. Andrienko Cyclic Voltammetry 2008.
[166] A. P. Singh, S. Kumari, R. Shrivastav, S. Dass and V. R. Satsangi, J. Phys. D:
Appl. Phys., 2009, 42, 085303-085308.
[167] S. Ashton, Springer Theses (2012), ISBN 978-3-642-30550-4.
[168] P. Bogdano, N. Alonso-Vante, J. Electroanal. Chem.1994, 379, 415-421.
[169] T. G. Mendes-de-Sa, A. M. B. Goncalves, M. J. S. Matos, P. M. Coelho, R.
Magalhaes-Paniago, R. G. Lacerda, Nanotechnology 2012, 23, 475602-475610.
[170] Y. Waseda, E. Matsubara and K. Shinoda, X-ray diraction crystallography,
Springer, 201, DOI 10.1007/978-3-642-16635-8.
[171] Y. Li, D. Lu, C.P. Wong, Electrical conductive adhesives with nanotechnolgies,
Springer 2010.
[172] L. Scudiero, X-Ray PhotoelectronSpectroscopy (XPS)-2.
104
[173] T. Haile, Photoelectron Spectroscopy , XPS & UPS, Addis Ababa University, 2014.
[174]http://www.gla.ac.uk/schools/ges/research/researchfacilities/isaac/ services/scanningelectronmicroscopy/
[175] Herzog, Ultra-thin inorganic nanocomposites for photovoltaic applications 2008.
[176] L. Kronik, Y. Shapira, Surface Science reports 1999, 37, 1-206.
[177] B. Goldstein and D. Szostk, Surface Science, 1980, 99, 235-258.
[178] J. Lagowski, C. Balestra, H. Gatos, Surface Science 1971, 27, 547-558.
[179] T. Dittrich, S. Bonisch, P. Zabel and S. Dube, Rev. Sci. Instrum. 2008, 79, 113903.
[180] A. O. Juma, PhD thesis Copper diusion in
In2S3
and charged separation at
In2S3
/CuSCN and
TiO2
/
In2S3
interfaces FU Berlin 2013.
[181] J. Domenech, A. Prieto, J. Phys. Chem. 1986, 90 (6), 1123-1126.
[182] F. Mesa, W. Chamorro, W. Vallejo, R. Baier, Th. Dittrich, A. Grimm, M. C.
Lux-Steiner, S. Sadewasser, J. Nanotechnology 2012, 3, 277-284.
[183] A. Azarpira , M. Lublow , A. Steigert , P. Bogdano , D. Greiner , Ch. A. Kaufmann
, M. Krüger , U. Gernert , R. van de Krol , A. Fischer, Th. Schedel-Niedrig, Adv. Energy
Mater. 2015, 1402148.
[184] R. Caballero, Ch. A. Kaufmann, T. Eisenbarth, Th. Unold, R. Klenk, H. W.
Schock, Prog. Photovolt: Res. Appl. 2011, 19, 547-551.
[185] J. Choi, H. Park, M. R. Homann, J. Phys. Chem. C 2010, 114, 783-792.
[186] Due to the deposition temperature used for
Cu(In,Ga)Se2
deposition, it is ex-
pected, that a strong in-depth double Ga-gradient is present within the absorber layer.
Hence for the cut-o energy correlates rather to the optical band gap of the
Cu(In,Ga)Se2
near the
TiO2
/
Cu(In,Ga)Se2
interface.
[187] S. W. Boettcher, E. L. Warren, M. C. Putnam, E. A. Santori, D. Turner-Evans,
M. D. Kelzenberg, M. G. Walter, J. R. McKone, B. S. Brunschwig, H. A. Atwater, N. S.
Lewis, J. Am. Chem. Soc. 2011, 133, 1216-1219.
[188] A. G. Muñoz, C. Heine, M. Lublow, H. W. Klemm, N. Szabò, T. Hannappel, H. J.
Lewerenz, ECS Journal of Solid State Science and Technology, 2013, 2 (4) Q51-Q58.
[189] R. C. Kainthla, B. Zelenay, J. O'M. Bockris, J. Electrochem. Soc., 1987, 134,
841-845.
[190] M. J. Kenney, M. Gong, Y. Li, J. Z. Wu, J. Feng, M. Lanza, H. Dai, Science 2013,
342, 836-840.
[191] K. Sun, N. Park, Zh. Sun, J. Zhou, J. Wang, X. Pang, Sh. Shen, S. Y. Noh, Y.
Jing, S. Jin, P.K. L. Yua, D. Wang, Energy Environ. Sci. 2012, 5, 7872-7877.
[192] H. J. Lewerenz, J. Electroanal. Chem. 1993, 356, 121-143.
[193] E. J. Nemanick, P. T. Hurley, L. J. Webb, D. W. Knapp, D. J. Michalak, B. S.
Brunschwig, N. S. Lewis, J. Phys. Chem., 2006, 110, 14770-14778 .
[194] L. J. Webb, N. S. Lewis, J. Phys. Chem., 2003, 107, 5404-5412.
[195] K. T. Wong, N. S. Lewis, Acc. Chem. Res. 2014, 47, 3037-3044.
[196] A. Azarpira, J. Pfrommer, K. Olech, Ch. Höhn, M. Driess, B. Stannowski, Th.
Schedel-Niedriga and M. Lublow, J. Mater. Chem. A DOI: 10.1039/c5ta07329d.
[197] A. Juma, J. Kavalakkatt, P. Pistor, B. Latzel, K. Schwarzburg, T. Dittrich, Phys.
Status Solidi A, 2012, 209, 663-668.
[198] L. Macor, M. Gervaldo, F. Fungo, L. Otero, T. Dittrich, C. Y. Lin, L. C. Chi, F.-C.
Fang, S.-W. Lii, C.-H. Tsai, C.-C. Wu, RCS Advances, 2012, 2, 4869-4878.
[199] M. C. Biesinger, B. P. Paynec, A. P. Grosvenord, L. M. Laua, A. R. Gersonb, R.
St. C. Smart, Appl. Surf. Sci. 2011, 257, 2717.
[200] M. P. Seah, W. A. Dench, Surface and Interface Analysis 1, 2 (1979).
105
[201] S. Trasatti, J. Electroanal Chem. 1972, 39, 163, doi: 10.1016/ S0022-0728(72)80485-
6.
[202] A. Azarpira, Th. Schedel-Niedrig, H. J. Lewerenz, M. Lublow (submited to J. Adv.
Eng. Mat. 2015).
[203] T. Akaishi, K. Miyasaka, K. Ishikawa, H. Shirakawa, S. Ikeda, J. Pol. Sci., 1980,
18, 745-750.
[204] P. Robin, J. P. Puget, A. J. Epstein, Polymer, 1983, 24, 1558-1564.
[205] B. Xu, R. J. Madix, C. M. Friend, Phys. Chem. Chem. Phys., 2013, 15, 3179-3185.
[206] M. Aggour, K. Skorupska, T. Stempel Pereira, H. Jungblut, J. Grzanna, H. J.
Lewerenz, J. Electrochem. Soc. 2007, 154, H794-H797.
[207] M. Lublow, B. Bouabadi, S. Kubala, Sol. En. Mat. & Sol. Cells, 2012, 107, 56-62.
106
7.2 List of gures
List of Figures
1.1 solar to hydrogen pathways. . . . . . . . . . . . . . . . . . . . . . . . . . 1
2.1 Position of conduction and valence band minima,
CB
and
VB
, of semi-
conductors with respect to the thermodynamic potentials for
H2
and
O2
evolution, respectively [69]. . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Dierent congurations of PEC cell. . . . . . . . . . . . . . . . . . . . . 9
2.3 Activity of Various catalysts for HER and OER in acidic and alkaline
electrolytes. .................................. 10
2.4 Comparison of
VB
and
CB
positions of a metal, a semiconductor and an
insulator[3]. .................................. 11
2.5 Schematic diagram of the energy levels of an n-type semiconductor (a) and
a p-type semiconductor (b). . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 energy levels in (a) p-type semiconductor, (b) n-type semiconductor and
(c) redox electrolyte [113]. . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.7 Eect of the energy levels of semiconductor and electrolyte on the band
edges in the interior of an n-type (a,c) and a p-type (d,f) semiconductor
upon contact formation [114]. . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8 Ideal current-voltage behavior of an n-type semiconductor (a) and p-type
semiconductor (b) under illumination (solid blue curve) and in the dark
(dashedgreencurve).............................. 15
2.9 behavior of (a) n-type semiconductor, (b) p-type semiconductor under il-
lumination (solid blue curve) and with co-catalysts (dashed red curve). . 16
3.1 (a) Light and dark C-V curves for an ideal solar cell. (b) External quan-
tum eciency of CIGSe-based solar cell devices with layers with dierent
Cu/(In+Ga) ratios. (c) SEM cross section of a CIGSe solar cell. (d)
Schematic band diagram of a CIGS solar cell under zero-bias voltage [115]. 18
3.2
TiO2
band alignment [126]. . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1 Schematic of the Ion-Layer-Gas (ILGAR) set-up for deposition of
TiO2
. . 24
4.2 Schematic of the EPD process. . . . . . . . . . . . . . . . . . . . . . . . . 25
4.3 Schematic illustration of the electrophoretic deposition (EPD) set-up. . . 26
4.4 Typical cyclic voltammogram where
ipc
and
ipa
show cathodic and anodic
peak currents, respectively, for a reversible reaction [165]. . . . . . . . . . 27
4.5 Schematic of the photoelectrochemical cell. . . . . . . . . . . . . . . . . . 28
4.6 IPCEsetup[3]................................. 29
4.7 Schematic of the Dierential Electrochemical Mass Spectroscopy (DEMS)
setup. ..................................... 30
4.8 (a) X-ray diraction in reection mode. (b) Grazing incidence X-ray
diraction (GIXRD) mode. . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.9 comparison between UPS and XPS[170]. . . . . . . . . . . . . . . . . . . 32
4.10 Relation between electron binding energies and atomic number (Z)[173]. . 34
4.11 Interaction of the electron beam with a sample in SEM [174]. . . . . . . . 35
4.12 (a) Kelvin probe mode and (b) Fixed capacitor mode. . . . . . . . . . . . 38
5.1 Current-voltage behavior (a)
TiO2
:ZnO/CdS/CIGSe (solid line) and ZnO/CdS
/CIGSe (dashed line), (b)
TiO2
:CdS/CIGSe (solid line) and CdS/CIGSe
and (c)
TiO2
:CIGSe (solid line), bare CIGSe (dashed line). . . . . . . . . 41
107
5.2 Grazing incidence X-ray diraction (GIXRD) at 0.3
◦
angle of incidence
of ILGAR
TiO2
deposited on FTO at dierent deposition temperatures:
250
◦
C (black curve), 350
◦
C (red curve), 400
◦
C (green curve) and 450
◦
C
(bluecurve)................................... 42
5.3 Surface photovoltage measurements: (a) Normalized SPV amplitude and
(b) normalized SPV divided by photon ux for
TiO2:CIGSe
samples
prepared at dierent deposition temperatures: bare CIGSe (purple curve),
250
◦
C (black curve), 350
◦
C (red curve), 400
◦
C (green curve) and 450
◦
C (
bluecurve). .................................. 43
5.4 Eect of deposition temperature on the current-voltage behavior of the
TiO2/Cu(In,Ga)Se2
photocathode:
TiO2
sprayed at (a) 250
◦
C (blue
curve), 350
◦
C (red curve) and (b) 400
◦
C (green curve), 450
◦
C (orange
curve). Electrolyte: 0.5 M
H2SO4
, pH0.3. Photoelectrodes were measured
under near AM1.5 (100
mW/cm2
) conditions, with Pt as counter electrode
and Ag/AgCl as reference electrode. . . . . . . . . . . . . . . . . . . . . . 44
5.5 (a) Current-voltage behavior of undoped
TiO2:Cu(In,Ga)Se2
(black
curve) and Pt-implemented
TiO2:Cu(In,Ga)Se2
devices. The best
current-voltage behavior shows an anodically shifted onset potential at
+230 mV (green curve); Electrolyte: 0.5 M
H2SO4
, pH0.3. Photoelec-
trodes were measured under near AM1.5 (100 mW/
cm2
) conditions. . . . 45
5.6 Surface morphology (SEM, left) and Pt-distribution (TEM, right) of
TiO2
thin lms on CIGSe supports upon increasing Pt concentrations. Charac-
terization by SEM clearly shows an increase of the anatase crystallite size
for incremented Pt concentrations. For 5 and 10 vol.%, agglomeration of
Pt to nano-clusters is visible in the corresponding TEM images [183]. . . 47
5.7 Resistivity and thickness of undoped and Pt-doped
TiO2
.......... 48
5.8 (a) Grazing incidence X-ray diraction (GIXRD) at 0.3
◦
angle of incidence
of undoped ILGAR
TiO2
paste (black curve), 2 vol.% Pt doped
TiO2
(red
curve) and 5 vol.% Pt doped
TiO2
(green curve). All lms were prepared
onCIGSesubstrates.............................. 48
5.9 X-ray photoelectron spectroscopy of the Pt4 core level for the 5 vol. %
Pt-
TiO2
lm[183]. .............................. 49
5.10 X-ray photoelectron spectroscopy of the Ti 2p (a), O 1s (b) and N 1s (c)
signals measured for 5% Pt-doped
TiO2
on CIGSe [183]. . . . . . . . . . 49
5.11 XPS analysis of the C 1s signal measured for 5% Pt-doped
TiO2
on CIGSe.
(a) before electrochemistry, (b) after electrochemistry, (c) comparison of
the C 1s signals in (a) and (b)[183]. . . . . . . . . . . . . . . . . . . . . . 50
5.12 UV-VIS transmission (a) and reection (b) of undoped and Pt-doped
TiO2
/CIGSe(c) Determination of the optical band gap of undoped and
Pt-doped
TiO2
/CIGSe............................. 51
5.13 Surface photovoltage measurements (SPV): (a) Normalized SPV ampli-
tude, in-phase (blue curves) and phase shifted by 90
◦
signals (red curves)
for commercial
TiO2
paste (b), undoped
TiO2
(c) and 5%Pt:doped
TiO2
(d) prepared by the ILGAR process. . . . . . . . . . . . . . . . . . . . . 53
108
5.14 Incident-photon-to-charge-carrier conversion eciency (IPCE) data of four
dierent Pt-doped
TiO2−Cu(In,Ga)Se2
heterojunctions. (a) green squares:
5 vol.% Pt; (b) red squares: 2 vol.% Pt; (c) blue squares: 10 vol.% Pt
compared with the undoped
TiO2−Cu(In,Ga)Se2
heterojunction photo-
cathode (d) black squares. The IPCE data have been measured at -0.4 V
vs. RHE. Electrolyte: 0.5 M
H2SO4
, pH0.3 [183]. . . . . . . . . . . . . . 54
5.15 Hydrogen evolution behavior of the most ecient Pt-implemented
TiO2
-
Cu(In,Ga)Se2
photocathode under illumination (red curve) and in the
dark (blue curve) detected by dierential photoelectrochemical mass spec-
troscopy (DEMS). The inset shows the hydrogen evolution with respect to
CV curves under illumination. Electrolyte: 0.5 M
H2SO4
, pH0.3. . . . . . 55
5.16 Current-voltage behavior of 5 vol.% Pt doped
TiO2
-
Cu(In,Ga)Se2
com-
posite photocathode devices in dierent electrolytes:
H2SO4
0.5M pH 0.3
(green curve), KOH 1N pH 14 (blue curve) and KPi buer phosphate
solution pH7 (red curve). All measurements were carried out in a three-
electrode mode vs. Ag/AgCl reference electrode. . . . . . . . . . . . . . . 56
5.17 Photocurrent density time behavior over 25 (20) hours of the most e-
cient Pt-doped
TiO2/Cu(In,Ga)Se2
heterojunction photocathode in elec-
trolytes of pH0.3 (pH7). Photoelectrodes were measured under near AM1.5
(100
mW/cm2
) conditions at the hydrogen reduction potential 0 V vs.
RHE. The spikes correspond to light on/light-o in order to assess the
dark current density in a short time of several minutes. . . . . . . . . . . 57
5.18 (a) and (b) SEM / EDX image of a low ecient sample, with formation
of Pt clusters on the surface of a
Cu(In,Ga)Se2
, (c) SEM image of cross
section of the sample, (d) the top view SEM image of the sample shows
formation of an interface layer with Pt nanoparticles. . . . . . . . . . . . 58
5.19 Band alignment of a p-semiconductor with a metal for dierent work func-
tion relations (a)
Φm< Φp
, (b)
Φm> Φp
and (c) Proposed band
alignment of the 5%Pt-doped
TiO2−Cu(In,Ga)Se2
[183]. . . . . . . . . 58
5.20 Comparison of the current-voltage behavior of the undoped
TiO2−Cu(In,Ga)Se2
(black curve),
Pt :TiO2−Cu(In,Ga)Se2
compsoite photocathode with
Pt :TiO2−TiO2(ALD)−Cu(In,Ga)Se2
sample. The best current-voltage
behavior is observed for the sample with
TiO2
(by ALD) interface layer
and shows an anodically shifted onset potential at +400 mV (brown curve);
Electrolyte: 0.5 M
H2SO4
, pH0.3. Photoelectrodes were measured under
near AM1.5 (100 mW/
cm2
)conditions.................... 59
5.21 (a) Eect of the organic solvent and the deposition time on the thick-
ness and the density of the layers. (b) Role of the organic solvent on the
electrophoretic mobility of ZnO:Co particles in the suspension [196]. . . . 63
5.22 Diagram of forces inuencing the movement of the dispersed particle. . . 64
5.23 (a), (b) Eect of the organic solvent as a function of deposition time on
the electrophoretic mobility and the zeta potential of the ZnO:Co particles
[196]....................................... 65
5.24 (a) Current voltage behavior of ZnO:Co prepared in dierent organic sol-
vents ethanol (red line), acetone (blue line) and acetonitrile (green line)
and deposited on FTO substrates. (b) Galvanostatic measurements of the
layers corresponding to a current density of
1mAcm−2
. Electrolyte: KOH
1N........................................ 67
109
5.25 Dierential electrochemical mass spectroscopy (DEMS) measurements for
O2
evolution on ZnO:Co deposited on FTO, dissolved in acetonitrile (a)
and ethanol (b). Electrolyte: KOH 1N. . . . . . . . . . . . . . . . . . . . 68
5.26 Stability assessment of ZnO:Co, prepared from ethanol-containing suspen-
sions, at pH14. For comparison, the rapid degradation of a bare FTO
substrateisindicated. ............................ 69
5.27 optical characterization of ZnO:Co deposited on FTO substrates (a) Trans-
mission spectra as a function of wavelength. . . . . . . . . . . . . . . . . 70
5.28 GIXRD diagram of ZnO:Co prepared in ethanol (red line), acetone (blue
line) and acetonitrile (green line) (a) after electrophoretic deposition and
(b) after electrochemical measurements. . . . . . . . . . . . . . . . . . . . 71
5.29 EDX spectra of ZnO:Co deposited on FTO in acetonitrile (a), ethanol (b)
and acetone (c) before and after electrochemistry measurements. . . . . . 72
5.30 XPS analysis of Co 2p signal (a) before electrochemistry for samples pre-
pared in acetone (blue curve), acetonitrile (green curve) and ethanol (red).
(b) XPS results for the Co 2p signal for acetone prepared sample before
and after electrochemistry (c) and (d) comparative deconvolution of the
Co 2p and Zn 2p signals for samples prepared in acetone and ethanol,
respectively.[196]................................ 72
5.31 Microscopy characterization of ZnO:Co morphologies after EPD using ace-
tonitrile as solvent. (a) and (b) SEM surface view before and after electro-
chemistry, respectively. (c) and (d) corresponding TEM images of ZnO:Co
lamellae before and after electrochemistry, respectively. (e) and (f) Mag-
nied TEM images of ZnO:Co lamellae before and after electrochemistry,
respectively[196]................................ 75
5.32 Microscopy characterization of ZnO:Co morphologies after EPD using ace-
tone as solvent. (a) and (b) SEM surface view before and after electro-
chemistry, respectively. (c) and (d) corresponding TEM images of ZnO:Co
lamellae before and after electrochemistry, respectively. (e) and (f) Mag-
nied TEM images of ZnO:Co lamellae before and after electrochemistry,
respectively[196]................................ 76
5.33 Microscopy characterization of ZnO:Co morphologies after EPD using ethanol
as solvent. (a) and (b) SEM surface view before and after electrochemistry,
respectively. (c) and (d) corresponding TEM images of ZnO:Co lamellae
before and after electrochemistry, respectively. (e) and (f) Magnied TEM
images of ZnO:Co lamellae before and after electrochemistry, respectively
[196]....................................... 77
5.34 Schematic setup of the combined supports, an amorphous/microcrystalline
silicon solar cell (bottom) and a ZnO:Co/FTO heterojunction (top). In
the top-view, the O-ring is depicted in decreased size in order to stress the
photoactive area beneath. . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.35 (a) Photocurrent-voltage behavior of the silicon triple-junction solar cell
(black curve) and the stacking conguration with a top ZnO:Co/FTO
layer, prepared from ethanol (red curve). (b) Corresponding stacking con-
gurations with ZnO:Co/FTO layers prepared from acetone (green curve)
and acetonitrile (black curve). For comparison the solar cell behavior with-
out electrocatalysts is shown in red [196]. . . . . . . . . . . . . . . . . . 79
110
5.36 (a) Current-voltage behavior of ethanol (red curve), acetone (blue curve)
and acetonitrile (green curve) prepared samples in
H2SO4
0.5 M. (b) Ef-
fect of the organic solvent on the thickness and the density of the layers.
(c) Role of the organic solvent on the electrophoretic mobility of
RuO2
particles in the suspension. . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.37 (a-c) Current-voltage behavior of ethanol (red curve), acetone (blue curve)
and acetonitrile (green curve) prepared samples. Dashed lines refer to FTO
substrates, solid lines to the corresponding behavior for silicon substrates.
(d-f) corresponding XPS analysis of the carbon C 1s signal [202]. . . . . . 81
5.38 (a) TEM image of Si/Carbon/
RuO2
, (b) SEM image of a particle-free
region (resolution 100 nm) and (c) EDX mapping of a Si/
RuO2
edge [202]. 82
5.39 Device scheme and interfaces of Si/
SiO2
/Carbon(
sp3
)/Carbon (
sp2
)/
RuO2
. 83
5.40 Photocurrent-voltage behavior of Si(100) in 0.5 M
H2SO4
after electro-
chemical surface conditioning in ethanol and ethanol-iodine, respectively. 84
5.41 XPS C 1s, Si 2p and I 3d signals of Si(100) after electrochemical surface
conditioning in ethanol-iodine [202]. . . . . . . . . . . . . . . . . . . . . . 84
5.42 UPS analysis of the cut-o region of secondary electrons of (a) H-terminated
silicon and (b) ethanol-iodine-prepared silicon. . . . . . . . . . . . . . . 85
5.43 XPS analysis of the Ru 3d and C 1s core-level signals of (a)
RuO2
powder
after physical adsorption from aqueous particle suspensions onto silicon
and (b)
RuO2
deposited for 30 min onto silicon from ethanol-iodine con-
taining suspensions. The C 1s deconvoluted signals and UPS analysis of
the cut-o region of secondary electrons of each case are shown below [202]. 86
5.44 XPS analysis of the Ru 3d and C 1s core-level signals and UPS analysis
of
RuO2
deposited for 60 min onto silicon from ethanol-iodine containing
suspensions[202]................................ 87
5.45 FTIR analysis of the used solutions, ethanol (red), ethanol/iodine after
electrochemistry (blue) and ethanol/iodine/
RuO2
after electrochemistry
(green). .................................... 87
5.46 XPS Iodine I 3d analysis after deposition from ethanol- (a/b), acetone- (c)
and acetonitrile- (d) iodine suspensions of
RuO2
. ............. 88
5.47 XRD analysis of Si/
RuO2
, prepared from ethanol-iodine containing sus-
pensions before (blue) and after annealing (red). . . . . . . . . . . . . . . 89
5.48 Operation of the Si/
SiO2
/
RuO2
electrode for photo-assisted water split-
ting. (a) The photocurrent-voltage behavior is shown for two dierent
distributions of
RuO2
particles. (b) Constant potential measurements in
0.5M
H2SO4
under illumination (AM1.5)[202]. . . . . . . . . . . . . . . . 90
5.49 Comparative surface and IPCE analysis of Si/
RuO2
after deposition from
ethanol-iodine containing suspensions with (red) and without toluene ad-
dition (blue). SEM surface analysis (upper images) conrms that addition
of toluene results in particle aggregation with an overall more scattered
distribution................................... 92
5.50 XPS analysis of the Ru 3d and C 1s core-level signals of Si/
RuO2
after
deposition for 30 min onto silicon from ethanol-toluene-iodine containing
suspensions[202]................................ 92
5.51 Scheme for the
RuO2
-catalyzed formation of an organic interface. . . . . 93
5.52 Dierential electrochemical mass spectroscopy (DEMS) measurements for
O2
evolution on
RuO2
/ n-Si photoanode, electrolyte
H2SO4
0.5 M. . . . 93
111
5.53 (a) current-voltage behavior of
RuO2
/Si photoanode in
H2SO4
(red), Kpi
pH7 (green) and KOH pH14 (blue). (b) Constant potential measurements
of the photoanodes in the corresponding electrolytes. . . . . . . . . . . . 94
5.54 Operation of the
Si/SiO2/RuO2
electrode as photoelectrochemical solar
cell: In (a), two-electrode measurements in a iodide/triiodide electrolyte
are shown using a Pt-counter electrode. The sensitive electrolyte was
purged by nitrogen during both current-voltage cycling and long-time as-
sessment of the short-circuit photocurrent (b). . . . . . . . . . . . . . . . 95
5.55 SEM surface analysis of the Si photoelectrochemical solar cell. . . . . . . 96
112
7.3 List of publications
1.
Ecient and Stable
TiO2
:Pt-
Cu(In,Ga)Se2
Composite Photoelectrodes for Visible
Light Driven Hydrogen Evolution
Anahita Azarpira, Michael Lublow, Alexander Steigert, Peter Bogdano, Dieter
Greiner, Christian A. Kaufmann, Martin Krüger, Ullrich Gernert, Roel van de
Krol, Anna Fischer, Thomas Schedel-Niedrig (Adv. Eng. Mat. 2015, 1402148)
2.
Optimized immobilization of ZnO:Co electrocatalysts realizes 5% eciency in pho-
toassisted splitting of water
Anahita Azarpira, Johannes Pfrommer, Katarzyna Olech, Christian Höhn, Matthias
Driess, Bernd Stannowski, Thomas Schedel-Niedriga and Michael Lublow, J. Mater.
Chem. A (DOI: 10.1039/c5ta07329d)
3.
Sustained Water Oxidation by Direct Electrosynthesis of Ultrathin Organic Protec-
tion Films on Silicon
Anahita Azarpira, Thomas Schedel-Niedrig, Hans Joachim Lewerenz, Michael Lublow
(revised manuscript submitted to Adv. En. Mat. 2015)
4.
Solvolytically prepared Ni:ZnO from heterobimetallic precursors as precatalysts for
highly ecient and prolonged water oxidation catalysis
Johannes Pfrommer, Anahita Azarpira, Kasia Olech, Alexander Steigert, Thomas
Schedel-Niedrig, Matthias Driess (manuscript submitted to Angewandte Chemie
2015)
5.
A Molecular Approach to Self-Supported Cobalt Substituted ZnO Materials as Re-
markably Stable Electrocatalysts for Water Oxidation
Johannes Pfrommer, Michael Lublow, Anahita Azarpira, Caren Göbel, Marcel
Lücke, Alexander Steigert, Martin Pogrzeba, Prash-anth W. Menezes, Anna Fis-
cher, Thomas Schedel-Niedrig, Matthias Driess (Angewandte Chemie 53, 20, 5183-
5187 (2014))
6. D
evelopment of Silicon-Supported ZnO:Co Heterojunctions for Photo-Assisted Wa-
ter Splitting
Michael Lublow, Anna Fischer, Johannes Pfrommer, Anahita Azarpira, Alexander
Steigert, Thomas Schedel-Niedrig, Matthias Driess (to be submitted)
7.
Novel electrocatalytic protection layers for photoassisted evolution of oxygen: car-
bonaceous transition-metal hydroxide lms on silicon
Michael Lublow, I. Zaharieva, M. Driess, A. Azarpira, Th. Schedel-Niedrig, H. Dau,
A. Fischer (to be submitted)
113
7.4 Patents
1. Verfahren zur Herstellung von photoelektrokatalytischen Elektroden aus indus-
triellen Pigment-Pulvern TUB 1322 : M. Lublow, A. Azarpira, T. Schedel-Niedrig,
A. Fischer, A. Azarpira, M. Dreiss, M. Kanis
(Status: submitted to the Deutsches Patent und Markenamt (2015), in discussion
with the patent examiner.)
2. Anorganisch-organische Mehrschicht-Photoelektrode für die solare Energieumwand-
lung HZB 2015-1: Lublow, Azarpira, Schedel-Niedrig
(Status: in preparation with HZB experts for intellectual property rights. Envis-
aged submission date: February 2016.)
3. Verfahren zur Herstellung polymerartiger Schutzschichten auf Halbleiteroberächen
HZB 2015-2: Lublow, Azarpira, Schedel-Niedrig
(Status: in preparation with HZB experts for intellectual property rights. Envisaged
submission date: February 2016.)
7.5 Conference contributions
1. MRS Spring Meeting 2015, April 2015 in San Francisco
2. NanoGe Solar Fuel15, March 2015 in Mallorca
3. DFG Projekttreen, September 2014 in Darmstadt
4. IPS 20, Juli 2014 in Berlin
5. E-MRS Spring Meeting 2014, Mai 2014 in Lille
6. DFG Projekttreen, Oktober 2013 in Ellwangen
115
7.6 Curriculum Vitae
Personal Information
Name: Anahita Azarpira
Nationality: Iranian
Date of Birth: 10 September 1982
Education:
Technical University of Berlin (TU-Berlin) from March. 2013
PhD student
Place: Berlin, Germany
Thesis title: Synthesis and Characterization of Novel Composite Photoelectrodes
based on Chalcopyrite and Silicon for the Visible Light-driven Hydrogen and Oxy-
gen Evolution
Technical University of Berlin (TU-Berlin) Sep. 2010-Nov. 2012
Master of Global Production Engineering in Solar Energy Engineering
Place: Berlin, Germany
Thesis title: Proof of concept of ultra-thin nanocomposite solar cells based on
nanoporous
TiO2
and sulde (
In2S3
,
Sb2S3
) absorbers
Tehran University Sep. 2006-Feb. 2009
Master of Chemical Engineering in Catalysis
Place: Tehran, Iran
Thesis title: CFD modeling of slurry bubble column reactor
Iran university of science & Technology Sep. 2001-July 2006
Bachelor of Chemical Engineering in Oil & Gas Process Design
Place: Tehran, Iran
Thesis title: Modeling of FT synthesis by MATLAB
Nyayesh Highschool Sep. 1997 - July 2000
Place: Tehran, Iran
Title: Mathematics-Physics
Work Experience:
Helmholtz Zentrum Berlin March 2012-present
Place: Berlin, Germany
Title of Position: PhD student
Helmholtz Zentrum Berlin May 2012-November 2012
Place: Berlin, Germany
Title of Position: Master student
GFE Gesellschaft für Energieezienz mbH October 2011-February 2012
Place: Berlin, Germany
Title of Position: internship
117
Parsikan Iran Consultant Engineers Co. 2009-2010
Place: Tehran, Iran
Title of Position: Safety Engineer / Process Senior Engineer
(Client: Iranian Oil Pipeline and Telecommunication Co.)
Faranegar Sanaat Design & Engineering Co. 2008 - 2009
Place: Tehran, Iran
Title of Position: Process Senior Engineer (Client: Iranian Petrochemical Co.)
118
