s3688078 Mendes J Thesis

What man sees depends both upon what he looks at and also upon
what his previous visual-conception experience has taught him to

see.
- Thomas Kuhn, The Structure of Scientific Revolutions

Declaration
I certify that except where due acknowledgement has been made, this research is that
of the author alone; the content of this research submission is the result of work which has
been carried out since the official commencement date of the approved research program;
any editorial work, paid or unpaid, carried out by a third party is acknowledged; and, ethics
procedures and guidelines have been followed.
In addition, I certify that this submission contains no material previously submitted for
award of any qualification at any other university or institution, unless approved for a joint-
award with another institution, and acknowledge that no part of this work will, in the future,
be used in a submission in my name, for any other qualification in any university or other
tertiary institution without the prior approval of the University, and where applicable, any
partner institution responsible for the joint-award of this degree.
I acknowledge that copyright of any published works contained within this thesis resides
with the copyright holder(s) of those works.
I give permission for the digital version of my research submission to be made avail-
able on the web, via the University’s digital research repository, unless permission has been
granted by the University to restrict access for a period of time.
Joao Otavio de Sousa Mendes
10 March 2023
i
Acknowledgements
I would like to take this opportunity to express my sincere gratitude and appreciation to
all those who have helped me in the completion of this work.
I thank my supervisor, Dr Joel van Embden, for the valuable guidance and support
throughout this project. His insightful comments and criticism were invaluable in helping
me clarify my ideas and develop them further. I also thank my co-supervisor, Dr Enrico della
Gaspera, I am extremely grateful for his patience, enthusiasm and encouragement in helping
me complete this project. Their support was instrumental in ensuring that I stayed focused
on the task at hand during challenging times.
This work could not have been completed without the assistance of many individuals who
have provided help at various stages of this research. My sincerest thanks to all members of
staff at RMIT University who have directly or indirectly contributed in one way or another
to the success of this project. I am particularly grateful to the staff at MNRF and RMMF for
maintaining the facilities, helping with experiments and providing prompt support under-
standing the complexities of the techniques. I most also thank Student Counsellor Yvonne
Yoong for the philosophical discussions that helped me understand my own complexities.
I would like to express my deepest appreciation to friends who have provided me with
unwavering support throughout my academic journey. Thanks to Fabio for coming from
Brazil and getting stuck in Melbourne for 3 months, truly a unique experience, and sure it
helped getting through the begging of the pandemic. Furthermore, I owe a great thanks to
my colleagues in the laboratory for helpful discussion and fun moments.
Finally, I wish to thank my mother Ides, my sister Janaı́na and my niece Maria Fernanda
for their resolute encouragement throughout my live; without their care it would not have
been possible for me to complete this step of the journey.
ii
Table of contents
Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Table of contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii
List of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii
List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiv
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Broad context and scope of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1 Electronic structure and modelling of solar cells 8
1.1 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Electronic structure of solid materials . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 Energy levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.2 Energy band and band gap . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.3 Metals, semiconductors and insulators . . . . . . . . . . . . . . . . . . . . 14
1.2.4 Density of states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.5 Density of states in semiconductors . . . . . . . . . . . . . . . . . . . . . . 18
1.2.6 Fermi-Dirac statistics and Fermi Level . . . . . . . . . . . . . . . . . . . . . 19
1.2.7 Carrier density and effective density of states . . . . . . . . . . . . . . . . 20
1.2.8 Shallow and deep energy levels . . . . . . . . . . . . . . . . . . . . . . . . 22
1.2.9 Intrinsic, n-type and p-type semiconductors . . . . . . . . . . . . . . . . . 23
1.3 Carrier generation, recombination and transport . . . . . . . . . . . . . . . . . . . 24
1.3.1 Generation and recombination mechanisms . . . . . . . . . . . . . . . . . 24
1.3.2 Regions of recombinations . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.3.3 Continuity equation and diffusion length . . . . . . . . . . . . . . . . . . . 27
1.3.4 Carrier transport in semiconductors . . . . . . . . . . . . . . . . . . . . . . 29
1.4 The pn junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.4.1 Electric field, potential and built-in voltage . . . . . . . . . . . . . . . . . . 31
iii
1.4.2 Built-in voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1.4.3 Junction under forward and reverse bias . . . . . . . . . . . . . . . . . . . 34
1.5 Thin film solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.5.1 Physical structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.5.2 Band structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.6 The current density vs voltage (JV) curve . . . . . . . . . . . . . . . . . . . . . . . 38
1.6.1 Measuring solar cell efficiency . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.6.2 Diode parameters and parasitic resistances . . . . . . . . . . . . . . . . . . 40
1.6.3 Dark JV curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.6.4 Graphical analysis of JV curves . . . . . . . . . . . . . . . . . . . . . . . . . 43
1.6.5 Modelling complex JV curves . . . . . . . . . . . . . . . . . . . . . . . . . . 45
1.7 Electrochemical impedance spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 47
1.7.1 Equivalent circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
1.7.2 Built-in potential and doping density from EIS data . . . . . . . . . . . . 50
1.8 External quantum efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
1.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2 Review of Sb2 Se3 properties and thin film fabrication methods 54
2.1 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.2 Crystal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.3 Ribbon tilt and crystallographic orientation . . . . . . . . . . . . . . . . . . . . . . 60
2.4 Sb2 Se3 thin film deposition methods . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.4.1 Solution processed Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . . 61
2.4.2 Thermal evaporated Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . 62
2.4.3 Vapour transport deposition of Sb2 Se3 thin films . . . . . . . . . . . . . . 63
2.4.4 Other Sb2 Se3 thin film devices . . . . . . . . . . . . . . . . . . . . . . . . . 64
2.5 Growth of oriented Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.5.1 Epitaxy growth and (hk0) Van der Waals layers . . . . . . . . . . . . . . . 65
2.5.2 Evolution selection principle and (hk1) growth . . . . . . . . . . . . . . . . 67
2.5.3 Seed layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.5.4 Substrate chemistry and evaporation parameters . . . . . . . . . . . . . . 69
iv
2.5.5 Other strategies to obtain (hk1) Sb2 Se3 layers . . . . . . . . . . . . . . . . . 72
2.5.6 (001) orientated Sb2 Se3 films . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.5.7 Graphoepitaxy and importance of substrate nanostructure on film ori-
entation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.6 Optoelectronic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.6.1 Band gap and absorption coefficient . . . . . . . . . . . . . . . . . . . . . . 74
2.6.2 Defect structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.6.3 Anisotropy of charge transport . . . . . . . . . . . . . . . . . . . . . . . . . 77
2.6.4 Contacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.6.5 Front contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.6.6 Back contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3 Experimental procedures 86
3.1 List of materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2 Film deposition methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2.1 Substrate washing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2.2 Sol-gel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2.3 Chemical bath deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.2.4 Additional film deposition methods used in this thesis . . . . . . . . . . . 90
3.3 Thermal evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.3.1 Furnace setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.3.2 Heating program and deposition rate . . . . . . . . . . . . . . . . . . . . . 94
3.4 Devices and materials characterisation . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.4.1 Instrumentation for physicochemical characterisation . . . . . . . . . . . . 95
3.4.2 Device fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.4.3 Photoelectrochemical measurements and instrumentation . . . . . . . . . 96
3.4.4 Sb2 Se3 devices and efficiency calibration . . . . . . . . . . . . . . . . . . . 97
3.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
A 1 Substrate washing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
v
A 2 ZnO SG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
A 2.1 ZnO SG step-by-step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
A 2.2 ZnO SG comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
A 3 TiO2 SG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
A 3.1 TiO2 SG step-by-step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
A 3.2 TiO2 SG comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
A 4 CdS CBD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
A 4.1 CdS washing (step-by-step) . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
A 4.2 H2 O2 treatment (H2 O2 30%) (step-by-step) . . . . . . . . . . . . . . . . . . 107
A 4.3 CdCl2 treatment (CdCl2 20 mg/mL in methanol) (step-by-step) . . . . . . 109
A 4.4 CdS CBD comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
A 5 ZnO CBD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
A 6 Sb2 Se3 layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
A 6.1 Sb2 Se3 deposition (step-by-step) . . . . . . . . . . . . . . . . . . . . . . . . 114
A 6.2 Furnace cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4 High-resistance metal oxide layers for improved front contact in CdS/Sb2 Se3 super-
strate solar cells 117
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
4.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.2.1 Solar cell efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.2.2 Device stability and current leak (shunt) models . . . . . . . . . . . . . . . 126
4.2.3 Electron barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5 Influence of substrate nanostructure on the growth of evaporated Sb2 Se3 layers 143
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.2.1 Experimental design and understanding ribbon orientation . . . . . . . . 146
5.2.2 Growth of Sb2 Se3 films on nanostructured and flat substrates . . . . . . . 148
5.2.3 Orientation control of Sb2 Se3 films . . . . . . . . . . . . . . . . . . . . . . . 152
vi
5.2.4 Growth mechanism of Sb2 Se3 on nanostructured substrates . . . . . . . . 155
5.2.5 Surface chemistry of the nanostructured substrate . . . . . . . . . . . . . . 157
5.3 Comment on the analysis of ribbon tilt . . . . . . . . . . . . . . . . . . . . . . . . 160
5.3.1 Texture coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.3.2 Total (hk1) texture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
5.3.3 Texture weighted average ribbon tilt . . . . . . . . . . . . . . . . . . . . . . 161
5.3.4 Ribbon tilt and substrate nanostructure . . . . . . . . . . . . . . . . . . . . 163
5.4 Comment on ZnONW/Sb2 Se3 photodetectors . . . . . . . . . . . . . . . . . . . . 166
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
6 Transferable highly (001) oriented Sb2 Se3 nanorod layer for innovative device archi-
tectures 172
6.1 Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
6.3.1 Template growth of the (001) Sb2 Se3 NR layer . . . . . . . . . . . . . . . . 175
6.3.2 Sb2 Se3 NR photodetectors from transferred layers . . . . . . . . . . . . . . 180
6.3.3 Flexible photodetectors and heart rate monitor . . . . . . . . . . . . . . . 187
6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
6.5 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
7 Conclusions and perspectives 196
7.1 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
vii
List of figures
1 Renewable electricity installed capacity by source in Australia over 10 years. . . 2
2 a) Best efficiency Sb2 Se3 solar cell by year and b) number of publications per
year related to Sb2 Se3 since 2015. *Research impacted by COVID-19 pandemic. . 4
1.1 Illustration of an infinite square potential well model, where the particles (such
as an electron) can only exist within region II (U ( x ) = 0). . . . . . . . . . . . . . 10
1.2 Allowed energies for an electron confined in a 1 nm quantum well . . . . . . . . 11
1.3 Illustration of a periodic potential where particles such as electrons may tunnel
through the potential barriers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 Illustration of allowed energy bands and band gaps (forbidden energy regions)
for a particle subject to a periodic potential. . . . . . . . . . . . . . . . . . . . . . . 13
1.5 a) Illustration of allowed energy zones as a function of the wave vector (k) and
correspondent b) energy band representation. . . . . . . . . . . . . . . . . . . . . 14
1.6 Illustration of atomic energy states splitting in a solid material and the corre-
sponding band diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.7 Energy band diagram representation for solid materials in equilibrium at 0 K
a) in metals, such as Au, Ag, Cu and Al, a lower energy band is only half
filled, while in Mg, Zn and Pb, for example, a filled band overlaps an empty
one. Insulators in b) have a filled valence band (VB), an empty conduction band
(CB) and a band gap that is relatively large compared to that of semiconductors
shown in (c). Electrons cannot have energies within the band gap region. . . . . 16
1.8 Infinite potential quantum box. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.9 Spherical space containing all possible energy numbers (n E ). . . . . . . . . . . . 17
1.10 Density of states and schematic representation of the state distribution within
the CB and VB of a semiconductor. . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.11 Fermi function, or Fermi distribution, that describes the distribution of electrons
or holes within an energy band. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.12 Carrier density in the conduction and valence bands as a result of ρe ( E) =
gC ( E) f ( E) and ρh ( E) = gV ( E)(1 − f ( E)), respectively. . . . . . . . . . . . . . . . 20
viii
1.13 Region in the band gap where deep defects (traps) are formed. Process in the
trap: I) and III) are the capture and re-emission of an electron and II) and IV)
are the capture and re-emission of a hole. . . . . . . . . . . . . . . . . . . . . . . . 22
1.14 Schematic representation of crystal structure and energy band diagrams of el-
emental semiconductor (Si - grey) a) without impurities and with b) donor
impurity (P - magenta) and b) acceptor impurity (B). . . . . . . . . . . . . . . . . 23
1.15 Schematic representation of crystal structure and energy band diagrams of a
semiconductor a) in the dark and b) illuminated with light with energy higher
than Eg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.16 Schematic representation of a) e− /h+ pair generation, b) band-to-band recom-
bination, c)Auger recombination and d) recombination via trap state. Where γ
represents a photon and R is the recombination rate. . . . . . . . . . . . . . . . . 26
1.17 Schematic representation of recombination in different regions of the solar cell
1) at the interface between the light absorber and electron transport layer (ETL),
2) in the bulk of the light absorber, 3) at the interface between the light absorber
and the back contact, 4) within light absorber space charge region (SCR), 5) in
the bulk of ETL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.18 Schematic representation of continuity equation for electrons in a solar cell
where Je ( x ) and J ( x + dx ) is the flux of incoming electrons to and from the
boundary element, Re is the recombination rate and Ge is the generation rate. . 28
1.19 Schematic representation of electron transport in a crystal under a) no electric
field (no net movement) and b) upon the application of an electric field (with
net movement). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.20 Scheme of a p–n junction, highlighting the depletion region formed between
p-type and n-type semiconductors. . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.21 Representation of a) pn-junction, and the b) charge density (ρcarrier ) and total
charge (Q), c) electric field (E ), d) potential φ( x ), where φi is the local built-in
potential and e) band diagram within the junction. . . . . . . . . . . . . . . . . . 32
1.22 Band diagram within a pn junction under a) forward and b) reverse bias. . . . . 34
ix
1.23 Basic structure of a thin film solar cell with electron transport layer (ETL), hole
transport layer (HTL), light absorber layer, transparent conductive oxide (TCO
- front contact) and metal back contact. . . . . . . . . . . . . . . . . . . . . . . . . 35
1.24 Band alignment between different semiconductor materials a) before being joined
together, b) at equilibrium after joining and c) band bending at the interface. . . 37
1.25 Band alignment between different layers in a solar cell, a) before being joined
together, b) at equilibrium after joining, where bc = back contact, ETL = electron
transport layer, TCO = transparent conductive oxide. . . . . . . . . . . . . . . . . 37
1.26 a) Simulated current density voltage (JV) curve of a solar cell under illumina-
tion and derived power density curve (P = JV). The areaA corresponds to
the maximum power density output (Pmax = Jmax Vmax ), and areaB is given by
J × VOC . The fill factor FF is defined as the ratio between areaA and areaB
SC
areaA

FF = and has a maximum value of 1 or 100 %. b) standard AM1.5G
areaB
solar spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
1.27 Equivalent circuit for an ideal solar cell under illumination, where J p is the
photocurrent, Jdark is the dark current and V is the voltage. . . . . . . . . . . . . 40
1.28 Equivalent circuit including parasitic resistances, RS and Rsh , for a solar cell
under illumination. J p is the photocurrent, Jdark is the dark current and V is the
voltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
1.29 Behaviour of JV curve of a cell with a) increasing series resistance, RS , and b)
decreasing shunt resistance, Rsh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.30 Simulated dark JV curve of a cell with a) increasing dark current, where J0 is
one order of magnitude higher in the dashed curve, and b) increasing diode
factor, n = 1 (full line) and n = 2 (dashed line). . . . . . . . . . . . . . . . . . . . 43
1.31 Example of using graphical methods to obtain solar cell parameters, a) simu-
dV dV
lated JV curve, b) vs V to find Rsh , c) vs ( J + JSC − GV ) to find RS and
dJ dJ
n, and d) using RS value obtained in the previous graph ln( J + J p − GV ) vs
(V − RS J ) is plotted to find J0 and n. . . . . . . . . . . . . . . . . . . . . . . . . . . 44
x
1.32 Current leak in a thin film solar cell, a) scheme of possible alternative pathways
for current leak and b) equivalent circuit model, where J1 and J2 are the satura-
tion current from diode elements (main-junction and weak diode), R1 and Rsh
are parallel non-Ohmic and Ohmic resistive elements, respectively, and RS is a
series resistive element. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.33 Effect of a current leak in logarithm JV curve, a) shoulder caused by the presence
of weak diode and b) effect of Ohmic and non-Ohmic shunt leaks. . . . . . . . . 47
1.34 a) Applied voltage and measured current response and b) correspondent impedance
plot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
1.35 Time and frequency scale of dielectric permittivity (ε) response for typical pro-
cesses in electronic and electrochemical systems. . . . . . . . . . . . . . . . . . . . 48
1.36 a) Simulated Nyquist plot and its b) equivalent circuit. . . . . . . . . . . . . . . . 50
1.37 a) EQE showing an ideal (dash-dot line) and typical (full line) quantum effi-
ciency; the effects that causes typical curves to deviate from the ideal behaviour
are indicated in the graph. An EQE curve for a cell with an electron transport
layer with a small band gap presenting a ‘shoulder’ (doted line) is also shown. . 52
2.1 Unity cell, charge transport and expanded structure of materials with a) 3D
(CdTe) b) quasi-2D (SnSe) and c) quasi-1D (Sb2 Se3 ) crystal structure. . . . . . . . 57
2.2 Representation of charge transport along Sb2 Se3 a) [001] and b) [120] directions. 59
2.3 The three possible ‘sets’ of ribbon orientation and their corresponding crystal-
lographic notations: a) (hk0), b) (hk1) and c) (001). . . . . . . . . . . . . . . . . . . 60
2.4 Efficiency evolution of Sb2 Se3 solar cells with absorber layer fabricated via so-
lution processing, thermal evaporation (TE) and vapour transport deposition
(VTD) methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.5 Typical vapour transport deposition equipment used to deposit Sb2 Se3 films. . . 63
2.6 Vapour transport deposition equipment used to deposit Sb2 (S, Se)3 films. . . . . 64
2.7 Epitaxy growth on with a) small and b) large mismatch. Growth of c) generic
quasi-1D or quasi-2D and d) Sb2 Se3 a van der Waals layers. . . . . . . . . . . . . 66
xi
2.8 Proposed growth mechanism for Sb2 Se3 on a flat substrate following the evo-
lution selection principle. A higher population of Sb2 Se3 grains nucleate with
(hk0) orientation, but the fewer (hk1) grains grow faster. Eventually, (hk1) grains
predominate in the crystal population at the surface of the film and start to
define the preferred crystallographic texture. . . . . . . . . . . . . . . . . . . . . . 68
2.9 a-k) schematic representation of the seed screening process. . . . . . . . . . . . . 69
2.10 a) summary of ‘growth rate’ model and b) ‘nanoepitaxy’ of Sb2 Se3 thin films. . . 71
2.11 a) Representation of electron transitions in a quasi-direct band gap semiconduc-
tor, where thicker arrows indicate the preferential the flow of electrons and b)
absorption spectrum of Sb2 Se3 thin film. . . . . . . . . . . . . . . . . . . . . . . . 75
2.12 a) Non-equivalent sites on (Sb4 Se6 )n ribbon, b) calculated defect formation en-
ergies and c) calculated and experimental defect positions. . . . . . . . . . . . . . 76
2.13 Angle-dependent photocurrent measurements with a) 12 electrode arrange-
ment, illustration of charge transport along the ribbon (90°) and inter-ribbon
(0°) and b) response under polarised light. . . . . . . . . . . . . . . . . . . . . . . 77
2.14 Band position of quasi-1D antimony chalcogenides cathode (TCO), ELTs, HTLs
and metal contacts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.15 Representation of band structure of CdS/Sb2 Se3 a) without and b) with a hole
transport layer (HTL). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1 Photograph of dual zone furnace used for VTD of Sb2 Se3 films. Z1 and Z2
indicate heating zones 1 and 2. Valves: V1 (tube valve), V2 (fast N2 input), V3
(fine N2 input) and V4 (back N2 input). T: thermocouple, P: pump and QT:
quartz tube. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.2 a) photograph and schemes showing b) front and c) side of the graphite holder
designed for this thesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.3 Schematic representation of the furnace configuration used in this work. . . . . 93
3.4 Temperatures measured in Zones 1, Zone 2 and holder/substrate. The graphite
holder with substrates was positioned in Zone 2 at exactly 10 cm from the
furnace centre. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
xii
3.5 Photograph of furnace tube after evaporation at 0.5 Pa showing the deposition
of Sb2 Se3 on tube wall of Zone 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.6 a) Scheme of the stack of layers used in the devices b) back and c) front of devices. 98
3.7 a) PCE, b) VOC , c) Jsc and FF d) for glass/ITO/CdS-120 nm/Sb2 Se3 /Au solar
cells. A 650 mg source was used for the series obtained at varying substrate
(Zone 2) temperature. For the thickness series (varying Sb2 Se3 mass in the
source) Zone 2 temperature was kept constant at 250 °C. . . . . . . . . . . . . . . 99
A1 ZnO sol-gel solution preparation (flowchart) . . . . . . . . . . . . . . . . . . . . . 102
A2 Photographs and top view SEM images of ZnO SG films spun for 30s at 3000RPM
from a 0.35 M solution and annealed at a) 200 °C and b) 200 °C for ≈60 min. . . 103
A3 TiO2 sol-gel solution preparation (flowchart) . . . . . . . . . . . . . . . . . . . . . 104
A4 a) Photography and b) top view SEM image of TiO2 SG film spun for 30s at
3000RPM from a 0.3 M solution and annealed at 500 °C for at least ≈60 min. . . 105
A5 CdS CBD procedure (flowchart) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
A6 Apparatus used to perform CdS CBD. 1. metal support, 2. polyethylene (PE)
holder, 3. substrates taped with polyimide (PI) tape, 4. chemical bath, 5. sup-
port for chemical bath + stirring bar (not visible), 6. external water bath covered
with Al foil, 7. heating + stirring, 8. flat dish (chemical bath lid) and 9. Milli-Q
water bath to quench the reaction. . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
A7 Schematic representation of CdS CBD; 1) apparatus during deposition and 2)
quench the reaction in a clean Milli-Q water bath. . . . . . . . . . . . . . . . . . . 108
A8 Photographs of CdS films a) washed, b) heated at 400 °C for 5 min, c) H2 O2
treated and d) CdCl2 treated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
A9 Top view SEM images of CdS deposited on glass/FTO substrate a) washed, b)
heated at 400 °C for 5 min, c) H2 O2 treated and CdCl2 treated . . . . . . . . . . . 111
A10 XRD patterns of CdS film washed (control - blue line) and treated at 400 °C
(grey line), H2 O2 (red line) and CdCl2 (black line). . . . . . . . . . . . . . . . . . . 111
A11 ZnO CBD procedure (flowchart) for deposition of a) ZnO Flat and b) ZnO NW
films. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
xiii
A12 Apparatus used to perform ZnO CBD. 1. chemical bath, 2. external water bath
wrapped in Al foil, 3. set of supports and stirring bar, 4. heating/stirring plate
and 5. Milli-Q water bath to stop reaction. . . . . . . . . . . . . . . . . . . . . . . 113
A13 Schematic representation of ZnO CBD; 1) apparatus during deposition and 2)
quench the reaction by dipping the substrates in a clean Milli-Q water bath. . . 113
A14 Mott–Schottky (MS) plot for a typical Sb2 Se3 layer. The MS curve shows a
positive inclination typical of a p-type film. In the literature Sb2 Se3 is typically
reported as a p-type material, however this character can change according to
the fabrication method and doping. MS was obtained from a Sb2 Se3 film in a
0.5 M Na2 SO4 solution pH = 5 at 1 kHz using a potentiostat Ivium Vertex.1A.EIS.114
4.1 a) schematic of the device architecture used in this study; b) schematic of
HRMO and CdS buffer deposition using sol-gel and chemical bath deposition
methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.2 Scanning electron microscope (SEM) images of FTO coated with a) TiO2 and b)
ZnO layers obtained via sol-gel method from precursor solutions with concen-
tration varying from 0.1 to 0.7 M. X-ray photoelectron spectra (XPS) for 0.3 M c)
TiO2 and d) ZnO. Thickness calibration for e) TiO2 and f) ZnO films measured
determined via profilometry and checked by ellipsometry when necessary. . . . 121
4.3 Top view scanning electron microscope (SEM) images of CdS deposited on a)
FTO, b) FTO/ZnO and c) FTO/TiO2 ; d) X-ray diffraction patterns from ≈60
nm CdS layers deposited directly on FTO (green), on ZnO layer (red) and on
ceTiO2 layer (black) showing that the CdS films have hexagonal wurtzite crystal
structure and slightly higher (100) preferential orientation on FTO when com-
pared to the CdS films deposited on ZnO or TiO2 . In e) the X-ray photoelectron
spectra from CdS layers on FTO (green), ZnO (red) and TiO2 (black) show the
typical signatures of Cd 3d (420 - 395 eV) and S 2s (226 eV) and 2p (162 eV). . . 122
xiv
4.4 Top view SEM images of Sb2 Se3 deposited on the respective CdS substrate, a)
no high resistance metal oxide (HRMO), b) ZnO HRMO and c) TiO2 HRMO and
cross sections of complete solar cell with gold electrodes, d) no, f) ZnO and g)
TiO2 layers deposited before CdS bath deposition. The average thickness of the
Sb2 Se3 layers is ≈550 nm. Importantly, the Sb2 Se3 grains present similar mor-
phology on all substrates, which allows us to rule out any major morphological
effects on the studies done in this work. . . . . . . . . . . . . . . . . . . . . . . . . 123
4.5 Device parameters a)PEC, b) JSC , c) FF and d VOC for Sb2 Se3 solar cells with ei-
ther ZnO or TiO2 HRMO layers deposited from 0.1, 0.3, 0.5 and 0.7 M precursor
solutions compared with CdS-only control devices. . . . . . . . . . . . . . . . . . 124
4.6 Nyquist plots for a) champions devices with different HRMO interlayers com-
pared to control and b) detail in the control device plot. . . . . . . . . . . . . . . 125
4.7 a) diagram of Sb2 Se3 crystal structure in the Pbnm space group viewed along the
c-axis ([001] direction), highlighting the unfavourable vertical charge transport
between (Sb4 Se6 )n ribbons within a device and b) relationship between HRMO
layer type and measured relative (120) crystal orientation for layers deposited
from 0.1, 0.3, 0.5 and 0.7 M precursor solutions compared to control. . . . . . . . 126
4.8 a) dark JV curves taken over time from the champion solar cells shown with
different HRMO layers; b) shunt resistance (Rsh ) values over time in champion
devices with and without HRMO layer deposition highlighting the large vari-
ability in devices without an appropriate interlayer; c) relative cell efficiencies
over time during ageing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.9 a) reverse bias region of dJ/dV curves highlighting the effective use of ZnO
or TiO2 to prevent the increased current leak observed during voltage sweeps
of CdS-only control devices; b) schematic of Ohmic and non-Ohmic shunting
mechanisms in Sb2 Se3 solar cells; c) equivalent circuit models; d) fitted dark
JV curves for the various solar cells used in this study. e) Ohmic and non-
Ohmic current leaks in freshly prepared solar cells, highlighting the removal of
non-Ohmic shunt pathways when using ZnO or TiO2 interlayers. . . . . . . . . . 128
xv
4.10 Assessment of dark shunt conductance of fresh Sb2 Se3 solar cells before any
direct sun light exposure submitted to sweeps in increasing voltage range: -0.5
to +0.5 V, -0.8 to +0.8 V and -1.1 to +1.1 V compared to dark measurements
obtained after sweep in the same ranges but under 1 sun of 1.5G simulated sun
light. In a) solar cells with no high resistance metal oxide (HRMO) layer, b)
TiO2 HRMO and c) ZnO HRMO applied before CdS deposition. . . . . . . . . . 129
4.11 Comparison of the reverse bias region of dJ/dV curve for Sb2 Se3 devices with a
60 nm CdS buffer layer (green dashed), 100 nm CdS buffer (green full line) and
60 nm CdS buffer deposited on the ZnO high resistance metal oxide (HRMO)
(red) highlighting that the preliminary deposition of ZnO HRMO layer prevents
the non-Ohmic shunt leakage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.12 a) dark and light JV curves for CdS-only solar cells compared to devices where
a TiO2 or ZnO HRMO layer was used. b) series resistance as a function of
increasing light power for typical solar cell devices. . . . . . . . . . . . . . . . . . 133
4.13 a) JV curves for devices with an TiO2 HRMO as a function illumination wave-
length of between 400–800 nm. b) corresponding dV/dJ curves highlight-
ing current blocking at high forward bias; c) schematic mechanism showing
how negatively charged acceptor levels at a defective interface may be neutral-
ized via capture of photogenerated holes. d) SCAPS simulation of solar cells
containing a 10 nm HRMO window layer with a variable density of acceptor
(trap) states. Specifically, a high density at the interface (1.75 × 1012 cm−2 , vis.
TiO2 /CdS) resulting in the observation of cross over behaviour of the JV curves
and a relatively low density of traps for (0.85 × 1012 cm−2 , vis. ZnO/CdS) with
no cross over observed in the simulated voltage range. . . . . . . . . . . . . . . . 134
4.14 a) Valence band X-ray photoelectron spectra and b) band gap determination via
UV-Visible spectroscopy for ZnO, TiO2 , CdS and Sb2 Se3 films (γ = 1/2 for all
materials except TiO2 where γ = 2). c) conduction (CBO) and valence band
offset (VBO) diagram highlighting the band alignment between the materials
after Fermi level alignment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
xvi
4.15 Current density vs voltage (JV) curves obtained from Sb2 Se3 solar cells made
with the deposition of a TiO2 high resistance metal oxide (HRMO) layer com-
pared to devices where no HRMO and ZnO HRMO layer was applied before
CdS deposition. The curves were measured in the dark, under illumination
with red, green or blue LEDs and under 1 sun of AM1.5G simulated solar light.
A power output of ≈2 mWcm2 was used for all measures with the red (625 nm),
green (530 nm) and blue (455 nm) LEDs. . . . . . . . . . . . . . . . . . . . . . . . 136
4.16 Capacitance-voltage measurements for Sb2 Se3 solar cells with different HRMO
interlayers; a) C−2 vs V curves at 50, 100, 150 and 200 kHz and b) correspondent
doping profile derived at 100 kHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.1 a) Schematic of existing strategies to achieve vertical ribbon orientation in Sb2 Se3
thin films. b) Proposed methodology in this work, wherein surface nanostruc-
ture is used to direct crystal growth and orientation. . . . . . . . . . . . . . . . . 145
5.2 a) Theoretical XRD pattern of Sb2 Se3 (ICSD #16680) color coded by ribbon ori-
entation group (horizontal (hk0), blue; tilted (hk1), orange/red; vertical (001),
brown). b) Relationship between Sb2 Se3 crystal planes analyzed in this paper
and (Sb4 Se6 )n ribbon tilt angle. c) Schematic of the VTD apparatus and the ‘flat’
and ‘nanostructured’ substrates employed in this work. . . . . . . . . . . . . . . . 148
5.3 a) Schematic of ZnO chemical bath deposition in the presence and absence of
ammonium citrate and corresponding AFM data of the resultant films, b) Top
view and crosssection SEM images showing the ZnO substrate film morphol-
ogy. It is clear the presence of ammonium citrate drastically alters the growth
of ZnO and enables the formation of continuous flat ZnO substrates (ZnO Flat),
c) X-ray diffraction (XRD) patterns of ZnO Flat and ZnO NW substrates, high-
lighting that in both cases the films have the same crystal orientation. d) Survey
X-ray photoelectron spectroscopy (XPS) highlighting the similar surface chem-
istry of ZnO-Flat (in red) and ZnO NW (in black) substrates, e) High resolution
XPS of oxygen 1s, f) High resolution XPS of zinc 2p. The scale bars in the SEM
images correspond to 500 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
xvii
5.4 X-ray diffraction patterns of Sb2 Se3 films deposited using an increasing mass of
powder in the source on a) nanostructured (ZnO NW) and b) flat (ZnO-Flat )
substrates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.5 SEM images and corresponding texture coefficients color coded by ribbon tilt
angle for Sb2 Se3 thin films as a function of Sb2 Se3 source mass (film thick-
ness).a) Sb2 Se3 deposited ZnO-Flat substrates and b) Sb2 Se3 deposited ZnO-Flat
substrates. Scale bars are 500 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.6 a) Top view SEM images of ZnO-NW coated with 0, 5, 20, 40, and 60 nm ZnO
via ALD layer, respectively. b) Cross section SEM images of Sb2 Se3 films ob-
tained on the substrates shown in the upper panels. c) XRD patterns of Sb2 Se3
films as a function of ZnO ALD coating. d) Texture coefficient as a function of
ZnO ALD coating thickness. e) Reflectance measurement of corresponding sam-
ples, wherein lighter colors indicate Sb2 Se3 films deposited on flatter substrates.
f) Average reflectance (400 - 900 nm) as a function of Sb2 Se3 (002) texture. g)
Photographs of Sb2 Se3 films on flat and nanostructured substrates, highlighting
light-trapping by the oriented Sb2 Se3 NR film. Scale bars correspond to 500 nm. 153
5.7 a) Highly reflective Sb2 Se3 film obtained on smooth FTO/CdS substrate and b)
dark (001) Sb2 Se3 NR film on ZnO NW substrate. . . . . . . . . . . . . . . . . . . 154
5.8 Reflectance measurement of Sb2 Se3 deposited on flat FTO/CdS substrate, the
400 - 900 nm average reflectance is 32.9%. . . . . . . . . . . . . . . . . . . . . . . . 154
5.9 a) Proposed mechanism of Sb2 Se3 crystal growth on ZnO-NW to form highly
(001) oriented films. b) Backscattered SEM (BS-SEM) image of an Sb2 Se3 thin
film in the early stages of deposition, highlighting trapped Sb2 Se3 crystals. c)
BS-SEM image of a mature (thicker) Sb2 Se3 film. d) TEM image of a single
crystal physically extracted from (c). e) HRTEM of a single Sb2 Se3 NR showing
the lattice planes. f) Crystal structure of a single (Sb4 Se6 )n ribbon and Sb2 Se3
unit cell. g) Selected-area electron diffraction (SAED) pattern of (d). Unless
indicated, scale bars correspond to 500 nm. . . . . . . . . . . . . . . . . . . . . . . 156
xviii
5.10 Characterization of coated ZnO nanowires (ZnO NW) substrates a) scanning
electron microscope (SEM) image, X-ray photoelectron spectroscopy (XPS), X-
ray diffraction (XRD) and backscattered SEM (BS-SEM) of ZnO-NW array coated
with ≈5 nm of TiO2 layer obtained via atomic layer deposition (ALD), the
smooth TiO2 coat is not distinguishable in the SEM; b) SEM, XPS, XRD and BS-
SEM of ZnO-NW array coated with ≈10 nm CdS deposited via successive ionic
layer adsorption and reaction (SILAR) method, the CdS layer appears brighter
than the ZnO on the BS-SEM image. The scale bars correspond to 500 nm. The
inset shows hi-resolution SEM of CdS dots on ZnO-NW (scale bar is 100 nm).
The Sb2 Se3 crystals trapped in the NW array are clearly visible (in lighter color)
in both BS-SEM images. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.11 a) Schematic of coated nanostructured and flat substrates. b) XRD patterns of
Sb2 Se3 films with CdS, Al2 O3 , and TiO2 coatings on flat FTO and ZnO-NW
substrates , c) (002) textures calculated from the XRD patterns shown in (b), d)
cross-sectional SEM images of Sb2 Se3 films on the indicated substrates. Scale
bars correspond to 500 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5.12 Top view of Sb2 Se3 deposited on ZnO-NW coated with a) Al2 O3 , b) TiO2 and c)
CdS; a source with 500 mg of Sb2 Se3 powder was used for the evaporation. The
stale bars correspond to 500 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.13 Definition of texture coefficient, where fractional values (0 - 1) have the same
weight for suppressed orientations as values from 1 to n for oriented samples. . 161
5.14 a) Distribution of texture coefficients and ribbon tilt for Sb2 Se3 deposited on
substrates with increasing degree of nanostructuring; c) total (hk1) and (hk0)
texture and d) texture averaged ribbon tilt angle (θr(avg) ). . . . . . . . . . . . . . . 164
5.15 a) and b) Distribution of texture coefficients and tilt for (hk1) reflections of
Sb2 Se3 deposited on flat and nanostructure substrates, respectively, with in-
creasing mass of Sb2 Se3 in the source; c) crystal growth mechanisms on flat and
nanostructured substrates; d) total (hk1) and (hk0) texture coefficients and e)
texture averaged ribbon tilt angle (θr(avg) ) of Sb2 Se3 on ZnO NW and ZnO Flat
substrates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
xix
5.16 Scheme showing the procedure used to fabricate ZnO NW/Sb2 Se3 NR photode-
tectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.17 a) Cross section SEM images of ZnO NW/Sb2 Se3 NR-PMMA exposed to O2
plasma for different times and b) surface of devices. . . . . . . . . . . . . . . . . . 167
5.18 Photoresponse of ZnO NW/Sb2 Se3 NR-PMMA exposed to O2 plasma for dif-
ferent etching times under illumination with 625 nm red LED at 80 mW cm−2
and no bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
5.19 a) Photoresponse and b) power curve for ZnONW/Sb2 Se3 NR-PMMA(180s plasma),
ZnONW/CdS/Sb2 Se3 NR-PMMA(180s plasma) and ZnOFlat/Sb2 Se3 detectors.
A 625 nm LED was used as the illumination source. Photoresponse was ob-
tained at 8 mW cm−2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.1 a) Scanning electron microscope (SEM) cross section image of Sb2 Se3 NR on
ZnONW and b) scheme of transferring process from ZnONW to a new sub-
strate. The scale bar in the SEM corresponds to 1 µm. . . . . . . . . . . . . . . . . 176
6.2 Steps of Sb2 Se3 NR/PMMA transferring process. . . . . . . . . . . . . . . . . . . . 177
6.3 a) Scanning electron microscope (SEM) images of ZnONW/Sb2 Se3 NR/PMMA
(template) and layer Sb2 Se3 NR/PMMA transferred onto FTO; b) X-ray diffrac-
tion (XRD) pattern highlight the relative intensity of (002) reflection and c) tex-
ture coefficient of Sb2 Se3 for selected reflections before and after transfer; d)
photography of transferred Sb2 Se3 NR layer onto glass/FTO substrate. * ZnO
and ∨ FTO. The scale bars in the SEM correspond to 1 µm. . . . . . . . . . . . . 177
6.4 a) Full set of analyzed texture coefficients (TC) and b) TC vs ribbon tilt (θr ) for
(hk1) reflections of template and Sb2 Se3 NR transferred onto FTO. . . . . . . . . . 179
6.5 Reflectance of Sb2 Se3 NR layer/PMMA (90s plasma etching) for template (blue
line) and transferred (red line) layers. . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.6 a) Top-view backscattered scanning electron microscope (BS-SEM) images and
b) correspondent X-ray photoelectron spectra (XPS) of the Sb and Se regions of
Sb2 Se3 NR before and after O2 plasma exposure. The scale bars in the BS-SEM
correspond to 1 µm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
xx
6.7 a) Schematic representation of the stack of layers in the devices and cross sec-
tion SEM images of Sb2 Se3 NR/PMMA layer transferred onto b) glass/FTO,
c) glass/FTO/CdS and d) glass/FTO/SnO2 . Comparison of e) photocurrent
density (J ph ) and f) Responsivity (R)/Detectivity (D ∗ ) for devices built on the
template (FTO/ZnONW/Sb2 Se3 NR/PMMA) and on Sb2 Se3 NR/PMMA trans-
ferred layers. d) -3dB frequencies for transferred devices. Devices were an-
nealed at 175 °C in air. A red light emitting diode (LED), 625 nm at 115 W cm−2
irradiation was used as the illumination source for photocurrent measurements.
The scale bars correspond to 500 nm. . . . . . . . . . . . . . . . . . . . . . . . . . 181
6.8 X-ray photoelectron spectroscopy (XPS) of the Sb and Se regions of Sb2 Se3 /PMMA
layer etched for 90s in O2 plasma on the template (before transfer, red line) and
on the receiving substrate (after transfer, blue line). . . . . . . . . . . . . . . . . . 182
6.9 Energy dispersive spectroscopy (EDS) analysis showing that Sb2 Se3 NR are slightly
Se deficient (expected Sb2 Se3 At%: Se 60% and Sb 40%). . . . . . . . . . . . . . . 182
6.10 a) Scheme showing O2 plasma treatment used to expose the tips of Sb2 Se3 NR
by partially etching the PMMA film to electrically contact the drop casted back
electrode; b) X-ray photoelectron spectroscopy (XPS) of Sb2 Se3 /PMMA layer
before and after plasma etching; c) scanning electron microscope (SEM) image
of Sb2 Se3 /PMMA etched film. The scale bar corresponds to 1 µm. . . . . . . . . 183
6.11 Photocurrent density (J ph ) as a function of annealing temperature for a) tem-
plate device (ZnONW/Sb2 Se3 NR) and devices with Sb2 Se3 NR transferred to b)
glass/FTO, c) glass/FTO/CdS and d) glass/FTO/SnO2 . Devices were annealed
in air for 10min before each measurement. . . . . . . . . . . . . . . . . . . . . . . 184
6.12 a) Responsivity (R) and b) Detectivity (D ∗ ) as a function of annealing tempera-
ture for devices with Sb2 Se3 NR transferred to glass/FTO, glass/FTO/CdS and
glass/FTO/SnO2 compared to control (FTO/ZnONW). Both R and D ∗ were
calculated from 5s ON/OFF pulses of red light emitting diode (LED), 625 nm
at 115 W cm−2 irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.13 Relative frequency photoresponse for a) control device (FTO/ZnONW/Sb2 Se3 NR)
and for devices with a Sb2 Se3 NR transferred on b) FTO, c) FTO/SnO2 and d)
FTO/CdS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
xxi
6.14 a) Top view and b) cross section scanning electron microscope (SEM) images of
Sb2 Se3 NR film transferred onto FTO and annealed at 400 °C for 10 min under
N2 atmosphere. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.15 a) Scheme of Sb2 Se3 NR crystals on flat vs flexed substrate; b) flat and flexed
device at φ = 40° and φ = 54°. Annealed at 175 °C. . . . . . . . . . . . . . . . . . . 188
6.16 a) Photocurrent density (J ph ) as a function of annealing temperature for flexible
photodetector (PET/ITO/Sb2 Se3 NR) and b) as deposited film (not annealed)
with poor mechanical properties, broken upon flexing. . . . . . . . . . . . . . . . 188
6.17 a) PET/ITO/SnO2 /Sb2 Se3 NR flexible device, completed with drop cast conduc-
tive PEDOT back electrode. Normalized photocurrent (J ph ) b) up to 250x flexing
cycles at φ = 54° and upon flexing at φ = 0°, 40° and 54°. c) J ph response for 5s
pulses (LED, 625 nm, 115 W cm−2 ). d) Power curves for devices with and with-
out SnO2 ETL. e) Responsivity (R)/Detectivity (D ∗ ). f) Relative photoresponse
to frequency modulated light (-3dB). . . . . . . . . . . . . . . . . . . . . . . . . . . 189
6.18 Characterization of flexible photodetector at 0V showing the stability over long
light pulse. A light emitting diode (LED), 625 nm at 115 mW cm−2 irradiation
was used. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
6.19 Normalized photocurrent response for flexible photodetector under illumina-
tion with light emitting diodes (LED) with wavelength varying from 455 nm to
850 nm at 10 W cm−2 irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
6.20 a) Scheme highlighting the light transmitted through the biological tissue as the
blood volume in the arteries change; b) absorption spectrum of hemoglobin in
its oxygenated (HbO2 ) form; c) flexible Sb2 Se3 NR heart rate detector wrapped
around the finger; d) signal from pulse measurements and c) fast Fourier trans-
form (FFT) of the pulse signal. A red light emitting diode (LED), 625 nm at 100
W cm−2 irradiation was used as the illumination source. . . . . . . . . . . . . . . 192
6.21 Heart rate measured under infrared (730 nm) light using light emitting diode
(LED), at 100 mW cm−2 irradiation. . . . . . . . . . . . . . . . . . . . . . . . . . . 193
xxii
List of tables
1.1 Impedance equation for common circuit elements as a function of the angular
frequency of an applied potential wave (ω). . . . . . . . . . . . . . . . . . . . . . . 49
2.1 Sb2 Se3 crystallographic parameters and relevant properties. . . . . . . . . . . . . 58
2.2 Intensity ratio between (001)/(221) for highly oriented Sb2 Se3 films. . . . . . . . 73
3.1 Example of a typical heating program used for Sb2 Se3 evaporation. . . . . . . . . 93
3.2 Fitted parameters for champion devices as a function of Zone 2 temperature. . . 100
3.3 Fitted parameters for champion devices as a function of Sb2 Se3 mass in the source.100
A2 Average thickness for ZnO SG films deposited on glass/FTO substrates at 4000
RPM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
A4 Average thickness for TiO2 SG films deposited on glass/FTO substrates at 4000
RPM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
A7 Thickness measurements for CdS films deposited on FTO. . . . . . . . . . . . . . 107
A8 Relative atomic composition of CdS layers after post treatments via energy dis-
persive spectroscopy (EDS). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
A11 Root mean square (RMS) roughness of ZnO Flat and ZnO NW films . . . . . . . 112
4.1 Representative fitted diode parameters from dark JV curves with and without
HRMO deposition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.2 SCAPS parameters for Sb2 Se3 solar cell materials and CdS/HRMO interface. . . 139
5.1 Parameters for Sb2 Se3 photodetectors tested here. . . . . . . . . . . . . . . . . . . 168
6.1 Comparison of Sb2 Se3 photodetector performance. . . . . . . . . . . . . . . . . . 190
xxiii
List of abbreviations
JV current density vs voltage
PCE power conversion efficiency
VOC open circuit voltage
FF fill factor
JSC short circuit current density
VB valence band
CB conduction band
SCR space charge region
SCLC space charge limited current
TCO transparent conductive oxide
ETL electron transport layer
HTL hole transport layer
Eg band gap energy
EF Fermi level
EV valence band edge
EC conduction band edge
N doping density
ρe electron concentration
ρh hole concentration
Nd density of donor states
Na density of acceptor states
NC effective density of states in the conduction band
NV effective density of states in the valence band
KB Boltzmann constant
q electron charge
SCR space charge region
QNR quasi-neutral region
Rsh shunt resistance
RS series resistance
xxiv
n diode ideality factor
J0,1,2. . . reverse saturation current (diode current)
XRD x-ray diffraction
XPS x-ray photoelectron spectroscopy
SEM scanning electron microscope
TEM transmission electron microscope
EDS energy dispersive x-ray spectroscopy
EQE external quantum efficiency
EIS electrochemical impedance spectroscopy
AFM atomic force spectroscopy
HRMO high resistance metal oxide
SG sol-gel
VTD vapour transport deposition
RTE rapid thermal evaporation
CSS close space sublimation
CBD chemical bath deposition
SILAR successive ionic layer adsorption and reaction
NP nanoparticle
NW nanowire
NR nanorod
xxv
Abstract
The development of new materials and manufacturing technologies for photovoltaic (PV)
power generation is vital for transitioning to a renewable and cleaner energy economy. The
use of PV panels to produce electricity has spiked globally, rising by 12-fold in only 10 years.
Emerging photovoltaic materials, which can be processed via simple and low-cost solution or
evaporation methods, have the potential to significantly reduce manufacturing costs and in-
crease the efficiency of solar cells for the realisation of low-cost-high-efficiency commercial
solar panels. Novel PV materials have improved remarkably fast, within only a few years of
development perovskite-based devices achieved power conversion efficiencies (PCE) that are
on a par with the best single crystal silicon solar cells. Antimony chalcogenides (Sb2 (S, Se)3 ,
Sb2 S3 and Sb2 Se3 ) have recently emerged as an excellent light absorbing material for thin
film PVs. The selenide phase (Sb2 Se3 ) has garnered special attention as Sb2 Se3 solar cells
have recently surpassed the 10% efficiency benchmark. Still, this material technology requires
significant improvements to be competitive. Major challenges facing Sb2 Se3 devices are related
to poor interfaces and to the requirement to favourably align the quasi-1D (Sb4 Se6 )n ribbon
crystal structure. This thesis investigates and explores solutions to these issues over three
main experimental chapters.
Firstly, an in-depth experimental investigation into the effect of employing different high
resistance metal oxide (HRMO) layers on the quality of the front contact in superstrate Sb2 Se3
solar cells is presented. The crucial role of HRMO in preventing shunting and maintaining
device stability is verified, followed by an investigation into the generation/suppression of de-
fects and the presence of electron barriers. Second, it is shown for the first time that substrate
nanostructure plays a key role in driving the growth of (001) oriented Sb2 Se3 thin films. An
in-depth study on the mechanistic origin of oriented film growth on highly nanostructured
substrates is presented, wherein Sb2 Se3 orientation was found to be independent of substrate
surface chemistry. Finally, the templated growth of (001) Sb2 Se3 thin films as a viable method
for device fabrication is demonstrated through the fabrication of rigid and flexible photodetec-
tors and a proof-of-concept flexible heart rate detector. Overall, the investigations presented
in this thesis are essential to guide the further development of high-quality Sb2 Se3 solar cells
and to overcome fundamental challenges in the fabrication of high quality Sb2 Se3 layers with
controlled morphology and orientation.

1
Broad context and scope of the thesis
More than 60% of total electricity produced globally comes from coal, oil or natural gas.[1]
Greenhouse gases released by burning fossil fuels have generated unprecedented environmen-
tal impact, including global warming and climate change, which require urgent action. This
has led many countries to implement energy policies to integrate renewable sources into the
energy mix, gradually replacing fossil fuels. The use of solar photovoltaic (PV) power to
generate electricity has seen a sharp increase in recent years with global capacity rising from
0.29% in 2011 to 3.63% in 2021.[1] A 12-fold rise in only 10 years. Australia has been leading
the installation of solar photovoltaic panels globally and (as shown in Figure 1) solar PV is
the fastest growing renewable energy source for electricity generation in the country. In 2021,
11.7% of all electricity produced in Australia came from solar, compared to only 0.80% in
2011. The percentage of power from solar photovoltaics is expected to grow even faster in
the coming years reaching ≈32% by 2025.[2] Moreover, Australia has invested heavily in the
development of high-efficiency low-cost solar panels employing emerging technologies such
as perovskite/Si and chalcogenide/Si tandem solar cells.[3, 4]
Solar
Wind
25 TWh
Renewable electricity
in Australia
20 TWh
15 TWh Hydropower
10 TWh
5 TWh
Other renewable,
including bioenergy
0 TWh
1991 1995 2000 2005 2010 2015 2021
Figure 1: Renewable electricity installed capacity by source in Australia over 10 years. Adapted from
ref. [1] licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
2
Currently, the PV market is dominated by silicon-based solar cells owing to the rapid
reduction in the production cost of silicon PV panels.[5] Nevertheless, silicon panels are still
limited to large scale solar farms and mid/small scale roof-top applications. Other PV tech-
nologies, notably thin film solar cells based on cadmium telluride (CdTe) and copper indium
gallium selenide (Cu(In, Ga)Se2 - CIGS), have well established niche markets.[6] However, the
high toxicity of Cd as well as a low abundance of In and Ga present major problems for the
mass production of such devices on the same scale as Si devices.[7, 8]
New materials and manufacturing technologies can play a major role in boosting the in-
creasingly large global solar PV energy production capacity. Emerging PV thin film technolo-
gies have improved rapidly. Research prototypes of organic–inorganic halide perovskite cells
have exceeded the 25% power conversion efficiency (PCE) milestone and are now on a par with
record single crystal Si devices (26-27% PCE). Remarkably, perovskite/Si tandem solar cells
have reached extraordinary 32.9% PCE and perovskite/CIGS tandem devices achieved 25.2%
PCE.[6, 9] Although the low stability and high toxicity of Pb are currently major challenges,
massive research efforts and financial investment are being employed to make perovskite de-
vices available in the market in the next few years.[10–12] Other emerging PV technologies
include Copper zinc tin sulphide (CZTS), organic, and quantum dot solar cells, all of which
still require improvements in efficiency to become competitive in the market. In contrast to
the energy intensive and complex fabrication of Si panels, emerging thin film solar cells can
be processed via simple, low-cost, and relatively low-energy solution or evaporation methods
in a range of different device chemistries and architectures. These new technologies have the
potential to significantly reduce manufacturing costs, increase the efficient of solar cells and
expand PV panel implementation beyond solar farms and root-tops.[5, 7, 13]
3
a) b)
10.57% 200
10 Sb2Se3
*
Number of publications
solar cells
150
Efficiency (%)
* * *
100
5
50
0 0
2016 2018 2020 2022 2016 2018 2020 2022
Year Year
Figure 2: a) Best efficiency Sb2 Se3 solar cell by year and b) number of publications per year related to
Sb2 Se3 since 2015. *Research impacted by COVID-19 pandemic.
Antimony chalcogenides (Sb2 (S, Se)3 , Sb2 S3 and Sb2 Se3 ) have recently emerged as an ex-
cellent light absorbing material for thin film PVs and other photoelectronic applications.[14–
16] The selenide phase (Sb2 Se3 ) has received special attention as it is highly stable,[17] has
good electron and hole mobility (≈15 and ≈40 cm2 V−1 s−1 ),[18] and a high optical absorption
coefficient (> 105 cm−1 ) through a 1.2 eV direct band gap.[19] Moreover, facile processing
of thin films by thermal evaporation can be achieved owing to the low vapour pressure of
Sb2 Se3 and the absence of crystal polymorphism.[20] Remarkably, Sb2 Se3 solar cells broke the
10% efficiency milestone in only 10 years of development (Figure 2a).[21, 22] Considering its
theoretical potential, rapid efficiency evolution, and the intensive research efforts employed
towards device development (Figure 2b) antimony selenide is set to overcome stagnated tech-
nologies (e.g., CZTS and dye sensitized devices).
4
Thesis outline
Introduction and background
• (Chapter 1) Introduction to semiconductor electronic structure and modelling of solar
cells.
• (Chapter 2) Review of Sb2 Se3 properties and thin film fabrication methods.
• (Chapter 3) Detailed experimental procedures.
Scope of investigations
• (Chapter 4) This chapter presents a study on the effect of high resistance metal oxide
(HRMO) layers on the quality of interfaces at the front contact of HRMO/CdS/Sb2 Se3
superstrate solar cells. The effect of two HRMOs, ZnO, and TiO2 on the overall de-
vice performance is comprehensively studied to determine efficiency and stability gains
as well as identify the generation/suppression of defects and the presence of electron
barriers.
• (Chapter 5) This chapter shows for the first time that substrate nanostructure plays a key
role in driving the growth of (001) oriented Sb2 Se3 layers. A growth mechanism is pro-
posed, wherein the substrate physically restricts the development of undesirable (hk0)
crystallographic orientations. Remarkably, the growth of (001) Sb2 Se3 on nanostructured
substrates was independent of substrate’s surface chemistry. Insights from this work are
expected to guide the design of oriented layers of low-dimensional functional materials.
• (Chapter 6) This chapter presents a template growth method for (001) oriented Sb2 Se3
light absorber layers, applied as a viable strategy for device applications. Thin films
grown on the template are readily transferable to functional substrates via a polymer
assisted method. Proof-of-concept photodetectors show satisfactory response. Finally,
the method was used to make highly flexible self-powered heart rate detectors.
Conclusions
• (Chapter 7) Conclusions and perspectives.
5
References
[1] H. Ritchie, M. Roser, P. Rosado, Our World in Data: Energy, https://ourworldindata.org/
energy, accessed: May, 2023.
[2] DCCEEW, Australia’s Emissions Projections 2022, Canberra, 2022.
[3] ARENA, Cost-effective Si/Perovskite Tandem Modules on Passivating Contact Si Cells,
https://arena.gov.au/projects/efficient-stable-chalcogenide-si-tandem-cells-integrating-
pv-technologies/, accessed: May, 2023.
[4] ARENA, Efficient & Stable Chalcogenide-Si Tandem Cells: Integrating PV Technolo-
gies, https://arena.gov.au/projects/efficient-stable-chalcogenide-si-tandem-cells-integrating-
pv-technologies/, accessed: May, 2023.
[5] M. A. Green, A. Ho-Baillie, H. J. Snaith, Nat. Photonics 2014, 8, 506.
[6] NREL, Best Research-Cell Efficiency Chart, https://www.nrel.gov/pv/assets/pdfs/
best-research-cell-efficiencies.pdf, accessed: May, 2023.
[7] T. D. Lee, A. U. Ebong, Renewable Sustainable Energy Rev. 2017, 70, 1286.
[8] G. Niu, X. Guo, L. Wang, J. Mater. Chem. A 2015, 3, 8970.
[9] M. Jošt, E. Köhnen, A. Al-Ashouri, T. Bertram, Š. Tomšič, A. Magomedov, E. Kasparavi-
cius, T. Kodalle, B. Lipovšek, V. Getautis, R. Schlatmann, C. A. Kaufmann, S. Albrecht,
M. Topič, ACS Energy Lett. 2022, 7, 1298.
[10] G. M. Wilson, M. Al-Jassim, W. K. Metzger, S. W. Glunz, P. Verlinden, G. Xiong, L. M.
Mansfield, B. J. Stanbery, K. Zhu, Y. Yan, J. J. Berry, A. J. Ptak, F. Dimroth, B. M. Kayes,
A. C. Tamboli, R. Peibst, K. Catchpole, M. O. Reese, C. S. Klinga, P. Denholm, M.
Morjaria, M. G. Deceglie, J. M. Freeman, M. A. Mikofski, D. C. Jordan, G. TamizhMani,
D. B. Sulas-Kern, J. Phys. D: Appl. Phys. 2020, 53, 493001.
[11] P. Wu, D. Thrithamarassery Gangadharan, M. I. Saidaminov, H. Tan, ACS Cent. Sci.
2023, 9, 14.
[12] X. Zheng, A. Y. Alsalloum, Y. Hou, E. H. Sargent, O. M. Bakr, Acc. Mater. Res. 2020, 1,
63.
[13] F. A. Jhuma, M. Z. Shaily, M. J. Rashid, Mater. Renew. Sustain. Energy 2019, 8, 1.
[14] Z. Li, X. Liang, G. Li, H. Liu, H. Zhang, J. Guo, J. Chen, K. Shen, X. San, W. Yu, R. E. I.
Schropp, Y. Mai, Nat. Commun. 2019, 10, 125.
[15] L. Wang, D.-B. Li, K. Li, C. Chen, H.-X. Deng, L. Gao, Y. Zhao, F. Jiang, L. Li, F. Huang,
Y. He, H. Song, G. Niu, J. Tang, Nat. Energy 2017, 2, 17046.
[16] X. Wen, C. Chen, S. Lu, K. Li, R. Kondrotas, Y. Zhao, W. Chen, L. Gao, C. Wang, J.
Zhang, G. Niu, J. Tang, Nat. Commun. 2018, 9, 2179.
[17] Y. Zhou, L. Wang, S. Chen, S. Qin, X. Liu, J. Chen, D.-J. Xue, M. Luo, Y. Cao, Y. Cheng,
E. H. Sargent, J. Tang, Nat. Photonics 2015, 9, 409.
[18] S. Hadke, M. Huang, C. Chen, Y. F. Tay, S. Chen, J. Tang, L. Wong, Chem. Rev. 2022,
122, 10170.
[19] C. Chen, W. Li, Y. Zhou, C. Chen, M. Luo, X. Liu, K. Zeng, B. Yang, C. Zhang, J. Han,
J. Tang, Appl. Phys. Lett. 2015, 107, 043905.
6
[20] M. Luo, M. Leng, X. Liu, J. Chen, C. Chen, S. Qin, J. Tang, Appl. Phys. Lett. 2014, 104,
173904.
[21] Z. Duan, X. Liang, Y. Feng, H. Ma, B. Liang, Y. Wang, S. Luo, S. Wang, R. E. I. Schropp,
Y. Mai, Z. Li, Adv. Mater. 2022, 34, 2202969.
[22] Y. Zhao, S. Wang, C. Li, B. Che, X. Chen, H. Chen, R. Tang, X. Wang, G. Chen, T. Wang,
J. Gong, T. Chen, X. Xiao, J. Li, Energy Environ. Sci. 2022, 15, 5118.
7
Chapter 1
Electronic structure and
modelling of solar cells
8
1.1 Preface
This chapter outlines the fundamental electronic structure of semiconductors and devices,
focusing on Sb2 Se3 thin film solar cells. The chapter also introduces relevant concepts and
techniques that are commonly employed for solar cell characterisation, including a detailed
discussion on the modelling of current density vs voltage (JV) curves and an introduction
to relevant models in impedance spectroscopy. Modelling facilitates the assessment of the
causes of efficiency losses and gains in solar cells by relating fundamental electronic phenom-
ena to the microstructure of devices. Overall, this chapter serves as an introduction to the
fundamental concepts and techniques related to Sb2 Se3 solar cell research.
9
1.2 Electronic structure of solid materials
1.2.1 Energy levels
An electron that is subject to a constant energy field can assume any energy value. How-
ever, if the electron is in an energy field that varies in space it may only assume discrete energy
values. The possible energies that an electron can assume are called energy levels. As shown
in Figure 1.1, this situation can be modelled in 1D using an infinite quantum well model.[1]
U(x)
I II III
U=0
U→∞ U→∞
0 a x
Figure 1.1: Illustration of an infinite square potential well model, where the particles (such as an
electron) can only exist within region II (U ( x ) = 0).
The Schrödinger equation for the electron is
2m d2 Ψe
2
( E − U ( x ))Ψe + (1.1)
h̄ dx2
Region I and III (0 ≤ x ≥ a, U ( x ) → ∞): an electron cannot pass through the potential
barrier and will not exist in this region, so the only possible solution is Ψe = 0.
Region II (0 < x < a, U ( x ) = 0)
2m d2 Ψe
EΨe + =0 (1.2)
h̄2 dx2
the solution takes the form
Ψe = Aeαix + Be−αix (1.3)
10
√ where Ψe is a solution of the Schrödinger equation. It is also convenient to define: k =
2mE
, where m and E are the mass and energy of the electron, respectively.
h̄
The wave function Ψe should be continuous and satisfy all possible values of x within
Region II. Applying the boundary conditions and replacing k:
x = 0, Ψe = 0 ⇒ A = − B (1.4)
x = a, Ψe = 0 ⇒ A(ekia − e−kia )
This can be solved using a fundamental trigonometric relationship expressed by Euler’s
formula (eiθ = cos θ + i sin θ), then we get Ψ(ka) = 0. Given that Ψ = 0 at the boundaries and
A 6= 0, to avoid a trivial solution this condition is only satisfied when ka = n E π, where n E =
integer and a is the width of the quantum well. Finally, by recalling k and rearranging we get
Equation 1.5.
π 2 h̄2 2
E= n , n E = 1, 2, 3, 4... (1.5)
2ma2 E
This important result illustrates that energy is quantized for the electron subject to a
variable potential. Figure 1.2 shows the allowed energy levels (n E ) for the electron in a 1 nm
quantum well calculated using Equation 1.5.
E (eV)
nE =
100
80 15
60
40 10
20
5
0 1
Figure 1.2: Allowed energies for an electron confined in a 1 nm quantum well
11
1.2.2 Energy band and band gap
Inside a crystal, the electron will be subject to a potential that will vary periodically
following the crystal lattice.[2, 3] A rough simplification of this case is shown in Figure 1.3.
This simple periodic potential model can be used to illustrate an important aspect of the
electronic structure of solid materials, the existence of allowed and forbidden energy regions.
The electrons will only occupy energy levels inside the allowed energy bands. The forbidden
energy zones form band gap regions that electrons cannot occupy, i.e., no energy level is located
within the band gap of a perfect crystal. Energy levels can appear within the band gap
as a consequence of defects in the crystal and have important implications for the practical
application of materials.
U(x) U₀
I II I II I
U=0
-b 0 a x
Figure 1.3: Illustration of a periodic potential where particles such as electrons may tunnel through the
potential barriers.
The general form of Schrödinger equation for the electron, stated in the previous section,
takes the form:
Region I (U ( x ) = 0)
2m d2 Ψe
EΨe + =0 (1.6)
h̄2 dx2
Region II (U ( x ) = 0)
2m d2 Ψe
2
( E − U0 )Ψe + =0 (1.7)
h̄ dx2
12
These equations can be solved for any value of x. The solution takes the form Ψe =
u( x )eikx . The function u( x ) modulates the amplitude of the wave function and reflects the
periodicity of the potential barriers. It is also convenient to introduce a wave vector ~k that is
related to the momentum of the electron ~p by:
~p = h̄~k (1.8)
The variables are emphasised as vector quantities as they assume different values in
different crystallographic directions.
The wave equation Ψe = u( x )eikx , where u( x ) is the Bloch function that reflects the peri-
odicity of the potential, can be solved to give a general solution valid for all values of x in the
limit
sin ka
f (ka) = P + cos ka = cos ka, −1 ≥ cos ka ≤ 1 (1.9)
ka
with
maU0 b
P=
h̄2
The solution is illustrated in Figure 1.4, where P = 10 was chosen arbitrarily. The so-
lution extends symmetrically for negative values of x that are not shown. As highlighted
in the figure, one important consequence of this solution is that cos ka is defined only be-
tween ±1, giving origin to allowed and forbidden energy regions. The function f (ka) will be
π
discontinuous at k = n , where is an integer n = 1, 2, 3....
a
f(ka)
allowed
energies
forbidden
energies
+1
-1 ka
2π 4π
Figure 1.4: Illustration of allowed energy bands and band gaps (forbidden energy regions) for a particle
subject to a periodic potential.
13
In Figure 1.5a, the energy is plotted as a function of k (E(k )). Figure 1.5b shows the
correspondent simplified energy band representation, with the regions of electrons occupation
(energy bands) highlighted in grey. The forbidden energy region (band gap) is also shown.
a) b)
E E
energy
bands
band
gap
-2π -π π 2π ka
Figure 1.5: a) Illustration of allowed energy zones as a function of the wave vector (k) and correspon-
dent b) energy band representation.
1.2.3 Metals, semiconductors and insulators

According to their electrical conductivity properties, solid materials can be classified into
three categories: metals, insulators, and semiconductors. Metals are characterised by having
high electrical conductivities (107 Ω−1 m−1 ). Insulators sit at the other end of the spectrum
having very low conductivities (<10−10 Ω−1 m−1 ). Semiconductor materials have moderate
conductivity (≈10−6 Ω−1 m−1 to ≈104 Ω−1 m−1 ).[4]
The electrical properties of solids are determined by the arrangement of atoms and elec-
trons in the material. This is illustrated in Figure 1.6. In isolated atoms, e.g. in a gas, electrons
will occupy individual energy levels, shells (1, 2, 3, 4...) and subshells (s, p, d and f ), low en-
ergy states are occupied first giving the atom its characteristic electron configuration. If a set
of N atoms is brought together, the electronic outer shells of individual atoms start to interact.
The final number of energy states in the band will be the same as the number of states in
the N atoms that interact to form the band. Eventually, the atomic energy states split into
a set of energy levels closely packed forming the energy bands. This means that bands are
formed by a massive number of individual states, even for a small volume of material. Each
energy state in the band corresponds to a unique solution of the Schrödinger equation, thus
the Pauli exclusion principle is preserved when states are occupied by charge carriers. For or-
bitals, s contributes N states and p orbitals contribute 3N states, thus the band resulting from
14
the interaction of s and p orbitals will have 4N energy states. As atomic distance decreases,
the interaction between atomic orbitals increases and the splitting of energy bands becomes
more complex. Ultimately, this interaction determines the metallic, insulator or semiconductor
characteristic of materials.
4N states
Energy
split energy states band

representation
N states
1s
3N states
1p } band gap
individual energy states
equilibrium distance
solid Interatomic distance gas
Figure 1.6: Illustration of atomic energy states splitting in a solid material and the corresponding band
diagram.
Band energy diagrams are used to represent the electronic configuration of solid mate-
rials in equilibrium. Metals, insulators, and semiconductors have characteristic energy band
diagrams. Metals have half-filled bands and electrons can move almost freely between closely
spaced empty levels, resulting in their characteristic high conductivity. Insulators have a low
energy band that is completely filled, valence band (VB), which is separated by a large band
gap (>2 eV) from a high energy band that is completely empty, conduction band (CB). Electrons
will normally not overcome this gap and are not free to move between energy levels - result-
ing in electrical insulation. Similarly, semiconductors have a filled valence band but with a
lower band gap (<2 eV). Electrons in the VB of semiconductors are not free to move but can
overcome the band gap relatively easily if enough energy (E ≥ Eg ) is provided, e.g., as light
or thermal energy for example. After an electron is excited to the CB, a hole (h+ ) representing
the absence of an electron is left in the VB. Excited electrons in the CB, and holes in the VB,
are relatively free to move.
15
a) b) CB c) CB
band
overlap gap
empty states
filled states VB VB
Metals Insulators semiconductors
Figure 1.7: Energy band diagram representation for solid materials in equilibrium at 0 K a) in metals,
such as Au, Ag, Cu and Al, a lower energy band is only half filled, while in Mg, Zn and Pb, for
example, a filled band overlaps an empty one. Insulators in b) have a filled valence band (VB), an
empty conduction band (CB) and a band gap that is relatively large compared to that of semiconductors
shown in (c). Electrons cannot have energies within the band gap region.
1.2.4 Density of states

An electronic state corresponds to a unique solution of the Schrödinger equation. Each
energy level can be occupied by two electrons with opposite spins, which corresponds to two
distinct electronic states. The distribution of states within the energy band is described by a
density of states function. This function can be solved for a simple model, the electron in an
infinite quantum box, and has a more general implication.
The 3D equivalent to the electron in an infinite quantum well is the electron in an infinite
potential quantum square box. As illustrated in Figure 1.8, the potential outside the box is
infinite and the electron cannot exist in this region, i.e. Ψe = 0 outside the box. Inside the box,
the potential is zero and the electron is free.
z
a
U→∞
U=0 y
a
x
Figure 1.8: Infinite potential quantum box.
The Schrödinger equation for the 3D case can be written as:
h̄2 2
− 5 Ψe + UΨe = EΨe (1.10)
2m xyz
16
This equation can be solved in a similar way to the 1D case. The solution yields:
h̄2 π 2 2
En = (n + n2y + n2z ) (1.11)
2ma2 x
One important result that arises from this model is the total number of energy levels, up
to a certain level En . This can be found by defining a ‘quantum space’ containing all possible
(energy) numbers n inside the volume of a sphere (Figure 1.9).
ny
nE
nz nx
nE² = nx² + ny² + nz²
Figure 1.9: Spherical space containing all possible energy numbers (n E ).
The total number of states (η) that is contained in one quadrant of the sphere is:
14 3 π a 3
η= πn E = (2mEn )3/2 (1.12)
83 6 h̄π
It is convenient to define the number of energy levels by unity of volume and energy,
which is defined as the density of energy levels (NE ( E)):
3/2
1 dη 1 2m

NE ( E) = 3 = E1/2 (1.13)
a dE 4π h̄
Multiplying NE ( E) by two (each energy level can be occupied by two electrons with
opposite spins), gives the density of states g( E) or DOS (Equation 1.14).
3/2
1 dη 1 2m

g( E) = 3 = E1/2 , E ≥ 0 (1.14)
a dE 2π h̄
17
1.2.5 Density of states in semiconductors
Electrons in the CB of a semiconductor are somewhat free to move through the semicon-
ductor crystal and can be approximated using the infinite quantum box model. The density
of states of electrons in the CB of a semiconductor is given in Equation 1.15.
8π (2m∗e )3/2 p
gC ( E ) = E − EC , E ≥ EC (1.15)
3h3
where the minimum energy an electron can have in the conduction band is EC and m∗e effective
mass of electrons in the semiconductor.
The density of states for holes in the VB is expressed in Equation 1.16.
8π (2m∗h )3/2 p
gV ( E ) = EV − E, E ≤ EV (1.16)
3h3
where the maximum energy holes can have in the VB is EV and m∗h is the effective mass of the
hole in the semiconductor.
Figure 1.10 shows the density of states of electrons in the CB and holes in the VB and
illustrates the corresponding separation between each energy level. At the bottom of the CB,
electrons occupy energy levels separated by a relatively large energy, while higher energy
electrons (far from the band edge) can move between more closely spaced levels. It is also
convenient to define the band gap energy as Eg = EC − EV .
E
CB
EC
Eg
EV
VB
g(E)
Figure 1.10: Density of states and schematic representation of the state distribution within the CB and
VB of a semiconductor.
18
1.2.6 Fermi-Dirac statistics and Fermi Level
The occupation of states by electrons or holes occurs according to fundamental physical
principles, and can be summarised into a set of rules. First, the Pauli exclusion principle
is preserved, i.e., that electrons in the same energy level must have opposite spins. Second,
levels are defined by their energy only, not by their momentum. Finally, occupation leads to
minimum free energy, meaning that although lower energy levels tend to be filled first (at
T = 0K), some electrons can occupy higher energy levels if this increases the total entropy of
the system (at T > 0K).
The Fermi-Dirac function is a type of statistic of distribution that satisfies these rules.
This function, Equation 1.17, describes the distribution of electrons or holes within an energy
band, i.e. the probability of finding a charge carrier in a level with a given energy E.
1
f ( E) = (1.17)
1 + e ( E − EF ) / ( K B T )
where K B is the Boltzmann constant and T is the absolute temperature.
The term EF in Equation 1.17 is the Fermi level. As highlighted in Figure 1.11, EF
corresponds to the level where electrons would have 50 % probability to be found, that is
f ( E = EF ) ≈ 0.5, for f ( E EF ) ≈ 1 and f ( E EF ) ≈ 0.

f ( E)
EF
Figure 1.11: Fermi function, or Fermi distribution, that describes the distribution of electrons or holes
within an energy band.
19
1.2.7 Carrier density and effective density of states
The density of electrons per unity of volume at a given energy in the conduction band
(ρe ( E)) is the product of the density of states in the conduction band (gC ) and the probability
each state has to be occupied ( f ( E)):
ρ e ( E ) = gC ( E ) f ( E ) (1.18)
The density of holes at a given energy in the valence band (ρh ( E)) is proportional to the
probability of a state not being occupied:
ρh ( E) = gV ( E)(1 − f ( E)) (1.19)
where gV ( E) is the density of states in the valence band.
The density of charge carriers in the CB and VB is illustrated in Figure 1.12.
E E E
ρe
EC
Eg EF EF
EV
ρh
0 0.5 1
g(E) f(E) ρe or ρh(E)
Figure 1.12: Carrier density in the conduction and valence bands as a result of ρe ( E) = gC ( E) f ( E) and
ρh ( E) = gV ( E)(1 − f ( E)), respectively.
To find the total number of charge carriers in a band, electrons (ne ( E)) or holes (nh ( E)), it
is necessary to integrate ρe ( E) or ρh ( E) over all energy values within the band. For electrons
that is:
Z ∞ Z ∞
ne = ρe ( E)dE = gC ( E) f ( E)dE (1.20)
EC EC
and for holes:
20
Z EV Z EV
nh = ρh ( E)dE = gV ( E)(1 − f ( E))dE (1.21)
−∞ −∞
These integrals can be solved analytically when E − EF K B T. In this case, the Fermi
distribution can be simplified to a Boltzmann distribution ( f ( E) = e(E−EF )/KB T ). Solving the
integrals, the total number of electrons in the conduction band ne can be found:
EF − EC
ne = NC e K B T (1.22)
and the total number of holes in the valence band (nh ) is:
EV − EF
nh = NV e K B T (1.23)
The constants NC and NV are conveniently defined as the effective density of states in the
conduction and valence bands, respectively. They are expressed by Equation 1.24 and Equa-
tion 1.25.
3/2
2πm∗e kT

NC = 2 (1.24)
h2
3/2
2πm∗h kT

NV = 2 (1.25)
h2
where m∗e and m∗e are the effective mass of electrons and holes, respectively.
The effective density of states gives the total number of states per unit of space and has
unit of cm−3 in 3D.
In equilibrium at 0 K electrons will fill the lowest energy levels available (conduction
band). At higher temperatures, the thermal excitation of electrons will excite some electrons
to higher energy levels. Thermal activation will result in the same density of electrons and
holes in the conduction and valence bands, respectively. An important relation that arises
from the fact that ρe = ρh is the mass action law (Equation 1.26).
21
ρe ρh = ρ2i (1.26)
where ρi (cm−3 )is the carrier density that can be expressed in terms of NC and NV as:
−( EF − EC ) −( EV − EF ) p − Eg
ρi = NC e KB T
= NV e KB T
= NC NV e 2KB T (1.27)
1.2.8 Shallow and deep energy levels

In a perfect semiconductor crystal, there are no energy levels within the band gap. Impu-
rities and defects (including interstitial impurities, vacancies, antisites, dislocations and grain
boundaries) may be introduced in the semiconductor, deliberately or as a consequence of
faulty fabrication processes. The presence of these defects generates an excess of free charge
carriers in the semiconductor that introduces extra energy levels that can be located in between
the conduction and valence bands. These energy levels can be shallow or deep in energy. Shal-
low levels are located near either band, typically within several orders of K B T from the band
edge, so they are easily accessible upon thermal excitation. Deep levels are located in the
middle of the band gap, K B T from the band edge, and require higher energy to populate.
The shaded region of Figure 1.13[5] indicates the region where deep traps are formed. Deep
defects can capture electrons and holes, hence the name traps. A high concentration of deep
traps leads to extensive charge carrier recombination and efficiency loss in solar cells.
EC
I III
Etrap
II IV EV
Figure 1.13: Region in the band gap where deep defects (traps) are formed. Process in the trap: I) and
III) are the capture and re-emission of an electron and II) and IV) are the capture and re-emission of a
hole. Adapted with permission from ref. [5]. Copyright 2011 Wiley-VCH.
22
1.2.9 Intrinsic, n-type and p-type semiconductors
An intrinsic (undoped) semiconductor has no energy levels within the band gap and the
concentration of electrons is equal to the concentration of holes in equilibrium (Equation 1.26),
EF is located at the centre of the band gap. Semiconductors with shallow energy levels within
the band gap are classified as n-type (excess electrons) or p-type (excess holes). Energy levels
within the band gap that are occupied by electrons can donate their electrons and are called
donor levels, conversely, unoccupied levels can accept electrons and are called acceptor levels.
Doping consists of the deliberate introduction of impurities or defects to make a semi-
conductor n-type or p-type. Figure 1.14 illustrates Si (grey) doping with P (magenta) and B
(blue). Doping Si with P results in an n-type semiconductor. As P is located in group 15 in
the periodic table it has an extra electron in its outer shell in relation to Si (Figure 1.14a). This
generates an excess of free electrons in the crystal, while holes are minority charge carriers,
and introduces a donor level (ED ) just below the conduction band edge. In n-type semicon-
ductors, EF shifts closer to the conduction band edge. Yet doping Si with B results in a p-type
semiconductor. As B (located in group 13) has one less electron in this outer shell in relation
to Si (Figure 1.14b) it introduces acceptor energy levels (E A ) just above the valence band edge,
which generates an excess of holes while electrons are minority charge carriers. In p-type
semiconductors, EF shifts closer to the valence band edge.
a) b) c)
CB CB CB
EC EC EC
ED EF
Eg EF Eg Eg
EA
EV EV EF
EV
VB VB VB
intrinsic n-type p-type
Figure 1.14: Schematic representation of crystal structure and energy band diagrams of elemental
semiconductor (Si - grey) a) without impurities and with b) donor impurity (P - magenta) and b)
acceptor impurity (B).
In highly doped semiconductors the total concentration of carriers in the conduction and
valence will be close to the effective density of acceptor (Na ) and donor (Nd ) states:
23
ρe = Nd (1.28)
ρh = Na (1.29)
It is worth noting that in semiconductors with a more complex composition, e.g. metal
oxides or chalcogenides, the lack or excess of an element can affect the carrier concentration
and define the n or p-type characteristic of the semiconductor. In metal oxide semiconductors,
the non-stoichiometric excess or lack of oxygen can make the semiconductor present more n
or p-type characteristics, e.g. the excess of oxygen in ZnO gives it a strong n-type character.[6]
Impurities introduced during the processing of semiconductor layers (i.e., ligands from solu-
tion processing methods or impurities from chemicals) can also affect carrier concentration,
e.g. presence of Cl− was shown to give an n-type character to Sb2 Se3 .[7]
1.3 Carrier generation, recombination and transport

1.3.1 Generation and recombination mechanisms
Free carriers are generated in the semiconductor by processes that provide at least the
minimum ‘generation energy’ for the formation of an e− /h+ pair. In semiconductors with a
small band gap the thermal energy at room temperature, K B T, will be enough to excite some
electrons to the conduction band. The high energy particles from an electron or ion beam,
lattice vibration (phonons) or light (photons) can also provide the generation energy. In solar
cells, the excitation of electrons by light is the most important process for generating e− /h+
pairs. Light generated e− /h+ pairs (excitons) is also a vital process in working photodetectors
and photoelectrochemical cells.
The excitation of electrons by light is illustrated in Figure 1.15. In intrinsic semiconduc-
tors, the minimum generation energy is equal to the band gap energy (Eg ). Shallow levels in-
troduced in the semiconductor by dopants will decrease the energy needed for the formation
of an e− /h+ pair. Deep energy levels will trap electrons or holes leading to recombination.
24
a) b)
CB CB
EC EC
h⁺ e⁻
Eg
EV hν ≥ Eg EV
VB VB
Semiconductor in the dark Semiconductor under illumination
Figure 1.15: Schematic representation of crystal structure and energy band diagrams of a semiconduc-
tor a) in the dark and b) illuminated with light with energy higher than Eg .
Recombination annihilates e− /h+ pairs and reduces the number of free charge carriers
in the semiconductor and can be detrimental to solar cell performance. The mechanisms of
recombination are shown in Figure 1.16. First, Figure 1.16a illustrates e− /h+ pair generation
by absorption of a photon. The band-to-band (radiative) recombination (Figure 1.16b) is the
reverse of this generation process. The electron relaxes back to its original energy level, re-
combines with a hole, with both being annihilated in the process. Energy is then released
as a photon with energy equal to the energy difference between the two levels. Auger re-
combination (Figure 1.16c) also involves a band-to-band process, but instead of creating a
photon, energy is transferred to another particle that may then dissipate its energy by collid-
ing with the crystal network, creating phonons.[8] Nonetheless, recombination via trap states
(Shockley-Read-Hall recombination, SRH) can significantly impact device efficiency.
Recombination via trap states (Figure 1.16d) is the most relevant for working solar cells.
This type of recombination is related to defects in the bulk crystal lattice and interfaces that
generate deep energy levels. These recombination sites are in most cases detrimental to the
performance of semiconductor devices that rely on a high concentration of free charge carrier
to operate.
25
a) b)
EC EC
Energy
Energy
Gγ
hν = ΔECV hν = ΔECV
G R
EV EV
γ → e- + h+ e- + h+ → γ
e /h pair generation
- +
band-to-band recombination
c) d)
EC
Energy
Energy
Re
Erec
R
G Rh
EV
k
Auger recombination via trap state
Figure 1.16: Schematic representation of a) e− /h+ pair generation, b) band-to-band recombination,

c)Auger recombination and d) recombination via trap state. Where γ represents a photon and R is the
recombination rate.
back
light absorber ETL TCO
contact
Figure 1.17: Schematic representation of recombination in different regions of the solar cell 1) at the
interface between light absorber and electron transport layer (ETL), 2) in the bulk of light absorber, 3) at
the interface between light absorber and the back contact, 4) within light absorber space charge region
(SCR), 5) in the bulk of ETL. Adapted with permission from ref. [5]. Copyright 2011 Wiley-VCH.
26
1.3.2 Regions of recombinations
As illustrated in Figure 1.17,[5] deep trap states, that act as path ways for recombination,
may be present in different regions of a solar cell.
Bulk defects may be intrinsic to the material or arise due to faulty fabrication processes
that introduce defects such as impurities, vacancies or antisites. Bulk recombination might
be reduced by improving the quality of the film, via an improved deposition method or by
means of adequate post-treatment, e.g. annealing.
Interfaces are especially prone to recombination. Surfaces and grain boundaries usually
have a high concentration of defects due to the termination of the crystal lattice leaving ‘dan-
gling bonds’ or due do the fact that impurities are adsorbed relatively easily at interfaces. These
surface defects can generate levels in the forbidden band region that can be distributed in en-
ergy within the band gap (deep or shallow). A high concentration of deep interface defects
often leads to efficiency losses in solar cells. Interface recombination is a major issue in 3D
crystal structured materials (e.g. Si and GaAs). Interface passivation through the deposition of
a passivation layer, such as SiO2 on Si, is crucial to reduce recombination by avoiding abrupt
termination of the 3D crystal lattice. Recombination may also be reduced in highly doped
semiconductors that have a concentration of shallow levels significantly higher than deep de-
fects. In low dimensional materials, such as Sb2 Se3 which has a crystal structure composed
of quasi-1D, intrinsically passivated grain boundaries can be achieved via orientation control,
with the grain boundaries of orientated grains having no dandling bonds. However, interfaces
between the different materials that constitute the solar cells are still a major concern in Sb2 Se3
devices. Optimal front contact interfaces in Sb2 Se3 is the subject of Chapter 4 of this thesis.
1.3.3 Continuity equation and diffusion length

A continuity equation describes the transport of a quantity. It is based on the fact that
quantities are locally conserved and move in a continuous flow. Equation 1.30 is its differ-
ential form for a generic particle. The concentration (ci ) of particles i in the volume element
x increases over time (t) at a generation rate (Gi ) and decreases via recombination rate Ri .
Particles can also escape and enter the volume element, which is accounted in the resultant
current density of particles in the source term 5 j.
27
∂ci ( x )
= Gi ( x ) − Ri ( x ) − 5 ji (1.30)
∂t
In a solar cell under illumination the concentration of the carriers, e.g. electrons (ce ),
change as a function of the generation rate (Ge ), recombination rate (Re ) and current (incoming
flux of electrons Je ( x ) and outgoing flux of electrons Je ( x + dx). The same applies to h+ .
Je(x) Re Je(x+dx)
Ge
Figure 1.18: Schematic representation of continuity equation for electrons in a solar cell where Je ( x ) and
J ( x + dx ) is the flux of incoming electrons to and from the boundary element, Re is the recombination
rate and Ge is the generation rate.
An important result that arises from this equation is the diffusion length of carriers (Le
for electrons, Lh for holes). The diffusion length is the average distance travelled by minority
charge carriers inside a semiconductor from the moment it is generated (t = 0) to the moment
it recombines in the bulk of the semiconductor after some time (∆t). The time between carrier
generation and recombination is the carrier lifetime (τ).
The diffusion current density of electrons (je ) in the x direction can be written in terms of
the diffusion coefficient (De ) as
dne d2 ne
je = − De ⇒ 5 je = − De 2 (1.31)
dx dx
ne ne ( x )
and the generation and recombination rates as Ge (0) = and Re ( x ) = .
τe τe
The continuity equation can be solved to give Le in terms of De and τe , and it takes the
form of Equation 1.32 for electrons (in the p-type semiconductor),
√
Le = De τe (1.32)
and Equation 1.33 for holes (in the n-type semiconductor).
28
p
Lh = Dh τh (1.33)
Materials with high diffusion lengths are preferable since it increases the probability that
carriers will be transported through the bulk of the material and collected by the respec-
tive transport layers or electrodes. High levels of recombination, especially via trap states,
will dramatically reduce the carrier diffusion length. Highly doped materials have a smaller
Lcarrier since the number of defects is higher than in an intrinsic semiconductor. The diffu-
sion coefficient, and diffusion length, can also vary within the same material along different
crystallographic directions, anisotropic diffusion length is significant in materials with a low
symmetry crystal structure. In the case of quasi-1D materials, such as Sb2 Se3 , the diffusion
coefficient is significantly higher along the ribbons, similarly in quasi-2D layered materials
diffusion coefficient is higher within the layers.
1.3.4 Carrier transport in semiconductors

In a perfect crystal, charge carriers do not interact with the crystal network and can
accelerate freely through the crystal lattice. In reality, as illustrated in Figure 1.19, lattice im-
perfections and impurities will scatter electrons and holes. The random movement of charges
due to associated thermal energy is called diffusion. In a semiconductor with a uniform car-
rier density, no net (directional) transport of charges is observed due to thermal excitation. If
a charge gradient is present, a random thermal motion will result in charge diffusion from
regions of higher carrier concentration to regions with lower carrier concentration until a uni-
form charge distribution is reached. Diffusion against the charge gradient concentration due
to thermal excitation gives rise to a diffusion current.
a) ℰ=0 b) ℰ
e⁻
e⁻
net distance net distance
Figure 1.19: Schematic representation of electron transport in a crystal under a) no electric field (no net
movement) and b) upon the application of an electric field (with net movement).
29
Under an applied field charges will accelerate; this type of transport is called drift. Elec-
trons will drift in the opposite direction of the applied field, due to the negative charge, and
holes will drift in the same direction as the field. The net movement of charges in the material
will generate an electrical current (drift current). Random scattering during drift will cause
charges to lose energy, limiting the velocity that carriers can reach. The terminal velocity or
drift velocity (νd ), expressed in Equation 1.34, is proportional to the applied electric field (ε).
The constant of proportionality (µ) is the charge carrier mobility (µe for electrons and µh for
holes, cm2 V−1 s−1 ). Materials with high mobility will present low resistance to the transport
of electrons or holes.
νd = µε (1.34)
The conductivity (σ) is a function of charge mobility. It has units of Ω m−1 and is defined as
the total current density (J) per unit of electric field (ε). Equation 1.35 shows the conductivity
written as a function of electron and hole mobilities.
J
= q ρe µe + ρ p µh (1.35)

σ=
ε
The resistivity (ρ, Ωm) is the inverse of conductivity, Equation 1.36.
1
ρ= (1.36)
σ
Resistivity may be complicated to determine by direct electrical measurement. Another
useful concept in thin film semiconductors is the sheet resistance (Rsheet ) expressed in Equa-
tion 1.37. It can be more readily determined via simple resistance measurement at the surface
of a thin film semiconductor. Rsheet has unit of Ω, but is commonly expressed as Ω/square
(Ω/sq or Ω/) in the context of sheet resistance.
ρ
Rsheet = (1.37)
t
where t represents the film thickness.
30
1.4 The pn junction
A junction is made when materials with different electronic characteristics are physically
joined together. When two semiconductors with p and n conductivity are joined, the result
is a pn junction. Figure 1.20 illustrates a pn junction formed by joining n- and p-type Si
(homojunction). Two semiconductors made of different materials can be put in contact to
form a heterojunction.[9]
E
- +
- +
p-type - +
n-type
- +
+
-
Space charge region
Figure 1.20: Scheme of a p–n junction, highlighting the depletion region formed between p-type and
n-type semiconductors.
The majority carriers will diffuse between semiconductors across the junction until equi-
librium is reached, when the Fermi level has the same value on both sides of the junction. This
results in regions with a deficiency of majority carrier in each side of the junction, collectively
these regions are called the depletion region or space charge region (SCR). This gives rise to
an electric field (~E) that contributes to charge separation within the junction, with holes being
directed to the p-type semiconductor and electrons to the n-type. Although this electric field
aids charge separation, it is not the main force driving carrier transport in heterojunctions.
In thin film solar cells, charge separation and transport relies more on the difference in the
chemical potential of electrons and holes in each different layer.
1.4.1 Electric field, potential and built-in voltage

A pn junction can be modelled using the full depletion approximation, Figure 1.21a.[10–
12] It assumes that the transition between the SCR and a quasi-neutral region (QNR) is sharp
within a semiconductor near the pn junction. The QNR is defined by the region in the bulk of
the semiconductor where the free carrier density is approximately close to the doping density,
which means that the electric field within the junction is close to zero. In summary, the
full depletion approximation assumes that the effective density of acceptor and donor levels
(N = Na or Nd ) is constant in the SCR, and drops abruptly to zero in the QNR.
31
a) -xp xd xn
p n
neutral depletion neutral
b) ρ
full
c) qNx E
qNa E (x) =
Qp depletion εε 0
xn -xp xn
x x
-xp
φi
Qn
real
qNd
Qp = -Qn dφ( x )
= −E ( x )
dx
d) φi − Va φ e) EC E
qφi
x EF
-xp xd EV
x
Emax xd
φi − Va = − V=0
2
Figure 1.21: Representation of a) pn-junction, and the b) charge density (ρcarrier ) and total charge (Q), c)
electric field (E ), d) potential φ( x ), where φi is the local built-in potential and e) band diagram within
the junction.
The charge density within the junction (ρcarrier = qN) is illustrated in Figure 1.21b for the
real case (red line) and for the full depletion approximation. The total charge Q will be equal
and have opposite signs on either side of the SCR.
The electric field in the SCR (Figure 1.21c) is derived from Gauss’s law. The derivation
gives rise to Poisson’s equation, which relates the electric field and charge density:
dE ρ qNx
= ⇒ E (x) = (1.38)
dx εε 0 εε 0
where ε 0 and ε are the vacuum permittivity and the permittivity constant of the material,
respectively.
The electrical potential within the junction (φ( x )) is illustrated in Figure 1.21d and is
given by:
32
dφ( x ) qNx2
= −E ( x ) ⇒ φ( x ) = − (1.39)
dx 2εε 0
The band diagram (Figure 1.21e) has the same form as the potential. It can be drawn by
calculating EC minimum and EV maximum at any given point (x) within the junction.
1.4.2 Built-in voltage

The electric field that builds up at the interface of a p-n junction when it is in equilib-
rium is sometimes refereed to as built-in voltage or built-in potential (Vbi ). It is the energy
difference between the conduction band energies ∆EC of p- and n-type semiconductors. The
built-in voltage can be found as a function of the effective density of acceptor and donor lev-
els (Na and Nd ) in the n-type and p-type semiconductors, respectively. Recall that in a doped
semiconductor, the concentration of carriers (ρe and ρh ) will be close to the effective density
of states in the conduction and valence bands. The concentration of electrons in the conduc-
tion and valence band can be written as a function of the effective density of states in the
conduction band (NC ):
For the n-type semiconductor:
−( EC − EF ) −qVn
ρe = NC e KB T
⇒ NC e KB T (1.40)
and for the p-type semiconductor:
−( Ec − EF ) n2i −qVp
ρh = NC e KB T
⇒ ρe = = NC e KB T (1.41)
Na
Here, two convenient substitutions were made, Vn and Vp , expressed as:
KB T NC
Vn = ln
q Nd
and
33
KB T NC Na
Vp = ln
q n2i
Finally, the built in voltage can be calculated as the difference between Vp and Vn (Vbi =
Vp − Vn ), taking the form of Equation 1.42.
KB T NC Na N K B T Nd Na

Vbi = ln 2
− ln C ⇒ Vbi = (1.42)
q ni Nd q Ni2
1.4.3 Junction under forward and reverse bias

The pn junction is under bias when a voltage is applied across the junction through an
external circuit. Figure 1.22 shows the band diagram of a pn junction under forward (a) and
reverse bias (b). To generate a forward bias a positive voltage is applied to the p side of
the junction and a negative voltage to the n side. Under forward bias, the applied voltage
weakens the local field (φi − Va ) in the junction and the electron barrier decreases. As a
consequence, the diffusion current across the junction becomes larger than the drift current
and excess electrons (those that do not recombine) are injected from the p-type semiconductor
into the external circuit. Similarly, a reverse bias is generated by inverting the polarity of the
external source, i.e., positive terminal connected to the n-type semiconductor. A reverse bias
strengthens the local field (φi + Va ) in the junction, so the electron barrier for the diffusion
current increases and blocks the current flow to the external circuit.
E
a) E b)
EC q(φi + Va )
EF,p
q(φi − Va ) EC
qVa
EF,p qV EF,n
a EF,n
EV
EV
x x
forward bias reverse bias
Figure 1.22: Band diagram within a pn junction under a) forward and b) reverse bias.
Devices can operate under forward or reverse bias. A solar cell under illumination gener-
ates a forward voltage (and current) which then powers components connected to the external
circuit. A sensor (or photodetector) works in a similar way, but an external forward bias is
34
usually needed to supply power to the device. Solar cells can act like self-powered photode-
tectors, i.e. they do not need an external power supply to work. In transistors diodes, current
flows in the circuit if a forward bias is applied to it, under reverse bias the diode blocks the
current. A forward bias can also be applied to reduce the activation energy and accelerate
hydrogen production during water splitting in photoelectrochemical cells or accelerate fuel
production during the photoelectrochemical conversion of CO2 .
1.5 Thin film solar cells

1.5.1 Physical structure
Figure 1.23 illustrates the structure of a superstrate thin film solar cell. This kind of device
is composed of a stack of thin layers (on the scale of nm – µm of semiconductor materials
tailored to convert light into electrical energy.
back
contact
HTL +
h+
Light e-
-
absorber
ETL
TCO
Glass
hν
Figure 1.23: Basic structure of a thin film solar cell with electron transport layer (ETL), hole transport
layer (HTL), light absorber layer, transparent conductive oxide (TCO - front contact) and metal back
contact.
The basic structure of a solar cell is made by physically joining an electron transport layer
(ETL, selective to e− only), a light absorber layer and a hole transport layer (HTL, selective
to h+ only). Power production requires efficient charge generation, separation, transport and
extraction when the cell is illuminated. It starts with the absorption of photons by the light
absorber layer and the generation of an e− /h+ pair. Then, charges separate and are trans-
ported through the layer to be extracted by the e− or h+ selective layers (ETL and HTL). The
35
structure of the solar cell is completed by joining adequate electric contacts (conductive ma-
terials such as metals, carbon, graphite, conductive polymers or conductive oxides). Finally,
when connected to an external circuit and illuminated, the solar cell generates work.
In the generation process, a photon with energy equal to or higher than the band gap
of the semiconductor (hν > Eg ) is absorbed and an e− /h+ pair is generated as previously
discussed. Photogenerated electrons in the conduction band of the semiconductor have a
‘lifetime’ before they recombine with a hole in the absorber or are extracted, while only under
specific conditions does charge separation and extraction occur. Excited e− are transported
to one edge of the semiconductor, driven by the chemical potential difference between the
layers, where they are captured by the ETL, while h+ are transported to the opposite end of
the semiconductor where they are captured by the HTL.
Efficient charge generation, separation, transport and extraction are the result of a com-
plex combination of factors. They depend not only on the chemical nature of the layers but
also on the adequate material combination of materials in the device. Moreover, in devices
light can also be absorbed (parasitic absorption), reflected, or transmitted by the many layers
that constitute the device. Carefully engineered solar cells should consider all these processes
since only light that is absorbed in the absorber layer will contribute to the generation of
carriers.
1.5.2 Band structure

Effective charge separation and extraction rely on the adequate combination of layers to
form effective band alignment, Figure 1.24a shows the bands of two semiconductors before
being joined together. Separated semiconductors will have their characteristic band positions
and Fermi level in relation to the vacuum level. When joined together (Figure 1.24b) the Fermi
level aligns after reaching equilibrium. The bands bend downwards or upwards (Figure 1.24c)
within the SCR between the two semiconductors. The amount of band bending will depend on
the doping level of each semiconductor. When the device is illuminated, electrons are exited
to the conduction band of the light absorber semiconductor and then driven towards the ELT
that, ideally, has a conduction band energy maximum at a lower energy level. Meanwhile,
holes in the valence band move in the opposite direction. If the energy bands are misaligned,
charge extraction will not be effective.
36
a) b) EC c)
Energy EC EC
EF EC EC
EF EF EF
EV
EF
EV EV
EV EV
Separated semiconductors Fermi level alignment band bending
(equilibrium)
Figure 1.24: Band alignment between different semiconductor materials a) before being joined together,
b) at equilibrium after joining and c) band bending at the interface.
As illustrated in Figure 1.24,[5] this band model can be extended to accommodate ETL,
absorber, interlayers, and contacts. Interlayers, such as the HRMO layers discussed in Chapter
4, can be introduced in the device to improve band alignment, provide better lattice matching,
prevent shunting and enhance overall charge extraction. However, sub optimal combination of
materials can lead to unsatisfactory band alignment or the formation of a defective interface.
A large energy offset results in the formation of electron barriers leading to efficiency losses.
Theory and experimental work show that a conduction band offset of ≈ 0.1 eV and 0.4 eV (
between light absorber/ETL or light absorber/interlayer) is generally favourable for the solar
cell.
a) interlayer b)
light
TCO
bc absorber
ETL
Figure 1.25: Band alignment between different layers in a solar cell, a) before being joined together, b)
at equilibrium after joining, where bc = back contact, ETL = electron transport layer, TCO = transparent
conductive oxide. Adapted with permission from [5]. Copyright 2011 Wiley-VCH.
37
1.6 The current density vs voltage (JV ) curve
1.6.1 Measuring solar cell efficiency
Solar cell efficiency is determined by measuring the current density response to an ap-
plied voltage under standard illumination.[13] Figure 1.26a shows an example of a current
density vs voltage curve (JV curve) highlighting important parameters. Figure 1.26b shows
the AM1.5G standard illumination spectrum used to determine solar cell efficiency.[14] Three
basic parameters: short circuit current density (JSC ), open circuit voltage (VOC ) and fill factor
(FF) can be directly extracted from the JV curve and used to determine its power conversion
efficiency (PCE or η).[15]
a) b)
visible range
JSC areaB Solar spectrum
1.5 AM1.5G
Irradiance (W m-2 nm-1)
Jmax
Current density (J)
Pmax
Sb2Se3 band gap
1
areaA
0.5
VOC
0 0
Vmax
0 1000 2000 3000 4000
Voltage (V) Wavelength (nm)
Figure 1.26: a) Simulated current density voltage (JV) curve of a solar cell under illumination and
derived power density curve (P = JV). The areaA corresponds to the maximum power density output
(Pmax = Jmax Vmax ), and areaB isgiven by JSC × VOC . The fill factor FF is defined as the ratio between
areaA
areaA and areaB FF = and has a maximum value of 1 or 100 %. b) standard AM1.5G solar
areaB
spectrum
The JSC is the current density across the cell when the voltage is zero (short circuited
device, J (0) = JSC ). The VOC is obtained when no current is generated by the device (J (V ) =
0). The VOC and JSC are related by Equation 1.43.
K T JSC

VOC = B ln +1 (1.43)
q J0
where J0 is the reverse saturation current (diode current).
38
The power density output of a solar cell (P) at a given value of J and V is given by P = JV.
It is worth noting that at the VOC or JSC no power can be delivered by the solar cell. The cell
will have a maximum power output Pmax at an optimal point, with a corresponding Vmax and
Jmax (Equation 1.44).
P = JV ⇒ Pmax = Jmax Vmax (1.44)
The FF indicates how much the JV curve deviates from the ideal (square) response. The
FF may be expressed as a function of Pmax , Equation 1.45.
Jmax Vmax Pmax

FF = = (1.45)
JSC VOC JSC VOC
The PCE is the ratio between the maximum power output of a solar cell (Pmax ) and the
power density of incident light (PL ), Equation 1.46.
Pmax
PCE = (1.46)
PL
Under standard incident light (PL = 1 sun), PCE can be calculated function of JSC , VOC
and FF (Equation 1.47).
PCE = JSC VOC FF (1.47)
39
1.6.2 Diode parameters and parasitic resistances
Features of the JV curve reflect the electronic and microstructure of layers and junctions
in the device. Beyond device efficiency, etc. the JV curve contains information about diode
parameters and parasitic resistances. A basic PCE analysis requires the determination of only
three points on the JV curve (JSC , VOC and Pmax ). Nevertheless, to extract other important
figures, the curve must be fitted to an equivalent circuit. Elements in this circuit should match
the characteristic electronic behaviour of the solar cell.
The equivalent circuit shown in Figure 1.27 can be used to model a simple solar cell.
Jp Jdark −
Figure 1.27: Equivalent circuit for an ideal solar cell under illumination, where J p is the photocurrent,
Jdark is the dark current and V is the voltage.
The Jdark is intrinsic to the diode component of the solar cell and can be modelled by the
Shockley equation (Equation 1.48).
qV
!
Jdark (V ) = J0 e KB T −1 (1.48)
Under illumination, a photocurrent (JP ) arises in the solar cell in the opposite direction of
Jdark . Thus, the total current density in the cell (J) under illumination is given by Equation 1.49.
qV
!
J (V ) = JP − Jdark (V ) = JP − J0 e KB T −1 (1.49)
To account for the non-ideal behaviour of a diode, an ideality factor (n) is inserted in the
Shockley equation, giving rise to Equation 1.50. This equation may be used to approximate
some basic solar cell behaviour, but the parasitic current loss still needs to be considered.
40
qV
!
J (V ) = J p − J0 e nKB T −1 (1.50)
Current losses are inherent to real devices, they may lead to efficiency losses, but can
often be significantly reduced by improving processing methods and device architecture. As
shown in Figure 1.28, parasitic current losses are modelled by introducing series (RS ) and
parallel, shunt (Rsh ), resistive elements into the equivalent circuit of an ideal solar cell. Series
resistance arises from the inherent resistivity of material and contacts and a low RS is generally
related to better device performance. Parallel resistance should be high to avoid current losses
from the diode. A low Rsh value indicates that current is leaking thought the solar cell and
is usually related to poorly manufactured devices, e.g. presence of pin holes in the ETL that
lead to direct contact between the light absorber and TCO.
RS
Rsh V
Jp Jdark −
Figure 1.28: Equivalent circuit including parasitic resistances, RS and Rsh , for a solar cell under illumi-
nation. J p is the photocurrent, Jdark is the dark current and V is the voltage.
The effect of RS and Rsh on JV curve of a solar cell under illumination is illustrated in
Figure 1.29. Both, high RS and low Rsh , will make the FF decrease and result in lower PCE.
41
(a ) (b )
R
C u r r e n t d e n s ity ( J )
↓ s h
↑ R S
0 0
0 0
V o lta g e ( V ) V o lta g e ( V )
Figure 1.29: Behaviour of JV curve of a cell with a) increasing series resistance, RS , and b) decreasing
shunt resistance, Rsh .
Accounting for parasitic resistances, the J (V ) can be re-written as Equation 1.51.
q(V + JRS )
!
V + JRS
J (V ) = J p − J0 e nKB T −1 − (1.51)
Rsh
For most ‘well behaved’ devices, parameters can be determined by fitting the JV curve to
Equation 1.51. To model more complex or unusual artefacts in the JV curve, more elements
may be added to the equivalent circuit. The new elements should still be related to the
physical and electronic characteristics of the device.
1.6.3 Dark JV curve

Dark JV curves reflect only the diode characteristics of the junctions forming the solar
cell device and are complementary to JV curves obtained under illumination. They can be
modelled to obtain diode parameters (J0 and n) as well as parasitic resistance components (RS
and Rsh ).[16]
As shown in Figure 1.30, dark JV curves are plotted in a logarithm scale to highlight
important features. A qualitative analysis of logarithm curves can the done to quickly assess
J0 and n. This might be useful to instantly compare similar devices, e.g. the effect of ETL with
different thicknesses. Figure 1.30a shows the effect of increased J0 (keeping all the other prop-
erties constant), where a higher J0 may indicate a suboptimal diode architecture. Figure 1.30b
42
shows the effect of n on the JV curve. The value of n is related to the type and region of
recombination, an n closer to 1 indicates standard band-to-band recombination in the bulk
of the ‘absorber’ semiconductor. Yet, an n closer to 2 may indicate the presence of deep trap
states in the SCR. Values between 1 and 2 can be related to interface recombination. A value of
n higher than 2 or lower than 1 is somewhat unusual and may be related to other phenomena
that affect recombination, such as tunnelling, or to the use of an unsuitable circuit model.[5,
17]
a ) b )
n = 1
n = 2
↑ n
J0* = 1 0 × J 0 ↑ d a rk c u rren t (J0)
J 0
0 0
Figure 1.30: Simulated dark JV curve of a cell with a) increasing dark current, where J0 is one order
of magnitude higher in the dashed curve, and b) increasing diode factor, n = 1 (full line) and n = 2
(dashed line).
1.6.4 Graphical analysis of JV curves

For relatively simple cases J0 , n, Rsh and RS can be extracted from the JV curves using
graphical methods. Figure 1.31 gives an example of the procedure for graphical analysis.[18]
43
b )
a )
R sh ( Ω· c m 2
)
d V /d J
0 0
c ) d )
- G V )
d V /d J
S C
ln (J + J
in te r c e p t = R S ( Ω· c m 2
) in te r c e p t = J0 (m A ·c m -2
)
s lo p e = n s lo p e = n
0 (J + J - G V ) -1
V - R J
S C S
Figure 1.31: Example of using graphical methods to obtain solar cell parameters, a) simulated JV curve,
dV dV
b) vs V to find Rsh , c) vs ( J + JSC − GV ) to find RS and n, and d) using RS value obtained in
dJ dJ
the previous graph ln( J + J p − GV ) vs (V − RS J ) is plotted to find J0 and n.
q(V + JRS )
Graphical analysis is usually more adequate when Rsh RS and e nK B T
1 in Equa-
tion 1.51, what gives Equation 1.52. This simplification should be valid for the majority of
cases where graphic analysis is possible.
q(V + JRS )
V
J (V ) = J0 e nK B T + − Jp (1.52)
Rsh
Figure 1.31b shows the procedure to obtain Rsh . Based on the simplified Equation 1.52,
the plot dJ/dV vs V gives Rsh at low voltages, since the contribution of RS is small in this
region.
44
Figure 1.31c illustrates the procedure to obtain RS and n by plotting dV/dJ vs ( J +
JSC ) − 1. The result of dV/dJ is shown in Equation 1.53, according to this, RS is obtained by
the interaction of this curve with the y axis (( J + JSC ) − 1 = 0) and n is calculated from the
slope of the curve.
dV nK B T
= RS + ( J + J p ) −1 (1.53)
dJ q
As illustrated in Figure 1.31d, the diode parameters J0 and n can be obtained from the
intercept and slope of the ln( J + J p − GV ) vs (V − RS J ) curve, (Equation 1.54). The value of n
obtained here can be compared with the value obtained from Figure 1.31c.
nK B T
ln( J + J p − GV ) = ln( J0 ) + (V − R S J ) (1.54)
q
If Rsh is small (significant shunt leak), ( J + JSC − GV ) − 1 can be used instead of ( J +
JSC ) − 1, where G = 1/Rsh . Subtracting the GV term helps reduce artefacts making the
determination of parameters more accurate.
1.6.5 Modelling complex JV curves

Current leaks may occur via alternative pathways and more elements may need to be
added to the equivalent circuit to account for such leaks. For example, a second diode-like
element (weak diode) may be introduced to model current leaks via a tunnelling mechanism.
A non-Ohmic resistive element may be used to model current leaks that occur due to the
formation of regions of space charge limited current (SCLC), where the main junction field is
weak due to direct contact between the light absorber and TCO.[17, 19] Figure 1.28a illustrates
some of these current leak pathways in a superstrate solar cell. Figure 1.28b shows the cor-
responding equivalent 2-diode circuit model (J1 and J2 ) with a non-Ohmic element (R1 ) used
to model the solar cell. The total current (J) is given by the sum of the currents in each of the
elements of the equivalent circuit (Equation 1.48).
45
J (V ) =
Jmain junction + Jweak diode + JOhmic shunt + Jnon−Ohmic shunt = (1.55)

q(V + JRS ) q(V + JRS )
! !
V + JRS
J1 e n1 K B T − 1 + J2 e n2 KB T − 1 + + k(V − JRS )m − J p
Rsh
where k and m are constants (discussed in more detail in Chapter 4).
a) main grain
back junction boundaries
contact b) RS
+
e-
h+ e-
absorber R1 Rsh V
layer e-
Jp J1 J2 −
ETL
TCO
glass ohmic non-ohmic
shunt shunt
Figure 1.32: Current leak in a thin film solar cell, a) scheme of possible alternative pathways for current
leak and b) equivalent circuit model, where J1 and J2 are the saturation current from diode elements
(main-junction and weak diode), R1 and Rsh are parallel non-Ohmic and Ohmic resistive elements,
respectively, and RS is a series resistive element.
Features on the JV curve indicate the presence of secondary elements (e.g. weak diodes
and non-Ohmic elements). The curve in Figure 1.33a shows a ‘shoulder’ at low voltages as
a result of the presence of a current leak in the device that behaves like a weak diode. Fig-
ure 1.33b illustrates the effect of the presence of non-Ohmic contacts, it will result in variable
dV
shunt resistance (variable in reverse bias).
dJ
46
a ) b )
o h m ic a n d n o n -o h m ic s h u n t
s h o u ld e r d u e to
w e a k d io d e
o h m ic s h u n t
c o n s ta n t
v a r ia b le
d V /d J
0 0
Figure 1.33: Effect of a current leak in logarithm JV curve, a) shoulder caused by the presence of weak
diode and b) effect of Ohmic and non-Ohmic shunt leaks.
I emphasise that care should be taken as elements in the equivalent circuit should always
be related to physical or electronic processes in the device. Moreover, adequate fitting requires
the collection of enough data points in an adequate voltage range where the analysis of the
elements is relevant (e.g. enough points should be collected at low voltages for adequate
determination of Rsh ).
1.7 Electrochemical impedance spectroscopy

Electrochemical impedance spectroscopy (EIS) is commonly applied to investigate elec-
tronic properties of solid-state devices and electrochemical cells. As shown in Figure 1.34a, a
potentiostatic EIS experiment is done by applying a potential wave (usually a sine wave) to
the working electrode (VWE ) at a given frequency ( f ) and amplitude (V0 ) oscillating around an
applied fixed voltage (Va ). The current response to the applied signal is measured in terms of
the current amplitude (I0 ) and phase (φ). Each measured point can be translated to a value of
V0
impedance (Z), defined as | Z | = . As illustrated in Figure 1.34b, Z is an imaginary number
I0
that may be broken to its real (Z 0 ) and imaginary (Z”) components and then plotted in the
imaginary plane as − Z” vs Z 0 .[20, 21]
47
a) b)
Z = Z 0 − iZ”
φ V0
|Z| =
Voltage − VWE
Current − IWE
V0 I0
I0
− Z”
Va |Z|
φ
0
VWE = Va + V0 sin(2π f t)
time 0
Z0
Figure 1.34: a) Applied voltage and measured current response and b) correspondent impedance plot.
Electronic processes in semiconductors and electrochemical systems can occur in a wide
range of time scales. As illustrated in Figure 1.35, physical phenomena, from slow SCR polar-
isation to fast electronic transition, can be probed and analysed using EIS by applying a signal
(voltage wave) in the adequate range of frequencies (time scale).[22, 23]
Time (s)
1 10 -3
10 -6
10-9 10-12 10-15
-++- -++-
ε’ +
- +- +
- +- +++ - -
++- -
+++ -- --
SCR
ε” polarisation
Charge
transfer dipole
ox+ polarisation
red-
ionic/atomic
polarisation electronic
polarisation/
transition
ε = ε’ + ε”
1 103 106 109 1012 1015

Frequency (Hz)
Figure 1.35: Time and frequency scale of dielectric permittivity (ε) response for typical processes in
electronic and electrochemical systems.
48
Table 1.1: Impedance equation for common circuit elements as a function of the angular frequency of
an applied potential wave (ω).
Element Symbol Impedance (Z)

R
Resistor ZR = R
C
1
Capacitor ZC =
iωC
Q
1
CPE ZCPE =
(iωQ)φ
1.7.1 Equivalent circuit

As previously discussed, JV curves can be fitted to an equation corresponding to an
equivalent circuit. Similarly, an equivalent circuit can be used to model the impedance re-
sponse of a device to an applied potential wave. Table 1.1 shows some basic circuit elements
used to model solar cells, which include resistors, capacitors, and CPE (constant phase ele-
ment). Each element has a characteristic impedance response to an applied potential wave of
angular frequency (ω).
The plot of the impedance in an imaginary plane ( − Z” vs Z 0 ) is known as the Nyquist
plot (or diagram). Figure 1.36 shows a simulated Nyquist plot of the RS ( R P /C ) circuit, this
simple circuit can be used to model a (well behaved) photovoltaic device. As indicated in the
figure, the plot yields a semicircle with a diameter equal to R P (that is equivalent to Rsh ) and
the first intersection with the x-axis is equal to RS .
49
RS
-Z "
RP
0
R S R S + R P
Z '
Figure 1.36: a) Simulated Nyquist plot and its b) equivalent circuit.
1.7.2 Built-in potential and doping density from EIS data

The capacitance of the pn junction can be obtained using EIS data and modelled to get the
built-in potential (Vbi ) and charge density profile (doping profile) of a semiconductor. For this,
the SCR of the diode is modelled as a parallel plate capacitor. From Table 1.1, the complex
1
impedance of a parallel plate capacitor is ZC = iωC . For the capacitive element, the phase of
the current is zero (φ = 0), thus the real component of the complex impedance is independent
of the frequency of the applied potential wave (Z 0 = 0 in Z = Z 0 − iZ”). Therefore, Z = iZ”
gives Equation 1.56.[24]
1
= ωZ” (1.56)
CSCR
where CSCR is the capacitance in the SCR.
Considering the definition of capacitance, we can write:
dQ(Va )
CSCR (Va ) = (1.57)
dVa
where Q is the total charge in the SCR (see also ).
The Mott-Schottky relation (Equation 1.58) arises by solving the differential equation
stated above.
50
1 2 KB T

2
= Va − Vbi − (1.58)
CSCR εε 0 NA2 q
where ε is the dielectric constant of the semiconductor, ε 0 is the vacuum permittivity, N is the
doping density and A is de area of the electrode.
1
A plot of 2
CSCR
vs Va is the Mott-Schottky plot. It can be used to determine Vbi at the
intersection of the curve with the x-axis by extrapolation. The doping density (N) is given by
Equation 1.59.
2 1
N=− (1.59)
qεε 0 A2 2
d(1/CSCR )
dVa
The value of N can be plotted as a function of the width of the SCR (xSCR , Equation 1.60)
to obtain the doping density profile along the pn junction (N( xSCR ) vs xSCR ).
εε 0 A
xSCR = (1.60)
CSCR
1.8 External quantum efficiency

The ratio between the number of electrons generated by a cell and the number of photons
produced by the light source at a given energy is termed the external quantum efficiency
(EQE).[13] A typical EQE plot is shown in Figure 1.37, highlighting factors that influence
the shape of the curve. Ideally, each incident photon would be converted to current by the
solar cell. In practice, photons are not converted for various reasons. Loss of light, due
to reflection and transmission, will reduce EQE and the subsequent power of the solar cell.
Moreover, front surface recombination will reduce cell response to absorbed light with short
wavelengths, while rear surface recombination, and the low absorption of light with longer
wavelengths reduce cell response to red/infrared light. Normally, light with energy lower
than the band gap of the semiconductor is not absorbed, thus no cell response is expected for
this range of wavelengths. Moreover, light that is absorbed in layers other than the absorber
51
layer usually does not contribute to power generation. This can be a problem in cells that use
layers (other than the absorber layer) with a small band gap, such as devices containing a CdS
layer.
The EQE experimental equipment consists of a light source, monochromator, calibrated
reference solar cell, and a sourcemeter to measure the current. The current response from the
sample is monitored under illumination with light of known power and wavelength. This re-
sponse is then expressed as a relative quantity from 0 to 100% by comparing it to the reference
solar cell. Integration of the EQE over the spectrum gives the expected JSC for the device. The
standard AM1.5G spectrum is normally used for integration. Other light spectra, such as one
from LEDs or tungsten light sources, may be more adequate for indoors applications.
Ideal quantum efficiency
Front Reflection and transmission

recombination
EQE (%)
Rear
recombination
(low absorption
of IR light)
λ with energy
lower than the
bandgap is not
absorbed
Wavelength (nm)
Figure 1.37: a) EQE showing an ideal (dash-dot line) and typical (full line) quantum efficiency; the
effects that causes typical curves to deviate from the ideal behaviour are indicated in the graph. An
EQE curve for a cell with an electron transport layer with a small band gap presenting a ‘shoulder’
(doted line) is also shown.
1.9 References
[1] R. Eisberg, R. Resnick, Quantum Physics of Atoms, Molecules, Solids and Particles, 2nd,
John Wiley & Sons, 1985.
[2] D. A. Neamen, Semiconductor Physics and Devices: Basic Principles, 3rd, McGraw-Hill,
New York, NY, 2003.
[3] M. A. Omar, Elementary Solid State Physics: Principles and Applications, Addison-Wesley
Publishing Company, 1975.
[4] W. D. Callister, D. G. Rethwisch, Materials Science and Engineering: An Introduction,
9th ed., John Wiley & Sons, 2013.
[5] R. Scheer, H. Schock, Chalcogenide Photovoltaics: Physics, Technologies, and Thin Film De-
vices, WILEY-VCH Verlag & Co. KGaA, Weinheim, Germany, 2011.
52
[6] C. N. R. Rao, G. V. Subba Rao, physica status solidi (a) 1970, 1, 597.
[7] T. D. C. Hobson, L. J. Phillips, O. S. Hutter, H. Shiel, J. E. N. Swallow, C. N. Savory,
P. K. Nayak, S. Mariotti, B. Das, L. Bowen, L. A. H. Jones, T. J. Featherstone, M. J.
Smiles, M. A. Farnworth, G. Zoppi, P. K. Thakur, T.-L. Lee, H. J. Snaith, C. Leighton,
D. O. Scanlon, V. R. Dhanak, K. Durose, T. D. Veal, J. D. Major, Chem. Mater. 2020, 32,
2621.
[8] P. Würfel, Physics of Solar Cells: From Principles to New Concepts, 3rd, Wiley-VCH Verlag
GmbH & Co., Weinheim, Germany, 2005.
[9] W. Shockley, Bcl] Syst. Tech. J. 1949, 28, 435.
[10] C. C. Hu, Modern Semiconductor Devices for Integrated Circuits, Pearson/Prentice Hall,
New Jersey, 2010.
[11] B. V. Zeghbroeck, Principles of Semiconductor Devices, http://truenano.com/PSD20/,
accessed: May, 2023.
[12] G. W. Neudeck, The PN junction diode, 2nd, Addisson-Wesley Publishing Company,
1989.
[13] T. Kirchartz, K. Ding, U. Rau in Advanced Characterization Techniques for Thin Film Solar
Cells, 2011, pp. 33–60.
[14] ASTM, ASTM G173-03(2020): Standard Tables for Reference Solar Spectral Irradiances: Direct
Normal and Hemispherical on 37° Tilted Surface, 2020.
[15] J. Nelson, The Physics of Solar Cells, Imperial College Press, Covent Garden, London,
2003.
[16] S. Suckow, T. M. Pletzer, H. Kurz, Prog. Photovoltaics 2014, 22, 494.
[17] B. L. Williams, S. Smit, B. J. Kniknie, K. J. Bakker, W. Keuning, W. M. M. Kessels, R. E. I.
Schropp, M. Creatore, Prog. Photovoltaics 2015, 23, 1516.
[18] S. S. Hegedus, W. N. Shafarman, Prog. Photovoltaics 2004, 12, 155.
[19] A. Rose, Phys. Rev. 1955, 97, 1538.
[20] J. R. Macdonald, W. B. Johnson in Impedance Spectroscopy, (Eds.: E. Barsoukov, J. R.
Macdonald), John Wiley & Sons, Inc., 2005, pp. 1–26.
[21] M. E. Orazem, B. Tribolllet, Electrochemical Impedance Spectroscopy, John Wiley & Sons,
Inc., Hoboken, New Jersey, 2008.
[22] E. von Hauff, The Journal of Physical Chemistry C 2019, 123, 11329.
[23] B. Brown, D. Hess, V. Desai, M. J. Deen, The Electrochemical Society Interface 2006, 15, 28.
[24] J. Heath, P. Zabierowski in Advanced Characterization Techniques for Thin Film Solar Cells,
(Eds.: D. Abou-Ras, T. Kirchartz, U. Rau), WILEY-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, 2011, Chapter 4, pp. 81–105.
53
Chapter 2
Review of Sb2Se3 properties and
thin film fabrication methods
54
2.1 Preface
The design of thin film solar cells is versatile. A vast range of materials can be combined
into different architectures to achieve optimal charge generation, transport, and extraction.
Charge generation and bulk charge transport are directly linked to the quality of the light
absorbing layer; thus, this layer is often regarded to determine solar cell efficiency. In fact,
device performance is the result of a complex interaction between the many layers the make
up that solar cell. An optimal combination of light absorber, buffer (ETL), and window (TCO,
HR layers) layers is crucial to generate high-quality interfaces for efficient charge extraction
and interlayer transport. The properties of these layers can vary widely depending on device
architecture and application. This chapter gives an overview of Sb2 Se3 as a light absorber layer
for optoelectronic device applications with particular consideration given to its peculiar quasi-
1D crystal structure. I review the Sb2 Se3 structural properties and discuss the implications of
its quasi-1D ribbon structure on devices considering film crystallographic orientation and
morphology. Finally, I review the interaction of Sb2 Se3 with the many layers in a solar cell
and briefly discuss the development of photodetectors, photoelectrochemical cells and battery
electrodes based on Sb2 Se3 .
55
2.2 Crystal structure
Established solar cell materials, such as Si, GaAs and CdTe, have high symmetry crystal
structures with atoms connected by strong covalent or ionic bonds in all 3 dimensions of
space. This 3D crystal structure, Figure 2.1a, ensures these materials have isotropic optical and
electronic properties. This also guarantees the isotropic properties of polycrystalline films.
However, dangling bonds at grain boundaries and at the surface of 3D materials have the
propensity to create deep traps that lead to carrier recombination and efficiency loss. Thus,
the use of expensive processing techniques is required to create monocrystalline layers and
the passivation of films to obtain devices with satisfactory efficiency. This has led to the search
for alternative light absorbing materials with the potential to mitigate recombination losses.[1]
Materials with low-dimensional crystal structure, composed of quasi-2D layer (e.g., SnS,
GeSe, TiS2 and CuSbS2 ) or quasi-1D ribbons (e.g. Sb2 Se3 , Sb2 S3 or Sb2 (Sex , S1 – x )3 have been
extensively explored in emerging device technologies.[1–5] In these compounds, layers (Fig-
ure 2.1b) or ribbons (Figure 2.1c) interact only via weak Van der Waals forces leaving no
dangling bonds at the edge of oriented crystals. This means that, in contrast to typical 3D
materials (i.e. silicon, CdTe, and organo-metal perovskites), polycrystalline films of low-
dimensional materials have intrinsically passivated grain boundaries, as long as the grains
are suitably oriented relative to the contacts.
Recently, special attention has been given to the Sb2 Se3 phase with high efficiencies being
achieved in a diverse range of applications, including photovoltaics,[6–11] water splitting,[12–
15] energy storage,[16–20] and photodetectors.[21–24] Nevertheless, low symmetry crystal
structure results in highly anisotropic properties requiring refined crystallographic orientation
control in order to obtain highly efficient devices.
56
Figure 2.1: Unity cell, charge transport and expanded structure of materials with a) 3D (CdTe) b) quasi-
2D (SnSe) and c) quasi-1D (Sb2 Se3 ) crystal structure.
57
Table 2.1: Sb2 Se3 Crystallographic parameters[25] and relevant properties.[6, 26]
Formula Sb2Se3
Formula weight 480.4
Crystal system Orthorhombic
Space group 62 (Pbnm)
Unit cell dimensions (Å) a = 11.62(1)
b = 11.77(1)
c = 3.962(7)
General reflection conditions 0kl: k+l=2n
hk0: h=2n
h00: h=2n
0k0: k=2n
00l: l=2n
Volume (Å3) 541.872
Z 4
Density (calculated) (g/cm3) 5.89
Melting point (°C) 611
Absorption coefficient (cm-1) >105
Direct band gap (eV) 1.2
Mobility (cm2 V-1 s-1) µe [100] > 16.9
µh [100] 1.17
[010] 0.69
[001] 2.59
Le (µm) [221] 0.29
[001] 1.7
58
Antimony selenide crystallizes with an orthorhombic lattice in the space group #62.
Throughout this thesis, the Pbnm setting is used for the indexing of planes and crystallo-
graphic directions.[27] Table 2.1 contains the crystallographic parameters for Pbnm Sb2 Se3 .[25]
Sb2 Se3 has a quasi-1D crystal structure, whereby Sb atoms are bounded by strong covalent
bonds to Se atoms forming (Sb4 Se6 )n ribbons that extend along the [001] direction. In direc-
tions that are perpendicular to the [001], e.g. the [120] direction, ribbons are connected only
by weak van der Waals forces.[28] As illustrated in Figure 2.2a, this ribbon assembly results
in efficient charge transport along the covalently bound ribbon in the [001] direction. Mean-
while, poor charge transport is observed between the ribbons in directions such as the [120]
(Figure 2.2b).[29] Additionally, the formation of grain boundaries along the planes of the [001]
zone axis leaves no dangling bonds, which results in intrinsically passivated grain boundaries.
Due to its peculiar quasi-1D structure, (Sb4 Se6 )n ribbons must be properly oriented relative
to the substrate to obtain high performing thin film devices. This implies that, for a vertical
(sandwich) device architecture, Sb2 Se3 thin films with a high degree of (001) orientation is
desirable.
Figure 2.2: Representation of charge transport along Sb2 Se3 a) [001] and b) [120] directions.
59
2.3 Ribbon tilt and crystallographic orientation
In this thesis, a ribbon tilt angle (θr ) is defined in relation to the substrate, and then
grouped into 3 ‘sets’ or ‘types’ of crystallographic orientations. Figure 2.3 illustrates (Sb4 Se6 )n
ribbons in each set of orientations defined here. Figure 2.3a, all (hk0) orientations are those
with ribbons that are parallel to the substrate and have θr = 0°. Figure 2.3b, the (hk1) set of
orientations contains tilted ribbons 0° < θr < 90°. Finally, in Figure 2.3c, the (001) orientation
contains ribbons that extend vertically from the substrate θr = 90°.
Figure 2.3: The three possible ‘sets’ of ribbon orientation and their corresponding crystallographic
notations: a) (hk0), b) (hk1) and c) (001).
In vertical devices, charges can be more readily transported within ribbons that extend
perpendicular from the substrate ‘connecting’ the contact layers. It is well established that
minimizing (hk0) orientations is beneficial to the efficiency of solar and photoelectrochemical
cells.[28, 30] In entirely (hk0) oriented thin films, (Sb4 Se6 )n ribbons lie parallel to the substrate
surface and therefore result in compromised electronic properties. Thin films that have (hk1)
preferential orientation contain a high concentration of tilted ribbons and show improved
charge transport properties.[31] Nevertheless, the most optimal transport between contact
layers is observed for (001) oriented thin films.[32, 33] Thus, a perfect (001) alignment is the
most desirable to fully exploit facile charge transport along the ribbons.
60
2.4 Sb2Se3 thin film deposition methods
Various deposition methods have been explored to grow Sb2 Se3 thin films with improved
crystal orientation. These approaches can be broadly divided into methods based on solution
processing and thermal evaporation techniques (Figure 2.4).
10 TE-substrate
TE-superstrate
Solution
VTD
Efficiency (%)
0
2010 2013 2016 2019 2022
Year
Figure 2.4: Efficiency evolution of Sb2 Se3 solar cells with absorber layer fabricated via solution pro-
cessing, thermal evaporation (TE) and vapour transport deposition (VTD) methods.[6, 11, 32]
2.4.1 Solution processed Sb2 Se3 thin films

Solution processing techniques for thin film deposition can involve methods such as spin
coating, successive ionic layer absorption and reaction (SILAR), spray pyrolysis, or chemical
bath deposition (CBD). These are relatively simple techniques with low equipment require-
ments that allow additives and doping agents to be easily processed in the precursor solution
to adjust thin film optoelectronic properties. However, these methods often result in films
that contain impurities, such as organic ligands from the precursor solutions that can impact
device film performance. Low crystallinity, small grain size and the formation of pin holes
and discontinuities can also be a concern in solution processed thin films.
61
Thin films of Sb2 Se3 have been made via solution processed techniques since the 90s for
various device applications. The first Sb2 Se3 solar cells were made in 2009 via solution pro-
cessing techniques. Employing CBD to deposit a Sb2 Se3 :Sb2 O3 layer, this device presented
low efficiency of only 0.66%.[34] In 2013, the first solar cell with phase pure Sb2 Se3 was
achieved via spin coating of molecular precursor with efficiency up to 3.21% (dye sensitized
solar cell).[35] Shortly after that, in 2014, the first phase pure planar device was fabricated
thin films deposited via spin coating of Sb and Se in hydrazine solution,[36] this solution
processed layer was highly (hk0) oriented and demonstrated a PCE of only 2.26%. Thermal
evaporation has since replaced solution proceed methods as the preferred method to achieve
high quality Sb2 Se3 thin films and high efficiency devices. It is worth noting however that
recently, in October 2022, a solar cell with a record 10.57% efficiency was achieved via CBD
with the use of additives to regulate film deposition kinetics.[11]
2.4.2 Thermal evaporated Sb2 Se3 thin films

Thermal evaporation methods are well established for industrial production of commer-
cial CdTe solar cells. The vapour phase, formed by heating a source containing the material,
re-condenses on a substrate to form the thin film. High vapour pressure and absence of crystal
polymorphism favour facile processing of Sb2 Se3 thin films via thermal evaporation methods,
mimicking those used in robust large-scale CdTe processing. High quality, phase pure and
highly crystalline thin films can be achieved via thermal evaporation.
Thermal evaporation is the current paradigm to achieve high quality Sb2 Se3 thin films
and has been applied for nearly all best efficient solar cell devices. The first thermally evap-
orated Sb2 Se3 solar cells were reported in 2014, wherein the films had a high degree of (hk0)
orientation and devices presented only 2.1% efficiency.[37] In 2015, rapid thermal evaporation
(RTE) was employed to fabricate a Sb2 Se3 layer with improved (hk1) orientation, achieving
5.6% efficiency.[28] After that, Sb2 Se3 solar cell efficiency improved rapidly. Using RTE and a
PbS HTL, 6.5% PCE was achieved in 2017 [38] and in 2018 vapour transport deposition (VTD)
was implemented to obtain efficiencies up to 7.6%.[31] In a notable development, highly (001)
oriented Sb2 Se3 layers allowed significant efficiency improvements. In 2019, an efficiency of
62
9.2% was achieved with an evaporated (001) Sb2 Se3 layer.[33] Recently, in June 2022, ‘Injection
Vapor Deposition’ technique was developed to yield highly (001) Sb2 Se3 layer in solar cells
that exceeded the 10% efficiency benchmark reaching up to 10.12% PCE.[32]
2.4.3 Vapour transport deposition of Sb2 Se3 thin films

In this thesis, an adapted VTD method is implemented to deposit Sb2 Se3 layers. VTD
is a type of thermal evaporation technique where a vapour phase is formed in a hot zone,
then transported and re-condensed within a colder zone containing (device) substrates cul-
minating in film deposition. This technique was first implemented as a robust high turnover
technique to manufacture high-quality CdTe thin film for commercial solar cells.[39] In some
VTD methods, a transport gas can be used to aid the transport of the vapour phase. However,
in laboratory scale equipment the temperature gradient is enough to drive vapour flow via
diffusion from the hot to the cold zone, with no need for a transport gas. In comparison
to other thermal evaporation techniques, VTD generally allows more control over thin film
growth conditions.
Typical laboratory scale VTD equipment uses a one zone tube furnace for film deposi-
tion.[31] As illustrated in Figure 2.5, a source containing Sb2 Se3 powder is posited within
the heating zone where the temperature is raised to evaporate the powder. The holder with
substrates is positioned in a colder edge of the tube, outside the heating zone. Fine tuning
deposition parameters in a one zone furnace can be difficult. Substrate temperature (at the
moment of deposition) is a direct function of the evaporation temperature in the hot zone.
Although substrate temperature can be adjusted by changing the distance between the holder
and the edge of the hot zone, even millimetre changes to the holder position drastically alters
the deposition dynamics, thus requiring extensive recalibration.
Figure 2.5: Typical vapour transport deposition equipment used to deposit Sb2 Se3 films.
63
Two-zone furnaces have also been used for the VTD deposition of thin films of Sb2 Se3 and
especially Sb2 (S, Se)3 .[40, 41] A scheme of a two zone furnace is shown in Figure 2.6. In this
case, sources with Sb2 Se3 powder (and Sb2 S3 for Sb2 (S, Se)3 ) are put within different heating
zones. The temperature and heating rate of each zone is controlled individually to produce
desirable evaporation rates for each source. In current reports, the two-zone furnace is used
in much the same way as the one-zone furnace, meaning that substrates are put in a cooler
area outside the heating zones.
Figure 2.6: Vapour transport deposition equipment used to deposit Sb2 (S, Se)3 films.
For this thesis, a slight adaptation was done to the classic VTD process. The method
uses a two-zones furnace and is discussed in detail in the Experimental Methods (chapter 3).
Here, a holder with substrates is put within zone 2 (right) of the furnace, instead of outside
the zones (like the classic process) while the source is kept in the middle of heating zone 1
(left). This allows decoupling of the source and substrate temperatures to achieve fine control
over film thickness. These adjustments were crucial to achieving solar cells with satisfactory
efficiencies (chapter 4) and to studying crystal growth in highly (001) oriented Sb2 Se3 films
(chapter 5).
2.4.4 Other Sb2 Se3 thin film devices

Photodetectors: Alongside its application in photovoltaic devices, Sb2 Se3 has also been widely
explored as a light absorbing material in photodetectors. Initially, lateral device architectures
were developed to benefit from horizontal Sb2 Se3 wires deposited parallel to the substrate.[42]
Sb2 Se3 wires were prepared in solution and then deposited on a substrate such as Si/SiO2 , Au
was deposited later to complete the lateral contacts.[43] Lateral devices are still being used in
polarimetric photodetectors.[29, 44, 45] Vertical Sb2 Se3 photodetectors have proved more ef-
ficient, as illustrated by the ultra-flexible Sb2 Se3 heart rate photodetector developed by Chao
64
Chen, et al. [21] that works in normal ambient light conditions. Another remarkable char-
acteristic of Sb2 Se3 is the (ultra)broad band detection capacity, up to ≈1000 nm via its ≈1.2
eV band gap and 1000-1650 nm via extrinsic excitation.[46] Proof-of-concept photodetectors
containing an unprecedented highly (001) Sb2 Se3 transferred layer are prepared in chapter 6
of this thesis, including a flexible heart rate detector.
Photoelectrochemical cells: Photoelectrochemical cells for water splitting with Sb2 Se3 photo-
cathodes have been highly successful.[12] In the record device, Sb2 Se3 /perovskite tandem, the
Sb2 Se3 layer was fabricated via close space sublimation (CSS) to achieve devices with up to
10% efficiency for H2 evolution and 2h of stability.[47] The best Sb2 Se3 single junction pho-
tocathode has a highly (001) oriented absorber layer (also deposited via CSS) and presents
higher stability, up to 10h, but significantly lower efficiencies ≈2.5% for H2 evolution.[48]
Batteries: Sb2 Se3 have also been applied for battery electrodes. A Sb2 Se3 vertical nanowire
array, deposited on Ti foil via CSS, was demonstrated to have high capacity (471 mA h g−1 )
as battery anode for storage of Li+ ions under an impressive ultra-high current density 21440
mA g−1 .[49] High capacity (up to 532.6 mA h g−1 ) has also been demonstrated for storage of
Na+ ion in Sb2 Se3 thin film battery anode.[50] This excellent storage performance has been
attributed to reactions involving ion intercalation, conversion, and alloying with Sb2 Se3 .
2.5 Growth of oriented Sb2Se3 thin films

2.5.1 Epitaxy growth and (hk0) Van der Waals layers
Epitaxy refers to the growth of films with well defined crystallographic orientation on
a crystalline substrate.[51, 52] Epitaxial film growth implies the deposition of a high-quality,
often single crystal, oriented layer on a lattice matching substrate. Epitaxy occurs naturally to
reduce the free energy of the resultant system. High lattice match between crystallographic
planes of the substrate and epilayer results in a lower surface energy interface and highly
ordered growth. Epitaxial growth results in fewer defects, such as local disorder, dangling
bonds, lattice strain, and dislocations and significantly reduces the internal energy of the final
system, compensating for the loss of entropy compared with more disordered (non-epitaxial)
65
Figure 2.7: Epitaxy growth on with a) small and b) large mismatch. Growth of c) generic quasi-1D or
quasi-2D and d) Sb2 Se3 a van der Waals layers.
growth. If the lattice mismatch is relatively large the surface energy for interface formation
is high, which increases the internal energy of the final system and epitaxial growth does not
occur. Nevertheless, lattice mismatch can be accommodated in several ways.
Figure 2.7 illustrates mechanisms for accommodating lattice mismatch. In 3D materials,
lattice mismatch is accommodated by lattice strain (Figure 2.7a). Any small lattice strain
increases the internal energy in the bulk of the film but can be maintained by strong covalent
bonds across the interface. Large lattice strain will eventually cause dislocations along some
crystallographic planes to accommodate the mismatch (Figure 2.7b).[52]
Another way to accommodate the lattice mismatch is via the formation of a van der
Waals layer. This kind of defect accommodation is common in 1D and 2D materials. As
shown in Figure 2.7c, only weak Van der Waals interactions occur between substrate and film
that ‘floats’ on top of the substrate. Quasi-1D materials tend to accommodate any lattice
mismatch by growing with (hk0) planes exposed on the surface of the substrate (Figure 2.7d).
This way the film interacts only weakly with the substrate permitting a lower concentration
of defects, given that no covalent bonds need to be broken or any dislocations need to form.
This reduces both, the energy of interface formation and the bulk internal energy of the film.
66
As such, under typical deposition conditions Sb2 Se3 films are highly (hk0) oriented, see for
example the first Sb2 Se3 planar solar cells,[37, 53] and the perfectly (hk0) orientated layer
obtained on mica.[24]
As expected, the surface energy of (hk0) Sb2 Se3 crystal surfaces is lower than those sur-
faces containing (hk1) planes. Exposing (hk0) planes at the surface requires no breaking of
covalent bonds between Sb and Se atoms. Yet the formation of surfaces containing (hk1)
planes necessitates breaking these strong covalent bonds. Specifically, (010), (120) and (110)
have the lowest surface energies at 0.25, 0.32, 0.33 J m−2 respectively; compared to (001), (211)
and (221) that have surface energies of 0.46, 0.56 and 0.53 J m−1 , respectively. Thus, Sb2 Se3
films are expected to have a high concentration of grains exposing the lower energy (010),
(120) or (110) crystallographic planes parallel to the substrate.[28]
2.5.2 Evolution selection principle and (hk1) growth

The evolution selection principle is a simple model proposed to explain the formation of
highly oriented films from randomly oriented crystal nuclei during physical vapour deposi-
tion.[54] Figure 2.8 illustrates the development of (hk1) texture in an Sb2 Se3 thin film according
to this model. Initially, a high concentration of (hk0) oriented grains nucleates on the surface
of the substrate (favoured by low energy interface formation - as discussed above). However,
the fewer crystals that nucleate exposing (hk1) planes grow vertically at a faster rate,[55–57]
due to the strong covalent bonds formed along the c-axis. Over time, these crystals overgrow
those with (hk0) orientations, which have a slower vertical growth rate, and an (hk1) film
texture emerges. Eventually, if deposition continues, orientation will be dominated by the
crystallographic direction with the highest vertical growth rate.
67
Sb2Se3: (hk0) (hk1)
preferential (hk0) rapid (hk1) predominance of

nucleation crystal growth (hk1) crystals
Deposition time
Figure 2.8: Proposed growth mechanism for Sb2 Se3 on a flat substrate following the evolution selection
principle. A higher population of Sb2 Se3 grains nucleate with (hk0) orientation, but the fewer (hk1)
grains grow faster. Eventually, (hk1) grains predominate in the crystal population at the surface of the
film and start to define the preferred crystallographic texture.
The fastest growth occurs along the Sb2 Se3 crystallographic c-axis ([001] direction). As a
consequence, the unencumbered growth of Sb2 Se3 (such as in solution) tends to yield elon-
gated structures with wire or rod morphology.[22, 58] Vertical growth of Sb2 Se3 along [001]
and [hk1] directions (on the surface of a substrate) can be achieved more effectively if the
deposition conditions favour the initial nucleation of crystals exposing (001) and (hk1) planes.
Once these crystals nucleate oriented growth is kinetically sustained and self-catalysed, the
major barrier being the high initial nucleation energy. Some of the strategies that have been
explored to lower this initial energy barrier and obtain oriented films are discussed in the
following section.
2.5.3 Seed layers

A seed layer is a very thin film of crystalline material used to direct the growth of a
secondary layer (deposited on top). A seed layer can reduce the energy for the formation of
(hk1) interfaces and favour (hk1) preferential nucleation. If during the early stages of the de-
position the substrate contains enough (hk1) oriented Sb2 Se3 crystals, the ensuing kinetics will
favour rapid vertical growth along [hk1] directions, as summarised by the evolution selection
principle.
Various works have studied the effect of seed layers on the growth of Sb2 Se3 films. In a
notable example, Sb2 Se3 was shown to grow with ≈2x higher (002) texture on Mo seeded with
a thin Sb2 Se3 layer, compared to films grown on bare Mo substrate. In this study, substrate
devices with an efficiency up to 8.5% were obtained.[59] Mo substrates pre-seed with MoSe2
68
Figure 2.9: a-k) schematic representation of seed screening process. Reproduced with permission from
ref. [56]. Copyright 2019 Wiley-VCH.
were also shown to induce high (002) texture.[60] Even more remarkable is the ‘seed screen-
ing’ treatment, schematically illustrated in Figure 2.9[56] It consists of selective re-evaporation
of (hk0) oriented seeds. As (hk0) crystals interact with the substrate only via weak van der
Waals forces they are more prone to re-evaporate than covalently bound (hk1) crystals. The
controlled heat treatment of substrates with Sb2 Se3 seeds results in a layer with a high con-
centration of (hk1) seeds. Sb2 Se3 films grown on top of seed screened layers were shown to
be more (hk1) oriented than films grown on Sb2 Se3 without the post-treatment. This strat-
egy has yielded the record efficiency (7.61%) for an evaporated Sb2 Se3 device in superstrate
configuration.
2.5.4 Substrate chemistry and evaporation parameters

The surface chemistry of substrates, i.e., substrate material and exposed crystal faces, is
well known to affect the orientation and morphology of Sb2 Se3 thin films obtained via evap-
oration techniques. Under the same evaporation conditions, different substrates can drive the
69
growth of Sb2 Se3 films with completely distinct characteristics. Conversely, evaporation pa-
rameters can be adjusted to achieve a distinct film morphology and orientation on substrates
with different surface chemistry.
A ‘rule of thumb’ to determine the preferential orientation of Sb2 Se3 crystals during the
initial nucleation is the bond dissociation energy of substrate’s material.[55, 56] Relatively
inert substrates are those made of materials with high dissociation energy, such as TiO2 (Ti-O
= 667 kJ mol−1 ). These inert substrates tend to drive the growth of (hk0) oriented Sb2 Se3 since
the growing (hk0) layer will only interact only via weak vdW with the substrates. Substrates’
materials with lower dissociation energy can more easily allow the formation of covalent
bonds and required for the growth of (hk1) oriented layers.”
A ‘rule of thumb’ to determine the preferential orientation of Sb2 Se3 crystals during the
initial nucleation is the activation energy required to break chemical bonds in the substrate in
order to form new bonds with growing (Sb4 Se6 )n ribbons.[55, 56] Relatively inert substrates
are those made of materials with a high dissociation energy, such as TiO2 (Ti-O = 667 kJ
mol−1 ). These substrates tend to drive the growth of (hk0) oriented Sb2 Se3 , since no covalent
bonds need to be broken in the substrate. In contrast, substrates with lower bond dissociation
energies, e.g., CdS (Cd-S = 208 kJ mol−1 ), can more easily allow the formation of covalent
bonds that are required for the growth of (hk1) oriented layers, and drive the growth of (hk1)
layers.
A more elaborate growth model was developed by Rokas Kondrotas, et al. [55]. It de-
scribes how substrate material and evaporation parameters affect the growth of Sb2 Se3 films.
In their work, Sb2 Se3 thin films were grown via VTD on various substrates under controlled
conditions wherein the growth rate (G) was identified as a key variable that determines the
type of film growth. They defined G = αc R, where R is the deposition rate which is a function
of source and substrate temperatures and αc is a ‘sticking coefficient’ which is an intrinsic sub-
strate parameter. Essentially, the ‘sticking coefficient’ is related to the empirical observation of
the orientation of the Sb2 Se3 crystals during the initial nucleation phase. The αc coefficient is
defined as ‘high’ for substrates that induce the nucleation of (hk1) Sb2 Se3 crystals and as ‘low’
for substrates that lead to the nucleation of (hk0) crystals.
70
Figure 2.10: a) summary of ‘growth rate’ model and b) ‘Nanoepitaxy’ of Sb2 Se3 thin films. Reproduced
with permission.[55, 61] Copyright 2019 Elsevier and 2021 Wiley-VCH.
The effect of G on Sb2 Se3 film orientation and morphology is summarized in Figure 2.10a.
High growth rates (high source temperature, low substrate temperature or high αc ) tend to
lead to the growth of a double layer consisting of random rods on top of a compact layer. While
low growth rates (low source temperature, high substrate temperature or low αc ) develop
mostly horizontal rods or sheets with (hk0) preferential orientation. Moderate growth rates,
achieved at moderate source and substrate temperatures, leads to close packed grains with
(hk1) preferential orientation. This compact grain morphology is preferred for some device
applications. However, the deposition rate (source and substrate temperature) will need to
be calibrated to achieve ‘moderate’ growth rates on substrates having different αc . The initial
Sb2 Se3 seed layer plays a crucial role in this model.
71
The effect of deposition parameters on the orientation and morphology of Sb2 Se3 films
is also clear in the work done by Xiaoyang Liang, et al. [61]. Figure 2.10b shows the ‘na-
noepitaxy’ strategy for orientation control. It consists of manipulating deposition parameters
to induce (hk1) preferential nucleation. As indicated in the figure, low temperatures favour
the preferential nucleation of (hk0) or randomly oriented crystals that results in films that are
not ideal for device applications. On the other hand, high temperatures lead to the preferen-
tial deposition of (hk1) crystals that subsequently drive the growth of a highly (hk1) oriented
Sb2 Se3 NR layer (yielding superstrate solar cells up to 9.0% PCE).
2.5.5 Other strategies to obtain (hk1) Sb2 Se3 layers

A substrate’s crystallographic orientation can also affect Sb2 Se3 film growth. Liang Wang,
et al. [30] showed that (001)-ZnO substrates lead to (hk0)-Sb2 Se3 films, while randomly ori-
ented ZnO induces the growth of (hk1)-Sb2 Se3 layers. Remarkably, the authors verified that
efficiency almost doubles for (hk1)-Sb2 Se3 devices compared to those with (hk0)-Sb2 Se3 films,
4.81% on random ZnO and 2.82% on oriented ZnO.
Interlayers that are not in direct contact with Sb2 Se3 also affect film orientation. This
was demonstrated for Sb2 Se3 films deposited on CdS for substrates with and without a
SnO2 interlayer. Devices with an interlayer (ITO/SnO2 /CdS/Sb2 Se3 ) have improved (hk1)-
Sb2 Se3 orientation and higher efficiency when compared to devices that did not contain SnO2
(ITO/CdS/Sb2 Se3 ). In this case, efficiency increases from 5.4% (no interlayer) to 7.5% (SnO2
interlayer).[62]
2.5.6 (001) orientated Sb2 Se3 films

Reports of Sb2 Se3 films with a (001) preferential orientation are relatively rare. Table 2.2
shows I(001)/(221) ratios for some highly (001) oriented films found in the literature (calculated
from XRD patterns provided in the respective publications). High (001) orientation may be
obtained via close space sublimation (CSS) under very specific evaporation conditions.[63]
Highly (001) oriented Sb2 Se3 films have also been obtained via the selenization of a metallic
Sb layer.[64] More recently, (001) oriented films were obtained via a new deposition method
named ‘Injection Vapor Deposition’ (IVD) to fabricate a 10.12% solar cell in substrate config-
uration.[32] Although these Sb2 Se3 films do not have a perfect (001) orientation, they have a
72
significant concentration of vertically oriented ribbons, i.e. (Sb4 Se6 )n that are perpendicular to
the substate. As indicated in Table 2, a simple way to compare the level of (001) orientation
from different works is using the intensity ratio between (001) and (221) peaks in the XRD
(I(001)/(221) ).
Table 2.2: Intensity ratio between (001)/(221) for highly oriented Sb2 Se3 films.
I(001)/(221) thickness (µm) Method Substrate Reference

0.72 1 CSS Mo/MoSe2 [33]
0.90 ≈2.6 IVD Mo/MoSe2 [32]
2.56 0.75 Selenization of Sb Mo [65]
1.46 0.75 Selenization of Sb Mo [64]
1.87 1.15 CSS (selenized) Mo [63]
4.38 1.6 CSS ZnO:B [66]
2.59 ≈2 VTD ZnO NW this work (chapter 5)
2.5.7 Graphoepitaxy and importance of substrate nanostructure on film orientation

The deposition of highly oriented films is often desired on substrate materials that do
not favour (001) or (hk1) orientation. The creation of a pattern on the surface of the substrate
may also be a way to manipulate film crystallographic texture. This is possible as long as the
development of an oriented film is still thermodynamically or kinetically favourable. In this
case, micro- or nano- features on the surface of the substrate guide the preferential growth
of certain crystallographic orientations. Graphoepitaxy has been used to grow oriented 3D
materials such as Si and ZnS on amorphous or highly mismatching substrates and has been
widely explored to obtain oriented organic semiconductors.[67–69] In contrast to conventional
epitaxy, graphoepitaxy does not require a lattice matched substrate. The ‘wetting’ charac-
teristics of the substrate, selective nucleation, and the preferential exposure of fast-growing
crystallographic planes (induced by substrate nanostructure) have been reported as part of
the mechanism by which texturing is achieved via graphoepitaxy.
Graphoepitaxy is yet to be explored for the oriented growth of 1D materials. In fact,
the role of substrate morphology and nanostructure on Sb2 Se3 thin film growth has been
vastly overlooked as a whole. In one study Sb2 Se3 has been electrochemically grown on TiO2
nanotube substrates. Initially, octahedral Sb2 Se3 nanostructures were obtained. Under op-
timal deposition conditions, nanorod like structures grew from inside the nanotubes with
73
a (hk1) preferential orientation. To explain the formation of the Sb2 Se3 nanorods, the au-
thors proposed a mechanism based on the ‘space-confined effect’ of TiO2 nanotubes. Notably,
TiO2 /nanorod-Sb2 Se3 showed up to 3.02 mA cm−2 as a photoanode for water splitting, a cur-
rent density ≈4x higher than the TiO2 /octahedral-Sb2 Se3 heterostructure.[70, 71] This work
indicates that substrate nanostructure may be used to drive Sb2 Se3 film orientation. How-
ever, detailed growth mechanisms were not explored in this work arousing the need for more
comprehensive studies relating substrate nanostructure and Sb2 Se3 film orientation.
2.6 Optoelectronic properties

2.6.1 Band gap and absorption coefficient
Antimony selenide is a quasi-direct band gap semiconductor, it has a direct band ≈1.2
eV and an indirect band gap ≈1.0 eV.[72] Figure 2.11a is a schematic representation of energy
levels in a quasi-direct band gap semiconductor.[3] Sb2 Se3 high light absorption coefficient,
≈ 105 cm−1 ( Figure 2.11b),[37] can be attributed to absorption through its direct band gap
(orange). A smaller percentage of photons may also be absorbed through indirect band gap
(purple) pathways. Exited electrons preferably relax to the lower energy level (indirect band
gap, dark green), and eventually recombine via the indirect pathway (light green). Quasi-
direct band gap Sb2 Se3 presents a high absorption coefficient (through direct band gap) and
slow recombination (via indirect band gap).
74
Figure 2.11: a) Representation of electron transitions in a quasi-direct band gap semiconductor, where
thicker arrow correspond to preferential flow of electrons and b) absorption spectrum of Sb2 Se3 thin
film. Reproduced with permission.[37] Copyright 2014 American Chemical Society.
2.6.2 Defect structure

Compared to other binary compounds, such as CdTe, Sb2 Se3 has a surprisingly complex
defect structure. Such complexity emerges as a result of the low symmetry of its crystal
structure, whereby all Sb and Se atoms are located in non-equivalent crystallographic sites
(Figure 2.12a). This means that point defects at any of the two Sb and three Se non-equivalent
sites will have distinct properties. Moreover, unconventional antisite defects, such as SbSe ,
SeSb , and 2 SeSb , can be quite easily formed due to due to the weak interaction and large
space between ribbons.[73] Figure 2.12b shows the calculated formation energy for acceptor
and donor defects under Se and Sb rich conditions. It can be implied that p-type conductivity
is expected under Se rich conditions due to the low formation energy of acceptor levels SeSb2
and 2 SeSb2 . Under Sb rich conditions the formation energy of these acceptor levels increases
significantly. Conversely, n-type conductivity is expected under Sb rich conditions as all donor
defects have a lower formation energy, below 1 eV, compared to acceptor levels.[73] Although
reports of p-type Sb2 Se3 are more commonly found in the literature, n-type Sb2 Se3 films have
also been reported.[74, 75]
75
Figure 2.12: a) Non-equivalent sites on (Sb4 Se6 )n ribbon, b) calculated defect formation energies and
c) calculated and experimental defect positions. Reproduced with permission.[73, 76] Copyright 2019
and 2020 American Chemical Society.
This unconventional defect structure can result in the formation of deep traps. Under
Se-rich conditions, defects that are deep in the band gap, e.g., SeSb1 , VSe2 , and VSe3 , have low
formation energies. Experimental defect analyses, Figure 2.12c,[76] show the highly defective
structure of Sb2 Se3 , supporting the results from DFT calculations. Deep traps are believed to
be the main cause of the low Voc observed experimentally, typically ≈0.4 V for highly efficient
solar cells, which is far from the Shockley-Queisser limit of ≈ 0.8 - 0.9 V considering a 1.1 -
1.2 eV band gap.
76
Figure 2.13: Angle-dependent photocurrent measurements with a) 12 electrode arrangement, illustra-
tion of charge transport along the ribbon (90°) and inter-ribbon (0°) and b) response under polarised
light. (a) Reprinted with permission from ref. [29]. Copyright 2017 Wiley-VCH. (b) Reproduced from
ref. [44] licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
2.6.3 Anisotropy of charge transport

As discussed before, Sb2 Se3 has highly anisotropic electronic properties due to its quasi-
1D ribbon structure. The fact that charge transport properties are significantly improved in
the direction of the ribbons can be directly verified experimentally. The diffusion length has
been measured at 1.7 µm in the [001] direction, i.e., along the (Sb4 Se6 )n ribbons, compared to
only 0.29 µm in the [221] direction. Moreover, the mobility of minority carriers along a, b and
c crystallographic axes have been measured at µh = 1.17, 0.69, and 2.59 cm2 V−1 s−2 , respec-
tively.[26] The charge transport properties can also be verified via angle-dependent photocur-
rent measurements.[45] Figure 2.13a show the 12 electrode arrangement,[29] where the highest
responsivity is along the [001] crystallographic direction (90° as indicated). The photocurrent
also responds to polarized light, being highest for polarizations that match the direction of
the ribbon (Figure 2.13b), making Sb2 Se3 suitable for polarimetric imaging applications.[44]
These results illustrate the highly anisotropic character of Sb2 Se3 and show the importance of
considering the crystallographic orientation control of thin films for device applications.
77
Figure 2.14: Band position of quasi-1D antimony chalcogenides cathode (TCO), ELTs,
HTLs and metal contacts. Adapted from ref. [77] licensed under CC BY 4.0
(https://creativecommons.org/licenses/by/4.0/).
2.6.4 Contacts
High-quality interfaces and optimal band alignment between different layers of a solar
cell can significantly improve the efficiency of the device. Ideally, front and back contact
interfaces will have a low concentration of deep traps and reduced interface recombination.
Recombination is increased by a large conduction band offset (CBO) between ELT/absorber or
a large valence band offset (VBO) between absorber/HTL. For reference, Figure 2.14 shows the
band position of Sb2 Se3 (and other quasi-1D antimony chalcogenides) compared to materials
that can be used to make the TCO, ELT, HTL and back contact layers.[77]
2.6.5 Front contact

Various materials have been tested as ETLs in Sb2 Se3 solar cells (Figure 2.14). Ideally,
contact materials should have a conduction band position that provides satisfactory alignment
with Sb2 Se3 (≈-4 eV). CdS and TiO2 are the most commonly used to make efficient devices.
These layers can be processed via facile solution-based methods and are widely applied in
well stabilised commercial solar cells (CdTe and CIGS). CdS has a compatible band position,
a high electron mobility (350 cm2 V−1 s−1 ), and is still the most preferred layer for achieving
highly efficient Sb2 Se3 devices.[78–80] Champion Sb2 Se3 solar cells, in superstrate and sub-
strate device architectures, have employed CdS as an ETL. A thin CdS layer is desirable to
78
minimize parasitic absorption of short wavelength light. However, reducing its thickness to
less than ≈50-60 nm is not practical as this inevitably results in film discontinuities and a
higher density of pin holes that yield devices with low shunt resistance.[81–83] This is further
compounded by the fact that CdS can be consumed by ion diffusion at CdS/Sb2 Se3 inter-
face.[75, 78] A common strategy to address this has been to apply a high resistance metal
oxide (HRMO) layer at the front contact prior to CdS buffer deposition to provide shunt pro-
tection. In this approach, a layer consisting of a film of an HRMO with a wide band gap is
deposited between the transparent conductive oxide (TCO) and buffer layers. In CdTe and
Cu(In, Ga)Se2 solar cells interfacial layers have been shown to affect device performance,[84–
93] and the benefits of applying an HRMO interfacial layer in chalcogenide solar cells with
a CdS buffer are well established.[94–96] However, not all metal oxides are adequate as high
resistance layers; e.g. while applying a SnO2 HRMO improves the performance of CdTe solar
cells, using a Al2 O3 HRMO has proven to be detrimental.[97] HRMO interfacial layers have
only begun to be explored and implemented within antimony chalcogenide solar cell archi-
tectures. The few HRMO materials that have been tested experimentally are discussed in the
following sections.
SnO2 HRMOs: Interlayers, such as high resistance metal oxides (HRMO), can be incorpo-
rated in the front contact to help band alignment, prevent shunting and, consequently, im-
prove device efficiency. The most studied HRMO layer for Sb2 Se3 solar cells is SnO2 . In
addition to giving high shunt protection, a SnO2 layer has proven to contribute to more
optimal band alignment and to assist in the improvement of Sb2 Se3 grain orientation. Kai
Shen, et al. [82] showed that a SnO2 layer can suppress current leaks that occur via a non-
Ohmic mechanism, i.e., due to the formation of space-charge limited current (SCLC) regions,
in FTO/(SnO2 )/CdS/Sb2 Se3 /Au solar cells obtained via closed space sublimation. Moreover,
Ke Yang, et al. [98] reported that SnO2 may act as an anti-reflective layer in Sb2 Se3 solar cells
prepared by pulsed laser deposition. The work by Jiahua Tao, et al. [99] showed that an ultra-
thin SnO2 interfacial layer can passivate interface defects within the layers in a solar cell with
Sb2 Se3 films obtained via VTD. They also showed that SnO2 , even though not in direct contact
79
with the Sb2 Se3 layer, drives an improved (221) texture and higher grain crystallinity. Finally,
SnO2 was shown to be able to reduce the thickness of an amorphous layer at CdS/Sb2 Se3
interface, which improved the lattice match at the p-n junction an increased device efficiency.
ZnO and TiO2 HRMOs: The resistivity of SnO2 layers is typically too low (≈ 10−2 Ω cm) to
act as an ideal shunt screening element. Yet ZnO and TiO2 may provide layers with higher
resistivity (> 104 Ω cm) for use as an ultrathin HRMO layer that can effectively prevent
shunt leaks.[100] High resistance TiO2 , ZnO, and Zn(O,S) interlayers have been explored in
CdS/Sb2 Se3 (solution processed) and Sb2 (Sex , S1 – x )3 solar cells. In solution processed Sb2 Se3
solar cells, CdS relieves the rapid drop in the conduction band maximum between Sb2 Se3 and
TiO2 layers.[101] In this sense, CdS is an interlayer in a TiO2 /interlayer/Sb2 Se3 architecture.
Band alignment is also enhanced in ‘double buffer’ solar cells such as TiO2 /CdS/Sb2 (S, Se)3
fabricated using hydrothermal methods[10] and in ZnO/CdS/Sb2 (Se, S)3 obtained via RTE.[102]
Finally, a Zn(O,S) applied as an interlayer in solar cells with a Sb2 (S, Se)3 light absorber permit-
ted a reduction to the CdS buffer layer thickness, lowering parasitic light absorption at short
wavelengths, helping device stability, and improving the overall quality of Sb2 (S, Se)3 grains.
[103, 104] The effects of the application of ZnO and TiO2 HRMO layers in FTO/(HRMO)/CdS/
Sb2 Se3 /Au device architecture is comprehensively investigated for the first time in chapter 4
of this thesis.
2.6.6 Back contact

Various metallic contacts have been tested for the back contact of Sb2 Se3 , including Au,
Co, Ni, and Sb.[105] The metallic back contact maybe deposited on an HTL or directly on the
Sb2 Se3 layer. Solar cells with an Au metallic back contact usually present satisfactory efficien-
cies even without an HTL. As illustrated in Figure 2.14, various p-type layers have been ap-
plied to as HTL in Sb2 Se3 solar cells (e.g. PbS,[40] P3HT,[106] NiOx,[105] CuSCN,[107] Spiro-
OmeTAD[11]). These HTLs can help provide more adequate band alignment between the light
absorber layer and metallic back contact. Moreover, as illustrated in Figure 2.15 an effective
HTL assists in the separation and extraction of photogenerated carriers by blocking electron
diffusion towards the back contact. This prevents recombination and enhances hole collec-
80
tion. It has been determined via simulation, for a solar cell in the ITO/CdS/Sb2 Se3 /HTL/Au
architecture, that the ideal HTL should have hole density (p) higher than 1017 cm−3 and VB
off-set (∆EV ) within 0.1 and -0.2 eV for maximum PCE.[38]
Figure 2.15: Representation of band structure of CdS/Sb2 Se3 a) without and b) with a hole transport
layer (HTL).
2.7 References
[1] S. Hadke, M. Huang, C. Chen, Y. F. Tay, S. Chen, J. Tang, L. Wong, Chem. Rev. 2022,
122, 10170.
[2] J. He, Y. Wei, T. Zhai, H. Li, Materials Chemistry Frontiers 2018, 2, 437.
[3] R. Kondrotas, C. Chen, X. Liu, B. Yang, J. Tang, Journal of Semiconductors 2021, 42,
031701.
[4] Y. Wang, Y. Zhao, X. Ding, L. Qiao, J. Energy Chem. 2021, 60, 451.
[5] W. Yang, X. Zhang, S. D. Tilley, Chem. Mater. 2021, 33, 3467.
[6] S. Barthwal, R. Kumar, S. Pathak, ACS Appl. Energy Mater. 2022, 5, 6545.
[7] R. Kondrotas, C. Chen, J. Tang, Joule 2018, 2, 857.
[8] A. Mavlonov, T. Razykov, F. Raziq, J. Gan, J. Chantana, Y. Kawano, T. Nishimura, H.
Wei, A. Zakutayev, T. Minemoto, X. Zu, S. Li, L. Qiao, Sol. Energy 2020, 201, 227.
[9] U. A. Shah, S. Chen, G. M. G. Khalaf, Z. Jin, H. Song, Adv. Funct. Mater. 2021, 31,
2100265.
[10] X. Wang, R. Tang, C. Wu, C. Zhu, T. Chen, J. Energy Chem. 2018, 27, 713.
[11] Y. Zhao, S. Wang, C. Li, B. Che, X. Chen, H. Chen, R. Tang, X. Wang, G. Chen, T. Wang,
J. Gong, T. Chen, X. Xiao, J. Li, Energy Environ. Sci. 2022, 15, 5118.
[12] S. Chen, T. Liu, Z. Zheng, M. Ishaq, G. Liang, P. Fan, T. Chen, J. Tang, J. Energy Chem.
2022, 67, 508.
[13] R. R. Prabhakar, T. Moehl, S. Siol, J. Suh, S. D. Tilley, Chem. Mater. 2020, 32, 7247.
[14] W. Yang, J. H. Kim, O. S. Hutter, L. J. Phillips, J. Tan, J. Park, H. Lee, J. D. Major, J. S.
Lee, J. Moon, Nat. Commun. 2020, 11, 861.
[15] W. Yang, J. Moon, J. Mater. Chem. A 2019, 7, 20467.
[16] S. Dong, C. Li, X. Ge, Z. Li, X. Miao, L. Yin, ACS Nano 2017, 11, 6474.
[17] R. Gusmão, Z. Sofer, J. Luxa, M. Pumera, ACS Sustain. Chem. Eng. 2019, 7, 15790.
81
[18] P. V. Prikhodchenko, J. Gun, S. Sladkevich, A. A. Mikhaylov, O. Lev, Y. Y. Tay, S. K.
Batabyal, D. Y. W. Yu, Chem. Mater. 2012, 24, 4750.
[19] S. Wang, X. Yang, P.-K. Lee, A. L. Rogach, D. Y. W. Yu, Chem. Mater. 2019, 31, 2469.
[20] L. Xia, Z. Yang, B. Tang, F. Li, J. Wei, Z. Zhou, Small 2021, 17, 2006016.
[21] C. Chen, K. Li, F. Li, B. Wu, P. Jiang, H. Wu, S. Lu, G. Tu, Z. Liu, J. Tang, ACS Photonics
2020, 7, 352.
[22] G. Chen, W. Wang, C. Wang, T. Ding, Q. Yang, Adv. Sci. 2015, 2, 1500109.
[23] M. R. Hasan, E. S. Arinze, A. K. Singh, V. P. Oleshko, S. Guo, A. Rani, Y. Cheng, I.
Kalish, M. E. Zaghloul, M. V. Rao, N. V. Nguyen, A. Motayed, A. V. Davydov, S. M.
Thon, R. Debnath, Adv. Electron. Mater. 2016, 2, 1600182.
[24] X. Wen, Z. Lu, L. Valdman, G. C. Wang, M. Washington, T. M. Lu, ACS Appl. Mater.
Interfaces 2020, 12, 35222.
[25] N. W. Tideswell, F. H. Kruse, J. D. McCullough, Acta Crystallogr. 1957, 10, 99.
[26] C. Chen, D. C. Bobela, Y. Yang, S. Lu, K. Zeng, C. Ge, B. Yang, L. Gao, Y. Zhao, M. C.
Beard, J. Tang, Front. Optoelectron. 2017, 10, 18.
[27] T. D. C. Hobson, K. Durose, Mater. Sci. Semicond. Process. 2021, 127, 105691.
[29] H. Song, T. Li, J. Zhang, Y. Zhou, J. Luo, C. Chen, B. Yang, C. Ge, Y. Wu, J. Tang, Adv.
Mater. 2017, 29, 1700441.
[34] S. Messina, M. T. Nair, P. K. Nair, J. Electrochem. Soc. 2009, 156, H327.
[35] Y. C. Choi, T. N. Mandal, W. S. Yang, Y. H. Lee, S. H. Im, J. H. Noh, S. I. Seok, Angew.
Chem. Int. Ed. 2014, 53, 1329.
[36] Y. Zhou, M. Leng, Z. Xia, J. Zhong, H. Song, X. Liu, B. Yang, J. Zhang, J. Chen, K. Zhou,
J. Han, Y. Cheng, J. Tang, Adv. Energy Mater. 2014, 4, 1301846.
[37] X. Liu, J. Chen, M. Luo, M. Leng, Z. Xia, Y. Zhou, S. Qin, D. J. Xue, L. Lv, H. Huang,
D. Niu, J. Tang, ACS Appl. Mater. Interfaces 2014, 6, 10687.
[38] C. Chen, L. Wang, L. Gao, D. Nam, D. Li, K. Li, Y. Zhao, C. Ge, H. Cheong, H. Liu,
H. Song, J. Tang, ACS Energy Lett. 2017, 2, 2125.
[39] R. C. Powell, U. Jayamaha, G. L. Dorer, H. McMaster, AIP Conf. Proc. 1999, 462, 31.
[40] Y. Lu, K. Li, X. Yang, S. Lu, S. Li, J. Zheng, L. Fu, C. Chen, J. Tang, ACS Applied Materials
& Interfaces 2021, 13, 46858.
82
[41] Y. Pan, X. Hu, Y. Guo, X. Pan, F. Zhao, G. Weng, J. Tao, C. Zhao, J. Jiang, S. Chen, P.
Yang, J. Chu, Adv. Funct. Mater. 2021, 31, 2101476.
[42] T. Zhai, M. Ye, L. Li, X. Fang, M. Liao, Y. Li, Y. Koide, Y. Bando, D. Golberg, Adv. Mater.
2010, 22, 4530.
[43] D. Choi, Y. Jang, J. Lee, G. H. Jeong, D. Whang, S. W. Hwang, K. S. Cho, S. W. Kim, Sci
Rep 2014, 4, 6714.
[44] S. Zhang, H. Wang, M. M. Kirchner, J. Liu, H. Luo, Y. Ren, C. Yuan, H. T. Hattori, A. E.
Miroshnichenko, W. Lei, Advanced Materials Interfaces 2022, 9, 2200448.
[45] M. Zhao, J. Su, Y. Zhao, P. Luo, F. Wang, W. Han, Y. Li, X. Zu, L. Qiao, T. Zhai, Adv.
Funct. Mater. 2020, 30, 1909849.
[46] K. Li, X. Yang, F. Yang, J. He, G. Zhang, S. Jiang, C. Chen, J. Tang, Materials Today
Electronics 2022, 2, 100011.
[47] W. Yang, J. Park, H.-C. Kwon, O. S. Hutter, L. J. Phillips, J. Tan, H. Lee, J. Lee, S. D.
Tilley, J. D. Major, J. Moon, Energy Environ. Sci. 2020, 13, 4362.
[48] C. Liu, T. Liu, Y. Li, Z. Zhao, D. Zhou, W. Li, Y. Zhao, H. Yang, L. Sun, F. Li, Z. Li, J.
Mater. Chem. A 2020, 8, 23385.
[49] W. Li, L. Deng, X. Wang, J. Cao, Y. Xie, Q. Zhang, H. Zhang, H. Deng, S. Cheng,
Nanoscale 2021, 13, 9834.
[50] L. Xia, Z. Yang, B. Tang, F. Li, J. Wei, Z. Zhou, Small 2021, 17, 2006016.
[51] D. W. Pashley, Adv. Phys. 1956, 5, 173.
[52] D. Smith, Thin-Film Deposition: Principles and Practice, McGraw-Hill Education, 1995.
[53] M. Luo, M. Leng, X. Liu, J. Chen, C. Chen, S. Qin, J. Tang, Appl. Phys. Lett. 2014, 104,
173904.
[54] A. Vanderdrift, Philips Res. Rep. 1967, 22, 267.
[55] R. Kondrotas, J. Zhang, C. Wang, J. Tang, Sol. Energy Mater. Sol. Cells 2019, 199, 16.
[56] K. Li, C. Chen, S. Lu, C. Wang, S. Wang, Y. Lu, J. Tang, Adv. Mater. 2019, 31, 1903914.
[57] R. E. Williams, Q. M. Ramasse, K. P. McKenna, L. J. Phillips, P. J. Yates, O. S. Hutter,
K. Durose, J. D. Major, B. G. Mendis, ACS Appl. Mater. Interfaces 2020, 12, 21730.
[58] Z. Ma, S. Chai, Q. Feng, L. Li, X. Li, L. Huang, D. Liu, J. Sun, R. Jiang, T. Zhai, H. Xu,
Small 2019, 15, 1805307.
[59] S. Rijal, D.-B. Li, R. A. Awni, C. Xiao, S. S. Bista, M. K. Jamarkattel, M. J. Heben, C.-S.
Jiang, M. Al-Jassim, Z. Song, Y. Yan, Adv. Funct. Mater. 2022, 32, 2110032.
[60] S.-N. Park, S.-Y. Kim, S.-J. Lee, S.-J. Sung, K.-J. Yang, J.-K. Kang, D.-H. Kim, Mater. Adv.
2022, 3, 978.
[61] X. Liang, Z. Li, X. San, T. Liu, Y. Liu, K. Shen, S. Wang, R. E. I. Schropp, Y. Mai, Sol.
RRL 2021, 6, 2100869.
[62] J. Tao, X. Hu, Y. Guo, J. Hong, K. Li, J. Jiang, S. Chen, C. Jing, F. Yue, P. Yang, C. Zhang,
Z. Wu, J. Tang, J. Chu, Nano Energy 2019, 60, 802.
[63] P. Fan, G. J. Chen, S. Chen, Z. H. Zheng, M. Azam, N. Ahmad, Z. H. Su, G. X. Liang,
X. H. Zhang, Z. G. Chen, ACS Appl. Mater. Interfaces 2021, 13, 46671.
83
[64] H. Zhou, M. Feng, K. Song, B. Liao, Y. Wang, R. Liu, X. Gong, D. Zhang, L. Cao, S.
Chen, Nanoscale 2019, 11, 22871.
[65] H. Zhou, M. Feng, P. Li, X. Gong, D. Zhang, S. Chen, Sustainable Energy Fuels 2020, 4,
3943.
[66] Y. Cao, P. Qu, C. Wang, J. Zhou, M. Li, X. Yu, X. Yu, J. Pang, W. Zhou, H. Liu, G.
Cuniberti, Adv. Opt. Mater. 2022, 10, 2200816.
[67] J. Y. Cheng, C. A. Ross, H. I. Smith, E. L. Thomas, Adv. Mater. 2006, 18, 2505.
[68] M. W. Geis, D. C. Flanders, H. I. Smith, D. A. Antoniadis, Journal of Vacuum Science and
Technology 1979, 16, 1640.
[69] K. Zong, K. M. Asawa, A. Circelli, N. Sparta, C.-H. Choi, S. S. Lee, ACS Materials Letters
2020, 2, 721.
[70] H. Lu, S. Zhang, Z. Jiang, A. Tang, J. Alloys Compd. 2022, 912, 165201.
[71] S. Zhang, M. Long, P. Zhang, J. Wang, H. Lu, H. Xie, A. Tang, H. Yang, Chem. Eng. J.
2022, 429, 132091.
[72] C. Chen, W. Li, Y. Zhou, C. Chen, M. Luo, X. Liu, K. Zeng, B. Yang, C. Zhang, J. Han,
J. Tang, Appl. Phys. Lett. 2015, 107, 043905.
[73] M. Huang, P. Xu, D. Han, J. Tang, S. Chen, ACS Appl. Mater. Interfaces 2019, 11, 15564.
2621.
[75] Y. Zhou, Y. Li, J. Luo, D. Li, X. Liu, C. Chen, H. Song, J. Ma, D.-J. Xue, B. Yang, J. Tang,
Appl. Phys. Lett. 2017, 111, 013901.
[76] C. Chen, J. Tang, ACS Energy Lett. 2020, 5, 2294.
[77] J. Dong, Y. Liu, Z. Wang, Y. Zhang, Nano Sel. 2021, 2, 1818.
[78] S. Chen, M. Ishaq, W. Xiong, U. Ali Shah, U. Farooq, J. Luo, Z. Zheng, Z. Su, P. Fan,
X. Zhang, G. Liang, Sol. RRL 2021, 5, 2100419.
[79] H. Guo, X. Jia, S. H. Hadke, J. Zhang, W. Wang, C. Ma, J. Qiu, N. Yuan, L. H. Wong,
J. Ding, J. Mater. Chem. C 2020, 8, 17194.
[80] T. Liu, X. Liang, Y. Liu, X. Li, S. Wang, Y. Mai, Z. Li, Adv. Sci. 2021, 8, 2100868.
[81] K. Shen, Q. Li, D. Wang, R. Yang, Y. Deng, M.-J. Jeng, D. Wang, Sol. Energy Mater. Sol.
Cells 2016, 144, 472.
[82] K. Shen, C. Ou, T. Huang, H. Zhu, J. Li, Z. Li, Y. Mai, Sol. Energy Mater. Sol. Cells 2018,
186, 58.
[84] F. Bittau, E. Artegiani, A. Abbas, D. Menossi, A. Romeo, J. W. Bowers, J. M. Walls in
2017 IEEE 44th Photovoltaic Specialist Conference (PVSC), 2017, pp. 752–756.
[85] C.-H. Chung, B. Bob, T.-B. Song, Y. Yang, Sol. Energy Mater. Sol. Cells 2014, 120, 642.
84
[86] C. Doroody, K. S. Rahman, H. N. Rosly, M. N. Harif, F. Haque, S. K. Tiong, N. Amin,
J. Renewable Sustainable Energy 2020, 12, 023702.
[87] Y. G. Fedorenko, J. D. Major, A. Pressman, L. Phillips, K. Durose, MRS Advances 2016,
1, 937.
[88] C. S. Ferekides, U. Balasubramanian, R. Mamazza, V. Viswanathan, H. Zhao, D. L.
Morel, Sol. Energy 2004, 77, 823.
[89] S. Gayam, S. Bapanapalli, H. Zhao, L. Nemani, D. L. Morel, C. S. Ferekides, Thin Solid
Films 2007, 515, 6060.
[90] J. D. Major, K. Durose in 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC),
2013, pp. 1146–1149.
[91] U. Rau, M. Schmidt, Thin Solid Films 2001, 387, 141.
[92] R. E. Treharne, A. Clayton, L. J. Phillips, J. D. Major, S. J. C. Irvine, K. Durose in 2014
IEEE 40th Photovoltaic Specialist Conference (PVSC), 2014, pp. 1722–1725.
[93] G. Womack, P. M. Kaminski, J. M. Walls, Vacuum 2017, 139, 196.
[94] C. S. Ferekides, R. Mamazza, U. Balasubramanian, D. L. Morel, Thin Solid Films 2005,
480-481, 224.
[95] J. M. Kephart, J. W. McCamy, Z. Ma, A. Ganjoo, F. M. Alamgir, W. S. Sampath, Sol.
Energy Mater. Sol. Cells 2016, 157, 266.
[96] B. E. McCandless, K. D. Dobson, Sol. Energy 2004, 77, 839.
[97] X. Li, K. Shen, Q. Li, Y. Deng, P. Zhu, D. Wang, Sol. Energy 2018, 165, 27.
[98] K. Yang, B. Li, G. Zeng, Sol. Energy Mater. Sol. Cells 2020, 208, 110381.
[99] J. Tao, X. Hu, J. Xue, Y. Wang, G. Weng, S. Chen, Z. Zhu, J. Chu, Sol. Energy Mater. Sol.
Cells 2019, 197, 1.
[101] X. Wang, R. Tang, Y. Yin, H. Ju, S. Li, C. Zhu, T. Chen, Sol. Energy Mater. Sol. Cells 2019,
189, 5.
[102] M. Ishaq, H. Deng, S. Yuan, H. Zhang, J. Khan, U. Farooq, H. Song, J. Tang, Sol. RRL
2018, 2, 1800144.
[103] Y. Zhao, C. Li, J. Niu, Z. Zhi, G. Chen, J. Gong, J. Li, X. Xiao, J. Mater. Chem. A 2021, 9,
12644.
[104] Y. Zhao, S. Wang, C. Jiang, C. Li, P. Xiao, R. Tang, J. Gong, G. Chen, T. Chen, J. Li, X.
Xiao, Adv. Energy Mater. 2021, 2103015.
[105] J. Zhang, R. Kondrotas, S. Lu, C. Wang, C. Chen, J. Tang, Sol. Energy 2019, 182, 96.
[106] J. Zhou, X. Zhang, H. Chen, Z. Tang, D. Meng, K. Chi, Y. Cai, G. Song, Y. Cao, Z. Hu,
Appl. Surf. Sci. 2020, 534, 147632.
[107] K. Li, S. Wang, C. Chen, R. Kondrotas, M. Hu, S. Lu, C. Wang, W. Chen, J. Tang, J.
Mater. Chem. A 2019, 7, 9665.
85
Chapter 3
Experimental procedures
86
3.1 List of materials
Chemicals: titanium(IV) diisopropoxide bis(acetylacetonate) (75 wt.% in isopropanol), 1-butanol
(anhydrous, 99.8 %), zinc acetate dihydrate (99 %), 2-methoxyethanol (99.9 %), ethanolamine
(99 %), cadmium(II) acetate dihydrate (98 %), ammonium acetate (98 %), thiourea (99 %), cad-
mium(II) chloride hydrate (99.995 %), zinc nitrate hexahydrate (> 99.0%), ammonium citrate
dibasic (98%), cadmium nitrate tetrahydrate (98%), sodium sulphide nonahydrate (> 99.99%),
diethyl zinc (DEZ, >52 wt.% Zn), trimethylaluminium (TMA, 97%), and titanium tetrachlo-
ride (TTC, 99.995%), Poly(methyl methacrylate) (PMMA Mw ≈120,000) were acquired from
Sigma Aldrich. Ammonia solution (25 vol%) was purchased from ChemSupply Australia. An-
timony(III) Selenide 200 mesh (99.999 %) was purchased from China Rare Metal. All reagents
and solvents were used as received without further purification.
Conductive paints: electric carbon paint was purchased from Bare Conductive. PEDOT:PSS
(PH 1000) was acquired from Ossila.
TCO substrates: FTO-coated glass was bought from Latech Scientific Supply (14 Ω sq−1 ). ITO-
coated glass and ITO-coated PET were acquired commercially (15 Ω sq−1 ).
3.2 Film deposition methods

3.2.1 Substrate washing
All substrates were washed completely free of grease and dust before film deposition
by sonicating in soapy water, acetone and IPA (5-10 min each). Substrates were then dried
thoroughly with under N2 flow and put under UV/ozone for 20 min. The PET substrate was
washed with soapy water and IPA only.
3.2.2 Sol-gel
Sol-gel methods allow processing metal oxide thin films directly from solution.[1] It in-
volves the use of a precursor solution (sol) containing a source of metal ions. This precursor
solution is deposited on a substrate, via spin coating for example, to obtain a thin layer of
concentrated precursor (colloidal particles of complexed metal ions - gel). In a next step the
87
precursor is oxidised by heat treatment to form the metal oxide. Several sol-gel routes have
been developed for the deposition of high quality ZnO and TiO2 via spin coating. The proce-
dures described in this chapter have been adapted from the literature and optimised.
TiO2 and ZnO sol-gel: TiO2 and ZnO sol-gel solutions were deposited on clean glass/FTO
substrates via spin coating at 4000 RPM for 30 s, the layers were heated at 250 °C for 2 min
before final annealing at 500 °C for 60 min in air. For TiO2 , a 0.1 M solution of titanium
diisopropoxide bis(acetylacetonate), 75 wt % in isopropanol, in 1-butanol was prepared and
stirred in a glove box for 15 min. For ZnO, a 0.5 M zinc acetate dihydrate in 2-methoxyethanol
was prepared with the addition of ethanolamine (150 µL) and left to stir for approximately 2
h until a completely clear solution was obtained. All solutions were filtered through a 0.22 µm
PTFE filter syringe filter before use.
3.2.3 Chemical bath deposition

Chemical bath deposition (CBD) is a technique commonly used to deposit high quality
thin films of semiconductor materials for device application.[1] In CBD a thin film grows
on a substrate from a solution containing metal ions and a source of counter ions, typically
O2 – , OH – , S2 – or Se2 – . The material being deposited must be highly insoluble in the bath
solution, e.g. CdS in water (Ksp = 8 × 10−7 ).[2] The choice of adequate complexing agents
limits the concentration of free metal ions. Additionally, the counter ions are typically not
free in solution, e.g. use of thiourea as a source of S2 – . This results in the slow release of
ions in solution and the controlled growth of a thin layer on the surface of the substrate.
Such controlled growth allows for the deposition of high quality and pin hole free thin films,
ranging from a few nanometers up to a few micrometres.
CdS CBD: This method was adapted from the literature.[3] Substrates were immersed in
a glass container with 750 mL aqueous solution of cadmium acetate dihydrate (1.3 g) and
ammonium acetate (1.5 g) at 65 °C with magnetic stirring. CdS film thickness was controlled
by adjusting the deposition time. For ≈60 nm CdS, the deposition was started by adding a
thiourea solution (0.74 g in 32.5 mL of 25 % ammonia) and stopped after 10 min by removing
the substrates and placing them in Milli-Q water. This ≈60 nm CdS deposition was used as
the standard for all devices compared in this manuscript. An additional 100 nm CdS film,
88
used for comparative work, was achieved by increasing the deposition time to 16 min. Films
were then ultrasonicated for 60 s in Milli-Q water and then for a further 30 s in HCl (pH 3) to
remove surface residues and finally rinsed with milli-Q and isopropanol before drying with
N2 gun.
CdS post-treatment: After deposition, CdS was treated with H2 O2 and CdCl2 . The H2 O2
treatment was done by dropping H2 O2 30 % (500 µL) on films spinning at 3000 RMP for 30 s.
Films were then rinsed with Milli-Q water and dried using an N2 gun. Next, the films were
treated with CdCl2 by spinning warm supersaturated solution of CdCl2 · H2 O in methanol
(100 µL, 20 mg mL−1 ) at 4000 RPM for 30 s followed by annealing for ≈5 min at 400 °C in air.
ZnO CBD: Nanostructured and flat ZnO were deposited via chemical bath deposition (CBD)
on glass/FTO (ZnO seeded) substrates.[4] First, substrates were cleaned in soapy water, ace-
tone and IPA ultrasonic baths for 10 min in each solution. ZnO seeds were then deposited
via a sol-gel method. Specifically, the sol-gel solution was prepared by adding ethanolamine
(105 µL) to 0.35 M zinc acetate dihydrate in 2-methoxyethanol (5 mL). The mixture was left to
stir for approximately 2 h until a completely clear solution was obtained followed by filtering
through a 0.22 µm PTFE syringe filter. This sol-gel solution was spun on clean glass/FTO
substrates at 4000 RPM for 30 s. Two layers were deposited to ensure full coverage (≈30 nm,
as measured by profilometry). Substrates were heated at 200 °C for ≈2 min between layers,
before final annealing at 250 °C for 60 min in air. To deposit ZnO-NW, the seeded substrates
were inclined inside a small (≈25 mL) beaker with the reactive (seeded) surface facing down-
wards. Aqueous Zn(NO3 )2 (20 mM in ≈0.8 vol% NH3 ) was then added to cover the substrate.
The beaker was then placed in a water bath at 80 °C for exactly 60 min. The deposition was
stopped by transferring substrates to a Milli-q water bath. Films were dried under a N2 gas
stream. To achieve ‘ZnO-Flat’ samples, ammonium citrate (2 mM) was added to the initial
growth solution. All ZnO-NW and ZnO-Flat substrates were annealed at 500 °C for ≈1 h
before being used.
89
3.2.4 Additional film deposition methods used in this thesis
ZnO-NW coatings: ZnO, TiO2 and Al2 O3 were deposited via atomic layer deposition (ALD).
The number of ALD cycles was calibrated (using ellipsometry) to yield the desired layer
thickness. CdS was deposited via successive ionic layer adsorption and reaction (SILAR).[5]
Atomic Layer Deposition: DEZ, TMA, and TTC were used as metal-organic precursors (MOP)
with H2 O as the oxidation source. MOPs and H2 O were kept at room temperature, whereas
the delivery manifold and the reactor were maintained at 80 °C and 100 °C, respectively.
High purity N2 (> 99.999% purity) was employed as the carrier and purge gas throughout
the deposition at a flow rate of 40 sccm. A typical ALD cycle consisted of the following 6
consecutive steps: MOP was pulsed for 200 ms, followed by a 3 s exposure and a 40 s purge.
H2 O was then pulsed for 200 ms, followed by a 6 s exposure and a 40 s purge. The number of
ALD cycles was calibrated (using ellipsometry) to yield the desired layer thickness. ZnO ALD
was conducted on a commercial Arradiance GEMStar benchtop XT-S thermal ALD reactor.
TiO2 was deposited on a thermal ALD system (TFS200, BENEQ) and Al2 O3 was deposited
using Oxford instruments PlasmaPro System100 Cluster with ALD chamber.
CdS Deposition: The SILAR deposition was started by dipping the ZnO-NW substrate in the
Cd2+ solution (Cd(NO3 )2 · 4 H2 O, 50 mM) for 20 s, followed by dipping in the S2 – solution
(Na2 S · 9 H2 O, 50 mM) for 20 s. Between each ionic solution, excess ions were removed by
immersing the substrate in a two-step Milli-Q water bath for 10 s each. This procedure was
repeated 30 times (30 SILAR cycles). Films were then washed with Milli-Q water and dried
using an N2 gun.
Sb2 Se3 film transfer: PMMA solution (10% in chlorobenzene) was spun on glass/FTO/ZnONW/
Sb2 Se3 NR substrates at 1000 RPM for 60 s and dried at 80 °C for 10 min. The glass/FTO/ZnONW/
Sb2 Se3 NR/PMMA substrate was put in a 1M H2 SO4 bath overnight until the ZnONW layer
was completely dissolved and the Sb2 Se3 NR/PMMA film detached from the substrate. The
detached film was moved through 2 clean milli-q water baths to wash out the acid. The float-
ing Sb2 Se3 NR/PMMA film is transferred to the receiving substrate directly from the water
bath and dried on a hot plate at 80 °C for ≈30 min.
90
3.3 Thermal evaporation
This method is based on the evaporation of the desired material from an evaporation
source, often under vacuum. The source is heated until sufficient vapour pressure is obtained,
to achieve an adequate evaporation rate. The thermally excited particles are transported from
the source to the substrate. Upon collision with the surface of the substrate, the high energy
particles lose kinetic energy and start to nucleate. Under ideal conditions, the nuclei will grow
giving rise to a thin film. The VTD method used to deposit Sb2 Se3 thin film in this thesis is
a type of physical vacuum deposition (PVD) with thermal excitation of the source. Other
PVD techniques use collision with high electron or ion beams to evaporate the material.[6]
However, thermal evaporation has proven to work well for deposition of Sb2 Se3 thin films
due to its high vapour pressure and absence of crystal polymorphism.[7]
Sb2 Se3 deposition: An optimized vapour transport deposition (VTD) method was used to
deposit the Sb2 Se3 films, as previous described.[8] In short, the Sb2 Se3 source (quartz crucible
with Sb2 Se3 powder) was put at the centre of the left heating zone of a dual zone furnace
(equipped with a quartz tube). The mass of Sb2 Se3 powder in the crucible was adjusted as
needed (500 mg in chapter 4, 50-1000 mg in chapter 5 and 2000 mg in chapter 6). Substrates
were mounted within the right heating zone on a near vertical graphite holder positioned 10
cm from the furnace centre. Substrates were mounted within the right heating zone on a near
vertical graphite holder positioned 10 cm from the furnace centre. Pressure was equilibrated
at 0.5 - 0.7 Pa before starting the heating program. Both zones were then heated at 250 °C for
30 min. The left zone, containing the Sb2 Se3 source, was then heated to 540 °C over the course
of 15 min (≈20 °C/min) and the temperature was held at 540 °C for 10 min. The right zone,
containing the substrates, was kept at 250 °C throughout the whole process. The furnace was
left to cool naturally for ≈2 h before ambient pressure was restored by bleeding N2 gas into
the tube. More details on the furnace setup used for deposition are in the next section.
3.3.1 Furnace setup

The Vapour Transport Deposition (VTD) of Sb2 Se3 was introduced in chapter 2. Here
a more efficient holder/source configuration is implemented in a two-zone furnace. This
configuration, which consists of a holder sitting in the middle of a heating zone (instead
91
of outside the zone), allows precise control over substrate temperature as well as over the
amount of material being deposited. Here, I describe this optimized VTD in detail and show
the efficiency calibration of Sb2 Se3 solar cells fabricated using this method.[9]
Figure 3.1 shows a photograph of the MTI dual zone furnace OTF-1200X-II with a quartz
tube (75 mm internal diameter). The furnace was adapted by first connecting a pump to the
tube to achieve a vacuum of ≈0.5 Pa. Second, valves connected to the N2 line were installed
to allow both, slow bleed of gas for fine control of pressure in the tube, as well as fast N2 flow
when required. A thermocouple was also installed to monitor substrate temperature during
deposition.
V4
Z1 Z2 QT P
V1
V2
V3
Figure 3.1: Photograph of dual zone furnace used for VTD of Sb2 Se3 films. Z1 and Z2 indicate heating
zones 1 and 2. Valves: V1 (tube valve), V2 (fast N2 input), V3 (fine N2 input) and V4 (back N2 input).
T: thermocouple, P: pump and QT: quartz tube.
A graphite holder, Figure 3.2a, was designed and custom-made. This holder has the ca-
pacity to hold up to four substrates (2.5 x 2.5 cm) per evaporation (Figure 3.2b). Additionally,
a 5° inclination allows the substrates to rest on the holder during the evaporation (Figure 3.2c).
On its back side, the holder allows a thermocouple (∅ = 1.5 mm) to be installed at 2 mm from
the surface of the holder and accurately monitor substrate temperature.
92
Table 3.1: Example of a typical heating program used for Sb2 Se3 evaporation.
Zone 1 Zone 2
Ti (°C) T f (°C) time (min) Ti (°C) T f (°C) time (min)
20 180 8 20 180 8
180 540 19 180 180 29
540 540 10
a) b) FRONT
c) SIDE
5°
~1.5 mm
~2 mm
73 mm
~2-3 mm ~3 mm
~26 mm
10 mm
~5 mm
~5 mm ~40 mm
Figure 3.2: a) photograph and schemes showing b) front and c) side of the graphite holder designed
for this thesis.
For clarity, Figure 3.3 shows a schematic representation of the furnace used here. The
source with the desired amount of Sb2 Se3 powder sits in the middle of zone 1 (12 cm form the
furnace centre) that is set to a high temperature (typically 540 °C). The graphite holder with
substrates is put within zone 2 (typically 10 cm from the furnace centre), with this zone set to
a lower temperature (typically 250 °C). This furnace configuration contrasts most methods de-
scribed in the literature that put the holder outside zone 2,[9]. The method presented here has
two main advantages. It allows precise control of film thickness by simply varying the amount
of Sb2 Se3 in the source and independently controlling Zone 1 and Zone 2 temperatures.
zone 1 zone 2
N2 pump
heating
Sb2Se3 center holder element
source
Figure 3.3: Schematic representation of the furnace configuration used in this work.
93
3.3.2 Heating program and deposition rate
Table 3.1 shows an example of heating steps for zones 1 and 2 of the furnace. Figure 3.4
shows the temperature curves of Zone 1, Zone 2 and substrates for the heating program in
Table 3.1. The first step of this heating program consists of heating both zones at the same
temperature for some time (typically 30 min) to equilibrate the temperature within the tube.
Then, the temperature is increased within zone 1, typically to 540 °C at a heating rate of
20 °C min−1 to evaporate the Sb2 Se3 in the source. The temperature of both zones is held for
some time (typically 10 min) before turning off the heat. Even though the temperature of Zone
2 is being held constant, the substrate temperature is slowing rising during the evaporation
due to irradiated heat from the hotter Zone 1. This problem could be partially solved by
increasing the distance between the holder and Zone 1 to minimize the absorption of the
irradiated heat; however, it results in deposition rates that are too slow and not practical for
the fabrication of devices.
%
Figure 3.4: Temperatures measured in Zones 1, Zone 2 and holder/substrate. The graphite holder with
substrates was positioned in Zone 2 at exactly 10 cm from the furnace centre.
Figure 3.5 shows a picture of the furnace tube after an evaporation. The temperature
gradient drives the Sb2 Se3 vapour created within Zone 1 to the colder Zone 2. In Zone 2 the
vapour condenses on the tube walls and its concentration decreases from the furnace centre
to the left end of the tube. This means that the deposition rate of Sb2 Se3 on the substrates
94
is inversely proportional to the distance substrates are from the centre of the furnace. The
Sb2 Se3 deposited on the tube walls can absorb heat or re-evaporate, thus it should be cleaned
completely before a new deposition batch by heating both zones at 560 °C under vacuum.
furnace Deposition
centre on tube walls
decreasing concentration
Zone 1 Zone 2
%
Figure 3.5: Photograph of furnace tube after evaporation at 0.5 Pa showing the deposition of Sb2 Se3 on
tube wall of Zone 2.
The range where the holder can be positioned within Zone 2 is limited by two factors (i)
high heat irradiation from Zone 1 and (ii) the deposition of material on tube walls (concentra-
tion gradient). First, heat irradiation from Zone 1 rises the temperature near the centre of the
furnace. The holder can be positioned further away from the centre, where the temperature
can be more easily controlled to yield quality films. However, pushing the holder too far
results in very slow deposition rates. Thus, an impractically large amount of Sb2 Se3 would be
required to achieve adequate film thickness (>500 nm), as the material deposits on the part of
the tube that is in front of the holder. For the work related to this thesis project, the distance
between the holder and the centre of the furnace was chosen at exactly 10 cm. This distance
was verified to yield a good balance between satisfactory temperature control, film quality,
and deposition rate.
3.4 Devices and materials characterisation

3.4.1 Instrumentation for physicochemical characterisation
SEM: the scanning electron microscope (SEM) images, backscattered SEM (BS-SEM) im-
ages, and energy dispersive spectra (EDS) were taken on a FEI Verios 460L XHR-SEM equipped
with in-lens secondary electron detector (TDL) and in-column backscattered detector (ICD).
A FEI Nova NanoSEM 200 was occasionally used to obtain some of the images in chapter 4.
TEM: Transmission electron microscope (TEM, HR-TEM) images and selected area electron
95
diffraction (SAED) patterns were taken on a JEOL JEM-2100 FEGTEM at 200 kV. XRD: X-ray
diffraction (XRD) patterns were obtained using a Bruker AXS D4 Endeavour diffractometer
with a Cu-Kff radiation source. XPS: X-ray photoelectron spectra were collected on a Kratos
AXIS Supra XPS with monochromated Aluminum Kff source (1486.7 eV). Absorbance: Ab-
sorbance and reflectance spectra were collected on a Cary 7000 UV-Vis-NIR spectrophotome-
ter equipped with an integrating sphere. AFM: Atomic Force Microscope (AFM) images were
obtained on an Asylum Research MFP-3D Infinity AFM. Film thickness: film thickness was
determined using a Tencor P 16+ Profiler or via ex-situ ellipsometry. Ellipsometric measure-
ments were conducted on a M-2000 Spectroscopic ellipsometer (J.A. Woollam). The analysis
was carried out at a fixed incident angle of 65° over a spectral range of 250-1500 nm. Curve
fitting and thickness modelling were conducted on Complete EASE software, employing a
Cauchy model to study the metal oxide layers.
3.4.2 Device fabrication

The Sb2 Se3 absorber layer was deposited on the various substrates specified in each chap-
ter (e.g. glass/FTO/(HRMO)/CdS, glass/FTO/ZnO-Flat, glass/FTO/ZnO-NW, etc) using
the optimised VTD method described in this chapter. To encapsulate the Sb2 Se3 NR layer, a
PMMA solution (10% in chlorobenzene) was spun on the substrates at 1000 RPM for 60 s and
dried at 80 °C for 10 min. The encapsulated Sb2 Se3 NR/PMMA layers were first etched in O2
plasma for 90 s using a Plasma Etch benchtop system (PE-50), 15 cc/min O2 flow. Devices
were completed by depositing a back contact: 100 nm gold layer using an e-beam evaporator
(Kurt J. Lesker PRO Line PVD 75) for solar cells in chapter 4 and photodetectors in chapter
5, C-paint (non-flexible detectors) or PEDOT (flexible detectors) that were drop-cast directly
on the etched layers and dried for ≈10 min at 70 °C for in chapter 6. Before deposition, the
PEDOT solution was doped with 5% DMSO to increase conductivity.
3.4.3 Photoelectrochemical measurements and instrumentation

Current density vs voltage curves and solar cell efficiency: current density vs voltage measure-
ments were conducted on a customized set up with Keithley 2401 Source Meter. Simulated
AM1.5G light was produced in an Oriel Sol3A Class AAA Solar Simulator, and solar cell effi-
ciency was determined following the guidelines for best measurement technique,[10] using a
laser cut mask (0.100 cm2 ) to define the device area.
96
Electrochemical measurements: Nyquist and Mott-Schottky plots were obtained from Electro-
chemical Impedance Spectroscopy (EIS) measurements done using an Ivium Vertex.1A.EIS
potentiostat at -0.8 V in a frequency interval varying from 200 kHz to 1 Hz.
External quantum efficiency: EQE was in a PVM QEX10 Solar Cell Quantum Efficiency Mea-
surement System the same equipment was used to generate quasi monochromatic light for
measurements in chapter 4.
Photodetector measurements: photoresponse was measured on a customized set up inside a
shielded dark box using a Keithley 2636B SourceMeter. ThorLabs LEDs 455 nm, 530 nm, 625
nm, 730 nm and 850 nm were used as illumination sources with a calibrated irradiation as
specified for each experiment. Rise/Fall times and -3dB curves were obtained using function
generator Stanford Research Systems (model DS345), low-noise current preamplifier Stanford
Research Systems (model SR570), ThorLabs laser diode control (model LDC205C) with a 530
nm laser diode and oscilloscope Tektronix (TDS1012B).
3.4.4 Sb2 Se3 devices and efficiency calibration

For clarity, Figure 3.6a shows a schematic representation of the stack of layers in the
devices used in this calibration. The Sb2 Se3 absorber layer was deposited on glass/FTO/CdS-
120 nm substrate and the device was completed by depositing a 100 nm gold layer as back
electric contact using an e-beam evaporator. The final device architecture is glass/FTO/CdS/
Sb2 Se3 /Au.
97
Figure 3.6: a) Scheme of the stack of layers used in the devices b) back and c) front of devices.
Figure 3.6b shows the back (Sb2 Se3 side) and front (glass side) of a Sb2 Se3 device. The
device consists of 8 isolated FTO islands. The 4 smaller islands contain the Au pads that will
touch contact with pins in the measurement box. Each of the 4 bigger islands contains 1 front
contact that is touching the FTO and 1 Au pad that constitutes the active back contact (pixel).
Each device contains 4 active pixels that are completely isolated from each other. The solar
cell efficiency is determined following guidelines of good practice,[10] and using a laser cut
mask (0.100 cm2 ) to define the illuminated device area, see Figure 3.6c for reference.
As shown in Figure 3.7 the efficiency of the solar cells was calibrated as a function of two
main parameters: substrate temperature and Sb2 Se3 film thickness. Substrate temperature
was controlled by varying the Zone 2 final temperature from 230 to 270 °C and film thickness
was controlled by changing the mass of Sb2 Se3 in the source.
98
a) b) 0.4
4
3 0.35
PCE (%)
Voc (V)
2
0.3
1
0 0.25
230 250 270 500 650 800 950 230 250 270 500 650 800 950
ZoneTemperature
2 temperature
(°C) (°C) SbThickness
2
Se3 mass(nm)
(mg) Zone 2 temperature
Temperature (°C) (°C) Sb Se3 mass
Thickness
2
(mg)
(nm)
c) d)
60 20
50 17
Jsc (mA·cm-2)
FF (%)
14
40
11
30
230 250 270 500 650 800 950 230 250 270 500 650 800 950
ZoneTemperature
2 temperature
(°C) (°C) 2
Se3 mass(nm)
SbThickness (mg) Zone 2 temperature
Temperature (°C) (°C) Sb Se3 mass
Thickness
2
(mg)
(nm)
Figure 3.7: a) PCE, b) VOC , c) Jsc and FF d) for glass/ITO/CdS-120 nm/Sb2 Se3 /Au solar cells. A 650
mg source was used for the series obtained at varying substrate (Zone 2) temperature. For the thickness
series (varying Sb2 Se3 mass in the source) Zone 2 temperature was kept constant at 250 °C.
First, Zone 2 temperature was adjusted using a fixed mass of Sb2 Se3 in the source (650
mg). Figure 3.7a reveals that the highest PCE was achieved for the layer deposited at 250 °C.
Devices obtained at this temperature showed significantly improved PCE and lower variability
for all parameters. Zone 2 temperature was then kept constant at 250 °C and the mass of
Sb2 Se3 was varied. Devices fabricated using 500 mg of powder in the source showed the
highest efficiency. It can be seen in Figure 3.7b and c that VOC and FF did not vary significantly
with film thickness and the improved PCE can the attributed to improved JSC .
The best device obtained here showed average PCE of 4.11%. This is comparable to aver-
age Sb2 Se3 solar cells reported in the literature, but still lower than the record 7.6% reported
for a device with the same architecture via VTD.[9] For reference Table 3.2 and Table 3.3
show parameters for the best performing solar cells. Additionally, the tables contain diode
parameters from fit of JV dark curves for the corresponding devices.
99
Table 3.2: Fitted parameters for champion devices as a function of Zone 2 temperature.
Temp. (°C) Jsc (mA cm−2 ) Voc (V) FF (%) PCE (%) n J0 (µA cm−2 ) Rsh (Ω cm−2 ) RS (Ω cm−2 )
230 14.6 0.362 50.4 2.67 1.49 0.954 134 4.42
250 17.3 0.382 57.6 3.80 1.61 1.63 324 2.28
270 15.5 0.392 57.8 3.52 1.65 1.32 283 2.33
Table 3.3: Fitted parameters for champion devices as a function of Sb2 Se3 mass in the source.
Thick. (nm) Jsc (mA cm−2 ) Voc (V) FF (%) PCE (%) n J0 (µA cm−2 ) Rsh (Ω cm−2 ) RS (Ω cm−2 )
500 18.7 0.389 57.2 4.17 1.53 0.874 273 2.60
650 17.3 0.382 57.6 3.80 1.61 1.63 324 2.28
800 16.9 0.380 57.5 3.71 1.57 1.31 313 2.44
950 17.2 0.390 57.6 3.87 1.52 0.773 347 2.85
The results show that the proposed VTD method can be used to make efficient Sb2 Se3 so-
lar cells. Optimised devices presented efficiency (≈4 %) that is comparable to average devices
reported in the literature. The method provided fine control over deposition parameters. This
fine tuning was crucial for the studies conducted in chapters 4, 5 and 6 of this thesis. Control-
ling the amount of Sb2 Se3 being deposited was especially important when studying Sb2 Se3
film growth in chapter 5. Source temperature, pressure and holder/furnace distance can still
be optimised in future work seeking higher PCE.
3.5 References
[1] P. K. Nair, M. T. S. Nair, V. M. Garcı́a, O. L. Arenas, A. C. Y. Peña, I. T. Ayala, O.
Gomezdaza, A. Sánchez, J. Campos, H. Hu, R. Suárez, M. E. Rincón, Sol. Energy Mater.
Sol. Cells 1998, 52, 313.
[2] D. Lide, W. Haynes, CRC Handbook of Chemistry and Physics: a Ready-reference Book of
Chemical and Physical Data, 89th, CRC Press/Taylor and Francis, Boca Raton, FL, 2009.
[3] J. van Embden, J. O. Mendes, J. J. Jasieniak, A. S. R. Chesman, E. Della Gaspera, ACS
Appl. Energy Mater. 2020, 3, 7885.
[4] E. Della Gaspera, D. F. Kennedy, J. van Embden, A. S. R. Chesman, T. R. Gengenbach,
K. Weber, J. J. Jasieniak, Adv. Funct. Mater. 2015, 25, 7263.
[5] Y. Tak, S. J. Hong, J. S. Lee, K. Yong, J. Mater. Chem. 2009, 19, 5945.
[6] H. Frey in Handbook of Thin-Film Technology, (Eds.: H. Frey, H. R. Khan), Springer Berlin
Heidelberg, Berlin, Heidelberg, 2015, Chapter 13-71.
[7] X. Liu, J. Chen, M. Luo, M. Leng, Z. Xia, Y. Zhou, S. Qin, D. J. Xue, L. Lv, H. Huang,
D. Niu, J. Tang, ACS Appl. Mater. Interfaces 2014, 6, 10687.
[8] J. O. Mendes, E. Della Gaspera, J. van Embden, Sol. RRL 2022, 6, 2200265.
[10] H. J. Snaith, Nat. Photonics 2012, 6, 337.
100
Appendix
To support the methods described in this chapter this appendix provides step-by-step
descriptions of the methods, flow diagrams, photos of apparatuses, schematic representation
of experiments and extra characterisation that provides insides into the quality of the films.
A1 Substrate washing
Washing substrates is a crucial step for adequate thin film growth. Substrates should
be completely free of grease and dust before film deposition. The substrates were cleaned
following the steps:
1. Sonicate in soapy water (5-10 min);
2. Rinse throughout with distilled water and dip in distilled water.
3. Rinse with acetone and sonicate in acetone (10 min).
4. Rinse with IPA and sonicate in IPA (10 min).
5. Dry thoroughly with N2 gun.
6. UV/ozone for 20 min.
Substrates should be used immediately after cleaning and should not be stored.
101
A2 ZnO SG
A 2.1 ZnO SG step-by-step
1. Set spin coater program to 30s@4000RMP, set hot plate to 200 °C and load the substrate
on the spin coater.
2. Drop ≈100 µL of solution on the substrate (adjust the volume according to the size of
the substrate) and spread it completely over the surface of the substrate before start
spinning.
3. Let the spinning program finish and transfer the film immediately to the 200 °C hot
plate.
4. If depositing more layers, remove the film from the hot plate, let it cool and repeat the
procedures above until the desired number of layers is achieved.
5. Anneal the film for ≈60 min at the desired temperature.
2-methoxyethanol
5 mL
0.22 μm PTFE
Vigorous stirring
Zinc Acetate ~2h at room temp. Spin coating
5mL
dihydrate 20mL clean vial Filter
clean vial
0.7683 g - 0.7 M Solution should be
completely clear
Use within a few hours
before next step
Ethanolamine
211 μL
(dropwise)
Figure A1: ZnO sol-gel solution preparation (flowchart)
102
A 2.2 ZnO SG comments
To gain insight into film quality, the visual aspect of the films was evaluated. Figure A2
shows ZnO SG films deposited on glass/FTO substrates annealed at 200 °C (a) and 500 °C
(b). Both films are highly transparent and pristine, absent of surface markings or scratches.
Annealing does not change the visual aspect of the films. The grain structure of ZnO in the
films was assessed by SEM. Figure A2a (bottom line) shows the top view SEM images of
ZnO SG deposited on glass/FTO substrate annealed at 200 °C, this relatively low temperature
annealing results in a layer composed of small ZnO crystal grains. Figure A2b (bottom line)
shows that considerable grain growth occurs after annealing the film at 500 °C.
Figure A2: Photographs and top view SEM images of ZnO SG films spun for 30s at 3000RPM from a
0.35 M solution and annealed at a) 200 °C and b) 200 °C for ≈60 min.
Table A2: Average thickness for ZnO SG films deposited on glass/FTO substrates at 4000 RPM.
Precursor solution concentration (M) 0.7 0.5 0.3 0.1

Layer thickness (nm) 41 ± 3 20.5 ± 0.5 11 ± 1 8±1
103
A3 TiO2 SG
A 3.1 TiO2 SG step-by-step
1. Set spin coater program to 30s@4000RPM, set hot plate to 200 °C and load the substrate
on the spin coater.
2. Drop ≈100 µL of solution on the substrate (adjust the volume according to the size of
the substrate) and spread it completely over the surface of the substrate before start
spinning.
3. Let the spinning program finish and transfer the film immediately to the 200 °C hot
plate.
4. If depositing more layers remove the film from the hot plate, let it cool and repeat the
procedures above until the desired number of layers is achieved.
5. Anneal the film for ≈60 min at 500 °C.
1. 1-butanol
3.4 mL
0.22 μm PTFE
mix
2. Titanium diisopropoxide ~5 min Spin coating
5mL
bis(acac) – 75wt.%IPA 5mL clean vial Filter
clean vial
~585 μL - 0.3 M Solution should be
completely clear
use within a few hours
before next step
Figure A3: TiO2 sol-gel solution preparation (flowchart)
104
A 3.2 TiO2 SG comments
To gain insight into film quality, the visual aspect of the films was evaluated. Figure A4a
shows a TiO2 SG film deposited on glass/FTO substrates annealed at 500 °C. The film is
highly transparent and pristine, absent of surface markings or scratches. The grain structure
of TiO2 in the films was assessed by SEM. Figure A4b shows the top view SEM image of TiO2
SG deposited on glass/FTO substrate annealed at 500 °C. The SEM indicates that full coverage
of substrate was obtained.
Figure A4: a) Photography and b) top view SEM image of TiO2 SG film spun for 30s at 3000RPM from
a 0.3 M solution and annealed at 500 °C for at least ≈60 min.
Table A4: Average thickness for TiO2 SG films deposited on glass/FTO substrates at 4000 RPM.
Concentration (M) 0.7 0.5 0.3 0.1

Thickness (nm) 48 ± 2 25 ± 3 16.7 ± 0.4 3.4 ± 0.4
105
A4 CdS CBD
cadmium acetate
dihydrate
1.30 g
2. 20 mL Milli-Q
vial 15 mL
Milli-Q remove
3. PE
750 mL bubbles
holder w/
substrates
Ammonium acetate
1. 1L container 4. 20 mL 1.50 g
vial Milli-Q
65 °C/300 RPM 15 mL
25% ammonia solution

5. 100 mL 30 mL
Erlenmeyer Thiourea
0.74 g (start timer)
Stop
reaction 7. Washing
6. Clean Milli-Q
8. post-treatments
Figure A5: CdS CBD procedure (flowchart)
106
A 4.1 CdS washing (step-by-step)
The deposition is immediately followed by washing the films. Remove the tape and:
1. Sonicate in milli-Q water (20 – 60s).
2. Sonicate in HCl solution pH 3 (30-60 s) and, immediately after, transfer the films to clean
Milli-Q water to wash out the acid.
3. Rinse with Milli-Q and IPA.
4. Dry with N2 gun.
Any CdS deposited on the back side of the films is removed by rubbing a tissue wet with
≈2 M HCl solution, followed by a final clean with an acetone tissue. After washing, CdS is
treated with H2 O2 [1] and CdCl2 [2], respectively.
A 4.2 H2 O2 treatment (H2 O2 30%) (step-by-step)

1. Set spin coater program to 5s@500RPM + 30s@3000RPM.
2. Start spinning and drop 500 µL of H2 O2 30% within the first 5s and let the spinning
program finish.
3. Immediately remove film from spin coater and rinse with plenty Milli-Q water.
4. Dry the films completely with N2 gun.
5. Clean spin coater with a tissue and acetone and let it dry completely before next step.
Table A7: Thickness measurements for CdS films deposited on FTO.
CBD time (min) Thickness (nm) Growth rate (nm/min)

14 90 6.4
16 100 6.4
Average growth rate: 6.4 nm/min
107
Figure A6: Apparatus used to perform CdS CBD. 1. metal support, 2. polyethylene (PE) holder, 3.
substrates taped with polyimide (PI) tape, 4. chemical bath, 5. support for chemical bath + stirring bar
(not visible), 6. external water bath covered with Al foil, 7. heating + stirring, 8. flat dish (chemical
bath lid) and 9. Milli-Q water bath to quench the reaction.
1) CdS deposition
flat dish (lid)
PE metal
holder support
PA
substrates tape 72°C 300RPM
external chemical
water bath bath 2) quench reaction
stirring support
bar
heating/
stirring
Figure A7: Schematic representation of CdS CBD; 1) apparatus during deposition and 2) quench the
reaction in a clean Milli-Q water bath.
108
A 4.3 CdCl2 treatment (CdCl2 20 mg/mL in methanol) (step-by-step)
1. Set spin coater program to 30s@4000RPM, set hot plate to 400 °C and put the CdS film
on spin coater.
2. Heat the CdCl2 solution with a hair dryer in the vortex mixer until it is completely clear.
3. Drop 100 µL of the warm CdCl2 solution on CdS film, you will see it spreading com-
pletely over the surface of CdS film. Start spinning immediately after it finishes spread-
ing.
4. Let the spinning program finish and transfer immediately to the 400 °C hot plate for 5-10
min.
5. Remove films from hot plate and let them cool down.
a) b) c) d)
heat only H2O2 H2O2 + CdCl2

washed
400°C only + heat
Figure A8: Photographs of CdS films a) washed, b) heated at 400 °C for 5 min, c) H2 O2 treated and d)
CdCl2 treated.
Table A8: Relative atomic composition of CdS layers after post treatments via energy dispersive spec-
troscopy (EDS).
Treatment Cd S Cl
Control 1.15 1 0
Heat 1.05 1 0
H2 O2 1.05 1 0
CdCl2 1.27 1 0.194
109
A 4.4 CdS CBD comments
To gain insight into film quality, the visual aspect of the films was evaluated. Figure A8
shows the surface of CdS films obtained via the present method. The films generally look nice,
clean, and even, with no marks on the surface. The untreated film is yellow (Figure A8a), and
if heated to 400 °C without CdCl2 treatment it will turn permanently orange (Figure A8b). It
is unclear why this colour transition occurs, two possibilities are (i) the formation of oxides or
(ii) the increase in the concentration of defects in the middle of the band gap which can change
the electronic and optical properties of the material. The yellow colour of CdS film remains
unchanged after H2 O2 treatment (Figure A8c). The CdCl2 treated film will turn orange at high
temperatures but recovers the characteristic bright yellow colouration when it cools down to
room temperature (Figure A8d). For the work reported in this thesis, CdS films were always
treated with CdCl2 before using in devices.
The grain structure of the CdS in the film was assessed by SEM. Figure A9a shows that
CdS deposits as an agglomerate of very small grains. Figure A9b and Figure A9c show that
this grain structure is preserved after heat or H2 O2 treatments. In Figure A9d noticeable grain
growth is observed after CdCl2 treatment. The improvement of solar cell performance after
CdCl2 treatment of the CdS layer is often attributed to this grain growth that can act to i)
improve film coverage by closing up pin holes and ii) improve charge transport by reducing
the concentration of grain boundaries.[1] It is also important to note that all films presented
excellent coverage and no pin holes were detected by SEM.
The crystalline phases were studied via XRD. Figure A10 shows the XRD patterns of
samples after the different treatments. It is unclear if the CdS deposits with a cubic or a highly
oriented hexagonal phase. This phase remains unchanged after heat or H2 O2 treatments. After
CdCl2 treatment the presence of the hexagonal phase is clear. The argument that a hexagonal
CdS buffer layer helps improve the performance of Sb2 Se3 solar cells has been supported on at
least one occasion.[3] Moreover, the extra peaks that are present in the pattern obtained from
CdCl2 treated CdS film are attributed to left over crystalline CdCl2 salt from the treatment.
110
Figure A9: Top view SEM images of CdS deposited on glass/FTO substrate a) washed, b) heated at
400 °C for 5 min, c) H2 O2 treated and CdCl2 treated
Treatment:
control
400 °C
H2O2
CdCl
CdCl2 * salt 2
* *
CdS
hexagonal
CdS
cubic
FTO
10 20 30 40 50 60
Diffraction angle (°2θ)
Figure A10: XRD patterns of CdS film washed (control - blue line) and treated at 400 °C (grey line),
H2 O2 (red line) and CdCl2 (black line).
111
A5 ZnO CBD
a) Zinc nitrate
ZnO Flat hexahydrate
(pillars) 1.78 g
8 mL 150 mL 2.5% Ammonia

Milli-Q
14.4 mL Erlenmeyer 100 mL
external ammonium citrate dibasic

water bath 1.6 mL 100.17 mg
25 mL beaker 20 mL vial
Milli-Q
15 mL
seeded
substrate
b) ZnO NW
Zinc nitrate
Milli-Q hexahydrate
14.4 mL 1.78 g
external
water bath 9.6 mL 150 mL 2.5% Ammonia
25 mL beaker
Erlenmeyer 100 mL
seeded
substrate
Figure A11: ZnO CBD procedure (flowchart) for deposition of a) ZnO Flat and b) ZnO NW films.
Table A11: Root mean square (RMS) roughness of ZnO Flat and ZnO NW films
Film RMS roughness (nm)

ZnO Flat 38.6
ZnO NW 9.56
112
1
2
3
Figure A12: Apparatus used to perform ZnO CBD. 1. chemical bath, 2. external water bath wrapped
in Al foil, 3. set of supports and stirring bar, 4. heating/stirring plate and 5. Milli-Q water bath to stop
reaction.
1) ZnO deposition
external chemical
water bath baths
stirring supports 80°C
bar
300RMP
heating/ 2) quench reaction

stirring
Figure A13: Schematic representation of ZnO CBD; 1) apparatus during deposition and 2) quench the
reaction by dipping the substrates in a clean Milli-Q water bath.
113
A6 Sb2Se3 layer
5E+11
4E+11
p-type
3E+11
1/C2 (1/F2)
2E+11
1E+11
0
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
Potential (V)
Figure A14: Mott–Schottky (MS) plot for a typical Sb2 Se3 layer. The MS curve shows a positive inclina-
tion typical of a p-type film. In the literature Sb2 Se3 is typically reported as a p-type material, however
this character can change according to the fabrication method and doping.[4–6] MS was obtained from
a Sb2 Se3 film in a 0.5 M Na2 SO4 solution pH = 5 at 1 kHz using a potentiostat Ivium Vertex.1A.EIS.
A 6.1 Sb2 Se3 deposition (step-by-step)

1. Open furnace’s lid and check if tube is clean before starting the evaporation (if the tube
is not clean see the Furnace Cleaning procedure below).
2. Put substrates in the holder and check if they are resting nicely and won’t fall out, up to
4 substrates 2.5 x 2.5 cm can be mounted in the holder at a time.
3. Open the right end of the furnace.
4. Gently put the holder with substrates in the tube and slide to zone 2.
5. For a typical deposition the holder should be positioned in zone 2 at exactly 10 cm from
the furnace centre.
6. Position the crucible (Sb2 Se3 source) exactly in the middle of zone 1 (12 cm from the
furnace centre), 500 mg of power is used in a typical evaporation.
114
7. Close the end of the furnace by tightening the screws evenly.
8. Close the lid and press the green bottom to activate the furnace.
9. Make sure all tube valves are fully closed and turn on the pump and wait for the pressure
to go down ≈10 to 30 Pa.
10. Bleed N2 in the tube using valve V3 (as indicated in Figure 3.1) for a 30-60 s and close
the valve.
11. Repeat the previous item two more times
12. Wait for the pressure to go down (<10 µbar) and become stable 5-10 min.
13. If the pressure does not go below ≈10 µbar repeat item 10 and check if the cap is well
closed with all 3 screws fastened evenly.
14. Start the heating program (see heating program example in Section 3.3.2), for a typical
deposition set Zone 2 final temperature to 250 °C.
15. Wait for the program to finish.
16. Let the furnace cool down naturally with the lid closed.
17. When the furnace reaches ≈300 °C open the lid and inject N2 through the back side of
the holder (valve V4).
18. Let the furnace cool down to ≈100 °C, close the lid, close valve V4 and open the right
end of the furnace.
19. Remove the holder from the tube and let substrates cool down.
115
A 6.2 Furnace cleaning
The quartz tube needs to be cleaned after every evaporation to remove Sb2 Se3 deposited
on tube walls. This is done by heating the furnace to 560 °C under vacuum for ≈10 min.
1. Put the empty holder and quartz crucible in the middle of Zone 1 and Zone 2 respec-
tively.
2. Close the cap, valves and vacuum the tube to ≈10 Pa .
3. Set a temperature ramp from 20 °C to 560 °C and hold at 560 °C for at least 10 min.
4. Start the heating program, let it finish and wait for the furnace to cool down.
5. Check if tube is clean. No Sb2 Se3 should remain in the middle of the tube.
[2] L. Wang, M. Luo, S. Qin, X. Liu, J. Chen, B. Yang, M. Leng, D.-J. Xue, Y. Zhou, L. Gao,
H. Song, J. Tang, Appl. Phys. Lett. 2015, 107, 143902.
[3] H. Guo, X. Jia, S. H. Hadke, J. Zhang, W. Wang, C. Ma, J. Qiu, N. Yuan, L. H. Wong,
J. Ding, J. Mater. Chem. C 2020, 8, 17194.
[4] Y. B. Kim, J. S. Kim, S. K. Baek, Y. D. Yun, S. W. Cho, S. H. Jung, H. K. Cho, Energy
Environ. Sci. 2018, 11, 2540.
Appl. Phys. Lett. 2017, 111, 013901.
2621.
116
Chapter 4
High-resistance metal oxide layers for
improved front contact in CdS/Sb2Se3
superstrate solar cells
Some of the contents of this chapter have been published as:
Mendes, J.O., Della Gaspera, E. and van Embden, J. (2022), High-resistance metal oxide window
layers for optimal front contact interfaces in Sb2 Se3 solar cells. Sol. RRL, 6: 2200265.
https://doi.org/10.1002/solr.202200265
117
Abstract
Here I present an in-depth experimental investigation into the effect of employing differ-
ent high resistance metal oxide (HRMO) layers on the quality of the front contact in solar cells
with an FTO/(HRMO)/CdS/Sb2 Se3 /Au device architecture. The application of ZnO or TiO2
HRMO layers between FTO substrates and CdS, improves overall device performance. Short
circuit current gains of ≈20 %, orders of magnitude higher shunt resistances (≈104 Ω cm2 ),
and greatly improved device stabilities – maintaining over 95 % of their initial efficiency over
137 days, are observed. A suppression of the unfavorable (120) orientation of the photoactive
Sb2 Se3 layer is observed in devices with HRMO interlayers. The application of HRMO layers
is crucial to prevent both Ohmic and non-Ohmic current leaks and maintain device stability
over time. Cross over in the JV curves observed in the case of TiO2 indicates the presence of a
high barrier for the diode current in these devices. Wavelength dependent JV curves coupled
with capacitance measurements and simulations show that this barrier can be attributed to a
high density of acceptor states. In contrast, ZnO deposition was found to reduce interfacial
defects and enhance the quality of the front contact, while boosting device performance and
increasing device longevity.
118
4.1 Introduction
Applying a high resistance metal oxide (HRMO) layer at the front contact prior to CdS
buffer deposition has been a common strategy to provide high shunt protection in chalco-
genide solar cells. The properties of most common HRMO in Sb2 Se3 devices have been re-
viewed in chapter 2. Here, it is worth emphasizing that device efficiency and overall per-
formance is affected by a complex dynamic between light-absorber, buffer, HRMO, and TCO
layer at the front contact. Thus, not all metal oxides are adequate as high resistance layers.
Poor material choice for the HRMO will affect cell stability, result in sub-optimal band align-
ment, provide inadequate shunt protection, or directly generate highly defective interfaces
and subsequent barriers for the diode current or photocurrent.[1–4] Device anomalies such as
cross over, roll over, and kinks in the dark and light JV curves may arise, signalling ineffective
carrier extraction.[5, 6]
As previously discussed, SnO2 is the most studied of such interfacial layers in Sb2 Se3
solar cells. Despite the progress provided by recent work, there remains a lack of detailed
studies on HRMO materials within antimony chalcogenide solar cells other than SnO2 . Ex-
isting studies into ZnO and TiO2 HRMOs have been somewhat cursory, focusing mainly on
overall efficiency gains. To date, there have been no in-depth studies focused on the origins of
these gains. Most importantly, it is still unclear what effect these HRMO have on both defect
generation/suppression and the resultant quality of the interfaces in these solar cells. This
shortfall in knowledge within the field of antimony chalcogenide photovoltaics represents
an opportunity to develop and contribute critical knowledge around the role of key HRMO
(window) layers within this system. These investigations are essential to guide the further
development of high-quality front contacts in these devices and provide rational decisions on
the most optimal window layer to employ in Sb2 Se3 solar cells.
In this chapter, I investigate how the deposition of ZnO and TiO2 layers affect the qual-
ity of the front contact in Sb2 Se3 solar cells containing a ≈60 nm CdS buffer. The given
HRMO was deposited via reliable and inexpensive solution processed methods.[7–10] The
effect of each HRMO on device efficiency, Sb2 Se3 grain orientation, parasitic resistances, and
119
front contact interface quality is comprehensively investigated and contrasted. For these stud-
ies, optimized devices were fabricated via vapour transport deposition (VTD) and analysed
by various techniques, including X-ray diffraction (XRD), X-ray photoelectron spectroscopy
(XPS), scanning electron microscopy (SEM), impedance spectroscopy, SCAPS simulations, as
well as general current-voltage analyses under various operational conditions.
4.2 Results and discussion

The schematic in Figure 4.1a displays the device architecture used in this study consisting
of FTO/(HRMO)/CdS/Sb2 Se3 /Au. For details on device preparation and deposition condi-
tions of the various layers used in this chapter see the Experimental procedures (chapter 3).
Briefly, either a ZnO or TiO2 layer was deposited via sol-gel method on glass/FTO substrates
from precursor solutions with concentration varying from 0.1 to 0.7 M, as discussed in detail
in the Experimental procedures (chapter 3). Figure 4.2(a to e) show the characterisation of
ZnO and TiO2 layers. Subsequently, devices with these layers, as well as control substrates
(glass/FTO), were used for deposition of a CdS buffer via chemical bath deposition (CBD, Fig-
ure 4.1b).Figure 4.3(a to e) show the characterisation of treated CdS layers. Next, the Sb2 Se3
absorber layer was fabricated via an optimized VTD method (Experimental Procedures, chap-
ter 3). Figure 4.4(a to e) show the characterisation of Sb2 Se3 champion solar cells with a ZnO,
TiO2 or no HRMO layer. The thin ZnO and TiO2 layers are not visible in the SEM. Finally,
gold back contacts were deposited by thermal evaporation to complete the devices.
a) b) HRMO + buffer deposition

Au contact ZnO or TiO2
sol-gel
Sb2Se3
Buffer +
HRMO
FTO
Glass Spin coating Chemical bath
ZnO or TiO2 CdS buffer
Figure 4.1: a) schematic of the device architecture used in this study; b) schematic of HRMO and CdS
buffer deposition using sol-gel and chemical bath deposition methods.
120
Figure 4.2: Scanning electron microscope (SEM) images of FTO coated with a) TiO2 and b) ZnO layers
obtained via sol-gel method from precursor solutions with concentration varying from 0.1 to 0.7 M.
X-ray photoelectron spectra (XPS) for 0.3 M c) TiO2 and d) ZnO. Thickness calibration for e) TiO2 and
f) ZnO films measured determined via profilometry and checked by ellipsometry when necessary.
121
Figure 4.3: Top view scanning electron microscope (SEM) images of CdS deposited on a) FTO, b)
FTO/ZnO and c) FTO/TiO2 ; d) X-ray diffraction patterns from ≈60 nm CdS layers deposited directly
on FTO (green), on ZnO layer (red) and on ceTiO2 layer (black) showing that the CdS films have
hexagonal wurtzite crystal structure and slightly higher (100) preferential orientation on FTO when
compared to the CdS films deposited on ZnO or TiO2 . In e) the X-ray photoelectron spectra from CdS
layers on FTO (green), ZnO (red) and TiO2 (black) show the typical signatures of Cd 3d (420 - 395 eV)
and S 2s (226 eV) and 2p (162 eV).
4.2.1 Solar cell efficiency

As shown in Figure 4.5a, the effect of ZnO and TiO2 HRMO layers on device efficiency was
assessed. It is clear that the use of either, ZnO or TiO2 , HRMO layer contribute to improving
the PCE of Sb2 Se3 solar cells reported here. The highest PCE was 4.39 %, achieved using a ZnO
HRMO from a 0.5 M precursor solution. An improvement of 19.9 % compared to the best CdS-
only control devices (CdS deposited directly on FTO) that shows efficiency of only 3.66 %. For
the TiO2 HRMO layer, the best efficiency was 3.89 %, obtained when using a 0.1 M precursor
solution. This corresponds to a PCE gain of 12.1 % compared to the control. Closer inspection
revels that PEC improvement took place mostly through an enhanced short circuit current,
Figure 4.5b, with a JSC gain of 17.4 % and 9.6 % for the champion devices containing a ZnO
and TiO2 layer respectively (CdS-only (no HRMO) = 16.6 mAcm−2 , TiO2 = 18.2 mAcm−2 , ZnO
122
= 19.5 mAcm−2 ). Equally important to overall PCE improvements is the significant reduction
in the variability of the solar cell device parameters observed when employing an HRMO
interlayer, including a vastly improved and less variable FF, Figure 4.5c. Optimised HRMO
layers also had no detrimental effect on the Voc , Figure 4.5d.
Figure 4.4: Top view SEM images of Sb2 Se3 deposited on the respective CdS substrate, a) no high
resistance metal oxide (HRMO), b) ZnO HRMO and c) TiO2 HRMO and cross sections of complete
solar cell with gold electrodes, d) no, f) ZnO and g) TiO2 layers deposited before CdS bath deposition.
The average thickness of the Sb2 Se3 layers is ≈550 nm. Importantly, the Sb2 Se3 grains present similar
morphology on all substrates, which allows us to rule out any major morphological effects on the
studies done in this work.
Notably, cell parameters were largely independent of the concentration of ZnO precursor
solution while a noticeable drop in PCE is observed with increasing concentration of TiO2
precursor solution used to deposit the HRMO layer. This is a first indication of the superior
performance of ZnO HRMO layer in CdS/Sb2 Se3 solar cells. Moreover, I notice that the ZnO is
partially etched during CdS CBD, what can contribute to a ZnO layer with reduced thickness
in the cell, and might explain the lack of variation in the cell parameters.
Impedance spectroscopy was used to investigate the electrical properties and differences
in the capacitive and resistive behaviour of thin film materials within devices. Figure 4.6 shows
Nyquist plots taken from champion devices with either a ZnO or TiO2 HRMO layer compared
123
a) 4.5 b)
ZnO/CdS TiO2/CdS CdS
0.1M 0.3M
20
4 0.5M
0.7M 0.3M
0.5M
JSC (mA cm-2)

0.1M
0.1M 0.7M
PCE (%)
3.5
0.3M
18
0.1M
0.3M
0.5M 0.7M
3 16 0.5M
ZnO/CdS TiO2/CdS CdS 0.7M
2.5
0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7 - 0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7 -
Concentration of sol-gel solution (mol L-1) Concentration of sol-gel solution (mol L-1)
c) 60 d)
ZnO/CdS TiO2/CdS CdS 0.4 ZnO/CdS TiO2/CdS CdS
0.1M
0.3M
VOC (V)
0.5M 0.7M
FF (%)
0.1M 0.3M
0.1M
55 0.1M
0.3M
0.3M
0.5M 0.7M
0.7M
0.5M
0.5M
0.7M
0.35
50
0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7 - 0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7 -
Concentration of sol-gel solution (mol L-1) Concentration of sol-gel solution (mol L-1)
Figure 4.5: Device parameters a)PEC, b) JSC , c) FF and d VOC for Sb2 Se3 solar cells with either ZnO or
TiO2 HRMO layers deposited from 0.1, 0.3, 0.5 and 0.7 M precursor solutions compared with CdS-only
control devices.
to CdS-only control devices. Devices with an HRMO layer deposition showed parallel re-
sistances of ≈104 Ω cm2 (TiO2 = 0.54 × 104 Ω cm2 , ZnO = 2.6 × 104 Ω cm2 ), which is over two
orders of magnitude higher than the control device (CdS-only (no HRMO)= 3.1 × 102 Ω cm2 ).
This confirms that the use of ZnO or TiO2 drastically improves the shunt resistance of the
Sb2 Se3 solar cells. Notably, the extracted series resistance remained unaffected by these HRMO
layers with TiO2 = 1.9 Ω cm2 , ZnO = 2.0 Ω cm2 , and CdS-only (no HRMO)= 2.2 Ω cm2 . This
improved shunt resistance leads to the significant JSC improvement in the cells containing a
HRMO layer.
124
a) b)
2 2
no HRMO
-Z' (×105 Ω) ZnO
-Z' (×103 Ω)
1 1
no HRMO
TiO2
0 0
0 1 2 3 0 1 2 3
Z" (×105 Ω) Z" (×103 Ω)
Sb2Se3 I(120)/I(221) (×10-2)
Figure 4.6: Nyquist plots for a) champions devices with different HRMO interlayers compared to
control and b) detail in the control device plot.
Superstrate Sb2 Se3 devices clearly receive benefits from the application of a metal oxide
interlayer between the FTO electrode and CdS. Depositing these layers also has implications
on Sb2 Se3 film texture. A schematic showing the crystal structure of Sb2 Se3 viewed along the
c-axis ([001] direction) was shown in the Literature Review (chapter 2, Figure 2.2a). The crystal
structure of Sb2 Se3 is comprised of covalently bonded 1D (Sb4 Se6 )n ribbons running along the
c-axis of the crystal that are weakly held together by Van der Waals forces. As highlighted
in this figure charge transport is not favoured in directions orthogonal to the c-axis, i.e., [hk0]
directions such as the [120], due to inefficient carrier transport between adjacent ribbons. It
is well established that an inverse relationship exists between the degree of (120) orientation
and cell efficiency.[11–13] In light of its 1D crystal structure, the intensity ratio between the
(120) and the (221) peaks in the XRD, I(120) /I(221) , provide a means to assess the expected
current losses within a thin film device. Although it may be somewhat unexpected that an
interlayer induces orientation changes in the non-adjacent Sb2 Se3 layer, this phenomenon has
been previously reported with the use of a SnO2 interlayer in FTO/SnO2 /CdS/Sb2 Se3 thin
films,[14] and can be explained by stress release on the CdS that is deposited on an HRMO
due to HRMO smoother surface or better lattice match. In Figure 4.7a, the I(120) /I(221) ratio
for my devices was calculated and plotted for comparison, a reduction in the (120) orientation
of the Sb2 Se3 layer is observed in both, ZnO and TiO2 , cases, with a near 50 % reduction
in unfavourable (120) orientation occurring when using ZnO HRMO interlayers. Moreover,
while for ZnO HRMO layers no significant variation in peak intensity is observed for varying
125
concentrations of precursor solution, the relative intensity of (120) decreases for lower TiO2
sol-gel solution concentration. Figure 4.7b highlights this decrease in (120) peak intensity in
the XRD patterns. This reduction in (120) orientation undoubtedly contributes to improving
JSC . However, other more dominant factors such as parasitic resistances and the presence of
electron barriers are of more paramount importance and are discussed next.
a) b) ZnO
HRMO
TiO2
HRMO
20
Ratio I(120) / I(221) from Sb2Se3 XRD (×10-2)
(221)
Sol-gel
(120)
(120)
concentration: Decreasing FTO/CdS/Sb2Se3/Au
(120)
15 (120) orientation .. *
0.7 M
Intensity (a.u)
0.5 M
0.3 M
0.1M 0.1M
FTO/TiO2/CdS/Sb2Se3/Au
..
0.1 M
*
0.7 M
10
0.3 M
0.1 M
0.3M 0.3M
0.5 M
0.5M 0.5M
. CdS FTO/ZnO/CdS/Sb2Se3/Au
..
*
5 * Au
0.7M 0.7M
0 16 17 18 16 17 18 Sb2Se3 (hkl)
CdS TiO2/CdS ZnO/CdS 2θ (°)
Front contact Detail on (120) peak
10 20 30 40 50 60
2θ (°)
Figure 4.7: a) diagram of Sb2 Se3 crystal structure in the Pbnm space group viewed along the c-axis
([001] direction), highlighting the unfavourable vertical charge transport between (Sb4 Se6 )n ribbons
within a device and b) relationship between HRMO layer type and measured relative (120) crystal
orientation for layers deposited from 0.1, 0.3, 0.5 and 0.7 M precursor solutions compared to control.
4.2.2 Device stability and current leak (shunt) models

First, parasitic current losses in devices with different interlayer architectures were eval-
uated over time. For this, the shunt resistance and efficiency of Sb2 Se3 champion devices
were measured over time for devices stored in a dry cabinet (<4 % relative humidity). In
Figure 4.8a, the dark JV curves taken over time from my best performing solar cells show that
a shoulder develops in the reverse bias region of CdS-only devices, which occurs due to the
reduction in shunt resistance upon ageing. No such changes are observed in the reverse bias
region of dark JV curves with the use of ZnO or TiO2 . Figure 4.8b shows the fitted Rsh values
for these devices. The Rsh of newly prepared devices was significantly more irregular in the
case of CdS-only devices. In addition to their initial variability, the average Rsh of all CdS-only
devices decreased significantly (more than two orders of magnitude over 100 days of aging)
stabilizing at ≈10 × 102 Ω cm2 . In contrast, the average Rsh remained remarkably stable over
126
time with the use of ZnO and decreased only slightly when TiO2 was used. Furthermore,
as shown in Figure 4.8c, control devices dropped to ≈80 % of their initial efficiency after 135
days while devices with HRMO interlayers remained highly stable, retaining more than 95 %
of their initial PCE over the same period.
a) b)
103 105
Current density (mA cm-2)
Relative eff. Rsh (Ω cm2)

no HRMO ZnO TiO2
101 day 137 103
day
day 137 101 no HRMO ZnO TiO2
10 -1 day
106 137 day 1 c)
day 1
day 106
10 -3 day 1 106 0.9
0.8
day 1
0.7 no HRMO ZnO TiO2
10-5
-0.5 0 0.5 -0.5 0 0.5 -0.5 0 0.5 1 106 137 1 106 137 1 106 137
Voltage (V) Days in dry cabinet
Figure 4.8: a) dark JV curves taken over time from the champion solar cells shown with different
HRMO layers; b) shunt resistance (Rsh ) values over time in champion devices with and without HRMO
layer deposition highlighting the large variability in devices without an appropriate interlayer; c) rela-
tive cell efficiencies over time during ageing.
To gain deeper insight into the effect of ZnO and TiO2 HRMO layers on device dynamics
their voltage and time dependent behaviour was investigated. Solar cells were cycled from
–0.6 to +0.8 V under AM1.5 light with decreasing illumination intensities from 1 to 0 suns.
Figure 4.9 shows the dJ/dV curves derived from these measurements. The results show that
CdS-only devices were stable for only the first few cycles (1, 0.9 and 0.8 suns). Additional
cycling caused the conductance to increase significantly. A final dark dJ/dV, taken after
the illumination cycles, revealed that the shunt conductance increased more than 2 orders of
magnitude compared to the initial dark current measurement. Further measurements showed
that this increase in shunt conductance is irreversible indicating that chemical changes have
occurred at the interface. Additional tests were also conducted to decouple the effect of light
and bias. It was found that a significant increase in conductance occurred when the CdS-only
cells were subjected to higher voltage bias (>0.5 V).
127
1
a) 10 no b)
main grain
ZnO TiO2 junction boundaries
HRMO
dJ/dV(×10-4 S cm-2)
8
Au
6 1st 1st Sb2Se3
sweep sweep
Suns
e- e-
h+
4 1st sweep
e-
2 CdS
before light FTO
sweep
0 glass
-0 .5 0 -0 .5 0 -0 .5 0 0 ohmic non-ohmic
Voltage (V) shunt shunt
10-1 no
c) J J0 d) Current density (mA cm2)
ZnO TiO2 e) 4
0 HRMO
D1 D1 10-3
Rnh 3
dJ/dV (µS cm-2)

Rsh Rsh 10 -5 ohmic +
2
RS RS non-ohmic
10-7 1 no HRMO
− V + − V +
TiO2
ohmic ohmic +
non-ohmic 10-9 0 ZnO
mostly ohmic
-0.6 0 0.6 -0.6 0 0.6 -0.6 0 0.6
Equivalent circut models Voltage (V) -1 -0.5 0
Data Fit D1 Rsh Rnh Voltage (V)
Figure 4.9: a) reverse bias region of dJ/dV curves highlighting the effective use of ZnO or TiO2 to
prevent the increased current leak observed during voltage sweeps of CdS-only control devices; b)
schematic of Ohmic and non-Ohmic shunting mechanisms in Sb2 Se3 solar cells; c) equivalent circuit
models; d) fitted dark JV curves for the various solar cells used in this study. e) Ohmic and non-Ohmic
current leaks in freshly prepared solar cells, highlighting the removal of non-Ohmic shunt pathways
when using ZnO or TiO2 interlayers.
Furthermore, it is evident that both light and voltage are required to expedite the break-
down of the junction, with the most rapid degradation occurring at higher biases, Figure 4.10.
The permanent nature of the degradation proves that CdS-only devices are susceptible to
light/bias induced chemical (physical) changes under operational conditions. For cells where
a ZnO or TiO2 HRMO layer was deposited no significant change was noted in the dark shunt
conductance before and after light cycling. The behaviour of these devices was far more ideal,
simply showing a steady decrease in conductance with decreasing incident light power. Fur-
thermore, no significant change in the shunt conductance was observed in devices with a ZnO
or TiO2 HRMO, even for broad voltages sweeps done from -1.1 to 1.1 V. Collectively, these re-
sults highlight the precarious nature of the FTO/CdS interface in such devices and the need
to mitigate the pitfalls of this junction though the application of a metal oxide interlayer.
128
Figure 4.10: Assessment of dark shunt conductance of fresh Sb2 Se3 solar cells before any direct sun
light exposure submitted to sweeps in increasing voltage range: -0.5 to +0.5 V, -0.8 to +0.8 V and -1.1
to +1.1 V compared to dark measurements obtained after sweep in the same ranges but under 1 sun of
1.5G simulated sun light. In a) solar cells with no high resistance metal oxide (HRMO) layer, b) TiO2
HRMO and c) ZnO HRMO applied before CdS deposition.
129
In the case of the Sb2 Se3 devices (and thin film solar cell devices in general), shunt leaks
may be divided into two categories. Specifically, an Ohmic component, which remains con-
stant with voltage variation, and a non-Ohmic component which varies with voltage.[15–17]
A physical representation of device phenomena that give rise to both these components is il-
lustrated in Figure 4.9b. Ohmic shunts occur due to low shunt resistance pathways in the CdS
buffer or absorber layer (highly conductive grain boundaries, regions of thinner CdS, ion mi-
gration at the main junction).[18] Conversely, non-Ohmic shunts predominantly arise within
regions of space charge limited current (SCLC) where the main junction field is weak.[15]
Such SCLC regions may form where the rough FTO grains disrupt the thin CdS film and
permit the Sb2 Se3 to contact the FTO yielding a localized FTO/Sb2 Se3 /Au junction. Likewise,
in cases where significant migration of Cd2+ or S2 – occurs the Sb2 Se3 layer will consume the
CdS leading to a localized degradation of the p-n junction and the generation of non-Ohmic
shunts.[19, 20] As such, determining the type of shunt leak provides direct insights into local
defects at the interfaces.
To determine the type of shunt leaks in superstrate Sb2 Se3 devices, dark JV curves may
be fitted to the equivalent circuit models shown in Figure 4.9c. The total current density, J, in
the circuits is given by the sum of two current density terms, Jmain junction and JOhmic shunt . An
extra term, Jnon−omic shunt , is added to model the current density to account for the presence
of SCLC regions.[21] The total current density, J = Jmain junction + JOhmic shunt + Jnon−omic shunt ,
expressed as a function of voltage is given by Equation 4.1.[17]
q(V + JRS )
V + JRS

J (V ) = J 0 e nK B T
−1 + + k (V − JRS )m (4.1)
Rsh
where J0 is the reverse saturation current, q is the elementary charge, RS is the series
resistance, n is the diode factor, K B is the Boltzmann constant, T is the temperature and Rsh is
the shunt resistance. In the non-Ohmic shunt term, k, is associated to the trap distribution, the
conductivity, and the length of the transport path, while m is related to the absence (m ≈ 2)
or presence (m > 2) of a high density of deep traps in the SCLC region.
130
Table 4.1: Representative fitted diode parameters from dark JV curves with and without HRMO depo-
sition.
HRMO Shunt model Rsh k m RS n J0

(Ω cm−2 ) (×10−6 ) (Ω cm−2 ) (×10−6 mA cm−2 )
none Ohmic + 35749 39 3.45 2.29 1.74 2.65
non-Ohmic
ZnO Ohmic 32623 - - 2.21 1.48 1.67
TiO2 Ohmic + 68684 1.0 3.36 30.1 1.59 0.599
non-Ohmic
Figure 4.9d shows typical dark JV curves obtained from pristine devices (that had not
been aged or exposed to repeated light/voltage cycles). These curves were fitted to the equiv-
alent circuit models and the associated parameters are presented in Table 4.1. Importantly, the
dark JV curve for the cell, where a ZnO HRMO layer was deposited, fits an equivalent circuit
where only the Ohmic parallel resistance element, Rsh , is needed. A secondary parallel resis-
tance element, Rnh , must be added to the circuit model to correctly account for the formation
of SCLC regions and the associated non-Ohmic current in both CdS-only and TiO2 /CdS de-
vices. The presence of non-Ohmic shunt pathways present in these devices is best exemplified
by plotting dJ/dV for devices. Figure 4.9e shows typical dJ/dV curves for the different solar
cell architectures evaluated here, which clearly show the presence of non-Ohmic shunts for
CdS-only devices and their removal when employing HRMO interlayers.
In the case of CdS-only devices non-Ohmic shunts were constant across countless CdS
depositions and irrespective of the CdS thickness employed in the devices, further highlight-
ing that these pathways are inherent to the CdS layer itself and not due to the quality of the
deposition, Figure 4.11. For TiO2 /CdS devices the fitted non-Ohmic component is small, re-
ducing the value of k nearly 40 times compared to the value obtained for CdS-only devices.
This indicates a significant reduction (although not complete removal) of non-Ohmic current
leakage. Although this may be acceptable, TiO2 layers also introduce a large series resistance
in devices as well as yielding a m > 2, which indicates the presence of a high density of deep
traps that are present in the TiO2 layer (vide infra).
131
Figure 4.11: Comparison of the reverse bias region of dJ/dV curve for Sb2 Se3 devices with a 60 nm
CdS buffer layer (green dashed), 100 nm CdS buffer (green full line) and 60 nm CdS buffer deposited
on the ZnO high resistance metal oxide (HRMO) (red) highlighting that the preliminary deposition of
ZnO HRMO layer prevents the non-Ohmic shunt leakage.
4.2.3 Electron barriers

Employing an HRMO in Sb2 Se3 devices is paramount to avoid issues inherent to the
CdS/Sb2 Se3 architecture. When selecting an optimal HRMO interlayer the quality of newly
generated interfaces must also be carefully considered. Defective interfaces and non-ideal
band alignment at the front contact can result in the formation of energy barriers for electron
transfer and increased recombination.[22, 23] Charge transport in solar cells is highly influ-
enced by these barriers, and can lead to anomalies in the JV characteristics of a solar cell, such
as cross over between dark and illuminated curves as well as kinks and roll-over in the dark
or light JV curves, ultimately reducing device lifetime and efficiency.[5, 6]
132
a) 75 b) 7
Current density (mA cm-2) TiO2
6
50
5
RS (Ω cm2)
25 4
ZnO 3 ZnO
0 TiO2
no HRMO no HRMO 2
-25 no HRMO
1
-0.5 0 0.5 -0.5 0 0.5 0 0.5 1
Voltage (V) Suns
Figure 4.12: a) dark and light JV curves for CdS-only solar cells compared to devices where a TiO2
or ZnO HRMO layer was used. b) series resistance as a function of increasing light power for typical
solar cell devices.
Figure 4.12a shows the light and dark JV curves for a typical solar cell with a TiO2
HRMO interlayer. Clear cross over behaviour is observed is these devices as the diode current
in the dark JV curve is significantly impeded. This reveals that, in the dark, a high barrier
for the diode current is introduced at the front contact by the TiO2 . Moreover, the data from
Figure 4.12b show that the series resistance of these solar cells drops rapidly with increasing
light intensity. This light dependent barrier is responsible for the cross over anomaly in devices
where TiO2 is deposited. In contrast, both ZnO and CdS-only devices show no cross over
behaviour. This implies a satisfactory combination of material properties at the front contact
and indicates suitable band energy alignment across the junctions. Likewise, similar to the
CdS-only control devices, the series resistance of devices with a ZnO HRMO interlayer does
not change significantly with light intensity. Therefore, unlike TiO2 , ZnO may be used to
preserve junction quality, prevent device degradation, and improve device efficiency without
introducing a defective interface through a light dependent electron barrier.
133
a) 40 b) 100 800 nm
dark
Current density (mA cm-2) decreasing 750 nm
diode current, 80 1.55eV 700 nm
increasing RS 1.65eV 650 nm
dV/dJ (Ω cm2)
20 60 1.77eV 600 nm
1.91eV
550 nm
2.07eV
dark 40 500 nm
2.25eV
475 nm
0 20 2.48eV 450 nm
425 nm
400 nm
0
0.2 0.4 0.6 0.8 1 0.4 0.6 0.8 1
Voltage (V) Voltage (V)
e-
1 1
c)
e-
d) diode current diode current
dark

80 dark 80
Eg 0 light 0 light
e-
Sb2Se3 Sb2Se3
-1 40 -1 40
Ec (eV)
Ec (eV)
h+
cross
capture SCR over SCR
CdS -2 -2
e-
CdS 0 0
h+
hν ≥ Eg dark CdS dark
HRMO -3 HRMO -3 HRMO
charged neutral level FTO -40 light FTO -40 light
acceptor level (recombined) -4 -4
500 600 0 0.5 500 600 0 0.5
in the dark under blue light
x (nm) Voltage (V) x (nm) Voltage (V)
High density of acceptor Low density of acceptor
CdS/HRMO interface
states at the interface states at the interface
Figure 4.13: a) JV curves for devices with an TiO2 HRMO as a function illumination wavelength of
between 400–800 nm. b) corresponding dV/dJ curves highlighting current blocking at high forward
bias; c) schematic mechanism showing how negatively charged acceptor levels at a defective interface
may be neutralized via capture of photogenerated holes. d) SCAPS simulation of solar cells containing
a 10 nm HRMO window layer with a variable density of acceptor (trap) states. Specifically, a high
density at the interface (1.75 × 1012 cm−2 , vis. TiO2 /CdS) resulting in the observation of cross over
behaviour of the JV curves and a relatively low density of traps for (0.85 × 1012 cm−2 , vis. ZnO/CdS)
with no cross over observed in the simulated voltage range.
To understand why ZnO (and not TiO2 ) acts as an effective interlayer the origin of the
electron barrier in cells with a TiO2 HRMO layer was explored in greater detail. Figure 4.13a
shows JV curves taken under quasi monochromatic light illumination. As the wavelength
of light is decreased from 800 to 400 nm a concomitant decrease in diode current blocking
is observed. At ≈500 nm (2.48 eV), which is just above the bulk bandgap of CdS (2.42 eV), a
sudden change in dynamics occurs. This can be better visualized on the series resistance curve,
dV/dJ vs V (Figure 4.13b) where we see a shoulder at ≈0.5 V that is virtually eliminated when
134
the device is illuminated with photons having energy of 2.48 eV or higher. This indicates that
the barrier height is significantly lowered as soon as light starts being absorbed by the CdS
layer. Barrier height modulation by light absorbed in the CdS layer is commonly observed in,
CuInSe2 , Cu(In, Ga)Se2 , and CdTe, and has been attributed to a high positive potential step
(barrier) for the diode current across buffer/window layers. [2, 4, 24–28] In Figure 4.14(a to
c) It was verified that the band alignment between ZnO, TiO2 , CdS, and Sb2 Se3 films was
satisfactory, indicating that the barrier is not caused by an intrinsic misalignment between
HRMO and CdS conduction bands.[10, 29] Rather, deep acceptor levels (traps) being present
at the buffer/window interface generate the barrier. Notably, cross over is significantly less
pronounced when ZnO HRMO is used.
a)
Normalized intensity (a.u.)
ZnO TiO2 CdS Sb2Se3

(3.26 eV) (3.11 eV) (2.21 eV) (0.74 eV)
c) 2
1 CBO 0.46
0.23
0.03 0.13
E - EFERMI (eV)
5 4 3 2 5 4 3 2 4 3 2 1 3 2 1 0 0 EFERMI
Binding energy (eV) -1
-0.74
b) -2
Normalized (αhν)1/γ
ZnO TiO2 CdS Sb2Se3

-3 VBO -2.21
(3.29 eV) (3.23 eV) (2.44 eV) (1.20 eV)
-3.26 -3.11
-4
ZnO TiO2 CdS Sb2Se3
3.3 3.4 3.5 2 2.5 3 3.5 2.4 2.5 2.6 1.1 1.2 1.3 1.4
hν (eV)
Figure 4.14: a) Valence band X-ray photoelectron spectra and b) band gap determination via UV-Visible
spectroscopy for ZnO, TiO2 , CdS and Sb2 Se3 films (γ = 1/2 for all materials except TiO2 where γ = 2).
c) conduction (CBO) and valence band offset (VBO) diagram highlighting the band alignment between
the materials after Fermi level alignment.
Figure 4.15 shows the JV curves obtained under low power, ≈2 mWcm2 , illumination
with red, green and blue LEDs for devices with TiO2 , no and ZnO HRMO layer. The cross
over is significantly reduced as the device is illuminated with lower wavelength light. This
indicates a reduction in the electron barrier caused by light via a mechanism that is discussed
next.
135
Figure 4.15: Current density vs voltage (JV) curves obtained from Sb2 Se3 solar cells made with the
deposition of a TiO2 high resistance metal oxide (HRMO) layer compared to devices where no HRMO
and ZnO HRMO layer was applied before CdS deposition. The curves were measured in the dark,
under illumination with red, green or blue LEDs and under 1 sun of AM1.5G simulated solar light. A
power output of ≈2 mWcm2 was used for all measures with the red (625 nm), green (530 nm) and blue
(455 nm) LEDs.
The mechanism of barrier formation is detailed in Figure 4.13c.[1] In the dark, electrons
from an adjacent defective layer (TiO2 ) fill the acceptor states at the interface and generate an
area of negative charge. Consequently, a positive space charge region is induced in the bulk
of these layers for charge balance. If the surface density of negatively charged acceptor levels
at this interface is comparable to the charge density in the buffer or window, the conduction
band at the interface is perturbed toward the vacuum level leading to a sharp drop in potential
over the HRMO layer and the creation of a high energy barrier for the diode current. Based
on penetration depth as the wavelength of light decreases an increasing density of charge is
excited closer to the Sb2 Se3 front contact within the space charge region. In the case of a
TiO2 HRMO, negatively charged acceptor levels at the CdS/TiO2 interface may more readily
capture photogenerated holes in the adjacent Sb2 Se3 buffer and neutralize the defect due to the
increased cross-section for charge transfer. This process is facile under blue light (wavelength
< 500 nm) given the generation of photogenerated holes in the directly adjacent CdS buffer.
Consistent with the experimental data, neutral defects exert no effect on the band diagram
and the barrier is virtually eliminated.[2]
136
To link the trap density to band perturbations and model the observed experimental be-
haviour, interfaces with varying defect densities were simulated using Solar Cell Capacitance
Simulator (SCAPS) to model the types of solar cell behavior observed in the devices.[30] A
Sb2 Se3 solar cell model was created using the parameters shown in the Table 4.2,[31–33] and an
interface containing negatively charged acceptor states was introduced. Figure 4.13d show the
simulated energy bands diagrams for Sb2 Se3 solar cells with relatively high (1.75 × 1012 cm−2 )
and low (0.85 × 1012 cm−2 ) density of acceptor levels at the buffer/window interface, to simu-
late the behaviour of TiO2 and ZnO HRMO solar cells respectively. The resulting JV curves in
the dark and under illumination are also shown in each case. In the case of a high density of
acceptor states a steep barrier for the diode current in the dark is observed. Under illumina-
tion the barrier is reduced, and the dark/light JV curves cross over at ≈0.4 V. In contrast, for
a low density of acceptor states no significant change is observed in the simulated band dia-
grams both in the dark and under illumination and subsequently no cross over is observed in
the simulated voltage range. The JV curves simulated here are consistent with experimental
results observed for each type of HRMO interlayer.
a) TiO2 ZnO b) 4
1
N(x) (×1017 cm-3)
2 ZnO
C-2 (×1017 F-2)
TiO2
0
0.5 2 MHz -2
-4
-6
50 kHz
0 0 100 200 300 400 500
0 0.5 0 0.5
x (nm)
Voltage (V)
Figure 4.16: Capacitance-voltage measurements for Sb2 Se3 solar cells with different HRMO interlayers;
a) C−2 vs V curves at 50, 100, 150 and 200 kHz and b) correspondent doping profile derived at 100
kHz.
The density of acceptor states as well as their nature (shallow or deep) may be investi-
gated using capacitance/voltage measurements obtained over a range of frequencies. Interface
states introduce a frequency dependency on the capacitance.[34] At sufficiently high frequen-
cies only near free carriers can follow the voltage change. Conversely, at lower frequencies,
137
free carriers as well as trapped carriers in deep level trap states can affect the capacitance
characteristics of a device. When interface acceptor states dominate, the curve 1/C2 vs V and
apparent built-in voltage will be frequency dependent. Figure 4.16a shows the plots 1/C2 vs
V taken at 50, 100, 150 and 200 kHz for devices with different HRMOs. Here, we can clearly
see a frequency dependence of 1/C2 vs V curves in devices where a TiO2 layer was used.
In contrast, no significant changes with frequency are observed with ZnO. This confirms
that the high density of deep level states is introduced by the TiO2 while the interfaces gen-
erated by ZnO remain largely trap free. Overall, a high density of interface states increases
recombination, which contributes to the reduced JSC observed in solar cells using the TiO2
layer. Likewise, the relatively high JSC observed for ZnO solar cells indicates that the deposi-
tion of this layer assists in the passivation of defects at buffer/window interface.
Density of states profiles were derived from the C vs V measurement at 100 kHz, using
Equation 4.2[35]
# −1
d C −2
"
2
N=− (4.2)
qε r A2 dV
where A is the electrode area, C is capacitance and ε r the relative dielectric constant. The
width of the depletion region, x, is obtained from the relation x = (qε r A)/C and used to plot
N ( x ). A density of states profile obtained by C vs V corresponds to bulk and interfacial defects
(traps).[11] The distortions in N ( x ) observed in Figure 4.16b when using the TiO2 layer are
attributed to the interfacial defects and electron barrier introduced by TiO2 . These distortions
are not seen on the profiles obtained for devices employing a ZnO HRMO interlayer.
138
Table 4.2: SCAPS parameters for Sb2 Se3 solar cell materials and CdS/HRMO interface.[31–33]
Sb2Se3 CdS HRMO* SnO2
Thickness [nm] 500 60 10 350
Eg [eV] 1.1 2.4 3.2 3.6
χ [eV] 4.15 4 4.15 4.15
εr 19 10 9 9
NC [cm-3] 1018 2.2×1018 2.2×1020 2.2×1018
NV [cm-3] 1020 1.8×1019 1.8×1019 1.8×1019
vn [cm s-1] 107 107 107 107
vp [cm s-1] 107 107 107 107
µe [cm2V-1s-1] 10 100 100 100
µh [cm2V-1s-1] 1 25 25 25
Doping density
[cm-3] 1014 1.1×1018 5×1017 1017
Defects
Sb2Se3 CdS CdS/HRMO
Defect type Neutral Acceptor Acceptor
Et [eV] 0.55 1.2 1.2
Reference Above EV Above EV Above highest EV
σe [cm2] 1×10-13 1×10-17 1×10-17
σh [cm2] 4×10-14 1×10-12 1×10-12
Nt 1015 cm-3 1018 cm-3 1.75×1012 or 0.85×1012 cm-2
*Parameters for the high resistance metal oxide (HRMO) layer correspond to TiO 2.
Due to parameter similarity, analogous results were obtained by replacing TiO2 by
ZnO in the simulation.
139
4.3 Conclusion
The response of Sb2 Se3 solar cells upon application of key HRMO ZnO or TiO2 layers
was investigated. The deposition of these layers was found to enhance cell efficiency, improve
device stability, decrease parameter variability, improve Sb2 Se3 grain texture, and significantly
reduce current leaks largely through the suppression of non-Ohmic shunt mechanisms. Al-
though the application of either ZnO or TiO2 improved overall Sb2 Se3 solar cell performance,
TiO2 deposition creates a high density of acceptor states at the buffer/window interface that
leads to the formation of a high electron barrier for the diode current in the dark. As such,
the use of TiO2 should be avoided as an HRMO interlayer for superstrate CdS/Sb2 Se3 devices.
Depositing a ZnO layer proved to be optimal as devices received the benefits of reduced in-
terfacial trap densities, an improved device front contact, and overall increases to solar cell
performance and stability.
140
4.4 References
[1] C.-H. Chung, B. Bob, T.-B. Song, Y. Yang, Sol. Energy Mater. Sol. Cells 2014, 120, 642.
[2] I. L. Eisgruber, J. E. Granata, J. R. Sites, J. Hou, J. Kessler, Sol. Energy Mater. Sol. Cells
1998, 53, 367.
[3] X. Li, K. Shen, Q. Li, Y. Deng, P. Zhu, D. Wang, Sol. Energy 2018, 165, 27.
[4] A. O. Pudov, A. Kanevce, H. A. Al-Thani, J. R. Sites, F. S. Hasoon, J. Appl. Phys. 2005,
97, 064901.
[5] S. S. Hegedus, W. N. Shafarman, Prog. Photovoltaics 2004, 12, 155.
[7] M. Ishaq, H. Deng, S. Yuan, H. Zhang, J. Khan, U. Farooq, H. Song, J. Tang, Sol. RRL
2018, 2, 1800144.
[8] W. Wang, X. Wang, G. Chen, B. Chen, H. Cai, T. Chen, S. Chen, Z. Huang, C. Zhu, Y.
Zhang, Sol. RRL 2018, 2, 1800208.
[9] X. Wang, R. Tang, Y. Yin, H. Ju, S. Li, C. Zhu, T. Chen, Sol. Energy Mater. Sol. Cells 2019,
189, 5.
[10] B. Zhang, J. Zheng, X. Li, Y. Fang, L. W. Wang, Y. Lin, F. Pan, Chem. Commun. 2016, 52,
5706.
[14] J. Tao, X. Hu, Y. Guo, J. Hong, K. Li, J. Jiang, S. Chen, C. Jing, F. Yue, P. Yang, C. Zhang,
Z. Wu, J. Tang, J. Chu, Nano Energy 2019, 60, 802.
[15] S. Dongaonkar, J. D. Servaites, G. M. Ford, S. Loser, J. Moore, R. M. Gelfand, H.
Mohseni, H. W. Hillhouse, R. Agrawal, M. A. Ratner, T. J. Marks, M. S. Lundstrom,
M. A. Alam, J. Appl. Phys. 2010, 108, 124509.
[16] K. Shen, C. Ou, T. Huang, H. Zhu, J. Li, Z. Li, Y. Mai, Sol. Energy Mater. Sol. Cells 2018,
186, 58.
[18] K. Shen, Q. Li, D. Wang, R. Yang, Y. Deng, M.-J. Jeng, D. Wang, Sol. Energy Mater. Sol.
Cells 2016, 144, 472.
[19] S. Chen, M. Ishaq, W. Xiong, U. Ali Shah, U. Farooq, J. Luo, Z. Zheng, Z. Su, P. Fan,
X. Zhang, G. Liang, Sol. RRL 2021, 5, 2100419.
Appl. Phys. Lett. 2017, 111, 013901.
[21] A. Rose, Phys. Rev. 1955, 97, 1538.
141
[22] Y. Chen, Y. Wang, R. Wang, X. Hu, J. Tao, G.-E. Weng, C. Zhao, S. Chen, Z. Zhu, J. Chu,
H. Akiyama, ACS Appl. Energy Mater. 2020, 3, 10415.
[23] I. Gharibshahian, A. A. Orouji, S. Sharbati, Opt. Mater. 2021, 116, 111098.
[24] G. Agostinelli, D. L. Bätzner, M. Burgelman, Thin Solid Films 2003, 431-432, 407.
[25] G. Agostinelli, E. D. Dunlop, D. L. Batzner, A. N. Tiwari, P. Nollet, M. Burgelman, M.
Kontges in Proceedings of 3rd World Conference on Photovoltaic Energy Conversion,
Vol. 1, 2003, 356–359 Vol.1.
[26] T. J. McMahon, A. L. Fahrenbruch in Conference Record of the Twenty-Eighth IEEE
Photovoltaic Specialists Conference - 2000 (Cat. No.00CH37036), 2000, pp. 539–542.
[27] P. Nollet, M. Köntges, M. Burgelman, S. Degrave, R. Reineke-Koch, Thin Solid Films
2003, 431-432, 414.
[28] U. Rau, H. W. Schock, Appl. Phys. A 1999, 69, 131.
[29] Y.-L. Lee, Y.-S. Lo, Adv. Funct. Mater. 2009, 19, 604.
[30] M. Burgelman, K. Decock, S. Khelifi, A. Abass, Thin Solid Films 2013, 535, 296.
[32] B. Enright, D. Fitzmaurice, J. Phys. Chem. 1996, 100, 1027.
[33] K. Li, Y. Lu, X. Ke, S. Li, S. Lu, C. Wang, S. Wang, C. Chen, J. Tang, Sol. RRL 2020, 4,
2000220.
[34] J. P. Donnelly, A. G. Milnes, IEEE Trans. Electron Devices 1967, 14, 63.
[35] J. Heath, P. Zabierowski in Advanced Characterization Techniques for Thin Film Solar Cells,
(Eds.: D. Abou-Ras, T. Kirchartz, U. Rau), WILEY-VCH Verlag GmbH & Co. KGaA,
Weinheim, Germany, 2011, Chapter 4, pp. 81–105.
142
Chapter 5
Influence of substrate nanostructure on
the growth of evaporated Sb2Se3 layers
Some of the contents of this chapter have been accepted for publication as:
Mendes, J.O., Della Gaspera, E. and van Embden, J. (2023). Substrate morphology directs (001)
Sb2 Se3 thin film growth by crystallographic orientation filtering. Small.
https://doi.org/10.1002/smll.202302721
143
Abstract
Antimony chalcogenides, Sb2 X3 (X = S, Se or (Sx Se1 – x )), applications greatly benefit from
efficient charge transport along covalently bonded (001) oriented (Sb4 X6 )n ribbons, making
thin film orientation control highly desirable - although particularly hard to achieve exper-
imentally. Here, it is shown for the first time that substrate nanostructure plays a key role
in driving the growth of (001) oriented antimony chalcogenide thin films. Vapor Transport
Deposition of Sb2 Se3 thin films is conducted on ZnO substrates whose morphology is tuned
between highly nanostructured and flat. The extent of Sb2 Se3 (001) orientation is directly
correlated to the degree of substrate nanostructure. These data showcase nanostructure sub-
strate engineering as an effective tool to control the orientation and morphology of evaporated
Sb2 Se3 layers. The optimized samples demonstrate high (001) crystallographic orientation. A
growth mechanism for these films is proposed, wherein the substrate physically restricts the
development of undesirable (hk0) crystallographic orientations. Finally, it is shown that the
surface chemistry of the nanostructured substrates can be altered and still drive the growth of
(001) Sb2 Se3 thin films - not limiting this phenomenon to a particular substrate type. Insights
from this work are expected to guide the rational design of Sb2 X3 thin film devices and other
low-dimensional crystal-structured materials wherein performance is intrinsically linked to
morphology and orientation.
144
5.1 Introduction
Crystallographic orientation control (texturing) to achieve the optimal alignment of the
quasi-1D (Sb4 Se6 )n ribbons has been the focus of numerous research endeavours and has been
proven to be intrinsically linked to the performance of Sb2 Se3 devices.[1–5] As previously
discussed, poor charge transport is observed in directions orthogonal to the c-axis ([hk0] di-
rections) due to the weak atomic interactions between ribbons. Conversely, charge transport
is far more efficient along the covalently bonded c-axis of the ribbon structure. As such, thin
films with higher proportions of vertical ribbon orientation have improved charge transport
properties, with the most optimal transport (between electrodes) being observed for (001)
oriented thin films.[6]
Several studies have settled for suboptimal textured Sb2 Se3 films (usually with slightly
preferential orientation along the [221] or [211] crystallographic directions) as these tend to
present a compact morphology that is convenient for planar device applications.[7] A per-
fect (001) alignment is the most desirable to fully exploit facile charge transport along the
ribbons. Thin films with a substantial degree of (001) orientation are vital to achieve record
performances in Sb2 Se3 solar cells,[8, 9] photoelectrochemical cells for water splitting,[10]
photodetectors[11] and insertion battery electrodes.[12]
Although strategies to control the growth and orientation are key to designing efficient
Sb2 Se3 device, orientation control of Sb2 Se3 thin films is experimentally challenging. Some
of the current strategies to preferentially orient Sb2 Se3 films were discussed in chapter 2.
a) Existing strategies for oriented Sb Se b) This work:

2 3
new strategy for oriented Sb2Se3

(001)
seed layers deposition

parameters (hk1)
(hk0)
substrate nanostructuring
substrate
chemistry
Figure 5.1: a) Schematic of existing strategies to achieve vertical ribbon orientation in Sb2 Se3 thin films.
b) Proposed methodology in this work, wherein surface nanostructure is used to direct crystal growth
and orientation.
145
Such strategies usually involve extensive optimization of deposition parameters, which must
be performed for each different substrate type. The many factors influencing Sb2 Se3 film
orientation and morphology were also discussed in chapter 2. Here, it is worth reiterating
that the thermodynamic tendency of crystal faces to grow with low energy (hk0) planes, i.e.
(100), (010), (110), and (120),[13] parallel to the substrate significantly complicates orientation
control. Figure 5.1a summarizes the different categories of approaches currently employed to
orient Sb2 Se3 thin films, including the use of a seed layer, optimizing deposition parameters,
and tailoring substrate chemistry, as discussed in chapter 2.[14–22]
The morphology (nanostructure) of the substrate is another important parameter that
may influence Sb2 Se3 grain growth and orientation. However, to date, the effect of substrate
nanostructure on the morphology and orientation of evaporated Sb2 Se3 films has remained
completely unexplored. Controlled manipulation of substrate nanostructure (Figure 5.1c) con-
stitutes an alternative approach to tune Sb2 Se3 orientation. Such an avenue opens the possi-
bility of selecting the desired substrate material and then using the nanostructure to direct the
desired growth orientation and bulk morphology.
Here, the effect of substrate nanostructure on the growth of evaporated Sb2 Se3 films is
investigated for the first time. To accomplish this, Sb2 Se3 thin films were grown on substrates
possessing distinctly different physical surface structures. X-ray diffraction (XRD) and scan-
ning electron microscopy (SEM) reveal the crucial role played by substrate nanostructure on
final Sb2 Se3 film orientation and morphology. A novel strategy is presented to overcome fun-
damental challenges in the fabrication of Sb2 Se3 thin films with controlled morphology and
orientation.

5.2.1 Experimental design and understanding ribbon orientation
To study how substrate nanostructure affects film growth, I must first develop a clear un-
derstanding of crystal orientation in Sb2 Se3 thin films. As outlined previously, charge trans-
port is most efficient along the covalent length of the ribbon structure, and thus the physical
phenomenon to be controlled is the ribbon orientation itself. When investigating the orienta-
tion (texture) of Sb2 Se3 thin films, appropriate analysis requires that the orientation of a given
146
crystal plane is associated to its ribbon orientation.[23] To achieve this, the ribbon tilt angle (θr )
defined in chapter 2 is used. In summary, all (hk0) orientations are those with ribbons that are
parallel to the substrate and have θr = 0°. Randomly orientated (hkl) Sb2 Se3 thin films include
ribbons with a range of orientations with 0◦ < θr < 90◦ . Finally, (00l) orientations are those
with ribbons that extend vertically from the substrate, θr = 90◦ . Figure 5.2a shows the cal-
culated intensities of all reflections for Sb2 Se3 between 10-50 °2θ. For clarity, these have been
divided into the three groups defined above. Specifically, (hk0) in blue, (hkl) in orange/red,
and (00l) in brown.
For this work, I considered all peaks in the experimental XRD pattern between 10-50° 2θ
that do not overlap with substrate peaks and have signal-to-noise ratios that allow their fit-
ting. These comprised 25 reflections (11 (hk0), 13 (hkl), and the (002)), which are sufficient to
accurately assess the crystallographic orientation of Sb2 Se3 thin films and associated ribbons.
I note here that (by coincidence) after discounting the (hk0) and (002) planes, the remaining
(13) reflections of Sb2 Se3 are all (hk1), i.e., (101), (211), (221) etc. Figure 5.2b shows all crystal
planes relevant to Sb2 Se3 thin film orientation as a function of ribbon tilt angle, specifically
highlighting that (hk1) reflections span a range of ribbon tilt angles between 30-90°. Consid-
ering this, when specific reference is made to ‘(hk1) orientated’ thin films in this work it must
be remembered that these planes are spread across a large range of ribbon tilts, i.e., that (hk1)
orientated thin films are not necessarily synonymous with favorably orientated (high ribbon
tilt angle) films, as is sometimes implied in the literature.
Having examined the relationship between Sb2 Se3 crystal and ribbon orientation, the
experimental methodology is now discussed. Figure 5.2c illustrates the vapor transport de-
position (VTD) reactor used in this work. Physical evaporation is the most common method
employed for preparing Sb2 Se3 thin films for device applications.[20, 24] VTD comprises an
evaporation source with the desired amount of Sb2 Se3 powder (source) located in zone 1. The
substrate is placed in a graphite holder within zone 2 (set to a lower temperature). This con-
figuration allows precise control of film thickness simply by varying the amount of Sb2 Se3 in
the source (see Experimental for further details). Prior to these studies, the VTD reactor was
first optimized to achieve the controlled and reproducible growth of Sb2 Se3 thin films.[25]
147
a) (hk0) (hk1) (001)
θr = 0° b)
0° > θr > 90° θr = 90°
100
(211)
(221)
80
Intensity (%)
90
(120)
(002)
60
80 40
(101)
20
70
0
60 10 20 30 40 50
Ribbon tilt, θr (°)
Diffraction angle (°2θ)

50 c)
zone 1 zone 2
40 N2 pump
30
Sb2Se3 heating
source center holder element
20
10
substrates:
0
(002)
(501)
(431)
(341)
(421)
(411)
(141)
(231)
(311)
(301)
(221)
(211)
(021)
(101)
flat nanostructured
Figure 5.2: a) Theoretical XRD pattern of Sb2 Se3 (ICSD #16680) color coded by ribbon orientation group
(horizontal (hk0), blue; tilted (hk1), orange/red; vertical (001), brown). b) Relationship between Sb2 Se3
crystal planes analyzed in this paper and (Sb4 Se6 )n ribbon tilt angle. c) Schematic of the VTD apparatus
and the ‘flat’ and ‘nanostructured’ substrates employed in this work.
Well-defined substrates are required to effectively investigate how substrate nanostruc-
ture controls the crystal growth and orientation of Sb2 Se3 thin films. ZnO was selected as the
substrate and fabricated with two distinct morphologies: nanostructured films (ZnO-NW);
and compact flat films (ZnO-Flat). ZnO is an ideal substrate as both flat and nanostructured
thin films can be produced with the same orientation (same ZnO crystal planes at the surface)
using chemical bath deposition (CBD), enabling investigation into the impact of substrate
morphology alone on Sb2 Se3 film texture. Details on the fabrication and characterization of
ZnO substrates are provided in the Experimental Procedures (chapter 3). Figure 5.3(a to f)
show the characterisation of ZnO-NW and ZnO-Flat layers.
5.2.2 Growth of Sb2 Se3 films on nanostructured and flat substrates

Sb2 Se3 film growth on ZnO-Flat and ZnO-NW substrates was investigated by SEM and
XRD as a function of Sb2 Se3 film thickness. Isolating samples with different thicknesses (be-
ginning from ultra-thin films) provides snapshots of thin film growth dynamics. Notably,
148
Figure 5.3: a) Schematic of ZnO chemical bath deposition in the presence and absence of ammonium
citrate and corresponding AFM data of the resultant films, b) Top view and crosssection SEM images
showing the ZnO substrate film morphology. It is clear the presence of ammonium citrate drastically
alters the growth of ZnO and enables the formation of continuous flat ZnO substrates (ZnO Flat), c) X-
ray diffraction (XRD) patterns of ZnO Flat and ZnO NW substrates, highlighting that in both cases the
films have the same crystal orientation. d) Survey X-ray photoelectron spectroscopy (XPS) highlighting
the similar surface chemistry of ZnO-Flat (in red) and ZnO NW (in black) substrates, e) High resolution
XPS of oxygen 1s, f) High resolution XPS of zinc 2p. The scale bars in the SEM images correspond to
500 nm.
except for varying the mass of Sb2 Se3 powder in the evaporation source, layers corresponding
to the different stages of film growth were obtained under identical VTD conditions. Figure 5.4
shows XRD patterns as a function of Sb2 Se3 source mass on ZnO NW and ZnO Flat substrates.
Figure 5.5 shows cross-sectional SEM images as a function of Sb2 Se3 source mass, and the cor-
responding texture coefficients of 25 key crystallographic planes grouped into (hk0), (hk1), and
(001) orientations as previously defined.
Remarkably, these data reveal that Sb2 Se3 film morphology and crystallographic orien-
tation vary drastically between Sb2 Se3 layers grown on flat and nanostructured substrates.
Under otherwise identical VTD conditions, flat substrates result in the deposition of compact
relatively unoriented films, while the nanostructured substrate drives the growth of highly
149
Figure 5.4: X-ray diffraction patterns of Sb2 Se3 films deposited using an increasing mass of powder in
the source on a) nanostructured (ZnO NW) and b) flat (ZnO-Flat ) substrates.
nanostructured and (001) oriented Sb2 Se3 films. Figure 5.5a (panels 1 to 3) reveal a compact
Sb2 Se3 layer nucleating on the surface of the ZnO-Flat substrate. From the corresponding
crystallographic textures, Sb2 Se3 films initially grow with (hk0) preferential orientation (panel
1), evidenced by the vast majority of (hk0) texture coefficients (TC) above TC = 1, and the low
(hk1) textures that are mostly well below TC = 1. Notably, for the thinnest films (in the early
stages of film growth) vertical (001) crystal growth is highly suppressed by the flat substrate
(TC (002) = 0.23). As the film thickness increases, they form moderately (hk1) oriented films,
with a slight preference for those orientations at higher ribbon tilt angles.
Growth governed by the evolution selection principle is invoked to explain the change in
preferential texture observed for Sb2 Se3 films grown on ZnO-Flat substrates. This principle
states that during film deposition a preferential texture evolves when crystals with faster
vertical growth rates overgrow and progressively bury (suppress) crystals with slower growth
rates.[26] In Sb2 Se3 films, crystals oriented in a way that expose (hk1) crystallographic planes
will have higher vertical growth rates than (hk0) oriented crystals.[13, 27, 28] Hence, as the
Sb2 Se3 films grow (on flat substrates) the relative population density of (hk1) oriented crystals
will naturally tend to increase. This growth mechanism is illustrated in Figure 2.8 of the
Literature Review (chapter 2). As observed here, the Sb2 Se3 thin film starts growing slightly
150
Figure 5.5: SEM images and corresponding texture coefficients color coded by ribbon tilt angle for
Sb2 Se3 thin films as a function of Sb2 Se3 source mass (film thickness).a) Sb2 Se3 deposited ZnO-Flat
substrates and b) Sb2 Se3 deposited ZnO-Flat substrates. Scale bars are 500 nm.
(hk0) oriented, with only a few (hk1) oriented grains. However, those crystals with (hk1) planes
exposed at the surface eventually overgrow the other crystals due to their higher growth rates.
A shift in preferential orientation occurs (Figure 5.5a, panels 3 and 4) when the population
density of crystals at the surface becomes dominated by overgrown (hk1) grains, eventually
giving rise to the slight (hk1) orientation observed in mature thin films (Figure 5.5, panel 5).
In stark contrast, Figure 5.5b shows that under the same VTD conditions, the ZnO-NW
substrate drives the formation of vertically oriented Sb2 Se3 ribbons. For the thicker films
shown in Figure 5.5b (panels 4 and 5), exceptionally high textures are observed for (101) and
(002) crystal planes (71° and 90° tilt, respectively), with low-tilt (hk1) and (hk0) orientations
151
almost completely suppressed by the nanostructured substrate. The exceptional degree of
(001) orientation observed here may be traced back to the early stages of film growth. The
orientation of the thinnest films (Figure 5.5b, panels 1 and 2) already exhibits a slight (hk1)
preference at high tilt angles. Importantly, as the film evolves, only those textures at the
highest ribbon tilt angles rapidly increase, culminating in highly (001) oriented nanorod (NR)
array. Reports of Sb2 Se3 NR arrays preferentially aligned in [221] or [211] directions are
common in the literature.[9, 28] However, such films have a low proportion of vertical (001)
oriented Sb2 Se3 grains. The thicker films fabricated here (Figure 5.5b, panels 4 and 5) are
comparable to the highest (001) oriented films previously reported.[11, 23, 29]
5.2.3 Orientation control of Sb2 Se3 films

To confirm a direct correlation between Sb2 Se3 film orientation and substrate nanostruc-
ture, Sb2 Se3 thin films were deposited on substrates whose surfaces were gradually modified
from highly nanostructured to flat, by conformal coating of ZnO-NW substrates with ZnO via
ALD. The number of ALD cycles was progressively increased until the space between wires
in the ZnO-NW array was filled and a flat substrate was obtained. Figure 5.6a presents top
view SEM images of these substrates, beginning from untreated ZnO-NW (‘nanostructured’)
and proceeding to ZnO-NW coated with 5, 20, 40, and 60 nm ZnO layers via ALD. A final
control sample with an even smoother substrate was obtained by depositing 20 nm ZnO via
ALD directly on FTO (‘flat’). Sb2 Se3 thin films were then deposited on these substrates in a
single VTD deposition using the previously described parameters.
Analysis of the SEM images in Figure 5.6b evidence a change in Sb2 Se3 film morphology
from a vertically aligned Sb2 Se3 NR array to a dense compact thin film as substrate flattens.
Corresponding XRD patterns (Figure 5.6c) highlight a concomitant gradual disappearance
of the (002) reflection. Analysis of texture coefficients in Figure 5.6d confirms that Sb2 Se3
films deposited on highly nanostructured substrates are highly (001) oriented, display little
preference for (hk1) growth, and have greatly suppressed (hk0) orientations. As the substrate
flattens and becomes more uniform, (002) texture coefficient steadily decreases towards the
formation of randomly oriented ribbons (TC decreasing to ≈1). I make a comment about the
detailed analysis of ribbon tilt and its correlation with substrate structural parameters later
in this chapter. Collectively, these data indisputably confirm a direct link between substrate
152
Figure 5.6: a) Top view SEM images of ZnO-NW coated with 0, 5, 20, 40, and 60 nm ZnO via ALD
layer, respectively. b) Cross section SEM images of Sb2 Se3 films obtained on the substrates shown in the
upper panels. c) XRD patterns of Sb2 Se3 films as a function of ZnO ALD coating. d) Texture coefficient
as a function of ZnO ALD coating thickness. e) Reflectance measurement of corresponding samples,
wherein lighter colors indicate Sb2 Se3 films deposited on flatter substrates. f) Average reflectance
(400 - 900 nm) as a function of Sb2 Se3 (002) texture. g) Photographs of Sb2 Se3 films on flat and
nanostructured substrates, highlighting light-trapping by the oriented Sb2 Se3 NR film. Scale bars
correspond to 500 nm.
nanostructure and the growth of (001) oriented Sb2 Se3 thin films. To my knowledge, this is
the first example of bulk substrate nanostructure directing the oriented growth of Sb2 Se3 thin
films.
Considering the multitude of optoelectronic applications that exploit Sb2 Se3 thin films, a
noteworthy characteristic of the Sb2 Se3 NR films is their ability to harvest incident light. It
is well-known that highly nanostructured films possess light-trapping capacity, wherein light
that penetrates the surface becomes trapped and is eventually absorbed.[30] Figure 5.6e shows
153
Figure 5.7: a) Highly reflective Sb2 Se3 film obtained on smooth FTO/CdS substrate and b) dark (001)
Sb2 Se3 NR film on ZnO NW substrate.
Figure 5.8: Reflectance measurement of Sb2 Se3 deposited on flat FTO/CdS substrate, the 400 - 900 nm
average reflectance is 32.9%.
154
reflectance measurements of Sb2 Se3 films as a function of substrate nanostructure. A signifi-
cant decrease in reflectance across the entire visible spectral range is observed with increasing
(001) orientation. Figure 5.6f highlights an inverse correlation between the average reflectance
(400-900 nm) and the degree of Sb2 Se3 (002) texture. Highly (001) oriented Sb2 Se3 NR thin
films exhibit a remarkably low average reflectance of only 3.1%. This contrasts with (tradi-
tional) compact, randomly oriented Sb2 Se3 thin films deposited on flat substrates, wherein
the reflectance is more than 4x higher (12.8%). For reference, Sb2 Se3 films deposited on flat
CdS substrates have average film reflectance values as high as 33% (Figure 5.7 and Figure 5.8).
Extensive light-trapping by the Sb2 Se3 NR array is readily observed from photographs of the
as-prepared Sb2 Se3 films (Figure 5.6g). Light trapping is highly desirable for solar cells and
photodetectors, as well as photoelectrochemical and photocatalytic applications (such as wa-
ter splitting); hence substrate nanostructure should be a key design feature for Sb2 Se3 light
absorbers in such applications.
5.2.4 Growth mechanism of Sb2 Se3 on nanostructured substrates

Having established that substrate nanostructure strongly influences the preferential ori-
entation and morphology of Sb2 Se3 grains (and (Sb4 Se6 )n ribbons), the underpinning growth
mechanism is now considered. The evolution selection principle explains the general trends
in Sb2 Se3 thin film orientation on flat substrates, factoring in deposition conditions and the
substrate surface, but cannot account for the highly (001) oriented thin films generated in this
work. I, therefore, propose a film growth mechanism based on the physical selection of (001)
oriented crystals, whereby (hk0) Sb2 Se3 crystals are filtered out by the ZnO-NW array at the
early stages of film growth. This mechanism is schematically represented in Figure 5.9a and
consists of three main stages: (i) nucleation (initial deposition); (ii) confinement of (hk0) and
low tilt (hk1) oriented crystals growing between nanowires; and (iii) fast [001] crystal growth.
Note that the ZnO-NW film surface has significant gaps between the nanowires. As such,
in early stages of growth Sb2 Se3 vapor will penetrate the NW array and nucleate on the sides
of the ZnO NWs. From Figure 5.5b we recall that the early stages of Sb2 Se3 film growth are
characterized by an almost equal degree of (hk0) and (hk1) texture (TC ≈ 1), wherein further
growth resulted in a large increase in (hk1) textures at high tilt angles and the suppression
of (hk0) orientations. As such, the nanostructured substrate halts the growth of all (hk0) and
155
Figure 5.9: a) Proposed mechanism of Sb2 Se3 crystal growth on ZnO-NW to form highly (001) oriented
films. b) Backscattered SEM (BS-SEM) image of an Sb2 Se3 thin film in the early stages of deposition,
highlighting trapped Sb2 Se3 crystals. c) BS-SEM image of a mature (thicker) Sb2 Se3 film. d) TEM
image of a single crystal physically extracted from (c). e) HRTEM of a single Sb2 Se3 NR showing the
lattice planes. f) Crystal structure of a single (Sb4 Se6 )n ribbon and Sb2 Se3 unit cell. g) Selected-area
electron diffraction (SAED) pattern of (d). Unless indicated, scale bars correspond to 500 nm.
low tilt (hk1) oriented crystals very early during film deposition, i.e., the ZnO-NW array
must act to filter out the maturation of low ribbon tilt Sb2 Se3 crystals. This is evidenced
by the back-scattered SEM image in Figure 5.9b (obtained from an Sb2 Se3 film in the early
stage of growth, see Figure 5.5b, Panel 1), which reveals horizontal (hk0) and low-tilt (hk1)
Sb2 Se3 crystals trapped within the ZnO-NW array, and the evolution of near-vertical (001)
grains emerging from the NW surface. It follows that a NW array comprising wider wires
(with lower porosity) will be less effective in trapping crystals with low ribbon tilt, and hence
less effective in directing [001] growth, consistent with the data of Figure 5.5d wherein (002)
texture decreases with substrate infilling. Physical confinement of (hk0) and (hk1) oriented
crystals during nascent Sb2 Se3 film growth is followed by a standard growth regime, wherein
(001) oriented crystals that have been selected by orientation filtering dominate the surface,
rapidly consuming ‘Sb2 Se3 ’ vapor to yield highly (001) oriented thin films. Figure 5.9c shows
156
the back-scattered SEM image of a thick (≈ 1 µm) Sb2 Se3 film, showing highly (001) orientated
crystals and small underdeveloped (hk0) and low tilt (hk1) crystals trapped in the ZnO-NW
array.
HRTEM of a single Sb2 Se3 NR physically extracted from a mature Sb2 Se3 film deposited
on nanostructured ZnO (Figure 5.9d) has an interplanar spacing of 0.582 nm (Figure 5.9e),
matching the d-spacing between (200) planes of Pbnm Sb2 Se3 . These images confirm that the
covalent (Sb4 Se6 )n 1D ribbons (illustrated in Figure 5.9f) are aligned to the longer (vertical) axis
of the NR crystal. Figure 5.6g shows the selected area electron diffraction (SAED) of the rod
from panel Figure 5.9d. The relationship between (101) and (200) diffraction spots correspond
to a [010] crystallographic axis - orthogonal to the (001). Importantly, the discrete nature
of diffraction spots indicates the Sb2 Se3 NR is a single crystal - a property highly beneficial
to charge transport and associated applications. Collectively, SEM and HRTEM confirm the
orientation filtering process (from the initial stages of film development) and verify that Sb2 Se3
NR array is composed of (001) vertically aligned monocrystalline Sb2 Se3 NRs.
5.2.5 Surface chemistry of the nanostructured substrate

If orientation control is purely a consequence of physical confinement within a nanos-
tructured substrate, then Sb2 Se3 crystal growth should be largely independent of the sub-
strate surface chemistry (provided the substrate morphology can trap nascent crystals with
low ribbon tilt angles). To test this hypothesis, Sb2 Se3 was grown on ZnO-NW substrates
coated with Al2 O3 , TiO2 and CdS, which exhibit diverse bond dissociation energies: Ti-O 662
kJ mol−1 ; Al-O 512 kJ mol−1 ; and Cd-S 196 kJ mol−1 ; compared with Zn-O 284 kJ mol−1 .[31]
Such coated substrates will exhibit different abilities to bond with Sb2 Se3 , and hence different
orientation preferences.[27] The Al2 O3 , TiO2 (and ZnO) coatings were deposited via ALD (5
nm), and the CdS by SILAR. In all cases, a thin (≈10 nm) conformal coating was deposited
(Figure 5.11a) to preserve the underlying ZnO-NW morphology. These experiments include
a ZnO ALD coated ZnO-NW substrate as an internal standard as well as control samples
with identical coatings deposited on flat FTO substrates. SEM and XPS characterization of
substrates is provided in Figure 5.10, see the Experimental procedures (chapter 3) for details
on the deposition of the coatings.
157
Figure 5.10: Characterization of coated ZnO nanowires (ZnO NW) substrates a) scanning electron mi-
croscope (SEM) image, X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD) and backscat-
tered SEM (BS-SEM) of ZnO-NW array coated with ≈5 nm of TiO2 layer obtained via atomic layer
deposition (ALD), the smooth TiO2 coat is not distinguishable in the SEM; b) SEM, XPS, XRD and
BS-SEM of ZnO-NW array coated with ≈10 nm CdS deposited via successive ionic layer adsorption
and reaction (SILAR) method, the CdS layer appears brighter than the ZnO on the BS-SEM image. The
scale bars correspond to 500 nm. The inset shows hi-resolution SEM of CdS dots on ZnO-NW (scale
bar is 100 nm). The Sb2 Se3 crystals trapped in the NW array are clearly visible (in lighter color) in both
BS-SEM images.
XRD patterns of Sb2 Se3 samples exhibited higher intensity (002) reflections for all nanos-
tructured substrates regardless of their surface composition (Figure 5.11b). Corresponding
texture coefficients (Figure 5.11c) confirm that the degree of (002) texture is significantly higher
for Sb2 Se3 films deposited on coated ZnO-NW (red bars) compared to their flat counterparts
(blue bars), irrespective of the coating type. The somewhat lower (002) textural preference in
the case of CdS is attributed to the less uniform and thicker coating typical of SILAR depo-
sition versus highly conformal ALD coatings. This observation is consistent with previously
established data (from Figure 5.6), wherein any increase in the NW coating thickness reduces
158
the gaps between wires, and therefore the selectivity towards (001) oriented thin films. As seen
by comparing the SEM images in Figure 5.11d, despite significantly different surface chem-
istry (TiO2 , Al2 O3 , CdS) the nanostructured ZnO-NW substrate drives the growth of favorably
oriented vertical Sb2 Se3 NR arrays. Top-view SEM images further highlight the similar mor-
phology of Sb2 Se3 films deposited on TiO2 , Al2 O3 and CdS coated ZnO-NW (Figure 5.12).
Importantly, I stress that TiO2 , Al2 O3 and CdS coatings grown on flat substrates yield com-
pact Sb2 Se3 films in which the (001) orientation is suppressed (TC (002) 1). Collectively,
these results confirm that the structure of the underlying substrate dictates the deposition dy-
namics and may be used as a tool to obtain highly (001) oriented Sb2 Se3 films independent of
surface chemistry. This discovery enables flexibility in the future design of functional nanos-
tructured substrates and opens the door to unexplored substrate-film combinations for device
application.
Figure 5.11: a) Schematic of coated nanostructured and flat substrates. b) XRD patterns of Sb2 Se3 films
with CdS, Al2 O3 , and TiO2 coatings on flat FTO and ZnO-NW substrates , c) (002) textures calculated
from the XRD patterns shown in (b), d) cross-sectional SEM images of Sb2 Se3 films on the indicated
substrates. Scale bars correspond to 500 nm.
159
Figure 5.12: Top view of Sb2 Se3 deposited on ZnO-NW coated with a) Al2 O3 , b) TiO2 and c) CdS; a
source with 500 mg of Sb2 Se3 powder was used for the evaporation. The stale bars correspond to 500
nm.
5.3 Comment on the analysis of ribbon tilt

The qualitative description of (Sb4 Se6 )n ribbon tilt, employing terms as ‘high hk1 orien-
tation’, ‘high 001 orientation’ or ‘suppressed hk0’, is common place in the current literature.
However, (hk1) planes span over a wide range of tilt angles making this qualitative analysis
unsatisfactory for more precise comparison. This section provides a complete guide for the
accurate analysis of ribbon tilt via the definition of a texture weighted average ribbon tilt
(θ(avg) ) as a precise way to express ribbon tilt.
5.3.1 Texture coefficient

To calculate the crystallographic texture coefficient (TC), I considered 25 major (hkl) re-
flections of Pbnm Sb2 Se3 that did not overlap with reflections from ZnO or FTO substrates.
In addition to the (002), the chosen reflections were: 11 (hk0): (110), (020), (120), (130), (310),
(230), (320), (430), (250), (520) and (530) 13 (hk1): (501), (431), (341), (421), (411), (141), (231),
(311), (301), (221), (211), (021) and (101)
TC is calculated using:
n( I (hkl )/I0 (hkl ))

TC (hkl ) =
∑n ( I (hkl )/I0 (hkl ))
where I is the experimental intensity of a given (hkl) reflection, I0 is the standard intensity
of the (hkl) reflection calculated for a randomly oriented sample, and n is the number of
diffraction peaks being analyzed. The standard intensities for Sb2 Se3 Pbnm were calculated
160
from ICSD 16680 reference. TC = 0 indicates a completely suppressed (hkl) orientation,
TC = 1 is the standard TC value for completely randomly oriented grains, and n is a perfect
alignment along a specific [hkl].
5.3.2 Total (hk1) texture

To calculate the total texture coefficient for the set of (hk1) reflections the intensities of
individual reflections were replaced by the sum of intensities of reflections in the set:
n(∑ I (hk1)/ ∑ I0 (hkl ))

Total TC (hk1) =
∑n (∑ I (hkl )/ ∑ I0 (hkl ))
where ∑ I (hkl ) is the sum of experimental intensities of (hkl) reflections in a set, ∑ I0 (hkl )
is the sum of calculated intensities of (hkl) reflections in a set, and n is the total number of
sets, in this case n = 2 corresponds to the (hk1) and (hk0) sets. TC = 0 indicates a completely
suppressed (hk1) orientation, TC = 1 is the standard TC value for completely randomly
oriented grains and the TC = 2 is a perfect alignment along the [hk1]. Total TC(hk1) can be
divided by 2 and converted to a percentage value. In this case Total TC (hk1) + Total TC (hk0) =
2 = 100%. I note that, for this analysis, I include the (002) reflection in the set of hk1 reflections.
5.3.3 Texture weighted average ribbon tilt

(Sb4 Se6 )n ribbons in the crystal can assume any tilt from 0° (completely horizontally or
hk0 oriented) to 90° (completely vertically or 001 oriented). Also, in a completely random
oriented sample (all TC = 1), θ(avg) = 45°.
Texture
0 1 2 3 4 5 6 ... n
suppressed random oriented
-100% 0 100%
Alignment
Figure 5.13: Definition of texture coefficient, where fractional values (0 - 1) have the same weight for
suppressed orientations as values from 1 to n for oriented samples.
161
Due to the definition of TC (0 ≤ TC < 1 suppressed, TC = 1 random and 1 ≤ TC ≤ n
oriented), Figure 5.13, the tilt angle should be weighted by the texture coefficient for each
reflection:
If 0 ≤ TC ≤ 1
θr(weighted) = θr TC
If TC > 1
nhk1
θr(weighted) = θr TC
ntotal
where TC is the texture coefficient; θr is the ribbon tilt (equal to the inclination of the corre-
spondent hkl plane); nhk1 is the number of hk1 reflections being considered (14, including the
002 reflection) and ntotal is the total number of reflections (25 in the present case). The θr(avg)
is the average value of θr(weighted) . That is:
∑ θr(weighted)
θr(avg) =
ntotal
Films with high a concentration of high tilt ribbons have θr(avg) closer to 90°, θr(avg) = 45°
for completely randomly oriented ribbons and θr(avg) < 45° for samples that contain significant
concentration of horizontal or low tilt ribbons.
162
5.3.4 Ribbon tilt and substrate nanostructure
An accurate analysis of ribbon tilt requires the calculation of the texture weighted average
ribbon tilt (θr(avg) ) as defined above. Calculating this average ribbon tilt is necessary to reflect
the true inclination of (Sb4 Se6 )n ribbons in the film and precisely assess its correlation to
substrate parameters. This figure of merit, θr(avg) , is a precise way to express ribbon tilt and
should whenever possible replace qualitative terms, such as ‘highly hk1 oriented’.
A ‘highly hk1 oriented’ film does not necessarily contain a high concentration of vertical
ribbons. I start by discussing the orientation of films deposited on substrates with varying
nanostructure (ALD substrates). Figure 5.14a shows TCs plotted as a function of ribbon tilt.
It is clear that Sb2 Se3 films deposited on substrates with increasing degree of nanostructure
present a completely distinct TC distribution over θr . As illustrated in Figure 5.14b, the con-
centration ribbons with a high tilt angle increases with increasing substrate nanostructuring,
so the TC of high tilt crystallographic planes steadily increases while TC of low tilt planes
decreases. As shown in Figure 5.14c, it results in a near constant total (hk1) texture, 74% (or
1.48 out of 2) for the film obtained on the smoothest substrate (FTO/20nm ZnOALD), and
81% (1.62 out of 2) for the film obtained on the most nanostructured substrate (ZnO NW). It is
counter intuitive since we would expect the total (hk1) to be significantly higher on the nanos-
tructured substrate since significantly higher the concentration high tilt (Sb4 Se6 )n ribbons. For
an unambiguous assessment of ribbon tilt, it is necessary to use the θr(avg) , as previously de-
fined. Using the more representative θr(avg) , Figure 5.14d, revels that the film obtained on
the smoother substrate (FTO/20nm ZnOALD) has an θr(avg) of only 39.1°, while significantly
higher θr(avg) of 56.9° is observed for the Sb2 Se3 on ZnO-NW. This significant increase in the
θr(avg) (45.5%) makes it immediately obvious that the Sb2 Se3 deposited on ZnO NW is consid-
erably more vertically oriented than the film deposited on the smother substrates. Therefore,
when more in depth analysis is required, I would recommend the calculation of θr(avg) to re-
place the usual qualitative classifications such as ‘high hk1 orientation’, ‘high 001 orientation’
or ‘suppressed hk0’.
The same analysis can be performed for the films for Sb2 Se3 deposited on flat and nanos-
tructured substrates. Figure 5.15a show the distribution of texture coefficients as a function of
ribbon tilt for each case, Sb2 Se3 films deposited on ZnO Flat and ZnO NW in different stages
163
Figure 5.14: a) Distribution of texture coefficients and ribbon tilt for Sb2 Se3 deposited on substrates
with increasing degree of nanostructuring; c) total (hk1) and (hk0) texture and d) texture averaged
ribbon tilt angle (θr(avg) ).
of growth (increasing Sb2 Se3 mass in the source). Films deposited on highly nanostructured
substrates (ZnO NW 0 and 5 nm ALD) present significantly higher texture coefficient for the
(002) and (101) reflections that correspond to high ribbon tilt, 90° and 71° respectively. This
is expected for growth following the mechanism presented in Figure 5.15b. Despite the sig-
nificantly higher (002) and (101) textures obtained on ZnO NW substrates, the total (hk0) and
(hk1) textures are remarkably similar, see for example growth stage 5 in Figure 5.15c. No-
tably, Figure 5.15d clearly shows that, as expected, the θr(avg) is always considerably higher
for Sb2 Se3 films deposited on nanostructured substrates compared to the ones deposited on
flatter substrates. Thinner films on ZnO-Flat have θr(avg) far below the random threshold of
45° (25-30°) and θr(avg) of only 46.1° indicates that the thicker Sb2 Se3 film on the flat substrate
164
Figure 5.15: a) and b) Distribution of texture coefficients and tilt for (hk1) reflections of Sb2 Se3 deposited
on flat and nanostructure substrates, respectively, with increasing mass of Sb2 Se3 in the source; c)
crystal growth mechanisms on flat and nanostructured substrates; d) total (hk1) and (hk0) texture
coefficients and e) texture averaged ribbon tilt angle (θr(avg) ) of Sb2 Se3 on ZnO NW and ZnO Flat
substrates.
is virtually randomly oriented. In contrast, Sb2 Se3 films on ZnO-NW quickly surpass the ran-
dom orientation (45°) to reach θr(avg) up to 63.9° confirming the superior vertical orientation
of ribbons on the nanostructured substrate. This case study clearly shows the importance of
defining θr(avg) as a quantitative tool to access ribbon orientation when a more sophisticated
analysis is required.
165
5.4 Comment on ZnONW/Sb2Se3 photodetectors
The work in the present chapter was focused on the investigation of how Sb2 Se3 grain
morphology and orientation respond to substrate nanostructure. Highly (001) oriented Sb2 Se3
films were obtained on ZnO NW, while more compact and randomly oriented films were ob-
tained on flat substrates. Actual device performance will depend not only on Sb2 Se3 orien-
tation, but also on the quality of the interfaces, electric contacts, composition, doping, bulk
defect concentration and type. Thus, further studies are needed to correlate substrate nanos-
tructure and performance of devices in a specific application. A countless number of methods
for making nanostructured functional substrates are available in the literature, so it is expected
that the insights provided here will have a broad range of applications. The template growth
of (001)-Sb2 Se3 layers for flexible photodetector is explored in chapter 6. First, as a proof
of concept, ZnO NW/Sb2 Se3 photodetectors were designed and tested. If deposited straight
on the highly nanostructured Sb2 Se3 layer, the thin layer of metallic back electrode becomes
patchy and does not provide an efficient electric contact. Figure 5.16 shows a schematic of the
steps for making ZnO NW/Sb2 Se3 NR photodetectors. First, it was necessary to fill the space
between Sb2 Se3 NR with PMMA (deposited via spin coating). The Au back contact could be
deposited directly on the smooth PMMA layer, but this polymer forms a highly resistive bar-
rier on the Sb2 Se3 NRs. Thus, to expose the tips of the encapsulated Sb2 Se3 NRs, the PMMA is
partially etched in O2 plasma. This etching step to enhances the electrical contact with the Au
back contact electrode, which is deposited via e-beam evaporation in the last step to complete
the device.
Figure 5.16: Scheme showing the procedure used to fabricate ZnO NW/Sb2 Se3 NR photodetectors.
Figure 5.17a shows the cross section of SEM images of ZnO NW/Sb2 Se3 NR-PMMA
etched under O2 plasma for longer times. It is clear that PMMA effectively fills the space
between the Sb2 Se3 NR, and only the tips of the rods become increasingly more exposed with
166
increasing etching times. Figure 5.17b shows a photography of the surface of the devices
fabricated here. Relatively, smooth Au contacts are obtained on pristine PMMA and after 90s
etching. For devices where the polymer layer was etched for more than 180s the Au electrode
has a rougher (darker) appearance.
Figure 5.17: a) Cross section SEM images of ZnO NW/Sb2 Se3 NR-PMMA exposed to O2 plasma for
different times and b) surface of devices.
The response of Sb2 Se3 NR photodetectors was measured to assess the effectiveness of
PMMA deposition and etching. Figure 5.18 revels that the device with a pristine PMMA has
a very low photoresponse, ≈ 0.2 µA. This low photocurrent is expected and confirms that
the PMMA encapsulates the Sb2 Se3 NR effectively creating a highly resistive barrier between
the light absorber layer and Au contact. Very long etching times, 360 s exposure to the O2
plasma, results in low photocurrent and high dark current. This occurs because the PMMA
is etched too far and stops being effective at filling the space in the Sb2 Se3 NR layer. The Au
that is deposited on this rough surface acts very poorly as an electrode and is likely to pass
through the Sb2 Se3 layer and touch contact with the ZnO layer. When the PMMA is only
167
partially etched, 90 - 180 s exposure time, the photoresponse increases by almost two orders
of magnitude to from ≈ 0.2 µA to ≈ 10 µA. This indicates an ideal etching time that exposes
only the tip of Sb2 Se3 NR, resulting in a better electrical contact with the back electrode,
while maintaining a relatively smooth surface for the deposition of the thin Au film. Table 5.1
provides a summary of parameters for all photodetectors fabricated here. The photoresponses
obtained here are significantly lower than what is expected from an efficient solar cell, 20 - 30
mA cm−2 . This low current is attributed to the poor junction between Sb2 Se3 and the ZnO
NW.
Figure 5.18: Photoresponse of ZnO NW/Sb2 Se3 NR-PMMA exposed to O2 plasma for different etching
times under illumination with 625 nm red LED at 80 mW cm−2 and no bias.
Table 5.1: Parameters for Sb2 Se3 photodetectors tested here.
Substrate Etching time I ph Id R D∗

(s) (µA cm−2 ) (nA cm−2 ) (mA W−1 ) (×1010 Jones)
ZnO NW 0 0.391 0.340 4.89E-05 0.148
ZnO NW 90 5.94 0.299 7.43E-04 2.40
ZnO NW 180 10.2 0.386 1.28E-03 3.63
ZnO NW 360 0.341 11.8 4.26E-05 0.0219
ZnONW-5nmZnOALD 180 104 20.0 1.30E-02 5.14
ZnONW/CdS 180 486 3.42 6.08E-02 5.80
ZnO Film No PMMA 410 19.4 5.13E-02 20.6
168
Notably, a thin CdS layer (≈10 nm, SILAR) coating the ZnO NW increases the photore-
sponse of detectors significantly. Figure 5.19a revels that, for ZnO NW substrate that was
coated with CdS SILAR, the photocurrent increases by almost two orders of magnitude. The
performance is comparable with a ZnO Flat/Sb2 Se3 device that has a compact surface for
the deposition of a smooth Au contact. Figure 5.19b shows that all detectors have a similar
linear range, extending over more than 4 orders of magnitude. I note that the red light used
in the experiments, 625 nm, is not absorbed by the CdS. Therefore, it is possible to conclude
that the CdS coating is helping to form a more efficient junction with Sb2 Se3 NR, which is
more effective in extracting photogenerated charge carriers. This shows that exploring meth-
ods for achieving better junctions and electric contacts is an effective way to harvest the high
photocurrent that is expected from the highly (001)-Sb2 Se3 layers obtained here.
Figure 5.19: a) Photoresponse and b) power curve for ZnONW/Sb2 Se3 NR-PMMA(180s plasma),
ZnONW/CdS/Sb2 Se3 NR-PMMA(180s plasma) and ZnOFlat/Sb2 Se3 detectors. A 625 nm LED was
used as the illumination source. Photoresponse was obtained at 8 mW cm−2
.
5.5 Conclusions
A new method to obtain preferentially oriented Sb2 Se3 thin films via physical confine-
ment effects (orientation filtering) is demonstrated. The morphology and orientation of Sb2 Se3
films were investigated as a function of substrate nanostructure for films deposited via VTD.
Exceptionally highly (001) oriented Sb2 Se3 films were obtained on highly nanostructured sub-
169
strates, while more compact and randomly oriented films were obtained on flat substrates.
Fine control over Sb2 Se3 film morphology and orientation is achievable by gradually manipu-
lating the substrate nanostructure. A mechanism based on the preferential trapping (filtering
out) of (hk0) and low tilt (hk1) crystals within a substrate nanowire array accounts for the
high degree of (001) orientation obtained for Sb2 Se3 films. As this mechanism is physical, it
is expected it will translate to other 1D crystal structure materials (including other antimony
chalcogenides). Finally, it was shown that the surface chemistry of the nanostructured sub-
strates can be altered and still drive the growth of highly (001) oriented Sb2 Se3 thin films
- not limiting this phenomenon to a particular substrate type. Consequently, the thin films
developed herein should be directly implementable within a range of Sb2 Se3 devices that
require specific surface chemistries. Likewise, with additional efforts on substrate design (ge-
ometry) and composition, the methodology outlined here provides an avenue to new device
architectures based on template-driven crystallographic orientation. This work constitutes an
important advance in the development of antimony chalcogenide-based applications, such as
solar cells, photocatalytic devices, sensors, and batteries, wherein performance is intrinsically
linked to morphology and orientation.
5.6 References
[3] G. Liang, M. Chen, M. Ishaq, X. Li, R. Tang, Z. Zheng, Z. Su, P. Fan, X. Zhang, S. Chen,
Adv. Sci. 2022, 9, 2105142.
[4] T. Liu, X. Liang, Y. Liu, X. Li, S. Wang, Y. Mai, Z. Li, Adv. Sci. 2021, 8, 2100868.
[5] S.-N. Park, S.-Y. Kim, S.-J. Lee, S.-J. Sung, K.-J. Yang, J.-K. Kang, D.-H. Kim, J. Mater.
Chem. A 2019, 7, 25900.
[6] H. Song, T. Li, J. Zhang, Y. Zhou, J. Luo, C. Chen, B. Yang, C. Ge, Y. Wu, J. Tang, Adv.
Mater. 2017, 29, 1700441.
[7] C. Chen, K. Li, J. Tang, Sol. RRL 2022, 6, 2200094.
[10] C. Liu, T. Liu, Y. Li, Z. Zhao, D. Zhou, W. Li, Y. Zhao, H. Yang, L. Sun, F. Li, Z. Li, J.
Mater. Chem. A 2020, 8, 23385.
170
[12] W. Li, L. Deng, X. Wang, J. Cao, Y. Xie, Q. Zhang, H. Zhang, H. Deng, S. Cheng,
Nanoscale 2021, 13, 9834.
[14] H. Deng, Y. Zeng, M. Ishaq, S. Yuan, H. Zhang, X. Yang, M. Hou, U. Farooq, J. Huang,
K. Sun, R. Webster, H. Wu, Z. Chen, F. Yi, H. Song, X. Hao, J. Tang, Adv. Funct. Mater.
2019, 29, 1901720.
[15] X. Liang, C. Guo, T. Liu, Y. Liu, L. Yang, D. Song, K. Shen, R. E. I. Schropp, Z. Li, Y.
Mai, Sol. RRL 2020, 4, 2000294.
[16] Y. Pan, X. Hu, Y. Guo, X. Pan, F. Zhao, G. Weng, J. Tao, C. Zhao, J. Jiang, S. Chen, P.
Yang, J. Chu, Adv. Funct. Mater. 2021, 31, 2101476.
[17] S.-N. Park, S.-Y. Kim, S.-J. Lee, S.-J. Sung, K.-J. Yang, J.-K. Kang, D.-H. Kim, Mater. Adv.
2022, 3, 978.
[18] F. Pattini, S. Rampino, F. Mezzadri, D. Calestani, G. Spaggiari, M. Sidoli, D. Delmonte,
A. Sala, E. Gilioli, M. Mazzer, Sol. Energy Mater. Sol. Cells 2020, 218, 110724.
[19] W. Wang, Z. Cao, L. Wu, F. Liu, J. Ao, Y. Zhang, ACS Appl. Energy Mater. 2021, 4, 13335.
[21] S. Yao, J. Wang, J. Cheng, L. Fu, F. Xie, Y. Zhang, L. Li, ACS Appl. Mater. Interfaces 2020,
12, 24112.
[22] J. Zhou, H. Chen, X. Zhang, K. Chi, Y. Cai, Y. Cao, J. Pang, J. Alloys Compd. 2021, 862,
158703.
[23] H. Zhou, M. Feng, P. Li, X. Gong, D. Zhang, S. Chen, Sustainable Energy Fuels 2020, 4,
3943.
[24] Mamta, Y. Singh, K. K. Maurya, V. N. Singh, Sol. Energy Mater. Sol. Cells 2021, 230,
111223.
[26] A. Vanderdrift, Philips Res. Rep. 1967, 22, 267.
[28] X. Liang, Z. Li, X. San, T. Liu, Y. Liu, K. Shen, S. Wang, R. E. I. Schropp, Y. Mai, Sol.
RRL 2021, 6, 2100869.
[29] H. Zhou, M. Feng, K. Song, B. Liao, Y. Wang, R. Liu, X. Gong, D. Zhang, L. Cao, S.
Chen, Nanoscale 2019, 11, 22871.
[30] D. Smith, Thin-Film Deposition: Principles and Practice, McGraw-Hill Education, 1995.
[31] J. Dean, N. Lange, Lange’s Handbook of Chemistry, McGraw-Hill, 1992.
171
Chapter 6
Transferable highly (001) oriented
Sb2Se3 nanorod layer for innovative
device architectures
172
6.1 Preface
Thin layers of antimony chalcogenides, Sb2 X3 (X = S, Se or Sx Se1 – x ), comprised of (001)
oriented grains are highly desirable for optoelectronic applications owing the highly efficient
charge transport along the optimally aligned quasi-1D (Sb4 X6 )n ribbons. However, perfectly
(001) oriented Sb2 X3 films are hard to obtain in practice due to orientation being highly sensi-
tive to external parameters such as deposition conditions and substrates. Innovative nanofab-
rication techniques are critical to achieve (001) oriented layers and harvesting Sb2 Se3 ’s full
potential as an excellent light absorber material. Here, exceptionally highly (001) oriented
Sb2 Se3 nanorod (Sb2 Se3 NR) layers are obtained via template growth to overcome the difficul-
ties associated with oriented growth directly on the desired substrate. The (001) Sb2 Se3 NR
layers are obtained using substrate nanostructure driven orientation control (orientation filter-
ing). The oriented films are readily transferable to functional substrates via a polymer assisted
transfer method, enabling low-temperature process of flexible photodetectors (PDs). The flex-
ible device fabricated with transferred layers shows broad range photoresponse, satisfactory
responsivity (up 0.54 mW cm−2 ), and detectivity (1.4 1010 Jones), which inspired me to design
a proof-of-concept self-powered flexible heart rate monitor. The concept of transferable layers
introduced in this work is expected to be readily applicable to generate innovative (001) Sb2 Se3
device architectures that would otherwise not be achievable via direct deposition, including
photoelectrochemical water splitting, battery electrodes and solar cells.
173
6.2 Introduction
As discussed in the previous chapters, a perfectly (001) orientated Sb2 Se3 layer is the
most desirable for electronic device applications, since charge transport is vastly more ef-
ficient along the covalently bound ribbons.[1] However, Sb2 Se3 film growth is notoriously
sensitive to substrate type, i.e., layers deposited under the same conditions but on different
substrates frequently present completely distinct crystallographic alignment and morphol-
ogy.[2] Orientation control is further complicated by the tendency of ribbons to grow in less
desirable (hk0) orientation, exposing the lower energy Van der Walls planes on the surface
of the substrate.[3] To date, a comprehensive method for (001)-Sb2 Se3 growth on different
substrates has not been reported. The orientation filtering mechanism, discussed in chap-
ter 5, presents as a great platform to enable a universal method for obtaining (001) oriented
Sb2 Se3 films via template growth and transfer onto a range of functional substrates. Template
growth of (001)-Sb2 Se3 NR, combined with an optimized transfer method, is highly an attrac-
tive nanofabrication technique to take full advantage of efficient charge transport along the
vertically oriented (Sb4 Se3 )n ribbons on a variety of substrates, including flexible substrates.
Film transfer can overcome restraints related to oriented growth directly onto the desired
substrate.[4] Transferring is especially useful to fabricate flexible and scalable electronic de-
vices with nanoscale functional features or large patterned areas.[5–9] The transferable layer
is first grown on a donor substrate (template) that provides the conditions for growth of high
quality layers with an ideal orientation and morphology. Polymer assisted transfer techniques
are commonly employed to fabricate devices with quasi-2D functional layers.[10] The layer is
immobilized in a polymer matrix and then detached from the template in a solvent bath (e.g.,
water).[11] The basic requirement is that the immobilized layer is stable enough to be lifted-off
from the donor substrate,[12] hydrophobically or by dissolving an interlayer. The detached
layer can then be transferred to a receiving substrate directly from solution to build the de-
vice. Importantly, transfer techniques decouple growth requirements for donor and recipient
substrate and allow for the stacking of layers into architectures that would be impossible to
achieve via direct deposition on the final desired substrate.
174
The actual production of (001)-Sb2 Se3 NR transferable layers relies on the development
of a practical nanofabrication method, tailored to preserve the vertical alignment of NR from
the donor onto the receiving substrate. Here, I develop a robust method to transfer (001)
oriented Sb2 Se3 NR layers and fabricate proof-of-concept functional PDs. The Sb2 Se3 NR tem-
plate is grown on a donor substrate (ZnONW) optimized to drive (001) growth, immobilized
in PMMA and detached in aqueous solution. X-ray diffraction (XRD), X-ray photoelectron
spectroscopy (XPS) and scanning electron microscopy (SEM) show that the crystallographic
orientation, chemical composition and grain morphology are preserved after transferring.
PD performance is investigated for layers transferred onto glass/FTO, glass/FTO/SnO2 and
glass/FTO/CdS receiving substrates. Finally, a wearable heart rate detector is demonstrated
by transferring the Sb2 Se3 NR layer into flexible PET/ITO substrate.

6.3.1 Template growth of the (001) Sb2 Se3 NR layer
In chapter 5, it was demonstrated that a substrate’s physical nature (morphology and
nanostructure) can be tailored to direct the crystal orientation and morphology of evaporated
Sb2 Se3 films. An appropriately nanostructured substrate (ZnO nanowires (NWs)) can drive
the growth of highly (001) oriented Sb2 Se3 NR layers via an ‘orientation filtering’ mechanism,
wherein (hk0) crystals are ‘trapped’ by the ZnONW substrate in the early stages of growth,
yielding an Sb2 Se3 layer with a very high concentration of vertically aligned (001) oriented
crystals. Figure 6.3a shows the cross section scanning electron microscope (SEM) image of a
typical Sb2 Se3 NR layer obtained via this method. The detailed procedure is described in the
experimental section, in short, a chemical bath deposition (CBD) method[13] was used to ob-
tain the ZnONW on glass/FTO and the Sb2 Se3 was deposited via vapor transport deposition
(VTD).[14]
175
Figure 6.1: a) Scanning electron microscope (SEM) cross section image of Sb2 Se3 NR on ZnONW and b)
scheme of transferring process from ZnONW to a new substrate. The scale bar in the SEM corresponds
to 1 µm.
Film transfer was achieved using polymer assisted technique, as summarized in Fig-
ure 6.3b. First, Sb2 Se3 NR film is immobilized in a PMMA layer (deposited via spin coating),
the sample is then immersed in an 1M H2 SO4 solution to dissolve the ZnONW underlayer
and detach the Sb2 Se3 NR layer. The detached ‘floating’ Sb2 Se3 NR/PMMA encapsulated film
is then moved to a clean water bath to wash out the excess acid, followed by being transferred
to a new substrate (directly from the water bath) and finally dried on a hot plate. Care must
be taken during transferring as the detached film tends to coil relatively easily and must be
uncoiled on the receiving substrate (Figure 6.2). Although the Sb2 Se3 NR layer by itself is ex-
tremely fragile, the Sb2 Se3 NR/PMMA forms a fairly resistant composite membrane, making
the transfer method described here highly effective.
176
Figure 6.2: Steps of Sb2 Se3 NR/PMMA transferring process.
Figure 6.3a shows cross section SEM images of the Sb2 Se3 /PMMA layer before trans-
ferring on the ZnONW template and after transferring to a glass/FTO substrate. The SEM
images reveal that the Sb2 Se3 NR remain completely intact and undisturbed after transferring
to the receiving substrate. Importantly, the vertical alignment of the Sb2 Se3 NRs is completely
preserved. It is also clear that the transferred film makes close contact with the receiving
substrate, indicating that it is applicable to build functional devices.
Figure 6.3: a) Scanning electron microscope (SEM) images of ZnONW/Sb2 Se3 NR/PMMA (template)
and layer Sb2 Se3 NR/PMMA transferred onto FTO; b) X-ray diffraction (XRD) pattern highlight the
relative intensity of (002) reflection and c) texture coefficient of Sb2 Se3 for selected reflections before
and after transfer; d) photography of transferred Sb2 Se3 NR layer onto glass/FTO substrate. * ZnO and
∨ FTO. The scale bars in the SEM correspond to 1 µm.
177
To further assess the effectiveness of the transfer method, the crystallographic orientation
of the Sb2 Se3 NR was investigated before and after transfer. Figure 6.3b shows the XRD pat-
terns for the template and transferred Sb2 Se3 NR layers. A high intensity (002) reflection is
observed for the template thin film, which is indicative of the remarkable vertical alignment
of Sb2 Se3 grains in the [001] crystallographic direction. It can be seen that, after transfer the
intensity of the (002) reflection is preserved. Texture coefficient analysis is the most accurate
way to elucidate any fine changes in crystallographic orientation and thus the effectiveness
of the substrate transfer. Figure 6.3c shows the texture coefficient (TC) for key (hkl) reflec-
tions for thin films before and after transfer. Despite the delicate nature of the Sb2 Se3 NR
film, after transferring a remarkable 80% of (002) texture is retained (TC(002) = 12.7 before
and 10.1 after), while the (101) texture increases by ≈7% (TC(101) = 4.5 before and 4.8 after).
The full set of analyzed TCs is shown in Figure 6.4a/b, revealing that the average ribbon tilt
(θr(avg) ) is also well preserved, 73° before transfer and 68° after transfer. These data confirm
that crystallographic alignment is well preserved in the transferred layer. Figure 6.3d shows
a picture of the Sb2 Se3 NR film on the receiving glass/FTO substrate. The Sb2 Se3 NR layer is
seen to be black due to its remarkable light-trapping capacity, which helps effectively harvest
incident light. The transferred film has a very low reflectance of only 2.4% in the visible range
(Figure 6.5). Close inspection confirms that the transferred layer is flat and well adhered to the
substrate which, once again, demonstrates the efficacy of the transfer method and indicates
the applicability of transferred layers to functional electronic devices.
178
a) b)
(hk0) (hk1)
15 15
θr(avg) (template) = 73°
Texture coefficient (hk1)

Template θr(avg) (transferred) = 68°
Texture coefficient
10 10
Transferred
5 5
0 0
30 40 50 60 70 80 90
(110)
(020)
(120)
(130)
(310)
(230)
(320)
(430)
(250)
(520)
(530)
(501)
(431)
(341)
(421)
(411)
(141)
(231)
(311)
(301)
(221)
(211)
(021)
(101)
(002)
Ribbon tilt, r (°)
Figure 6.4: a) Full set of analyzed texture coefficients (TC) and b) TC vs ribbon tilt (θr ) for (hk1)
reflections of template and Sb2 Se3 NR transferred onto FTO.
5 0
4 0
R e fle c ta n c e ( % )
3 0
2 0
1 0
te m p la te
0 tra n s fe rre d
4 0 0 6 0 0 8 0 0 1 0 0 0
W a v e le n g th ( n m )
Figure 6.5: Reflectance of Sb2 Se3 NR layer/PMMA (90s plasma etching) for template (blue line) and
transferred (red line) layers.
179
6.3.2 Sb2 Se3 NR photodetectors from transferred layers
The polymer assisted transfer is highly effective at preserving the desired Sb2 Se3 NR mor-
phology and orientation, however PMMA is not conductive. As such, to fabricate the optoelec-
tronic devices from transferred layers it is necessary to make the films electronically accessible.
To achieve this, clean Sb2 Se3 surfaces must be obtained by etching the PMMA from atop the
encapsulated Sb2 Se3 NR thin film. For this, the film was treated under O2 plasma to expose
Sb2 Se3 NR tips (Figure 6.10a). In Figure 6.6a, the backscattered electron SEM (BS-SEM) images
show the encapsulated Sb2 Se3 NR/PMMA (before plasma) and the partially etched Sb2 Se3 NRs
(after plasma). Before O2 plasma treatment, the relatively low atomic weight PMMA film pro-
duces a low BS-SEM signal, yielding darker regions in the BS-SEM image. After plasma
etching, exposed Sb2 Se3 NR scatters more electrons due to higher atomic weight and produces
a stronger BS-SEM signal, resulting in the brighter regions.
Figure 6.6: a) Top-view backscattered scanning electron microscope (BS-SEM) images and b) corre-
spondent X-ray photoelectron spectra (XPS) of the Sb and Se regions of Sb2 Se3 NR before and after O2
plasma exposure. The scale bars in the BS-SEM correspond to 1 µm.
180
Figure 6.7: a) Schematic representation of the stack of layers in the devices and cross section
SEM images of Sb2 Se3 NR/PMMA layer transferred onto b) glass/FTO, c) glass/FTO/CdS and d)
glass/FTO/SnO2 . Comparison of e) photocurrent density (J ph ) and f) Responsivity (R)/Detectivity
(D ∗ ) for devices built on the template (FTO/ZnONW/Sb2 Se3 NR/PMMA) and on Sb2 Se3 NR/PMMA
transferred layers. d) -3dB frequencies for transferred devices. Devices were annealed at 175 °C in
air. A red light emitting diode (LED), 625 nm at 115 W cm−2 irradiation was used as the illumination
source for photocurrent measurements. The scale bars correspond to 500 nm.
The effectiveness of the etching process to expose Sb2 Se3 surfaces was also investigated
using XPS. The XPS spectra in Figure 6.6b show no detectable Sb or Se characteristic peaks
before etching and clear characteristic peaks after the film is exposed to plasma. These results
confirm that Sb2 Se3 NRs are well encapsulated by a PMMA film before plasma treatment, this
is valuable as the PMMA layer helps isolate the Sb2 Se3 NRs from the acid bath used for the
transferring process. Moreover, XPS analyses on the etched template and transferred layers
(Figure 6.8) confirm that Sb2 Se3 remain chemically unchanged after acid bath. The Sb 3d
peaks (3/2 at 540.0 eV and 5/2 at 530.9 eV) are attributed to Sb(III)/Sb(IV) surface oxides
formed upon exposure to O2 plasma.[15, 16] Energy dispersive spectroscopy (EDS) analysis
(Figure 6.9) confirms the layer is Se deficient. Crucially, I verified that after 90s O2 plasma
treatment, Sb2 Se3 NR tips became satisfactorily exposed, assuring that Sb2 Se3 makes effective
contact with the back electrode (Figure 6.10b/c).
181
Sb 3d3/2
Se 3d3/2
Se 3d5/2
Sb 3d5/2
Intensity (a.u.)
Intensity (a.u.)
545 540 535 530 525 62 60 58 56 54 52 50

Binding energy (eV) Binding energy (eV)
Sb2Se3/PMMA (90s O2 Plasma)
Template Transferred
Figure 6.8: X-ray photoelectron spectroscopy (XPS) of the Sb and Se regions of Sb2 Se3 /PMMA layer
etched for 90s in O2 plasma on the template (ZnONW/Sb2 Se3 /PMMA-90sPLASMA, before transfer,
red line) and on the receiving substrate (FTO/Sb2 Se3 /PMMA-90sPLASMA, after transfer, blue line).
The characteristic peaks on the XPS indicates that Sb2 Se3 remains chemically unchanged through the
acid bath, during the transferring process.
Se
Transferred Sb2Se3NR
(90s Plasma)
Intensity (a.u.)
Se 53%
At%:
Sb 47%
Sb
C Sb
Sb Se Sb Sb
O
1 2 3 4 5
Energy (keV)
Figure 6.9: Energy dispersive spectroscopy (EDS) analysis showing that Sb2 Se3 NR are slightly Se defi-
cient (expected Sb2 Se3 At%: Se 60% and Sb 40%).
182
Having resolved electronic access at the Sb2 Se3 contact I turn my attention to the sub-
strate contact. In Sb2 Se3 optoelectronic devices the quality and chemical composition of this
contact is known to strongly affect device performance.[14, 17] Likewise, the type of contact
is expected to affect the adherence of the transferred film. To evaluate different substrate con-
tacts, prototype PD devices were developed (Figure 6.7a). A solution processed carbon paint
electrode was employed (back contact), and the devices were constructed as summarized in
Figure 6.10a. Substrates were chosen based on direct transfer to a conductive substrate, as
well as common n-type contacts (electron transport layers, ETLs) known to be applicable to
Sb2 Se3 devices. Specifically, PD devices were fabricated by transferring the Sb2 Se3 NR onto
glass/FTO, glass/FTO/CdS and glass/FTO/SnO2 . Figure 6.7b-d shows SEM images of the
transferred films onto these layers highlighting the close contact of the Sb2 Se3 NR thin film.
The performance of these devices was first evaluated as a function of annealing tempera-
ture on each substrate to find the optimal temperature for substrate adherence and electrical
contact. It was found that relatively independent of substrate, devices annealed at 175 °C
provided the best performance (see Figure 6.11 and Figure 6.12).
Figure 6.10: a) Scheme showing O2 plasma treatment used to expose the tips of Sb2 Se3 NR by partially
etching the PMMA film to electrically contact the drop casted back electrode; b) X-ray photoelectron
spectroscopy (XPS) of Sb2 Se3 /PMMA layer before and after plasma etching; c) scanning electron mi-
croscope (SEM) image of Sb2 Se3 /PMMA etched film. The scale bar corresponds to 1 µm.
183
Annealing temperature (°C)
0.001
150 175 200 225 250 FTO (b)
1E-4 100 150 175
200
1E-5 80 225 250
Jph (A cm2) 1E-6
100
1E-7 80
1E-8
1E-9
Template (a)
1E-10
0 20 40 60 0 20 40 60
Time (s) Time (s)
0.001
FTO/CdS (c) 150 175
(d)
1E-4 200
150 175 200 225 80 225
1E-5
100
Jph (A cm2)
250 250
1E-6 80 100
1E-7
1E-8
1E-9 FTO/SnO2 NP
1E-10
0 20 40 60 0 20 40 60
Time (s) Time (s)
Figure 6.11: Photocurrent density (J ph ) as a function of annealing temperature for a) template device
(ZnONW/Sb2 Se3 NR) and devices with Sb2 Se3 NR transferred to b) glass/FTO, c) glass/FTO/CdS and
d) glass/FTO/SnO2 . Devices were annealed in air for 10min before each measurement.
a) b)
101 1011
FTO
control
Responsivity (mA W-1)
100 control 1010

Detectivity (Jones)
10-1 109
CdS
FTO SnO2
10-2 108
SnO2
-3
10 CdS 107
10-4 106
100 150 200 100 150 200
Annealing temperature (°C)
Figure 6.12: a) Responsivity (R) and b) Detectivity (D ∗ ) as a function of annealing temperature for
devices with Sb2 Se3 NR transferred to glass/FTO, glass/FTO/CdS and glass/FTO/SnO2 compared to
control (FTO/ZnONW). Both R and D ∗ were calculated from 5s ON/OFF pulses of red light emitting
diode (LED), 625 nm at 115 W cm−2 irradiation.
184
Figure 6.7e shows the photocurrent density (J ph ) of the self-powered (0 V bias) PDs on
each substrate. The device with a SnO2 ETL outperforms devices with CdS ETL (10.9 µA
cm−2 ), as well as the device built on the layer transferred directly on FTO (18.8 µA cm−2 ).
This indicates that SnO2 is an adequate ETL for this particular application. Notably, the
transferred PD with SnO2 ETL shows a J ph within 10% of that of the initial control device
(FTO/ZnONW/Sb2 Se3 NR/PMMA), 108 µA cm−2 (transferred) and 116 µA cm−2 (control).
As device contact, and hence photocurrent, is expected to be best when the Sb2 Se3 is directly
evaporated onto a substrate, this result indicates that intimate contact has been formed for the
optimized transferred layers with the SnO2 interface. Furthermore, the Sb2 Se3 NR electronic
properties are well preserved after transfer.
The performance of PDs was further characterized by calculating the responsivity (R) and
specific detectivity (D ∗ ) parameters of devices.[18] The R (mA W−1 cm−2 ) is a measure of the
PD efficiency and is calculated by the ratio between input power at a given light wavelength
(Pλ ) and output current density and is given by:
J ph
R=
Pλ
The parameter D ∗ (cm Hz1/2 W−1 or Jones) is a measurement of the sensitivity of a PD
and is defined as:
J ph
D∗ =
Pλ (2qJdark )1/2
where q is the elementary charge.
Figure 6.7f shows R and D ∗ values for the PDs built on transferred layers and the (un-
transferred) control. When compared to the detector built on the template, the R and D ∗ pa-
rameters confirm the excellent performance of the Sb2 Se3 NR layer transferred on FTO/SnO2
substrate. This transferred PD presented a performance close to the control; R = 0.96 mA W−1
cm−2 (vs 1.04 mA W−1 cm−2 , for the control) and D ∗ = 1.38 1010 Jones (vs 1.76 1010 Jones, for
the control).
185
a) b)
0 0
Relative response (dB)

-3dB
-3dB
13kHz
22.52Hz
-3 -3
-6 -6
control FTO
-9 -9
3 4 5
10 10 10 100 101 102 103
Light modulation frequency (Hz) Light modulation frequency (Hz)
c) d)
0

0 -3dB
-3dB 25.77Hz
73.72Hz
-3 -3
FTO/CdS
-6 -6
FTO/SnO2
-9 -9
100 101 102 103 100 101 102 103

Light modulation frequency (Hz) Light modulation frequency (Hz)
Figure 6.13: Relative frequency photoresponse for a) control device (FTO/ZnONW/Sb2 Se3 NR) and for
devices with a Sb2 Se3 NR transferred on b) FTO, c) FTO/SnO2 and d) FTO/CdS.
Figure 6.14: a) Top view and b) cross section scanning electron microscope (SEM) images of Sb2 Se3 NR
film transferred onto FTO and annealed at 400 °C for 10 min under N2 atmosphere.
186
To evaluate the extraction of photogenerated carriers in transferred devices with and
without an ELT, the response to frequency modulated light was assessed. Figure 6.7g reveals
that the -3dB bandwidth of devices with a SnO2 ETL was approximately ≈3x higher (73.7
Hz) compared to the device with FTO only (22.5 Hz) or FTO/CdS (25.8 Hz). The full -3dB
curves are plotted in Figure 6.13a-d. This result indicates that the photogenerated carriers are
extracted more effectively by the SnO2 ETL. It is worth noting that, the control device was con-
siderably faster, -3dB of 13 kHz, which indicates that the junctions forming the devices with
transferred layers can still be improved, for example by annealing at higher temperatures.
However, high annealing temperature may cause the Sb2 Se3 to degrade (Figure 6.14a/b).
Moreover, the mild optimal annealing temperature (175 °C) is ideal for the fabrication of
flexible devices using PET/ITO substrate.
6.3.3 Flexible photodetectors and heart rate monitor

Engineering flexible thin film devices present unique challenges. Firstly, the device must
be able to be fabricated at mild temperatures given that most flexible substrates have far lower
heat resistance than glass. As I have shown, transferred devices can successfully be processed
at mild temperatures (150-175 °C), ideal for flexible device fabrication with PET substrate.[19]
The second main requirement is that all layers must withstand the shear stress caused by
flexing, without fracturing or detaching from the substrate.[20] Flexible devices with a Sb2 Se3
absorber layer have been demonstrated on a few occasions.[21–27] Sb2 Se3 has lower Young’s
modulus along the crystallographic a- and b-axis, because ribbons are held together only
by weak van der Walls forces along these directions.[21] This means that vertically oriented
Sb2 Se3 films can easily deform on the ab-plane, which presents a great advantage for flexi-
ble device applications. However, if ribbons were laid horizontally on the substrate,[22, 27]
tension is applied directly along the strong covalent bonds within the (Sb4 Se3 )n chain when
substrate flexes. A compact grain arrangement may also struggle to accommodate tension, as
forces act to pull the grains apart upon flexing. A (001) Sb2 Se3 NR layer is composed by verti-
cally oriented grains (Figure 6.15a), that are in turn comprised of (Sb4 Se3 )n vertically aligned
ribbons, which makes the transferred (001)-Sb2 Se3 NR layer ideal to endure large deformation
stress. As such, transferrable (001) oriented Sb2 Se3 NR layers provide an avenue to fabricate
robust flexible devices.
187
Figure 6.15: a) Scheme of Sb2 Se3 NR crystals on flat vs flexed substrate; b) flat and flexed device at φ =
40° and φ = 54°. Annealed at 175 °C.
Figure 6.16: a) Photocurrent density (J ph ) as a function of annealing temperature for flexible photode-
tector (PET/ITO/Sb2 Se3 NR) and b) as deposited film (not annealed) with poor mechanical properties,
broken upon flexing.
188
Figure 6.17: a) PET/ITO/SnO2 /Sb2 Se3 NR flexible device, completed with drop cast conductive PEDOT
back electrode. Normalized photocurrent (J ph ) b) up to 250x flexing cycles at φ = 54° and upon flexing
at φ = 0°, 40° and 54°. c) J ph response for 5s pulses (LED, 625 nm, 115 W cm−2 ). d) Power curves for
devices with and without SnO2 ETL. e) Responsivity (R)/Detectivity (D ∗ ). f) Relative photoresponse
to frequency modulated light (-3dB).
Here, flexible devices were fabricated by transferring the Sb2 Se3 NR layers onto PET/ITO/
SnO2 substrate. Figure 6.17a shows the device upon flexing, highlighting that the transferred
Sb2 Se3 NR remains well attached to the substrate. I note that, optimal performance was also
obtained at ≈175 °C, with the as deposited film (dried at 80 °C) presenting both poor elec-
tronic and mechanical performance (Figure 6.16a-b). To assess the electronic performance of
the optimized device upon mechanical stress, the PD was exposed to numerous flexing cycles
and flexed at increasing angles (φ, Figure 6.15b). Stable performance was demonstrated over
the 250x flexing cycles. The device also shows great mechanical stability and no loss in pho-
tocurrent even at very high flexing angles (Figure 6.7b), in fact, a slightly increased J ph upon
flexing was observed. Furthermore, Figure 6.7c reveals that the self-powered flexible device is
stable over several on/off light cycles, and upon constant illumination (Figure 6.18).
189
The photoresponse of the flexible detector was further characterized as a function of light
irradiance. Fitting these data to a power law function allows important information to be
extracted on device dynamics. Devices with power law exponents 1 indicates the presence
of traps at key locations in the device, either at interfaces or in the bulk transport layer.[28]
Devices fabricated directly on ITO have a low current/power ratio, exhibiting a J ph ∼ P0.54
and thus possess a significant density of trap states. Devices with a SnO2 interlayer, on the
other hand, have near unity power law exponent (J ph ∼ P0.96 ) indicating the passivation of
traps at the ITO/Sb2 Se3 NR interface. Consequently, the device exhibits a linear dynamic range
(LDR) over 3 orders of magnitude. These data highlight that the Sb2 Se3 NRs themselves are
highly crystalline with few defects - otherwise current would scale sub-linearly regardless of
interface passivation. The flexible device shows respectable R and D ∗ values, up to 0.54 mA
W−1 and 1.37 1010 Jones, respectively for 0.115 mW cm−2 irradiance. The -3dB point at 118 Hz,
Figure 6.7f, is also adequate. However, compared to other Sb2 Se3 PDs reported in the literature
(Table 6.1), devices built on transferred Sb2 Se3 NR layers presented an average to low R and
D ∗ . Although the transfer method may still be optimized, this is a satisfactory preliminary
performance considering that Sb2 Se3 NR layer is not directly deposited on the substrate. These
preliminary results are excellent for simple, and potentially commercial applications, which
inspired me to test the flexible PD as a heart rate monitor.
Table 6.1: Comparison of Sb2 Se3 photodetector performance.
Bias D∗ R Wavelength Irradiance

Detector Ref.
(V) (×1010 Jones) (mA W−1 ) (nm) (mW cm−2 )
PET/ITO/SnO2 /
This
Sb2 Se3 NR/PEDOT 0 1.4 0.54 625 0.115
work
(transferred)
0 24.5 78 905 5
Au/ZnO/Sb2 Se3 /Au [29]
0 5.43 11 520 5
mica/Sb2 Se3 /Au
5 8.58 155 940 0.19
(epitaxy) [27]
Mica/Sb2 Se3 /Au
5 8.10 141 940 0.19
(non-epitaxy)
TiO2 /Sb2 Se3 0 0.331 0.13 400-750 100 [30]
ZnO:B/Sb2 Se3 0 22.5 173 625 10 [31]
190
10-4
10-5
Jph (A cm-2) 10-6
10-7
10-8
10-9
0 100 200 300 400
Time (s)
Figure 6.18: Characterization of flexible photodetector (PET/ITO/Sb2 Se3 NR) at 0V showing the stabil-
ity over long light pulse. A light emitting diode (LED), 625 nm at 115 mW cm−2 irradiation was used.
455nm
530nm
625nm
Normalized current
730nm
850nm
5s
Figure 6.19: Normalized photocurrent response for flexible photodetector (PET/ITO/SnO2 /Sb2 Se3 NR)
under illumination with light emitting diodes (LED) with wavelength varying from 455 nm to 850 nm
at 10 W cm−2 irradiation.
191
Figure 6.20: a) Scheme highlighting the light transmitted through the biological tissue as the blood
volume in the arteries change; b) absorption spectrum of hemoglobin in its oxygenated (HbO2 ) form;
c) flexible Sb2 Se3 NR heart rate detector wrapped around the finger; d) signal from pulse measurements
and c) fast Fourier transform (FFT) of the pulse signal. A red light emitting diode (LED), 625 nm at
100 W cm−2 irradiation was used as the illumination source.
Flexible PDs may be used to fabricate wearable heart rate monitors.[21, 32] Figure 6.20a
illustrates the working principle of a heart rate photosensor.[33] When biological tissue is
illuminated by a light source, most of the light is scattered or absorbed in the blood and other
tissues. The amount of light absorbed in arterial blood varies during the cardiac cycle due to
changes in the blood volume in the arteries. Light can be easily transmitted through biological
tissue at body extremities (e.g., fingertips or earlobes) and sensed by the PD. Transmission is
minimum when the heart contracts - pumping blood to the arteries (systole), and maximum
when the heart expands - pulling blood from arteries (diastole). As shown in Figure 6.20b,
hemoglobin in the blood absorbs light in the visible and near infrared range, which is well
192
within the Sb2 Se3 absorption range.[34] Furthermore, as the flexible detector performs well
within the 455 - 730 nm spectral range (Figure 6.19), it is suitable to probe the variation of
blood volume during the cardiac cycle (pulse).
To probe the pulse, a red LED was used as the illumination source. Figure 6.20c shows
the flexible PD wrapped around an index finger, the percentage of red light that is transmitted
through the finger is clearly visible in the image. Figure 6.20d shows the detector response
over time when placed over an individual’s finger. The detector senses the transmitted light,
and the electrical current output is measured as a pulse signal that is plotted over time. The
signal is maximum during diastole when transmitted light is maximum (low blood volume
in the arteries) and minimum during systole when transmitted light is minimum (high blood
volume in the arteries). By a simple count of the peaks up to 10 s, the initial pulse can be
estimated at ≈48 beats per minute (bpm), which will update after each subsequent pulse
(peak in photocurrent). Figure 6.20e shows the fast Fourier transform (FFT) of the signal,
with a clear peak with an easily distinguishable signal to noise ratio at the characteristic
frequency of 0.8125 Hz or 48.75 bpm. The detector also showed excellent performance under
infrared light (730 nm) illumination (Figure 6.21). These results clearly show that the robust
and flexible PD, integrated with a low energy red or infrared light source, may be used to
accurately monitor heart (pulse) rate.
730 nm
1 nA
0 10 20
Time (s)
Figure 6.21: Heart rate measured under infrared (730 nm) light using light emitting diode (LED), at
100 mW cm−2 irradiation.
193
6.4 Conclusion
A template growth method combined with an optimized polymer assisted transfer tech-
nique has been demonstrated as an effective way to obtain highly (001) oriented Sb2 Se3 NR
layers on various substrates. The exceptionally high (001) orientation is preserved after the
transfer process to the receiving substrate. To demonstrate the applicability of this pro-
cess to optoelectronic devices, PDs were fabricated, which showed excellent preliminary
performance. Optimized transfer and PMMA back-etching processes yielded transferred
FTO/SnO2 /Sb2 Se3 NR devices that performed as well as the control device (built directly on
the template, FTO/ZnONW/Sb2 Se3 NR) with < 10% current losses. The methods developed
here are robust and remove the constraints previously presented by direct deposition on the
desired substrate. Finally, to demonstrate the versatility of the method and of the Sb2 Se3 NR
films, flexible PD, and a proof-of-concept flexible heart rate monitor, was demonstrated by
transferring Sb2 Se3 NR onto a PET/ITO substrate. The ability to transfer delicate (001)-Sb2 Se3
nanorod thin films onto any substrate, preserving morphology and orientation, represents
an important development towards other innovative device architectures. It is expected that
the methodology outlined here may be used to fabricate highly efficient solar cells (especially
those in substrate device configurations) as well as photoelectrochemical cells, and micro bat-
teries.
6.5 References
[2] R. Kondrotas, J. Zhang, C. Wang, J. Tang, Sol. Energy Mater. Sol. Cells 2019, 199, 16.
[4] A. Carlson, A. M. Bowen, Y. Huang, R. G. Nuzzo, J. A. Rogers, Adv. Mater. 2012, 24,
5284.
[5] J. A. Rogers, R. G. Nuzzo, Mater. Today 2005, 8, 50.
[6] E. J. Smythe, M. D. Dickey, G. M. Whitesides, F. Capasso, ACS Nano 2009, 3, 59.
[7] Y. Xia, J. A. Rogers, K. E. Paul, G. M. Whitesides, Chem. Rev. 1999, 99, 1823.
[8] Y. Xia, G. M. Whitesides, Annu. Rev. Mater. Sci. 1998, 28, 153.
[9] Y. Xia, G. M. Whitesides, Angew. Chem. Int. Ed. 1998, 37, 550.
[10] M. Zhao, J. Su, Y. Zhao, P. Luo, F. Wang, W. Han, Y. Li, X. Zu, L. Qiao, T. Zhai, Adv.
Funct. Mater. 2020, 30, 1909849.
[11] P. V. Pham, S. C. Bodepudi, K. Shehzad, Y. Liu, Y. Xu, B. Yu, X. Duan, Chem Rev 2022,
122, 6514.
194
[12] Y. Zhang, B. Kim, Y. Gao, D. S. Wie, C. H. Lee, B. Xu, Int. J. Solids Struct. 2019, 180-181,
30.
[13] E. Della Gaspera, D. F. Kennedy, J. van Embden, A. S. R. Chesman, T. R. Gengenbach,
K. Weber, J. J. Jasieniak, Adv. Funct. Mater. 2015, 25, 7263.
[15] M. B. Costa, F. W. S. Lucas, M. Medina, L. H. Mascaro, ACS Appl. Energy Mater. 2020,
3, 9799.
[16] F. Garbassi, Surf. Interface Anal. 1980, 2, 165.
[17] Y. Chen, Y. Wang, R. Wang, X. Hu, J. Tao, G.-E. Weng, C. Zhao, S. Chen, Z. Zhu, J. Chu,
H. Akiyama, ACS Appl. Energy Mater. 2020, 3, 10415.
[18] Y. Fang, A. Armin, P. Meredith, J. Huang, Nat. Photonics 2019, 13, 1.
[19] V. Zardetto, T. M. Brown, A. Reale, A. Di Carlo, J. Polym. Sci. Part B: Polym. Phys. 2011,
49, 638.
[20] S.-I. Park, J.-H. Ahn, X. Feng, S. Wang, Y. Huang, J. A. Rogers, Adv. Funct. Mater. 2008,
18, 2673.
[21] C. Chen, K. Li, F. Li, B. Wu, P. Jiang, H. Wu, S. Lu, G. Tu, Z. Liu, J. Tang, ACS Photonics
2020, 7, 352.
[22] G. Chen, W. Wang, C. Wang, T. Ding, Q. Yang, Adv. Sci. 2015, 2, 1500109.
[23] K. Li, F. Li, C. Chen, P. Jiang, S. Lu, S. Wang, Y. Lu, G. Tu, J. Guo, L. Shui, Z. Liu, B.
Song, J. Tang, Nano Energy 2021, 86, 106101.
[24] K. Li, X. Yang, F. Yang, J. He, G. Zhang, S. Jiang, C. Chen, J. Tang, Materials Today
Electronics 2022, 2, 100011.
[25] X. Liang, Y. Feng, W. Dang, H. Huang, X. Wang, Y. Guo, K. Shen, R. E. I. Schropp, Z.
Li, Y. Mai, ACS Energy Lett. 2023, 8, 213.
[26] Y.-C. Shen, C.-Y. Lee, H.-H. Wang, M.-H. Kao, P.-C. Hou, Y.-Y. Chen, W.-H. Huang,
C.-H. Shen, Y.-L. Chueh, ACS Nano 2023, 17, 2019.
[27] X. Wen, Z. Lu, L. Valdman, G. C. Wang, M. Washington, T. M. Lu, ACS Appl. Mater.
Interfaces 2020, 12, 35222.
[28] S. Farooq, T. Feeney, J. O. Mendes, V. Krishnamurthi, S. Walia, E. Della Gaspera, J. van
Embden, Adv. Funct. Mater. 2021, 31, 2104788.
[29] J. Jiang, Y. Guo, X. Weng, F. Long, Y. Xin, Y. Lu, Z. Ye, S. Ruan, Y.-J. Zeng, J. Mater.
Chem. C 2021, 9, 4978.
[30] J. Wang, S. Li, T. Wang, F. Guan, L. Zhao, L. Li, J. Zhang, G. Qiao, ACS Appl. Mater.
Interfaces 2020, 12, 38341.
[32] W. Wu, L. Wang, Y. Yang, W. Du, W. Ji, Z. Fang, X. Hou, Q. Wu, C. Zhang, L. Li, Biosens.
Bioelectron. 2022, 210, 114328.
[33] C. M. Lochner, Y. Khan, A. Pierre, A. C. Arias, Nat. Commun. 2014, 5, 5745.
[34] S. Prahl, Tabulated Molar Extinction Coefficient for Hemoglobin in Water, https://
omlc.org/spectra/hemoglobin/summary.html, accessed: May, 2023.
195
Chapter 7
Conclusions and
perspectives
196
The potential of Sb2 Se3 as a light absorbing material has been comprehensively investi-
gated in this thesis considering the particularities of its quasi-1D crystal structure.
First, the deposition of key HRMO (ZnO or TiO2 ) was shown to significantly improve
device overall performance. The ZnO layer proved to be the most optimal, leading to reduced
interfacial trap densities, an improved device front contact, and overall increase in solar cell
efficiency and stability. Meanwhile, TiO2 deposition creates a high electron barrier and, as
such, it should be avoided as an HRMO interlayer for superstrate CdS/Sb2 Se3 devices.
Second, exceptionally highly (001) oriented Sb2 Se3 films were obtained via substrate
nanostructure engineering. This is the first time that substrate physical structure is demon-
strated to direct Sb2 Se3 preferential orientation. A trapping mechanism was proposed to ex-
plain the selection of (001) oriented crystals, while (hk0) and low tilt (hk1) crystals are ‘filtered
out’ within substrate’s nanowire array during early stages of film growth. Importantly, the
growth of (001) oriented Sb2 Se3 films on highly nanostructured substrates was demonstrated
to be independent of substrate surface chemistry.
Finally, the template growth of (001) oriented Sb2 Se3 NR layers was proposed as a viable
strategy for device fabrication. A polymer assisted transferring method was developed and
the (001) Sb2 Se3 transferable layers were used to fabricate functional photodetectors on FTO,
FTO/SnO2 and FTO/CdS substrates. After initial tests, a flexible heart rate detector was
fabricated into PET/ITO/SnO2 /transferred-Sb2 Se3 NR architecture.
The work presented in this thesis constitutes several important and novel advances in the
development of antimony chalcogenide-based applications, and the fabrication of thin films.
It is safe to conclude that antimony chalcogenide devices will soon become competitive with
emerging PV technologies that have higher efficiencies such as organic and quantum dot cells.
Upon improvement, large scale application of antimony chalcogenides will become practical
in tandem solar cells, the internet of things, and indoor photovoltaic devices. Furthermore,
solar to fuel photocatalytic energy conversation (water splitting and CO2 conversion) and
battery electrodes are other promising uses of Sb2 Se3 thin films and coatings. The fundamen-
tal studies presented here have provided insights for improving the application of quasi-1D
materials in all these areas.
197
Several studies point out directions to improve Sb2 Se3 device efficiency.[1–4] In future
studies, one may also consider the following suggestions based on the findings presented in
this thesis. Other Zn based materials may be explored as interlayers in TCO/(interlayer)/CdS/
Sb2 Se3 solar cells, such as Zn(O,S),[5, 6] Zn(Mg,S),[7] Zn(Sn,O),[8], Cd(Zn,S),[9] ZnSe,[10] and
also In2 S3 ,[11] which may also improve band aliment or direct the preferred (001) orientation.
The orientation filtering strategy proposed in chapter 5 should be considered for depositing
other quasi-1D layers (Sb2 S3 and Sb2 (S, Se3 )3 ). The effect of substrate nanostructure on the
deposition of quasi-2D layers may also be further explored. Conversely, one may explore the
oriented growth of low-dimensional materials on textured Si for tandem solar cells. Moreover,
other device architectures may be explored to construct devices on nanostructured Sb2 Se3 lay-
ers, i.e., substrate devices. Regarding the growth quasi-1D materials, I would also suggest
exploring the epitaxial growth of Sb2 Se3 on Sb2 S3 layers, and vice-versa; these materials have
an almost perfect crystallographic lattice match and the stacking of layers may be engineered
to achieve optimal band alignment.[12, 13] Finally, post selenization treatments have the po-
tential to drastically improve the quality of Sb2 Se3 NR layers fabricated here. This process
will help to reduce deep defects by adjusting the Se chemical concentration[14] and, equally
important, to promote grain growth and flatten the layer.[15, 16]
7.1 References
[1] S. Barthwal, R. Kumar, S. Pathak, ACS Appl. Energy Mater. 2022, 5, 6545.
[2] C. Chen, K. Li, J. Tang, Sol. RRL 2022, 6, 2200094.
[4] J. Dong, Y. Liu, Z. Wang, Y. Zhang, Nano Sel. 2021, 2, 1818.
[5] Y. Zhao, C. Li, J. Niu, Z. Zhi, G. Chen, J. Gong, J. Li, X. Xiao, J. Mater. Chem. A 2021, 9,
12644.
[6] Y. Zhao, S. Wang, C. Jiang, C. Li, P. Xiao, R. Tang, J. Gong, G. Chen, T. Chen, J. Li, X.
Xiao, Adv. Energy Mater. 2021, 2103015.
[7] R. E. Treharne, A. Clayton, L. J. Phillips, J. D. Major, S. J. C. Irvine, K. Durose in 2014
IEEE 40th Photovoltaic Specialist Conference (PVSC), 2014, pp. 1722–1725.
[8] J. D. Major, K. Durose in 2013 IEEE 39th Photovoltaic Specialists Conference (PVSC),
2013, pp. 1146–1149.
[9] G. Li, Z. Li, X. Liang, C. Guo, K. Shen, Y. Mai, ACS Appl. Mater. Interfaces 2019, 11, 828.
[10] R. Kumari, M. Mamta, R. Kumar, Y. Singh, V. N. Singh, ACS Omega 2023, 8, 1632.
[11] J. van Embden, J. O. Mendes, J. J. Jasieniak, A. S. R. Chesman, E. Della Gaspera, ACS
Appl. Energy Mater. 2020, 3, 7885.
198
[12] K. Li, Y. Lu, X. Ke, S. Li, S. Lu, C. Wang, S. Wang, C. Chen, J. Tang, Sol. RRL 2020, 4,
2000220.
[13] Y. Zhang, J. Li, G. Jiang, W. Liu, S. Yang, C. Zhu, T. Chen, Sol. RRL 2017, 1, 1700017.
[14] M. Leng, M. Luo, C. Chen, S. Qin, J. Chen, J. Zhong, J. Tang, Appl. Phys. Lett. 2014, 105,
083905.
[16] G. Liang, M. Chen, M. Ishaq, X. Li, R. Tang, Z. Zheng, Z. Su, P. Fan, X. Zhang, S. Chen,
Adv. Sci. 2022, 9, 2105142.
199

s3688078 Mendes J Thesis

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

s3688078 Mendes J Thesis

Uploaded by

Copyright:

Available Formats

What man sees depends both upon what he looks at and also upon

what his previous visual-conception experience has taught him to

- Thomas Kuhn, The Structure of Scientific Revolutions

procedures and guidelines have been followed.

partner institution responsible for the joint-award of this degree.

with the copyright holder(s) of those works.

granted by the University to restrict access for a period of time.

Joao Otavio de Sousa Mendes

all those who have helped me in the completion of this work.

on the task at hand during challenging times.

my colleagues in the laboratory for helpful discussion and fun moments.

been possible for me to complete this step of the journey.

Table of contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

List of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii

List of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii

List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiv

Broad context and scope of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1 Electronic structure and modelling of solar cells 8

1.2 Electronic structure of solid materials . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.2.1 Energy levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.2.2 Energy band and band gap . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.2.3 Metals, semiconductors and insulators . . . . . . . . . . . . . . . . . . . . 14

1.2.4 Density of states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.2.5 Density of states in semiconductors . . . . . . . . . . . . . . . . . . . . . . 18

1.2.6 Fermi-Dirac statistics and Fermi Level . . . . . . . . . . . . . . . . . . . . . 19

1.2.7 Carrier density and effective density of states . . . . . . . . . . . . . . . . 20

1.2.8 Shallow and deep energy levels . . . . . . . . . . . . . . . . . . . . . . . . 22

1.2.9 Intrinsic, n-type and p-type semiconductors . . . . . . . . . . . . . . . . . 23

1.3 Carrier generation, recombination and transport . . . . . . . . . . . . . . . . . . . 24

1.3.1 Generation and recombination mechanisms . . . . . . . . . . . . . . . . . 24

1.3.2 Regions of recombinations . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

1.3.3 Continuity equation and diffusion length . . . . . . . . . . . . . . . . . . . 27

1.3.4 Carrier transport in semiconductors . . . . . . . . . . . . . . . . . . . . . . 29

1.4 The pn junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

1.4.1 Electric field, potential and built-in voltage . . . . . . . . . . . . . . . . . . 31

1.4.3 Junction under forward and reverse bias . . . . . . . . . . . . . . . . . . . 34

1.5 Thin film solar cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1.5.1 Physical structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1.5.2 Band structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

1.6 The current density vs voltage (JV) curve . . . . . . . . . . . . . . . . . . . . . . . 38

1.6.1 Measuring solar cell efficiency . . . . . . . . . . . . . . . . . . . . . . . . . 38

1.6.2 Diode parameters and parasitic resistances . . . . . . . . . . . . . . . . . . 40

1.6.3 Dark JV curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

1.6.4 Graphical analysis of JV curves . . . . . . . . . . . . . . . . . . . . . . . . . 43

1.6.5 Modelling complex JV curves . . . . . . . . . . . . . . . . . . . . . . . . . . 45

1.7 Electrochemical impedance spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 47

1.7.1 Equivalent circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

1.7.2 Built-in potential and doping density from EIS data . . . . . . . . . . . . 50

1.8 External quantum efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

2 Review of Sb2 Se3 properties and thin film fabrication methods 54

2.2 Crystal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

2.3 Ribbon tilt and crystallographic orientation . . . . . . . . . . . . . . . . . . . . . . 60

2.4 Sb2 Se3 thin film deposition methods . . . . . . . . . . . . . . . . . . . . . . . . . . 61

2.4.1 Solution processed Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . . 61

2.4.2 Thermal evaporated Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . 62

2.4.3 Vapour transport deposition of Sb2 Se3 thin films . . . . . . . . . . . . . . 63

2.4.4 Other Sb2 Se3 thin film devices . . . . . . . . . . . . . . . . . . . . . . . . . 64

2.5 Growth of oriented Sb2 Se3 thin films . . . . . . . . . . . . . . . . . . . . . . . . . . 65

2.5.1 Epitaxy growth and (hk0) Van der Waals layers . . . . . . . . . . . . . . . 65

2.5.2 Evolution selection principle and (hk1) growth . . . . . . . . . . . . . . . . 67

2.5.3 Seed layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

2.5.4 Substrate chemistry and evaporation parameters . . . . . . . . . . . . . . 69

2.5.6 (001) orientated Sb2 Se3 films . . . . . . . . . . . . . . . . . . . . . . . . . . 72