Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Nanostructured III-V photoelectrodes for high performance, high durability solar water splitting
(USC Thesis Other)
Nanostructured III-V photoelectrodes for high performance, high durability solar water splitting
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Nanostructured III-V Photoelectrodes For
High Performance, High Durability Solar Water Splitting
By
Haneol Lim
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)
August 2019
Copyright 2019 Haneol Lim
i
To my beloved wife, Sohyeon
ii
Acknowledgement
First of all, I sincerely would like to expression the deepest appreciation to my advisor,
Professor Jongseung Yoon. He always has kindly supervised me and my research works all the
time and has generously gave me more opportunities as well as scientific knowledge to excel
myself to build multifaceted skills and to improve problem-solving ability. Under his guidance, I
have been able to acquire a lot of skills including critical thinking, self-learning, creative thinking,
self-driving motivation, and sense of duty. Moreover, he has gave me insightful comments,
suggestions, and encouragement which have been an enormous help to me. All his contributions
to my PhD life are invaluable for my future professional career.
I would also gratefully appreciate my former, current colleagues, and friends, Dr. Sung-
Min Lee, Dr. Dongseok Kang, Dr. Hyungwoo Choi, Dr. Boju Gai, Huandong Chen, Lauren Liaw,
Qian Lang Shen, Moon Chul Jung, Qian Yi, Yuzhou Xi, and Sang Bum Kim. I am grateful for
their assistance and supports in my research, and it has been an incredibly enjoyable and exciting
time for me with them. I am also deeply thankful to unforgettable collaborators Dr. James L.
Young, Dr. Todd G. Deutsch, Dr. John F. Geisz, Dr. Daniel J. Friedman, and Chase Aldridge in
National Renewable Energy Laboratory and Professor M. L. Lee and Yookun Sun at the University
of Illinois Urbana-Champaign for their valuable supports, suggestions, and comments. In addition,
I would like to give great thanks to Dr. Yuanrui Li and Hao Yang in Professor Wei Wu’s group
for the great collaborations and Dr. Fanqu Wu and Dr. Qingzhou Liu in Professor Chongwu Zhou’s
group for their experimental supports. Advice given by Dr. Nina Hong and James Hilfiker at
J.A.Woollam Co. has been also a great help in optical calculation of Ellipsometry analysis.
I would also like to offer my special thanks to my dissertation committees, Professor Wei
Wu and Professor Chongwu Zhou for their time to guide me and for giving me favorable comments
iii
to share their idea. I would like to express my gratitude to Donghai Zhu and Alfonso Jimenez for
their hard works to maintain all the equipment in a good condition at Keck Photonics Laboratory.
I would like to thank John Curulli and Dr. Andrew Clough for their help with XPS at Center for
Electron Microscope and Microanalysis at USC. In addition, I am grateful to Prof. Willie Ng for
helping me to use the Ellipsometer.
Last but not least, I would like to give my heartfelt appreciation and greatest gratitude to
my beloved wife, Sohyeon Lim. Without her endless encouragement, continued support, and
sincere comments, I could not pursue my PhD degree. She is a wonderful woman giving me best
of luck, and I would deeply appreciate her for being my better half.
iv
Table of Contents
Acknowledgement ………………………………………………………………………………………………. ii
Abstract …………………………………………………………………………………………….…..………… vi
List of Figures ……………………………………………………………………………………………….… viii
List of Tables ………………………………………………………………………………..………………… xvii
Chapter 1. Background ................................................................................................................ 1
1. 1. Storable renewable energy ...................................................................................... 1
1. 2. Electrolysis of water ............................................................................................... 7
1. 3. Solar water splitting ................................................................................................ 9
Chapter 2. Methods .................................................................................................................... 17
2. 1. Fabrication of monofacial nanostructured GaInP2 photocathodes........................ 17
2. 1. 1. Growth of GaInP2 by metal organic vapor phase epitaxy ............................ 17
2. 1. 2. Electroless deposition of silver nanoparticles .............................................. 18
2. 1. 3. Dry-etching by inductively coupled plasma-reactive ion etching ................. 20
2. 1. 4. Surface passivation by ammonium sulfide solution ...................................... 22
2. 1. 5. Deposition of amorphous molybdenum disulfide .......................................... 23
2. 2. Fabrication of bifacial ultrathin GaAs photocathodes .......................................... 24
2. 2. 1. Growth of ultrathin GaAs by molecular-beam epitaxy ................................. 24
2. 2. 2. Nanofabrication of hexagonally periodic TiO2 nanoposts ........................... 25
2. 2. 3. Fabrication of printed-assembly of bifacial GaAs photocathodes ............... 27
2. 3. Optical modeling and characterization ................................................................. 31
2. 3. 1. FDTD optical modeling for monofacial GaInP2 photocathodes .................. 31
2. 3. 2. FDTD optical modeling for bifacial GaAs photocathodes ........................... 32
2. 3. 3. UV-Vis, Ellipsometry, and Photoluminescence spectroscopy ...................... 33
2. 4. Photoelectrochemical and photovoltaic measurement .......................................... 36
v
2. 5. IPCE measurement................................................................................................ 37
2. 6. X-ray photoelectron spectroscopy ........................................................................ 39
2. 7. Faradaic efficiency measurement ......................................................................... 39
Chapter 3. Monofacial Nanostructured GaInP2 Photocathodes ............................................ 41
3. 1. Introduction ........................................................................................................... 41
3. 2. Fabrication ............................................................................................................ 46
3. 3. Results and discussion .......................................................................................... 49
3. 4. Conclusion ............................................................................................................ 69
Chapter 4. Bifacially Optimized, Nanostructured Ultrathin GaAs Photocathodes.............. 71
4. 1. Introduction ........................................................................................................... 71
4. 2. Fabrication ............................................................................................................ 79
4. 3. Results and Discussion ......................................................................................... 83
4. 4. Conclusion .......................................................................................................... 105
Chapter 5. Future works .......................................................................................................... 107
5. 1. Nanostructured, pn-junction GaInP2 photocathode ........................................... 107
5. 2. Multilayer-grown nanostructured, ultrathin GaAs photoanodes ........................ 113
5. 3. Conclusion .......................................................................................................... 115
References .................................................................................................................................. 116
vi
Abstract
High performance and high durability of photoelectrodes are the most critical factors to
realize practical utilization of III-V compound semiconductors in solar water splitting. Here we
analytically departmentalized such several factors and proposed different approaches to develop
those factors by categorizing the form factors of photoelectrodes into two configurations; one is
monofacial photoelectrodes, and another is bifacial photoelectrodes. In each category, unique and
noble strategies have been suggested to improve both photoelectrochemical performance and
durability, which are the first priority aims in this dissertation.
In the first category, we present a novel strategy for III-V monofacial photocatalysis that
can provide significant enhancement of catalytic efficiency through suppression of front-surface
reflection but also dramatically augment electrochemical durability in solar water splitting. The
density-graded surface of nanostructured GaInP2, when synergistically combined with sulfur
passivation, effectively suppressed the front-surface reflection without compromising the
collection and transport of photogenerated carriers via non-radiative surface recombination,
leading to the significantly increased electrode efficiency. The nanostructured, sulfur-passivated
GaInP2 also provided superior electrochemical durability compared to the bare, unmodified
GaInP2 due to the synergistic effects of enlarged surface area for reduced local current density and
incorporation of sulfate bonding on the electrode surface that can serve as excellent protective
coating. The resulting nanostructured GaInP2 permitted a ~4.3 % electrode efficiency in a three-
electrode configuration, and undiminished electrode performance of ~120 hours in hydrogen
evolution reaction of solar water splitting. As a future work, ultrathin GaAs photocathode and
photocathode cooperated with TiO2 nanostructures and bifacial design will be performed.
vii
In the second category, we focus on materials-cost, lifetime, light-management, and
charge-transfer and present a design of GaAs photocathodes for the hydrogen-evolution reaction
(HER) that can address all of these difficulties in conjunction with bifacially engineered electrode
configuration, nanoscale photon management, and multilayered composite protection medium.
Ultrathin GaAs photocathodes whose active layer is ~10 thinner than conventional devices were
integrated on a glass substrate by transfer printing to form a bifacial GaAs photocathode with
decoupled optical and catalytic interfaces. Titanium dioxide (TiO2) diffractive nanostructure
effectively increases absorption of ultrathin GaAs photocathodes to yield comparable performance
to wafer-based electrodes, while multifunctional protection layers composed of ohmic metals,
TiO2, and HER co-catalysts significantly improves the charge transfer efficiency to yield
photovoltaic-level fill-factor and current density but also augments the lifetime. By incorporating
all the advances reported here, we anticipate that all represent important challenges can be
overcome before realizing practical utilization of III-V compound semiconductors in high
performance, low cost solar water splitting.
viii
List of Figures
Figure 1. 1. (a) Carbon dioxide level in the last 650,000 years, with the abrupt end of the last ice age about 7,000 years
ago marking the beginning of the modern climate era and of human civilization. Source : NASA Global Climate
Change (https://climate.nasa.gov/evidence). (b) U.S. greenhouse gas emission in 2017. Source: U.S. Environmental
Protection Agency (2019). Inventory of U.S. Greenhouse Gas Emissions and Sinks 1990-2017. (c) The production
and absorption of extra CO 2, between 1900 and 2016, and it is evident human-made CO 2 has been increasing rapidly,
which has been mostly absorbed by atmosphere among planetary systems. Source: IPCC, “Climate Change 2014:
Mitigation of Climate Change” ..................................................................................................................................... 1
Figure 1. 2. (a) The greenhouse effect by the absorption of outgoing heat by greenhouse gas molecules. (b) The change
in global surface temperature relative to 1951-1980 average temperatures. (c) Satellite observations of Arctic sea ice
in 1984 and 2016. The declining rate of Arctic sea ice is now 12.8 percent per decade. (d) Sea height variation in mm
from 1993 until present with 90 mm of rise. The rate of change is +3.3 mm per year. Source : NASA Global Climate
Change (https://climate.nasa.gov/evidence). ................................................................................................................. 2
Figure 1. 3. Renewable energy sources (a) solar energy, (b) hydroelectric power, (c) wind power, (d) ocean energy,
(e) biofuel, (f) hydrogen gas. Source: National Renewable Energy Laboratory website
(https://www.nrel.gov/workingwithus/learning.html) ................................................................................................... 3
Figure 1. 4. (a) Generation and demand of solar and wind electricity power with the mismatch between the peak times
of solar supply and grid demand, which motivates a need to store the excess amount of solar power during the peak
time. Source : QIC (https://www.qic.com.au) (b) A satellite image of hurricane Maria (Puerto Rico blackout in
September 2017) which happened in the east of the Florida east coast, captured by NASA, (c) scattered solar paneled
by hurricane Maria. Source: CNN Business. Ahiza Garcia, March 6, 2019. ................................................................. 4
Figure 1. 5. (a) Comparisons of energy densities in terms of energy per mass between lithium-ion (Li-ion) battery,
hydrogen gas, and general fossil fuels (gasoline and diesel). (b) A comparison of power source between representative
electricity car (Tesla Model S) and fuel-cell car (Toyota Mirai fuel). (c) Illustration of lithium-ion battery packs at the
bottom of Tesla Model S. Source: Tesla (d) Illustration of hydrogen tanks in Toyota Mirai. Source : Toyota ............ 5
Figure 1. 6. (a) A factory for natural gas reforming. Source : Office of energy efficiency and renewable energy
(https://www.energy.gov/eere/fuelcells/) (b) A cycle of renewable hydrogen produced by an electrolysis of water with
renewable energies and fuel cells. Source: Hydroville (http://www.hydroville.be/) ..................................................... 6
Figure 1. 7. Fundamentals of electrochemical water splitting. Copyright (2017) Nature Reviews Chemistry. ............. 7
Figure 1. 8. (a) HER volcano plot for various metal and MoS 2 (b) TOF avg plots with linear sweep voltammograms of
various HER catalysts. (c) OER volcano plot for metal oxides. Copyrights (2017) Science. (d) Comprehensive plots
of catalytic activity, stability, and electrochemically active surface area for OER electrocatalysts in acidic (top) and
alkaline (bottom) solutions. Copyright (2013) Journal of the American Chemical Society. ......................................... 8
Figure 1. 9. (a) A simplified schematic illustration of the band diagram of a semiconductor photoelectrode and the
standard redox potential of hydrogen and oxygen. Copyright (2010) Chemical Reviews. (b) Bandgap energies and
band-edge positions of various semiconductor photoelectrodes relative to the NHE and the vacuum level at pH = 0.
..................................................................................................................................................................................... 10
Figure 1. 10. (a) Fraction of Shockley-Queisser detailed-balance limit for voltage and current achieved by record cells.
Copyright (2016) Science. (b) Bandgap energy and lattice constant of various III-V compound semiconductors. (Tien,
1988) ............................................................................................................................................................................ 11
Figure 1. 11. The comparison of photoelectrochemical stability (J-t) of (a) GaInP 2 photocathodes and (b) Si
photocathodes covered by Mo/MoS 2 as a protection layer and co-catalyst. Copyrights (2016) The Journal of Physical
Chemistry Letters and (2017) ACS Applied Materials & Interfaces. ........................................................................... 12
ix
Figure 1. 12. Photoelectrochemical current density (J-E) of various nanostructured photocathodes. (a) Si photocathode.
Copyright (2011) Energy & Environmental Science. (b) InP photocathodes. Copyright (2012) Angewandte Chemie.
(c) InGaN/Si photocathodes. Copyright (2015) Nano Letters. (d) ZnO/Si photocathode. Copyright (2012)
Nanotechnology. .......................................................................................................................................................... 13
Figure 1. 13. (a) Energetic positions of conduction band edge, valence band edge, and Fermi level (or quasi Fermi
levels) of photoanode in pre-equilibrium state and in equilibrium states without and with illumination. (b) Schematic
relating bandgaps and photovoltages for a tandem absorber. (c) Photovoltage benchmarks for PEC and PV materials
as a function of optical bandgap. Copyrights (2017) Current Opinion in Electrochemistry. ...................................... 14
Figure 1. 14. (a) Mott-Schottky plots of a p-type GaInP, sample at pH values of 2.0, 4.8, 9.1 and 12.0 (all at 5 kHz).
Copyright (1994) Journal of Electroanalytical Chemistry. (b) Conceptual band alignment and bending model for p-
GaInP and n/p-GaInP in acidic electrolyte. Copyright (2017) Nature Energy. (c) Conceptual band alignments of
monolithic tandem photoelectrode (GaAs/GaInP). Copyright (1998) Science. (d) Photocurrent vs. potential (J-E) of
surface modified GaInP photocathode by conjugate molecules. Copyright (2015) ACS Applied Materials & Interfaces.
..................................................................................................................................................................................... 15
Figure 2. 1. A schematic illustration of fabrication procedures for surface-tailored ‘black’ GaInP 2 photocathodes. .. 17
Figure 2. 2. (a) A cross-sectional SEM image of p-GaInP 2 grown on GaAs wafer including GaAs buffer layer. (b)
Energy gap (E g) of GaP, InP, GaAs, and InAs vs. lattice constant. Copyright (1981) Journal of Crystal Growth. .... 18
Figure 2. 3. Top-view SEM images of silver nanoparticles deposited for 1 min (a) directly without any pre-treatment,
(b) with UV-ozone pre-treatment (5 min), (c) with NH 4OH (1:10) pre-treatment (1 min), and (d) NH 4OH (1:10) pre-
treatment (1 min) followed by perchloric acid pre-treatment (30 sec) ........................................................................ 19
Figure 2. 4. (a) A schematic illustration of inductively coupled plasma-reactive ion etching with BCl/N2 as a reactive
etching/carrier gas, respectively. (b) Cross-sectional and top-view SEM images of the nanostructured GaInP 2 at
etching times of 1, 2, 3, and 4 min after removing remaining Ag nanoparticles. (d) Tilt-view and top-view SEM images
of nanostructured GaInP 2 before and after etching of Ag nanoparticles. ..................................................................... 21
Figure 2. 5. (a) Oxygen- and/or defect-induced inter-band defect states in III-V compound semiconductors. (b) A
conceptual illustration of mechanism of sulfur passivation on III-V compound semiconductors. (c) Schematic
illustration of surface treatment by using two-step (NH 4) 2S passivation. .................................................................... 22
Figure 2. 6. (a) electroless photochemical deposition of MoS 2 under white light illumination and (b) electrochemical
deposition of MoS 2 in dark condition with three-electrode configuration. .................................................................. 23
Figure 2. 7. (a) An epitaxial design of ultrathin GaAs photocathodes. (b) A cross-sectional SEM image of epitaxially
grown ultrathin GaAs photocathode on GaAs substrate. ............................................................................................. 24
Figure 2. 8. A schematic illustration of nanofabrication of hexagonally periodic TiO 2 nanoposts on GaAs
photocathodes. ............................................................................................................................................................. 25
Figure 2. 9. Cross-sectional SEM images of imprinted polymer layer (a) before and (b) after O 2 RIE dry etch for 150
sec. (c) A photograph image of PTFE/NOA/PET stamp. (d) A top-view SEM image of chrome nanoarrays after lift-
off of SU-8 2000.5 layer using acetone. ...................................................................................................................... 26
Figure 2. 10. A schematic illustration of fabrication procedures for printed-assembly of nanostructured, ultrathin GaAs
photocathodes via transfer printing.............................................................................................................................. 28
Figure 2. 11. Optical microscopic images of GaAs photocathodes at every process step after forming (a) n-type ohmic
contact, (b) TiO 2 NPs and isolation, (c) after undercut etch, (d) after picking up to the first PDMS stamp, (e) after
transfer to the second PDMS stamp, (f) after printing, (g) after MESA etching, exposing p-type ohmic metal, and
forming n-type ohmic metal, and (h) after Pt deposition (i.e. completed GaAs photocathodes). ................................ 29
x
Figure 2. 12. (a) Top-view SEM images of nanostructured GaInP 2 (without (NH4)2S-treatment) at various etching
times. (b) Tilt-view schematic illustration of constructed nanostructured surfaces by 3D modeling software
(Rhinoceros® ) using the SEM images in (a). .............................................................................................................. 30
Figure 2. 13. (c) Cross-sectional schematic illustrations of nanostructured GaInP 2 for the FDTD calculation of
reflectance and absorption spectra in water. The inset shows a nanostructured GaInP 2 implemented in Lumerical
TM
.
(d) Measured refractive index (n) and extinction coefficient (k) of GaInP 2 by spectroscopic ellipsometry, which were
used in optical calculations. ......................................................................................................................................... 31
Figure 2. 14. (a) Cross-sectional schematic illustration of a nanostructured bifacial GaAs photocathodes printed on a
glass substrate. (b) Schematic illustrations of a model for FDTD calculation with dimensional information of TiO 2
NPs and a plane of cross section for P abs. (c) Refractive index (n) and (d) extinction coefficient (k) of various materials
used in FDTD optical calculation. ............................................................................................................................... 33
Figure 2. 15. A schematic illustration of a homemade optics system for micro UV-visible spectroscopy. The inset
images show photographs of the homemade optics system, light source, objective lens, USB-type spectrometer, and
light collimator (i.e. doublet lens). ............................................................................................................................... 34
Figure 2. 16. (a) A photograph of VASE Ellipsometer (J.A.Woollam) and (b) measured optical constants of GaInP 2
comparing with reported values. .................................................................................................................................. 35
Figure 2. 17. A schematic illustration of photoelectrochemical measurement setup for GaInP 2 and GaAs photocathode
with a full-spectrum solar simulator in three-electrode configuration. ........................................................................ 37
Figure 2. 18. Optical microscopic images of GaAs photocathodes with titanium aperture captured from backside for
the IPCE purpose. Two separate GaAs cells were visible at the (a) out-focused and (b) in-focused depth through the
aperture and a transparent glass. (c) An optical microscopic image of GaAs photocathodes captured from frontside
under back illumination, where the bright region was the transmitted light of the back illumination. (d) Schematic
illustration of IPCE measurements for Ti-aperture-confined GaAs photocathodes. .................................................... 38
Figure 3. 1. (a) Bandgap energy and lattice constant of various III-V semiconductors. (b) Spectral irradiance of the
AM1.5 Global spectrum and absorption spectrum of GaInP 2 (672 nm). Raw spectrum data was downloaded from PV
Education (https://www.pveducation.org) (c) Bandgap energies and band-edge positions of various semiconductor
photoelectrodes relative to the NHE and the vacuum level at pH = 0. Copyright (2017) Sustainable Energy & Fuels.
(d) Band-edge alignment of p-type GaInP 2 photocathode relative to hydrogen redox potential and oxygen redox
potential in non-equilibrium state ................................................................................................................................ 42
Figure 3. 2. (a) Schematic of the monolithic GaInP 2/GaAs tandem PEC cell with 12.4% of STH and (b) its equilibrium
energy band diagram. Copyright (1998) Science. (c) Schematic of the monolithic GaInP 2/GaInAs tandem PEC cell
with 16.2% of STH. Copyright (2017) Nature Energy. (d) Schematic of the monolithic GaInP 2/GaInAs tandem PEC
cell with 19.3% of STH. Copyright (2018) ACS Energy Letters. ................................................................................ 43
Figure 3. 3. (a) Multiple roles of catalytic interface from monofacial photoelectrodes in light absorption,
electrocatalysis, as well as and corrosion protection. Previous approaches of (a) nanostructures to reduce reflection
loss (Copyright (2011) Energy & Environmental Science and (2012) Angewandte Chemie), (c) modulation of energetic
states to improve carrier separation (Copyright (2015) Science and (2017) Nature Energy), and (d) the addition of
protective layers to prevent photocorrosion of semiconductors (Copyright (2014) Science and (2017) Nature Energy), .
..................................................................................................................................................................................... 44
Figure 3. 4. Schematic illustration, morphological and optical properties of surface-tailored GaInP 2 photocathodes. a
Schematic illustration of fabrication procedures for surface-tailored ‘black’ GaInP 2 photocathodes. ........................ 46
Figure 3. 5. (a) A top-view SEM image of electrolessly deposited silver nanoparticles on GaInP 2 as a mask for dry
etching (scale bar: 1 m). (b) Tilt-view scanning electron microscope (SEM) image of nanostructured p-type GaInP 2
photocathodes after the dry etching and before the (NH 4) 2S-treatment (scale bar: 500 nm). (c) The photographic
xi
images of fully-functional bare (labeled as “Bare”) and nanostructured (labeled as “NS”) GaInP 2 photocathodes
mounted on a slide glass with epoxy encapsulation (scale bar: 5 mm), where the ‘black’ surface of nanostructured
GaInP 2 is evidently shown in contrast to the shiny surface of bare GaInP 2. ................................................................ 47
Figure 3. 6. (a) Schematic illustration of two-step (NH 4) 2S treatment. (b) XPS spectra of S 2p and Ga 3s before
(NH 4) 2S-treatment (left), after drying at 85 C (center), and after heating at 250 C in air (right), supporting the
formation of sulfide and sulfate species at each step. The second XPS spectra of S 2p was measured after washing the
sample with DI water to remove a thick polysulfide that shown on (c). (c) Scanning electron microscope (SEM) image
and results of energy dispersive spectroscopy (EDS) from the sample right after the casting at 85 C without washing
in DI water. Thick islands of (NH 4) 2S-based polysulfides are clearly observed. EDS spectra indicate the formation of
oxides and carbon-containing species after this first step of (NH 4) 2S-treatment. Because of the penetration depth (2~3
mm) of highly-energized electron beam (10 kV), arsenic was detected from GaAs substrate, and the atomic percentage
of gallium was slightly higher than indium. ................................................................................................................ 48
Figure 3. 7. (a) Cross-sectional SEM images of the nanostructured GaInP2 (yet without (NH4)2S-treatment) at etching
times of 1, 2, 3, and 4 min (scale bar: 300 nm) and (b) corresponding etching depth of measured from SEM images.
Error bars represent the range of values obtained from three separate measurements (n = 3). .................................... 49
Figure 3. 8. (a) Corresponding total (i.e. specular and diffuse) reflectance spectra of nanostructured GaInP 2 measured
on spectrophotometer equipped with an integrating sphere at an incidence angle of 8° and calculated (dotted line)
reflectance spectra obtained from FDTD-based numerical optical modeling matched well with the experimental (solid
line) spectra. (b) Zoomed-in, total reflectance spectra below 5% shown in a. (c) Calculated absorption spectra and (d)
integrated solar flux absorption (S_abs) of nanostructured GaInP 2 in water calculated using the numerical model
established in a. ........................................................................................................................................................... 50
Figure 3. 9. Photoelectrochemical performance of nanostructured GaInP 2 photocathodes for the HER. Representative
J-E curves of bare and nanostructured GaInP2 photocathodes driving the hydrogen evolution reaction (HER),
measured under simulated AM1.5G solar illumination (1000 W/m
2
). All samples were measured without (NH 4) 2S-
treatment. ..................................................................................................................................................................... 52
Figure 3. 10. (a) Corresponding steady-state photoluminescence (PL) spectra of bare and nanostructured GaInP 2
photocathodes without (NH4)2S passivation. (b) Steady-state PL spectra of as-received bare GaInP 2 before and after
the dry-etching (4 min) without silver nanoparticles (i.e. still plane surface but dry-damaged) measured at room
temperature. The etching was performed without silver nanoparticles using the same condition (BCl 3/N 2 (1.5/9.0 sccm),
100W/500W , 5 mTorr, 100°C) as for nanostructured GaInP 2 .................................................................................... 53
Figure 3. 11. (a) XPS spectra of Ga 2p 3/2 for bare and nanostructured GaInP 2. The measured spectra (black line)
matched quantitatively with fitted spectra (green dotted line) composed of deconvoluted Ga-O (blue line) and Ga-P
(red line) peaks. (b) XPS spectra of P 2p 1/2 and 2p 3/2 for bare and nanostructured GaInP 2. The fitted spectra were
deconvoluted to resolve P-Ga (red line) and P-O (blue line) peaks. (c) XPS spectra of In 3d 5/2 for bare and
nanostructured GaInP 2. The measured spectra (black line) matched quantitatively with the fitted spectra (green dotted
line) composed of deconvoluted In-O (blue line) and In-P (red line) peaks. All samples were yet without (NH 4) 2S-
treatment. (d) Inter-band defect states of oxygen- and/or crystalline-defect sites. ...................................................... 54
Figure 3. 12. Top-view and tilted-view SEM images of (a) nanostructured (4-min etching) GaInP 2 without (NH 4) 2S-
treatment nor MoS 2-deposition, (b) nanostructured GaInP 2 after (NH 4) 2S-treatment (15 min, yet without MoS 2-
deposition) and (c) nanostructured GaInP 2 after MoS 2-deposition (yet without (NH 4) 2S-treatment) All samples used
in (a), (b), and (c) were prepared together using one piece of wafer up to the process of dry-etching, and they were
cleaved into three pieces for each case. (d) Corresponding total reflectance spectra of nanostructured (4-min etching)
GaInP 2 without (NH 4) 2S-treatment nor MoS 2-deposition, nanostructured GaInP 2 after (NH 4) 2S-treatment (15 min, yet
without MoS 2-deposition), and nanostructured GaInP 2 after MoS 2-deposition (yet without (NH 4) 2S-treatment) ..... 56
Figure 3. 13. Representative J-E curves of bare and nanostructured GaInP 2 photocathodes for the HER after the
(NH 4) 2S-treatment of 3 and 5 min, measured under simulated AM1.5G solar illumination (1000 W m
-2
). ................ 57
xii
Figure 3. 14. (a) Photoluminescence intensities of nanostructured (4-min-etched) GaInP 2 before and after (NH 4) 2S-
treatment (5 min). XPS spectra of (b) Ga 2p 3/2 and (c) P 2p 1/2 and 2p 3/2, and (d) In 3d 5/2 for nanostructured GaInP 2
before and after the (NH 4) 2S-treatment (15 min). ........................................................................................................ 58
Figure 3. 15. (a) Representative J-E curves of bare GaInP 2 (black line), nanostructured GaInP 2 after (NH 4) 2S-treatment
(red line), and nanostructured GaInP 2 after (NH 4) 2S-treatment and MoS 2 deposition (blue line), measured in an acidic
electrolyte (0.5M H 2SO 4) under simulated AM 1.5G illumination. (b) XPS spectra of Mo 3d and S 2p of amorphous
molybdenum disulfide deposited on GaInP 2 as a HER co-catalyst. ............................................................................ 59
Figure 3. 16. Tapping-mode AFM images of (a) bare GaInP 2 and (b) MoS 2-deposited bare GaInP 2. (c) Zoomed-in
image of MoS 2-deposited bare GaInP 2. The inset shows the distribution of height on the image. (d) The height (z)
profiles corresponding to the scan lines in (c). ............................................................................................................ 60
Figure 3. 17. Electrochemical durability of surface-tailored GaInP 2 photocathodes performing the HER under bias.
Current density–time (J–t) plots of GaInP 2 photocathodes in an acidic electrolyte (0.5M H 2SO 4) for (a) short- and (b)
long-term measurements, at various materials configurations including bare GaInP 2 (black data), bare GaInP 2
deposited with MoS 2 (yet without (NH 4) 2S-treatment, red data), nanostructured GaInP 2 with (NH 4) 2S-treatment (yet
without MoS 2-deposition, blue data), and nanostructured GaInP 2 with MoS 2 (yet without (NH 4) 2S treatment, green
data), measured at an electrode potential of 0 V (vs. RHE) under simulated AM1.5G solar illumination. Dry etching
and (NH 4) 2S-treatment were performed for 4 min and 15 min, respectively. .............................................................. 61
Figure 3. 18. (a) J-E curves of nanostructured and (NH 4) 2S-treated GaInP 2 measured during the stability test (Figure
3. 17(b)). XPS spectra of (b) Ga 2p, (c) In 3d, and (d) P 2p obtained from (NH 4) 2S-treatd bare GaInP 2 photocathode
before and after the stability test performed for 124 h shown in Figure 3. 17(b). ....................................................... 63
Figure 3. 19. Pt-related XPS survey spectra of (NH 4) 2S-treated bare GaInP 2 photocathode before and after the stability
test (Fig. 4f). ................................................................................................................................................................ 64
Figure 3. 20. (a) Top-view SEM images of bare and nanostructured/(NH 4) 2S-treated GaInP 2 (scale bar: 500 nm) before
and after the stability test in b (i.e. ~1 h for bare, ~124 h for NS). (b) Reflectance spectra of nanostructured and
(NH 4) 2S-treated GaInP 2 photocathodes before and after the stability test in b. Insets show corresponding photographic
images of samples (scale bar: 5 mm). .......................................................................................................................... 65
Figure 3. 21. (a) XPS spectra of bare GaInP 2 before and after the (NH 4) 2S-treatment. (b) J-t plots of bare and
nanostructured GaInP 2 photocathodes with and without (NH 4) 2S-treatment, obtained under the same measurement
condition as in (a). (c) Representative J-E curves of (NH 4) 2S-treated bare GaInP 2 before (red) and after (blue) the 1-h
stability test (b), measured in an acidic electrolyte (0.5M H 2SO 4) under simulated AM 1.5G illumination. The J-E
curve from the bare GaInP 2 (i.e. without (NH 4) 2S-treatment) is also shown for comparison. (d) XPS spectra of S 2p
and Ga 3s from bare GaInP 2 photocathodes with (NH 4) 2S-treatment before and after the stability test (J-t) for 1 h
shown in (b) (cyan data). After the stability test, the relative area of sulfate-related peaks decreased. ....................... 66
Figure 3. 22. (a) XPS spectra of S 2p and Ga 3s measured from bare GaInP 2 photocathodes with and without thermal
annealing (250°C, 1 h, in air) during the (NH 4) 2S-treatment. (b) Representative J-E curves of bare GaInP 2
photocathodes with and without thermal annealing during the (NH 4) 2S-treatment, measured in an acidic electrolyte
(0.5M H 2SO 4) under simulated AM 1.5G illumination. .............................................................................................. 67
Figure 4. 1. A chart of the highest confirmed conversion efficiencies for research cells for a range of photovoltaic
technologies, plotted from 1976 to the present. Copyright (2019) National Renewable Energy Laboratory, Golden,
CO. .............................................................................................................................................................................. 71
Figure 4. 2. (a) Fraction of Shockley-Queisser detailed-balance limit for voltage and current achieved by record cells.
Copyright (2016) Science. (b) Comparison of realized limiting STH efficiencies and historic development in dual-
junction limiting efficiency. Copyright (2018) ACS Energy Letters. .......................................................................... 72
xiii
Figure 4. 3. (a) Cross-sectional illustration of an integrated GaAs photocathode fabricated by printing-based materials
assemblies. (b) Representative current density–potential (J–E) curves of integrated GaAs photocathodes. (c) Epitaxial
design of ‘p-on-n’ GaAs photocathodes. Copyright (2017) Nature Energy. (d) Cost estimation (~2.6 $/Watt) for
growing p-GaAs base layer (2.5 um) using MOVPE. Copyright (2013) National Renewable Energy Laboratory. ... 73
Figure 4. 4. (a) AM 1.5 solar spectrum and the absorption of solar energy in a 2-μm-thick crystalline Si film (assuming
single-pass absorption and no reflection). Copyright (2010) Nature Materials. (b) Short-circuit current density (J sc)
vs. thickness of light absorbers (GaAs and Si) with planar surface or Lambertian light trapping. Copyright (2012)
Journal of Applied Physics. (c) Epitaxial design of triple-stack ultrathin GaAs solar cells grown by MBE. (d)
Representative current density (J)−voltage (V) curves of ultrathin GaAs solar cells with (red) and without (orange)
TiO 2 NPs after the printing and on the growth wafer, respectively. The insets show an optical image of printed GaAs
solar cells with TiO 2 NPs and a SEM image of TiO 2 NPs, respectively. Copyright (2017) ACS Nano. ..................... 74
Figure 4. 5. Relationship between power conversion efficiency, module areal costs and cost per peak watt (in $/Wp).
Each of dashed black lines indicates a given $/Wp, and the light blue lines represents the current record-high efficiency
obtained from bulk crystal Si photovoltaic module, while the blue horizontal line is the Shockley-Queisser limit for
single-junction devices. For next-generation technologies the goal is to reach 0.03-0.05 $/kWh, to replace fossil fuels
such as gasoline (i.e. 0.03 $/kWh). Copyright (2014) Nature Nanotechnology. ......................................................... 75
Figure 4. 6. Benchmarking of reported electrocatalysts for hydrogen evolving reaction and oxygen evolving reaction
for solar water splitting devices. Copyright (2015) Journal of the American Chemical Society. ................................ 77
Figure 4. 7. Benchmarking of reported stabilities of photocathodes and protective materials for the HER, versus tested
pH condition, with resulting photocurrent and degradation rate indicated. Copyright (2017) Chemical Society Reviews.
..................................................................................................................................................................................... 77
Figure 4. 8. A schematic illustration of bifacial, nanostructured, ultrathin GaAs photocathodes with ohmic contact,
TiO 2, and HER co-catalyst. ......................................................................................................................................... 78
Figure 4. 9. (a) Epitaxial design of ‘p-on-n’ ultrathin GaAs photocathodes. (b) Cross-sectional view SEM image of
the ultrathin GaAs photocathode grown by MBE using Si and Be as n- and p-type dopants, respectively. ................ 79
Figure 4. 10. Schematic illustration of fabrication processes of bifacial, nanostructured GaAs photocathodes printed
on a glass substrate for solar-driven photoelectrochemical water splitting. ................................................................ 80
Figure 4. 11. Optical microscopic images of a nanostructured GaAs photocathodes integrated on a glass substrate on
(a) a catalytic interface and (b) light-absorbing interface. (c) Cross-sectional and (d) tilt-view scanning electron
microscope (SEM) images of TiO 2 NPs implemented onto the light-absorbing interface (i.e. Al 0.4Ga 0.6As) of
nanostructured GaAs photocathode. ............................................................................................................................ 81
Figure 4. 12. A photograph of a completed ultrathin GaAs photoelectrode cells integrated on a transparent glass,
encapsulated by polyimide. ......................................................................................................................................... 82
Figure 4. 13. (a) Schematic illustration of antireflection and light trapping, resulting from periodic nanostructures with
1D gratings (p ≤ λ). (b) Schematic of a 2D grating structure with p. (c) Allowed diffraction modes in k-space or
reciprocal space. Copyright (2012) Journal of Applied Physics. (d) Distribution of the forward diffraction efficiencies
(DET mn) of a 2D-grating. Copyright (2007) Progress in Photovoltaics: Research and Applications. ........................ 83
Figure 4. 14. Schematic illustration of (a) a bifacial, nanostructured GaAs photocathodes printed on a glass substrate,
(b) a bifacial, planar GaAs photocathodes printed on a glass substrate, and (c), (d) a monofacial, planar GaAs
photocathode on GaAs wafer with and without Pt co-catalyst, considered for FDTD optical modeling. ................... 85
Figure 4. 15. (a) A schematic illustration of optical model for periodic TiO 2 NPs on ultrathin GaAs photocathodes for
FDTD optical calculations. (b) A contour plot of the integrated solar flux absorption (S_abs) of a bifacial,
xiv
nanostructured GaAs photocathode in (a) as a function of diameter (D) and height (h) of TiO 2 NPs at the period (p) of
500 nm, where optical absorption in window, emitter, base, and back-surface field (BSF) layers were considered. . 86
Figure 4. 16. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes without bottom contact layer incorporating TiO 2 NPs and gold as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm. ............ 87
Figure 4. 17. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes without bottom contact layer incorporating TiO 2 NPs and ohmic metal as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm. ............ 88
Figure 4. 18. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes with bottom contact layer incorporating TiO 2 NPs and ohmic metal as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm. ............ 89
Figure 4. 19. Maximum values of calculated S_abs of bifacial, nanostructured GaAs photocathode as a function of NP
period at various materials configurations including (i) ultrathin GaAs with TiO2 NPs and ohmic metal contact after
the removal of bottom contact layer (red diamond), (ii) ultrathin GaAs with TiO 2 NPs and silver reflector after the
removal of bottom contact layer (black square), (iii) ultrathin GaAs with TiO 2 NPs and ohmic metal contact without
the removal of bottom contact layer (blue triangle), (iv) ultrathin GaAs with SLARC and ohmic metal contact after
the removal of bottom contact layer (orange dotted line), (v) ultrathin GaAs with ohmic metal contact after the removal
of bottom contact layer (cyan dotted line), and (vi) thick GaAs directly immersed in water (green dotted line). ....... 90
Figure 4. 20. (a) A schematic illustration of optical model for periodic TiO 2 NPs on ultrathin GaAs photocathodes
with the plane of incidence for FDTD optical calculations. (b) Contour plots of normalized absorbed power density of
bifacial ultrathin GaAs photocathodes at three different configurations including bare GaAs with a bottom contact
layer and metal contact (top, BC/OM), nanostructured GaAs (i.e. with TiO 2 NPs) with a bottom contact layer and
metal contact (middle, NP/BC/OM), and nanostructured GaAs with metal contact yet without a bottom contact
(bottom NP/OM).......................................................................................................................................................... 91
Figure 4. 21. Corresponding calculated reflectance and absorption spectra in water of bifacial, nanostructured GaAs
photocathodes, where optimum geometries of TiO 2 NPs (i.e. D: 300 nm, H: 400 nm, p: 500 nm) (green) were assumed.
..................................................................................................................................................................................... 92
Figure 4. 22. Maximum integrated solar flux from various thickness of GaAs photoelectrodes in different
configurations including optically thick monofacial GaAs photoelectrode without (green) and with co-catalyst (pink),
ultrathin, nanostructured GaAs photoelectrode with Au and without bottom contact (BC) (black), with ohmic metal
and without BC (red), and with ohmic metal and with BC (blue). Optical models shown in Figure S3 were used by
varying the thickness of GaAs emitter/base while maintaining the thickness of Al 0.4GaAs window and BSF layer. . 92
Figure 4. 23. (a) Measured reflectance spectra of ultrathin GaAs photocathodes in air at various materials
configurations at the light absorbing interface, including bare GaAs on wafer (black), bare GaAs printed on glass (red),
and GaAs with TiO 2 NP printed on glass (blue), where the light was illuminated through the glass substrate and ohmic
metal contact was incorporated for all printed samples. The dotted lines depict calculated reflectance spectra from
numerical optical modeling, while dashed lines show corresponding calculated absorption spectra in water. (b) A
schematic illustration of tapered TiO 2 NPs for FDTD optical calculations. ................................................................ 93
Figure 4. 24. Optical micrographs from light-absorbing interface of (a) bare GaAs on wafer, (b) bare GaAs on glass,
and (c) NS GaAs on a glass substrate after printing, measured under the same illumination condition. Fringe color
from (b) was originated from the effects of back-surface reflector (ohmic metal contact) and the antireflection effect
of printing medium (NOA 61). .................................................................................................................................... 94
xv
Figure 4. 25. Representative current density (J)-voltage (V) curves of ultrathin GaAs photocathodes at various
configurations including bare GaAs on wafer (black), bare GaAs printed on glass (red), and GaAs with TiO 2 NP
printed on glass (blue), under simulated AM1.5G solar illumination (1000 W/m
2
). ................................................... 96
Figure 4. 26. Representative current density (J)-potential (E) curves of ultrathin GaAs photocathodes printed on glass
in various configurations including bare GaAs (red) and GaAs with TiO 2 NP (blue), measured in a three-electrode
configuration with an acidic electrolyte (0.5M H 2SO 4) under simulated AM1.5G solar illumination (1000 W/m
2
). Pt
and Ag/AgCl were employed as counter and reference electrodes, respectively. ....................................................... 97
Figure 4. 27. (a) Cyclic voltammetric curves of platinum with various thickness of TiO 2 protective layer as a dark
HER electrode measured in 0.5M H 2SO 4. Onset potential from platinum catalyst deposited by ebeam evaporator (black)
showed a slightly higher overpotential than commercially purchased platinum rod (pink). (b) HER onset potentials at
-0.5 mA/cm
2
(black, triangle) and -10 mA/cm
2
(red, rectangular) captured from (a). It is note that onset potential after
adding TiO 2 protective layer still maintained onset potential. ..................................................................................... 98
Figure 4. 28. Comparisons between photovoltaic and photoelectrochemical performance of nanostructured, ultrathin
GaAs photocathodes with Pt co-catalyst without TiO 2 protective layer captured from Figure 3b and 3c. It is note that
all the properties from PV performance were maintained in PEC performance by the effect of bifacial design ; adopting
n-type ohmic contact metal and high performance HER catalyst. The small decrease of V oc to V onset would be originated
from the overpotential (~40 mV) and Tafel slope (~30 mV/dec)of ebeam-deposited Pt as HER catalyst. The higher
saturation current density from PEC might be due to a small fraction of light waveguide effect by the glass substrate
as well as PEC cell....................................................................................................................................................... 98
Figure 4. 29. Incident-photon-to-current efficiency (IPCE) spectra for the ultrathin GaAs photocathodes including
bare GaAs and GaAs with TiO 2 NP measured at 0 V (vs. RHE) under AM 1.5G illumination. ................................. 99
Figure 4. 30. Representative current density (J)-potential (E) curves of bifacial, nanostructured GaAs photocathodes
at various materials configurations at the catalytic interface, including GaAs with Pt, GaAs with n-type ohmic contact
and Pt, GaAs with n-type ohmic contact, Ti, TiO 2, and Pt, measured in a three-electrode configuration with an acidic
electrolyte (0.5M H 2SO 4) under simulated AM1.5G solar illumination (1000 W/m
2
). ............................................. 100
Figure 4. 31. (a) Thickness (nm) vs. deposition time (s) of MoS x film deposited on an ITO glass using cathodic
electrodeposition in an aqueous solution of (NH 4) 2MoS 4 (2 mM), 0.5 M of Na 2SO 4 (pH 6.4), and (NH 4) 2S ((NH 4) 2S:DI
water = 1: 750 by volume, Macron, 20.0-24.0 wt% in H 2O) at -0.5 V vs. RHE. (b) Corresponding transmittance
spectra of MoS x films measured by UV-Vis spectroscopy. (c) XPS spectra of Mo 3d and S 2p of MoS x films. (d)
Linear voltammetric curves of MoS x deposited on a ITO glass as a dark HER electrode measured in 0.5M H 2SO 4 102
Figure 4. 32. Corresponding current density (J)-time (t) plots of bifacial, nanostructured GaAs photocathodes,
measured under simulated AM1.5G illumination at an applied potential of 0 V (vs. RHE). .................................... 103
Figure 4. 33. Degradation of bare ultrathin GaAs with OM/Ti/Pt without TiO 2 protective layer (a) (b) before and (c)
(d) after photoelectrochemical stability test (~14 hours) in 0.5M H 2SO 4 under AM 1.5G illumination shown on. Left
images were microscopic images from the light-absorbing interface (i.e. TiO 2 nanoposts) and right images were from
the reactive interface (i.e. Pt co-catalyst.). The failure was caused by a crack formation of underlying ultrathin GaAs
photocathodes, which was led by the stress corrosion cracking of Ti. ...................................................................... 103
Figure 4. 34. energy band diagrams of bifacial, nanostructured, ultrathin GaAs photocathodes with ohmic contact,
TiO 2, and HER co-catalyst. ....................................................................................................................................... 104
Figure 4. 35. The electrochemical dissolution/delamination of MoS x layer from ohmic metals during stability test in
0.5 M H 2SO 4. Microscopic images (a) before depositing a-MoSx on ohmic metals, (b) after depositing MoS x on ohmic
metals, (c) and (d) after 25 hours of photoelectrochemical stability test shown in Figure 4(b) (cyan), where GaAs
photocathode still remained nearly intact without corrosion. .................................................................................... 105
xvi
Figure 5. 1. (a) Energy levels of semiconductor bands and energy distribution of the occupied and unoccupied states
of the redox acceptor. (b) Band bending in n-type (left side) and p-type (right side) semiconductor electrodes upon
equilibration of the Fermi levels of the semiconductor with the redox species. Copyright (1996) The Journal of
Physical Chemistry. ................................................................................................................................................... 107
Figure 5. 2. Positions of energy bands of various semiconductors in the dark (d) and in light (l) with respect to the
electrochemical scale. Copyright (1996) The Journal of Physical Chemistry. .......................................................... 108
Figure 5. 3. Effects of buried junction in photoelectrodes for solar water splitting. (a) Conceptual band alignments of
p-GaInP 2 and np-GaInP 2 photocathodes and (b) photo-current density vs. potential (J-E) of IMM (Inverted
metamorphic multi-junction) without and with ultrathin n-GaInP 2 buried junction (IMM-p and IMM-pn). Copyright
(2017) Nature Energy. Similarly, (c) Conceptual band alignments and (d) photo-current density vs. potential (J-E) of
p-Si and np-Si photocathodes, respectively. Copyright (2011) Journal of the American Chemical Society. ............ 109
Figure 5. 4. Cross-sectional SEM images of (a) as-grown pn-GaInP 2 and (b) nanostructured np-GaInP 2 by electroless
Ag deposition and dry-etching process. Inset image indicates zoom-in cross-sectional SEM image of nanostructured
surface........................................................................................................................................................................ 110
Figure 5. 5. (a) A schematic illustration of photo-induced electroless deposition of Ag nanoparticles for n-type GaInP 2,
where the electrolyte-junction layer is n-type GaInP 2. (b) A photograph of pn-GaInP 2 immersed in an aqueous AgNO 3
solution during photo-induced electroless deposition under white LED illuminated. ............................................... 111
Figure 5. 6. (a) reflectance spectra of bare p-GaInP 2 and nanostructured pn-GaInP 2 measured by UV-Vis spectroscopy.
(b) the current density-potential curves (J-E) of bare, NS pn-GaInP 2, and a-MoS 2 deposited NS pn-GaInP 2
photocathodes measured under 1-sun illumination. ................................................................................................... 112
Figure 5. 7. (a) Illustration of multi-epitaxial liftoff for multi-layer ultrathin GaAs photoelectrodes. (b) (b) Cost
estimation (~2.6 $/Watt) for growing p-GaAs base layer (2.5 um) using MOVPE. Copyright (2013) National
Renewable Energy Laboratory. (c) A cross-sectional SEM image of multilayer-grown 3-stack ultrathin GaAs
photoanodes. (b) A cross-sectional SEM image of top layer consisting of n
+
-GaAs contact, AlInP window, n-GaAs
emitter, p-GaAs base, AlGaAs BSF, p
+
-GaAs contact, and sacrificial layer. ............................................................ 113
Figure 5. 8. representative photovoltaic current density (J)-voltage (V) curves of GaAs photocathodes on wafer, where
isolated ultrathin GaAs photoelectrodes from top, middle, and bottom layers .......................................................... 114
xvii
List of Tables
Table 3. 1. A summary of photoelectrochemical (PEC) performance characteristics of nanostructured GaInP 2
photocathodes performing the HER, extracted from Figure 3. 9. ................................................................................ 52
Table 3. 2. A summary of photoelectrochemical (PEC) performance characteristics of bare and nanostructured GaInP 2
Photocathodes with and without (NH 4) 2S treatment, extracted from Figure 3. 13. ..................................................... 57
Table 3. 3. A summary of photoelectrochemical (PEC) performance characteristics of nanostructured GaInP 2
photocathodes deposited with MoS 2 performing the HER, extracted from Figure 3. 15. ............................................ 59
Table 3. 4. Values for determining faradaic efficiency of bare GaInP 2 photocathodes with and without (NH 4) 2S-
treatment. ..................................................................................................................................................................... 68
Table 3. 5. Summary of published stability data of III-V compound semiconductor photoelectrodes in solar water
splitting. ....................................................................................................................................................................... 69
Table 4. 1. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, obtained
from Figure 4. 16. The thickness of TiO 2 base was fixed at 50 nm. ............................................................................ 87
Table 4. 2. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, exported
from Figure 4. 17. The thickness of TiO 2 base was fixed at 50 nm. ............................................................................ 88
Table 4. 3. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, exported
from Figure 4. 18. The thickness of TiO 2 base was fixed at 50 nm. ............................................................................ 89
Table 4. 4. Comparisons of relative light absorption enhancement between measured short-circuit current densities
(J SC) from photovoltaic performance shown in Figure 3a and calculated integrated solar flux absorption (S_abs) in air
with different configurations including bare GaAs on wafer, bare GaAs on glass, and nanostructured GaAs on glass
with bottom contact layer and ohmic metal as back-side reflector. ............................................................................. 95
Table 4. 5. Summarized photovoltaic performance shown in Figure 4. 25. ................................................................ 96
Table 4. 6. Summarized photoelectrochemical performance shown in Figure 4. 30. ................................................ 100
1
Chapter 1. Background
1. 1. Storable renewable energy
Energy conservation and renewable energy sources have become increasingly since fossil
fuels have been being depleted and their continued usage has raised the density of carbon dioxide
Figure 1. 1. (a) Carbon dioxide level in the last 650,000 years, with the abrupt end of the last ice age about 7,000 years
ago marking the beginning of the modern climate era and of human civilization. Source : NASA Global Climate
Change (https://climate.nasa.gov/evidence). (b) U.S. greenhouse gas emission in 2017. Source: U.S. Environmental
Protection Agency (2019). Inventory of U.S. Greenhouse Gas Emissions and Sinks 1990-2017. (c) The production
and absorption of extra CO 2, between 1900 and 2016, and it is evident human-made CO 2 has been increasing rapidly,
which has been mostly absorbed by atmosphere among planetary systems. Source: IPCC, “Climate Change 2014:
Mitigation of Climate Change”
2
leading to global warming. For example, Figure 1. 1 shows the concentration of carbon dioxide
which had been fluctuating by the cycles of glacial advance and retreat within the range between
180 and 280 ppmv (Concentration unit referring parts per million by volume, 1 ppmv = 0.0001% )
over last 650,000 years. However, it has rapidly increased from 280 ppmv to 400 ppmv over the
last century, after the industrial revolution. Figure 1. 1(b) shows the types of greenhouse gases in
2017, where CO2 still comprised the largest portion among them. The problem of the increase of
CO2 was not significant until 1960’s, as its production had been mostly absorbed by ocean and
land, meaning that the Earth ecosystem could deal with the natural production rate of CO2 while
maintaining the global temperature. However, the rapid increase of CO2 after 1950’s exceeded the
absorption capabilities of ocean and land, and the absorption by atmosphere became dominant,
causing the severe greenhouse effect.
Figure 1. 2. (a) The greenhouse effect by the absorption of outgoing heat by greenhouse gas molecules. (b) The change
in global surface temperature relative to 1951-1980 average temperatures. (c) Satellite observations of Arctic sea ice
in 1984 and 2016. The declining rate of Arctic sea ice is now 12.8 percent per decade. (d) Sea height variation in mm
from 1993 until present with 90 mm of rise. The rate of change is +3.3 mm per year. Source : NASA Global Climate
Change (https://climate.nasa.gov/evidence).
3
Figure 1. 2(a) shows the greenhouse effects with the absorption of outgoing heat by
greenhouse gases . Due to the rapid increase of greenhouse gases, especially CO2 after 1950’s, the
change in global surface temperature (relative to average temperatures in 1951-1980) has kept
increasing until today with the 0.8 C° of the annual average temperature increase as shown in
Figure 1. 2(b). Figure 1. 2(c) illustrates the huge shrink of Arctic sea ice between 1984 and 2016
Figure 1. 3. Renewable energy sources (a) solar energy, (b) hydroelectric power, (c) wind power, (d) ocean energy,
(e) biofuel, (f) hydrogen gas. Source: National Renewable Energy Laboratory website
(https://www.nrel.gov/workingwithus/learning.html)
4
with 12.8 percent per decade of the declining rate of Arctic sea ice. It clearly shows that the sea
level has been increased by ~90 mm from 1993 with +3.3 mm per year as shown in Figure 1. 2(d).
To solve such issues of global warming and shortage of fossil fuels, various renewable
energy sources such as solar energy, wind power, or biofuel have been intensively developed and
started to gradually replace fossil fuels (see Figure 1. 3). Among them, solar energy has been
considered as the most promising and sustainable energy source because of its nearly infinite
amount and accessibility from the equator to the middle latitude. However, there are still problems
Figure 1. 4. (a) Generation and demand of solar and wind electricity power with the mismatch between the peak times
of solar supply and grid demand, which motivates a need to store the excess amount of solar power during the peak
time. Source : QIC (https://www.qic.com.au) (b) A satellite image of hurricane Maria (Puerto Rico blackout in
September 2017) which happened in the east of the Florida east coast, captured by NASA, (c) scattered solar paneled
by hurricane Maria. Source: CNN Business. Ahiza Garcia, March 6, 2019.
5
related with its nature that solar energy can be only collected during a moderately clear day. Figure
1. 4(a) shows a relative amount of generation and demand in solar and wind electricity power,
showing the clear mismatch between the peak times of solar supply and grid demands. Another
problem is that such renewable energies are susceptible to weather conditions. The problem
happened if solar and/or wind powers are dominant because of regional limitations in such as
Puerto Rico or Australia. Figure 1. 4(b) shows the satellite image of hurricane Maria which
occurred in Puerto Rico blackout in September 2017 and caused the biggest blackouts in the US
history by destroying the most of solar panels in the island (see Figure 1. 4(c)).
Figure 1. 5. (a) Comparisons of energy densities in terms of energy per mass between lithium-ion (Li-ion) battery,
hydrogen gas, and general fossil fuels (gasoline and diesel). (b) A comparison of power source between representative
electricity car (Tesla Model S) and fuel-cell car (Toyota Mirai fuel). (c) Illustration of lithium-ion battery packs at the
bottom of Tesla Model S. Source: Tesla (d) Illustration of hydrogen tanks in Toyota Mirai. Source : Toyota
6
Due to these problems, the storage of solar energy has been important, and lithium-ion
battery can be considered as an alternative way to store the excess power. However, there is still a
restriction of volume or mass because of the low energy density of lithium-ion battery (see Figure
1. 5(a)). In contrast to the energy density of lithium-ion battery, hydrogen has a very high energy
density per unit mass, which is even higher than the general fossil fuels. It is obvious that hydrogen
as a fuel does not require high volume or mass to store the similar level of energy. For example,
Tesla Model S (i.e. electric car) and Toyota Mirai (i.e. fuel-cell car) are equipped with 540 kg of
lithium-ion battery and 88 kg of hydrogen tank, respectively, to run ~310 miles as shown in Figure
1. 5(b), (c), and (d). Obviously, hydrogen can be considered as one of the most promising carriers
to store solar energy. Nevertheless, 95% of the hydrogen generated in the United States is currently
made by natural-gas reforming from existing natural gas delivery infrastructure (see Figure 1. 6(a)),
which still emits carbon dioxide during the steam-methane reforming reactions (i.e. CH4 + 2H2O
Figure 1. 6. (a) A factory for natural gas reforming. Source : Office of energy efficiency and renewable energy
(https://www.energy.gov/eere/fuelcells/) (b) A cycle of renewable hydrogen produced by an electrolysis of water with
renewable energies and fuel cells. Source: Hydroville (http://www.hydroville.be/)
7
+ heat → CO + 3H2 + H2O → CO2 + 4H2). This is why an electrolysis of water by sunlight can be
considered as an alternative pathway to produce hydrogen without emitting greenhouse gases as
shown in Figure 1. 6(b).
1. 2. Electrolysis of water
Electrolysis of water is a decomposition of water molecule into hydrogen and oxygen by
flowing electric current through cathode and anode in an electrolyte (see Figure 1. 7)
1
. In
thermodynamics, the Gibbs free energy for the conversion of one water molecule into one
hydrogen and a half oxygen is ∆G = 2.46 eV (i.e. 237.2 kJ/mol) in standard conditions.
𝐻 2
O + 2( ℎ
+
) →
1
2
𝑂 2
+ 2𝐻 +
(OER)
2𝐻 +
+ 2e
−
→ H
2
(HER)
𝐻 2
O →
1
2
𝑂 2
+ H
2
∆G = 2.46 eV
Figure 1. 7. Fundamentals of electrochemical water splitting. Copyright (2017) Nature Reviews Chemistry.
8
Because it is the energy corresponding to the two electron-hole pairs per molecule, one
electron-hole pair must have 1.23 eV to enable the reaction. Semiconductors absorb photons with
a higher energy than their bandgap and generate electron and hole pairs having an energy
corresponding to the bandgap.
Figure 1. 8. (a) HER volcano plot for various metal and MoS 2 (b) TOF avg plots with linear sweep voltammograms of
various HER catalysts. (c) OER volcano plot for metal oxides. Copyrights (2017) Science. (d) Comprehensive plots
of catalytic activity, stability, and electrochemically active surface area for OER electrocatalysts in acidic (top) and
alkaline (bottom) solutions. Copyright (2013) Journal of the American Chemical Society.
9
In reality, water electrolysis requires higher energy than 1.23 eV due a loss of energy by a
kinetic charge transfer between electrode and electrolyte, and the part that exceeds 1.23 eV is
called overpotential( ). Active electrocatalysts are required to minimize the overpotential
necessary to drive the hydrogen evolution and the oxygen evolution reactions to maximize
electrocatalytic efficiency. Figure 1. 8 shows the representative HER and OER catalysts and their
electrochemical properties in various conditions
2,3
. Among various catalysts, platinum (Pt) is well
known as the best catalyst for the HER with negligible overpotentials in acidic electrolytes.
However, the scarcity and high cost of Pt limits its wide and practical use. This triggers a need for
earth-abundant catalysts to replace Pt. Such efforts have developed a wide range of excellent and
competitive HER catalysts such as MoS2, MoP, and NiP. In contrast to HER catalysts, OER
catalysts still need to be further developed since the best catalyst which has an almost thermo-
neutral GH (i.e. thermoneutral Gibbs free energy of the adsorbed hydrogen) like Pt in HER has
not been found. IrOx or RuOx are known as the best OER catalyst so far despite their high
overpotential (>200 mV at 10 mA/cm
2
). This challenge has motivated significant efforts for
developing and improving novel OER catalysts over the past decade as shown in Figure 1. 8 (c)
and (d). Because of the presence of overpotentials from HER and OER, the required energy for
overall water splitting is typically 1.7 ~ 1.8 eV.
1. 3. Solar water splitting
An electrolysis of water enabled by an external energy using electricity is still considered as
a potentially possible pathway to replace methane reforming method for the production of
hydrogen. The main obstacle in electrolysis of water is the cost of electricity and environmental
aspects. Therefore, renewable energies such as sunlight have been considered as the source of
10
electricity to drive electrolysis of water. Through a series of intense R&D efforts, photovoltatics
have been applied in water splitting, which is called as photoelectrolysis of water or solar water
splitting by using semiconductors as both the chemical reactor and the light absorber. As a source
of energy, semiconductors must have a bandgap larger than 1.7 ~ 1.8 eV to drive the reaction as
discussed in the previous section (see Figure 1. 9(a)). In addition to this bandgap requirement for
the ideal semiconductor, its conduction and valence band edges must straddle the hydrogen and
oxygen redox potentials to enable unassisted solar water splitting and minimize overpotentials for
HER and OER, respectively. The energetic positions of a conduction band-edge energy (CB) and
a valence band-edge energy (VB) of various semiconductor electrodes are shown in Figure 1. 9(b).
They should be well aligned to the electrochemical redox potential of hydrogen E° (H
+
/H2) and
oxygen E° (O2/H2O) to carry out the hydrogen evolution reaction (HER) on a cathode and the
oxygen evolution reaction (OER) on an anode with minimum energy inputs. Among these
materials, metal oxide semiconductors have enough band gap energy for unassisted water splitting
and provide a durable photoelectrochemical lifetime with low production costs. However,
Figure 1. 9. (a) A simplified schematic illustration of the band diagram of a semiconductor photoelectrode and the
standard redox potential of hydrogen and oxygen. Copyright (2010) Chemical Reviews. (b) Bandgap energies and
band-edge positions of various semiconductor photoelectrodes relative to the NHE and the vacuum level at pH = 0.
11
photogenerated current densities from metal oxide semiconductors are generally very low (< 5
mA/cm
2
) because of their large bandgap, which severely limits their solar-to-hydrogen efficiencies.
III-V compound semiconductors such as GaAs or GaInP2 have relatively smaller bandgaps than
metal oxides, but still near ideal bandgaps for solar irradiance spectrum. In addition to the absolute
bandgap energy, III-V compound semiconductors have direct bandgap, showing a superior
photovoltaic performance (see Figure 1. 10(a))
4
. Besides, III-V compound semiconductors have a
capability to build epitaxial multilayers by changing their composition and element while carefully
controlling its lattice constant during the growth (see Figure 1. 10(b)). Using this technique, highly
efficient multijunction photoelectrodes have been developed for unassisted solar water splitting
without relying on any input of external energy
5-9
.
Despite these compelling advantages of III-V compound semiconductors as a potential
material for solar water splitting, there are still several issues that need to be solved before practical
applications of III-V photoelectrodes, including photoelectrochemical stability, production cost,
and efficiency. Therefore, it is critically important to develop new platforms of material and/or
Figure 1. 10. (a) Fraction of Shockley-Queisser detailed-balance limit for voltage and current achieved by record cells.
Copyright (2016) Science. (b) Bandgap energy and lattice constant of various III-V compound semiconductors. (Tien,
1988)
12
system for highly efficient and stable III-V photoelectrodes for solar water splitting. Figure 1. 11
depicts the photoelectrochemical durability of GaInP2 and Si photocathodes with the similar
protective layer (Mo/MoS2) developed by Jaramillo research group at Stanford, where GaInP2
photocathodes showed a relatively shorter durability (~60 hours) than Si photocathodes (~60 days).
Among various candidate materials in photocathodes, III-V materials typically demonstrated a
very short-term photoelectrochemical stability than other types of materials such as silicon, GaN,
or InGaN, which must be improved to compete with other candidates
10-14
. To extend
photoelectrochemical durability of III-V photocathodes, TiO2 has been widely adopted as a
protective layer because it is a n-type semiconducting material, while new types of MoS2 have
been developed (e.g. a graded MoSx/TiO2 or Mo/MoS2 bilayer)
15-21
. However, their lifetimes have
still not been sufficiently long to satisfy practical applications. In addition, photogenerated current
density is also a critical factor to further boost the advantages of III-V photocathodes for the higher
efficiency (e.g. >15% STH) solar water splitting. This is because photogenerated current density
is a direct parameter which determines the amount of hydrogen molecules per unit area of
Figure 1. 11. The comparison of photoelectrochemical stability (J-t) of (a) GaInP 2 photocathodes and (b) Si
photocathodes covered by Mo/MoS 2 as a protection layer and co-catalyst. Copyrights (2016) The Journal of Physical
Chemistry Letters and (2017) ACS Applied Materials & Interfaces.
13
photocathodes per unit time. Typically, antireflection coatings have been used in photovoltaic
system to reduce Fresnel reflection and to increase photogenerated carriers by employing single-
or double-layered high-index dielectric materials such as TiO2 or MgF2/ZnS. However, this
technique has a narrow window to be adopted in photoelectrode system because the charge
transport rates and photoelectrochemical durability of those materials must be simultaneously
optimized. That’s why only TiO2 has been widely used as a protective layer in the last decade. Due
to this limitation, researchers have alternatively modified the morphology of active
Figure 1. 12. Photoelectrochemical current density (J-E) of various nanostructured photocathodes. (a) Si
photocathode. Copyright (2011) Energy & Environmental Science. (b) InP photocathodes. Copyright (2012)
Angewandte Chemie. (c) InGaN/Si photocathodes. Copyright (2015) Nano Letters. (d) ZnO/Si photocathode.
Copyright (2012) Nanotechnology.
14
photoelectrodes. For example, Si, InP, InGaN, and ZnO, were implemented with nanostructured
interface to effectively increase photogenerated current densities as well as onset potentials (see
Figure 1. 12)
14,22-24
. Furthermore, photovoltage or onset potential is an important parameter to
provide a sufficient energy for an electrolysis of water. Onset potential for photocathodes is known
as the difference between photovoltage (i.e. built-in voltage) of photoelectrodes and overpotential
at the interface. Figure 1. 13 shows energetic positions of conduction band edge, valence band
edge, and Fermi level (or quasi Fermi levels) of photoanode in pre-equilibrium state and in
Figure 1. 13. (a) Energetic positions of conduction band edge, valence band edge, and Fermi level (or quasi Fermi
levels) of photoanode in pre-equilibrium state and in equilibrium states without and with illumination. (b) Schematic
relating bandgaps and photovoltages for a tandem absorber. (c) Photovoltage benchmarks for PEC and PV materials
as a function of optical bandgap. Copyrights (2017) Current Opinion in Electrochemistry.
15
equilibrium states without and with illumination. It is noted that the Fermi level of photoelectrode
at the interface always aligns with the equilibrium potential of HER in electrolyte, which is at 4.44
eV referenced to the vacuum level
25
. Due to this junction property, pristine photoelectrodes
without co-catalyst lose their photovoltage depending on how their band-edge positions are aligned
well to the redox potential of hydrogen or oxygen (Vloss), and how small overpotentials ( ηOER or
ηHER) are (see Figure 1. 9 and Figure 1. 13 (b)). That’s why PEC photovoltage is rarely similar to
Figure 1. 14. (a) Mott-Schottky plots of a p-type GaInP, sample at pH values of 2.0, 4.8, 9.1 and 12.0 (all at 5 kHz).
Copyright (1994) Journal of Electroanalytical Chemistry. (b) Conceptual band alignment and bending model for p-
GaInP and n/p-GaInP in acidic electrolyte. Copyright (2017) Nature Energy. (c) Conceptual band alignments of
monolithic tandem photoelectrode (GaAs/GaInP). Copyright (1998) Science. (d) Photocurrent vs. potential (J-E) of
surface modified GaInP photocathode by conjugate molecules. Copyright (2015) ACS Applied Materials & Interfaces.
16
or typically lower than PV photovoltage (see Figure 1. 13 (c)). Diverse pathways have been
demonstrated to improve PEC photovoltage or onset potential by using different pH of electrolytes,
buried junction (n/p-type GaInP2), monolithic tandem materials (GaAs/GaInP2), and conjugate
molecules (see Figure 1. 14)
5,7,26,27
.
Overall, there have been many divergent approaches in different aspects to improve each of
photoelectrochemical properties such as stability, photogenerated current density, and charge
transport rate. However, there have been lack of studies to accomplish improvements of all of such
properties by using a simple method and/or platform. In this thesis, we present two different
strategies for III-V photocathodes that can comprehensively handle those three of
photoelectrochemical properties (i.e. stability, current density, and onset potential) to enable high
performance and high durability solar water splitting by using monofacial nanostructured GaInP2
photocathodes and bifacial nanostructured ultrathin GaAs photocathodes.
17
Chapter 2. Methods
2. 1. Fabrication of monofacial nanostructured GaInP2 photocathodes
2. 1. 1. Growth of GaInP2 by metal organic vapor phase epitaxy
Gallium indium phosphide (Ga0.51In0.49P, referred to as GaInP2) epilayer was grown by a
group of Dr. John Geisz in National Renewable Energy Laboratory (NREL). p-type GaInP2 with
a thickness of 2.5 m was epitaxially grown on a (100) p-type GaAs substrate miscut 2° towards
the (110) at 700°C by an atmospheric pressure metal organic vapor phase epitaxy (MOVPE),
where zinc (2 × 10
17
cm
-3
) was used as a p-type dopant. Contact metals were deposited on the
backside of GaAs substrate by electron-beam evaporation with titanium/gold (10 nm/300 nm) . As
shown in Figure 2. 2, the epitaxial layer of p-GaInP2 was grown without a lattice mismatch and
any visible defects on GaAs substrate with a GaAs buffer layer (~100 nm)
28
. Figure 2. 1
schematically illustrates the overall fabrication procedures to form randomly nanostructured
Figure 2. 1. A schematic illustration of fabrication procedures for surface-tailored ‘black’ GaInP 2 photocathodes.
18
GaInP2 photocathodes, including electroless deposition of Ag nanoparticles, dry-etching, and
(NH4)2S passivation.
2. 1. 2. Electroless deposition of silver nanoparticles
To form a nanoscale etch mask for dry-etching process, silver nanoparticles were formed
on p-GaInP2 by using electroless deposition with a silver nitrite solution (AgNO3). The surface of
the as-grown p-type GaInP2 on GaAs substrate was cleaned with acetone, isopropyl alcohol (IPA),
and deionized (DI) water. Subsequently, the native oxide from the surface of GaInP2 was removed
by a dilute NH4OH solution (NH4OH (29%, EMD):DI water = 1:10, by volume, 2 min) and treated
by perchloric acid solution (CR-7, KMG, 30 s). It is noted that perchloric acid, which is known
as a strong oxidizing agent, effectively activates the surface to be favorable for the electroless
deposition of silver. After cleaning and activating the surface, silver nanoparticles were
electrolessly deposited on GaInP2 in an aqueous solution of silver nitrite (AgNO3, 10 mM) and
hydrofluoric acid (HF, 5 M). Figure 2. 3 shows top-view SEM images of electroless-deposited
Figure 2. 2. (a) A cross-sectional SEM image of p-GaInP 2 grown on GaAs wafer including GaAs buffer layer. (b)
Energy gap (E g) of GaP, InP, GaAs, and InAs vs. lattice constant. Copyright (1981) Journal of Crystal Growth.
19
silver nanoparticles on GaInP2 for 1 min with different pre-treatment conditions. The following
equations are the possible chemical reactions of electroless deposition of silver nanoparticles used
in this study.
1) GaInP2(s) + 6h
+
→ Ga
3+
(aq) + In
3+
(aq) + 2P
2) 6Ag
+
(aq) + 6e
-
+ 6H
+
(aq) + 2P → + 6Ag(s) + 2PH3(g)
Since the electroless deposition is easily influenced by the surface chemistry, it is important to
modify/activate the surface of GaInP2 for the more favorable reduction of silver ions. Figure 2. 3
shows how the surface chemistry affects the shape, density, and morphology of deposited silver
nanoparticles and why the activation of surface by the pre-treatment is important to accurately
Figure 2. 3. Top-view SEM images of silver nanoparticles deposited for 1 min (a) directly without any pre-treatment,
(b) with UV-ozone pre-treatment (5 min), (c) with NH 4OH (1:10) pre-treatment (1 min), and (d) NH 4OH (1:10) pre-
treatment (1 min) followed by perchloric acid pre-treatment (30 sec)
20
control the size, shape, and density of silver nanoparticles. Without any pre-treatment, silver
nanoparticles were not able to be grown uniformly in terms of size and shape onto as-grown GaInP2
(see Figure 2. 3(a)). After treating UV-ozone for 5 min on GaInP2, silver was grown like a flower
with a lower density (see Figure 2. 3(b)). We postulate that the presence of oxide layer and/or III-
Oxide terminating groups (e.g. Ga-O-In) on GaInP2 might inhibit the transport of electrons from
GaInP2 to silver ions, which was also confirmed by the deposition of silver nanoparticles on GaInP2
treated by UV-ozone for 5 min. After treating the surface of GaInP2 with NH4OH solution (1:10
in DI water) for 1 min, the size and shape of deposited silver nanoparticles became more uniform
and predictable because of the removal of the native oxide. However, the nanoparticles could not
cover the surface of GaInP2, and the size of them were still too large to form density-graded
nanoporous structures (see Figure 2. 3(c)). By additionally treating NH4OH-treated GaInP2 with
perchloric acid (CR-7), we obtained very uniform, small, and dense silver nanoparticles (see
Figure 2. 3(d)). We also confirmed that a pre-treatment only with perchloric acid also provided the
similar morphology of silver nanoparticles even without using NH4OH treatment. The size and
density could be also controlled by the deposition time or pre-treatment time of perchloric acid.
For example, nanoparticles become smaller and denser as pre-treatment time increases.
2. 1. 3. Dry-etching by inductively coupled plasma-reactive ion etching
Nanostructured GaInP2 photocathodes were formed by inductively coupled plasma reactive
ion etching (ICP-RIE, STS D-RIE system) with the as-deposited silver nanoparticles as an etch
mask. Figure 2. 4(a) shows a general schematic of ICP-RIE where the gases are introduced above
an inductive coil. Radio frequency (RF) power (13.56 MHz) is applied to both the coil and plate
(i.e. chuck) to create a plasma and to generate anisotropic etching. Using BCl3 and N2 (1.5 and
9 in sccm) as a reactive etching gas and carrier gas, respectively, GaInP2 was dry-etched in 5 mTorr
21
with the applied RF power (ICP/Platen = 100W/500W). The temperature of chuck was maintained
at 100 C to remove the non-volatile compounds at room temperature such as GaCl3 and InCl3
generated during the reactive etching process. As the etching time increased from 1 to 4 minutes,
the height of nanostructured GaInP2 gradually increased and the more detail information about the
height will be discussed in the Chapter 3 (see Figure 2. 4(b) and (c)). Since Ag nanoparticles were
very slowly etched under this condition and the shape of particles were rounded, the diameter of
nanopillars gradually decreased from the bottom to the tip, thereby creating a graded index of
Figure 2. 4. (a) A schematic illustration of inductively coupled plasma-reactive ion etching with BCl/N2 as a reactive
etching/carrier gas, respectively. (b) Cross-sectional and top-view SEM images of the nanostructured GaInP 2 at
etching times of 1, 2, 3, and 4 min after removing remaining Ag nanoparticles. (d) Tilt-view and top-view SEM images
of nanostructured GaInP 2 before and after etching of Ag nanoparticles.
22
refraction to suppress front-surface reflection. After the dry-etching, Ag nanoparticles were wet
chemically etched by using an aqueous solution of ammonium hydroxide (NH4OH:H2O2:DI water
=1:1:1, by volume). Figure 2. 4(d) shows tilt-view and top-view SEM images of nanostructured
GaInP2 before and after etching Ag nanoparticles (NPs).
2. 1. 4. Surface passivation by ammonium sulfide solution
After forming nanoporous morphology, the surface of GaInP2 became considerably
damaged because of plasma-induced crystalline defects and oxide at the nanostructured surface,
which could act as sites for non-radiative recombination in various III-V compound
Figure 2. 5. (a) Oxygen- and/or defect-induced inter-band defect states in III-V compound semiconductors. (b) A
conceptual illustration of mechanism of sulfur passivation on III-V compound semiconductors. (c) Schematic
illustration of surface treatment by using two-step (NH 4) 2S passivation.
23
semiconductors such as GaP, GaAs, and GaInP2
29-32
and and/or surface states in water (see Figure
2. 5(a)). To passivate the defect sites and surface states, chemical treatment using sulfur-containing
chemicals (e.g. (NH4)2S or Na2S solutions) were widely used in high-speed electronic and high-
power microwave III-V devices
33-35
. Figure 2. 5(b) shows a general mechanism of sulfur
passivation on III-V materials, where sulfur replaces oxygen sites and acts as a very shallow
donor
32
. In this study, the surface of nanostructured GaInP2 was soaked in an aqueous solution of
ammonium sulfide ((NH4)2S), followed by thermal annealing in air to incorporate corrosion-
resistant surface stoichiometry (see Figure 2. 5(c)). A pre-heated (NH4)2S solution (~61 C, ~0.77
M, prepared by adding 34 mL of DI water or isopropanol to 10 mL of as-received (NH4)2S solution
(Macron, 20.0-24.0 wt% in H2O)) was cast onto the surface of bare or nanostructured GaInP2
placed on a hot plate (~85 C) for 3~10 min. Subsequently, samples were further annealed at 250°C
in air for 1 h to oxidize sulfide and incorporate sulfate bonding.
2. 1. 5. Deposition of amorphous molybdenum disulfide
For the deposition of amorphous molybdenum disulfide (MoS2), the GaInP2 was immersed
in an aqueous solution of (NH4)2MoS4 (1 mM, Sigma Aldrich) and 0.5 M of Na2SO4 buffer (pH
Figure 2. 6. (a) electroless photochemical deposition of MoS 2 under white light illumination and (b) electrochemical
deposition of MoS 2 in dark condition with three-electrode configuration.
24
6.6) under white light (LED-6WD, AmScope) illumination (~20 mW/cm
2
) for 5 min at an open-
circuit condition, followed by rinsing with DI water and drying under N2 (see Figure 2. 6(a)). After
the photochemical deposition, the sample was thermally annealed at 250°C under N2 atmosphere
for 1 h. Alternatively, MoS2 was also electrochemically deposited at a cathodic cyclovoltammetry
condition by scanning 8-10 times of potential cycles (i.e. one cycle: -0.4 V → +0.15 V → -0.4 V)
using the same regents as in photochemical deposition under a dark condition (see Figure 2. 6(b)).
2. 2. Fabrication of bifacial ultrathin GaAs photocathodes
2. 2. 1. Growth of ultrathin GaAs by molecular-beam epitaxy
Ultrathin pn-GaAs epilayers were grown by Professor Minjoo L. Lee’s group at the
University of Illinois Urbana-Champaign. The epitaxial layers for ultrathin GaAs photocathodes
grown by molecular beam epitaxy (MBE) consist of p
+
-GaAs contact (Be-doped, 1 × 10
19
cm
-3
,
200 nm), p-Al0.4Ga0.6As window (Be-doped, 2 × 10
18
cm
-3
, 40 nm), p-GaAs emitter (Be-doped, 2
× 10
18
cm
-3
, 50 nm), n-GaAs base (Si-doped, 3 × 10
17
cm
-3
, 220 nm), n-Al0.4Ga0.6As back- surface
Figure 2. 7. (a) An epitaxial design of ultrathin GaAs photocathodes. (b) A cross-sectional SEM image of epitaxially
grown ultrathin GaAs photocathode on GaAs substrate.
25
field (Si-doped, 3 × 10
18
cm
-3
, 50 nm), n
+
-GaAs contact (Si-doped, 5 × 10
18
cm
-3
, 50 nm), and
Al0.9Ga0.1As sacrificial layer (undoped, 400 nm), grown on a semi-insulating (100) GaAs wafer
(see Figure 2. 7(a)). Figure 2. 7(b) shows a cross-sectional scanning electron microscope (SEM)
image of the MBE-grown ultrathin GaAs epilayers, where defect-free epitaxial layers including
sacrificial Al0.9Ga0.1As (the darkest layer) between device stack and GaAs substrate are clearly
identified as designed.
2. 2. 2. Nanofabrication of hexagonally periodic TiO2 nanoposts
Hexagonally periodic TiO2 nanoposts (NPs) were formed through successive fabrication
steps including nano-imprinting, dry-etching of polymer, ebeam-evaporation of chrome, lift-off,
and dry-etching of TiO2 as shown in Figure 2. 8
36,37
. The fabrication of bifacial ultrathin GaAs
photocathodes began with the deposition of a p-type ohmic contact by photolithography, electron
beam evaporation of metals (Cr/Au = 5 nm/70 nm), and wet chemical etching of the p
+
-GaAs
contact layer except the region covered by the metal. Titanium dioxide (TiO2) with a thickness of
Figure 2. 8. A schematic illustration of nanofabrication of hexagonally periodic TiO 2 nanoposts on GaAs
photocathodes.
26
~420 nm was deposited by radio-frequency (RF) magnetron sputtering (Orion 5, AJA International)
at ambient temperature. An inert gas Ar (15 sccm) was used as a sputtering gas, and the pressure
was maintained at 3 mTorr during the deposition. After the deposition of TiO2, the surface was
treated by UV-ozone for 10 min, and ultrathin (<~300 nm) layer of polymer (SU-8 2000.5 diluted
in SU-9 thinner at 40 wt%, Microchem) was directly spin-coated onto ozone-treated TiO2 and
imprinted by using perfluoropolyether (PFPE)/NOA61/Polyethylene terephthalate (PET) stamp at
95º C for 15 s and subsequently at 45º C for 5 min. In order to remove the residual layer of polymer
where chrome hardmask will be deposited in the following step, the residue of SU-8 2000.5 (~75
nm) by oxygen reactive-ion etching (O2 RIE; Oxford PlasmaPro 80 RIE, 50 W, 40 mTorr, O2 50
Figure 2. 9. Cross-sectional SEM images of imprinted polymer layer (a) before and (b) after O 2 RIE dry etch for 150
sec. (c) A photograph image of PTFE/NOA/PET stamp. (d) A top-view SEM image of chrome nanoarrays after lift-
off of SU-8 2000.5 layer using acetone.
27
sccm, 150 s, room temperature). Figure 2. 9 (a) and (b) show cross-sectional SEM images of the
imprinted SU-8 2000.5 layer on TiO2, showing that the residue of polymer within holes was clearly
removed after O2 RIE. In this step, it is critical to stop O2 RIE at the optimal time when the residue
within hole structures is just fully removed while still maintaining sufficient polymer barriers for
a lift-off purpose. If etching time is shorter than the optimal, all chrome will be lifted off together
with the remaining residue of polymer within holes and no chrome dots remain. By contrast, if
etching time is longer than the optimal time, the diameter of chrome dots (i.e. TiO2 nanoposts
finally) becomes larger as the size of polymer holes increases by the over etching. If polymer is
further etched, the width and height of polymer barriers are not thick enough to be stripped by a
solvent because of the coverage of e-beam deposited chrome, resulting in the failure of lift-off.
After O2 RIE, thin Cr (~12 nm) was deposited by electron beam evaporation (Temescal) and lifted
off in acetone for 15 min under a gentle sonication. Figure 2. 9 (c) and (d) show a photographic
image of PTFE/NOA/PET stamp and Cr nanodots after lift-off. Using Cr nanodots as a hardmask,
TiO2 layer was etched by a reactive-ion etching (RIE; Oxford PlasmaPro 80 RIE) (200 W, 40
mTorr, SF6 5 sccm, CHF3 20 sccm, O2 5sccm, ~900 s, room temperature), followed by a removal
of residual hydrocarbon deposits using O2 descume (Plasmalab; 100 W, 200 mTorr, 1 min) and an
wet-chemical etching of remaining chrome nanodots with an etchant (CR-7, Cyantek) for 10 s.
Finally, TiO2 NPs were lithographically defined within the cell area in a diluted HF (HF (48%,
EMD):DI water = 1:10, by volume) for 9 min.
2. 2. 3. Fabrication of printed-assembly of bifacial GaAs photocathodes
Figure 2. 10 shows fabrication procedures for printed-assembly of nanostructured, ultrathin
GaAs photocathodes via transfer printing. After forming TiO2 NPs as discussed in the previous
chapter, isolated arrays of microscale (~580 580 m
2
) photoelectrodes were delineated by
28
photolithography and wet chemical etching, followed by the partial etching of the sacrificial layer
(i.e. Al0.9Ga0.1As) and spin-coating of photoresist anchor structure. After the undercut etching in
dilute hydrochloric acid (HCl:H2O = 1:2 by volume), a selective set of GaAs photoelectrodes was
then detached from the growth substrate by an elastomeric stamp made of polydimethylsiloxane
(PDMS) and printed over a glass substrate through a two-step transfer process. The n
+
-GaAs
contact layer that was originally grown at the bottom of the epitaxial stack is exposed to the front
as an active reactive surface for catalyzing hydrogen evolution reaction (HER), while the side of
Al0.4Ga0.6As window implemented with TiO2 NPs was faced down and embedded in a polymeric
printing medium (i.e. NOA 61) coated on the glass substrate. It is also worthwhile to reuse the
GaAs growth substrate by using a transfer printing, therefore, the process and material cost for III-
V photoelectrode could be further reduced. After the printing, the n-type ohmic metal contact
(AuGe/Ni/Au/Ti = 100 nm/30 nm/100 nm/30 nm or Pd/Ge/Au/Ti = 5 nm/35 nm/100 nm/30 nm)
Figure 2. 10. A schematic illustration of fabrication procedures for printed-assembly of nanostructured, ultrathin GaAs
photocathodes via transfer printing.
29
was deposited within the active junction area. The whole GaAs epilayers from n
+
-type GaAs
contact through p
+
-type GaAs contact were etched to define a mesa structure and to open p-type
ohmic contact metal in a mixture of phosphoric acid, DI water and H2O2 (H3PO4 (85%, Fisher
Scientific):DI water:H2O2 = 1:12:13, by volume) with photolithography. After the exposure of the
p-type metal contact and MESA structure, polyimide (PI2525 (HD MicroSystems); 500 rpm/20 s,
5500 rpm/30 s) was spin-coated and thermally cured at 200° C for 1 hour under N2 atmosphere in
the glove box, followed by the formation of ‘via’ holes for the metal interconnections via RIE
(100W, 100 mTorr, O2 45 sccm, CF4 5 sccm, ~15 min, room temperature). Metal interconnects
(Cr/Ag/Au = 15 nm/1500 nm/100 nm) were deposited by photolithography (AZnLOF 2070
(Merck KGaA) and electron beam evaporation (Temescal) to electrically connect the p-type and
Figure 2. 11. Optical microscopic images of GaAs photocathodes at every process step after forming (a) n-type ohmic
contact, (b) TiO 2 NPs and isolation, (c) after undercut etch, (d) after picking up to the first PDMS stamp, (e) after
transfer to the second PDMS stamp, (f) after printing, (g) after MESA etching, exposing p-type ohmic metal, and
forming n-type ohmic metal, and (h) after Pt deposition (i.e. completed GaAs photocathodes).
30
n-type ohmic metals for photovoltaic and photoelectrochemical measurements. After the metal
interconnection, polyimide encapsulated the electrode surface except the region where protection
layers and/or HER co-catalysts were incorporated using the same method used in the first
polyimide encapsulation. As a protective and co-catalytic layer, TiO2 and Pt were sequentially
deposited by RF magnetron sputtering and electron beam evaporation with a liftoff, respectively.
As an alternative earth-abundant co-catalyst instead of using Pt, amorphous MoSx was
electrochemically deposited by applying cathodic constant potential at -500 mV vs. RHE to ohmic
contact metal in a precursor aqueous solution of (NH4)2MoS4 (2 mM, Sigma Aldrich), 0.5 M of
Na2SO4 buffer (pH 6.6), and (NH4)2S ((NH4)2S:DI water = 1: 750 by volume, Macron, 20.0-24.0
wt% in H2O) for 500 s. After the deposition of co-catalyst, the completed GaAs photocathodes
were annealed at 150°C for 30 min under N2 atmosphere.
Optical microscopic images of GaAs photocathodes at respective processing steps are
shown as insets in Figure 2. 11, including after forming (a) n-type ohmic contact, (b) TiO2 NPs
and isolation, (c) after undercut etch, (d) after picking up to the first PDMS stamp, (e) after transfer
Figure 2. 12. (a) Top-view SEM images of nanostructured GaInP 2 (without (NH4)2S-treatment) at various etching
times. (b) Tilt-view schematic illustration of constructed nanostructured surfaces by 3D modeling software
(Rhinoceros® ) using the SEM images in (a).
31
to the second PDMS stamp, (f) after printing, (g) after MESA etching, exposing p-type ohmic
metal, and forming n-type ohmic metal, and (h) after Pt deposition (i.e. completed GaAs
photocathodes).
2. 3. Optical modeling and characterization
2. 3. 1. FDTD optical modeling for monofacial GaInP2 photocathodes
Reflectance and absorption spectra of nanostructured GaInP2 photocathodes were
numerically calculated by finite-difference time-domain method (FDTD Lumerical
TM
). To
generate a 3D model of nanostructured surface, top-view SEM micrographs (~1.2 x 0.9 m
2
) of
the dry-etched GaInP2 at various etching times were imported to a 3D modeling software
(Rhinoceros® ), where lateral profiles of nanopillars were adjusted to closely match with those
observed experimentally (see Figure 2. 12). The created nanostructured surface was further
imported to the FDTD software (FDTD Solutions, Lumerical
TM
). Figure 2. 13(a) and (b) show the
cross-sectional schematic illustrations of nanostructured GaInP2 for the FDTD calculation and
Figure 2. 13. (c) Cross-sectional schematic illustrations of nanostructured GaInP 2 for the FDTD calculation of
reflectance and absorption spectra in water. The inset shows a nanostructured GaInP 2 implemented in Lumerical
TM
.
(d) Measured refractive index (n) and extinction coefficient (k) of GaInP 2 by spectroscopic ellipsometry, which were
used in optical calculations.
32
imported model of nanostructured GaInP2 in Lumerical
TM
, respectively. For calculation at normal
incidence, a 3D simulation volume was confined with periodic boundary conditions for the x- and
y-directions, and a perfectly matched layers (PML) boundary condition for the z-direction, where
a continuous plane wave that has a broad Gaussian frequency spectrum (270-750 THz or 400-1100
nm) was assumed as a light source. The absorption within GaInP2 was defined as A = T1 − T2 by
assuming that absorption in water is negligible. To obtain more accurate results, measured
refractive index (n) and extinction coefficient (k) of GaInP2 by spectroscopic ellipsometry were
used in optical calculations (see Figure 2. 13(c)).
2. 3. 2. FDTD optical modeling for bifacial GaAs photocathodes
Optical spectra of nanostructured, ultrathin GaAs photocathodes were numerically
modelled by FDTD Lumerical
TM
. As shown in Figure 2. 14, a simulation volume comprising
epitaxial layers identical to the experimental GaAs photocathodes (GaAs photocathode coupled
with TiO2 NPs and ohmic metal) was defined, where periodic boundary condition in x- and y-
directions, and a perfectly matched layers (PML) boundary condition for the z-direction, where a
continuous plane wave that has a broad Gaussian frequency spectrum (270-750 THz or 400-900
nm) was assumed as a light source. Hexagonal arrays of periodic TiO2 nanostructures were
generated onto the GaAs model within the confined boundary and refractive index (n) and
extinction coefficient (k) of sputtered TiO2 measured by spectroscopic ellipsometry (VASE
Ellipsometer, J.A.Woollam) were used for the optical modeling in FDTD calculation. Figure 2.
14(b) shows schematic illustrations of a model for FDTD calculation with dimensional information
of TiO2 NPs and a plane of cross section for normalized absorbed power density (Pabs). Refractive
index and extinction coefficients of all materials used in optical calculations including Al0.4GaAs,
GaAs, air, water, and NOA61 (i.e. polymeric media) were obtained from literature, web-based
33
open source (NSM n-k database, http://www.ioffe.ru/SVA/NSM/Semicond/), and material
provider, and they were depicted in Figure 2. 14(c) and (d).
2. 3. 3. UV-Vis, Ellipsometry, and Photoluminescence spectroscopy
Reflectance spectra of GaInP2 photocathodes were recorded using UV-Vis-NIR
spectroscopy (Lambda 950, Perkin-Elmer) at near-normal incidence (θ = 8°) in air, measured on a
spectrophotometer equipped with an integrating sphere using a Spectralon® as a 100% reflectance
standard. For the small-featured samples such as encapsulated GaInP2 and GaAs photocathodes,
Figure 2. 14. (a) Cross-sectional schematic illustration of a nanostructured bifacial GaAs photocathodes printed on a
glass substrate. (b) Schematic illustrations of a model for FDTD calculation with dimensional information of TiO 2
NPs and a plane of cross section for P abs. (c) Refractive index (n) and (d) extinction coefficient (k) of various materials
used in FDTD optical calculation.
34
reflectance spectra were recorded using a home-made optical set-up consisting of a white light
source (HL-2000, Ocean Optics) and a fiber-optic spectrometer (Flame-T-VIS-NIR, Ocean Optics)
as shown in Figure 2. 15. The source light was collimated by an achromatic doublet lens (f = 19
mm, N.A. = 0.42) and then focused on the cell region (beam diameter = ~50 m) through an
objective lens (20X, N.A. = 0.4), allowing micro-scale UV-Vis spectroscopy. The reflected light
was collected by the same objective lens and guided to the spectrometer through a multimode fiber.
A silver mirror deposited on fused silica (PF10-03-P01, Thorlabs) was used as a 100% calibration
standard.
As described in the previous chapter, refractive index (n) and extinction coefficient (k) of
GaInP2 and TiO2 were measured by spectroscopic ellipsometry shown in Figure 2. 16(a) (VASE
Ellipsometer, J.A.Woollam). For the measurement and optical calculation of GaInP2, complex
Figure 2. 15. A schematic illustration of a homemade optics system for micro UV-visible spectroscopy. The inset
images show photographs of the homemade optics system, light source, objective lens, USB-type spectrometer, and
light collimator (i.e. doublet lens).
35
models were used, and Cauchy model was initially used for transparent region beyond the optical
bandgap edge of GaInP2 (i.e. ~677 nm or 1.83 eV) to calculate the thickness of GaInP2 by
𝑛 ( 𝜆 ) = 𝐴 +
𝐵 𝜆 2
+
𝐶 𝜆 4
Cauchy is good for transparent region but is not good for absorbing region of wavelengths for
semiconductors below ~670 nm (GaInP2). Due to this, once confirming the thickness, two
Gaussian oscillators within the genosc layer were further used to calculate the refractive index and
extinction coefficient of GaInP2. The real and imaginary optical functions ( 1 and ) are physically
connected by the Kramers-Kronig relationship, and then the values of 1 can be calculated by
above equation by
𝜀 1
( E)= 1 +
2
𝜋 𝑃 ∫
𝐸 ′
𝜀 2
( 𝐸 ′
)
𝐸 ′2
− 𝐸 2
𝑑 𝐸 ′
∞
0
if all the spectral values of are known
38
. Figure 2. 16(b) shows the comparison of the measured
optical constants of GaInP2 with the reported value
39
, showing a well-matched behavior. To
Figure 2. 16. (a) A photograph of VASE Ellipsometer (J.A.Woollam) and (b) measured optical constants of GaInP 2
comparing with reported values.
36
calculate the refractive index and extinction coefficient of TiO2 film, Urbach equation are further
used by
𝑘 ( 𝜆 ) = 𝛼 𝑒 𝛽 ( 𝐸 −𝐸 𝑏 )
, where E are Eb the incident photon energy (
1.24
𝜆 𝜇𝑚
eV) and bandgap energy of TiO2 (
1.24
𝜆 𝑏 ,𝜇𝑚
eV).
Photoluminescence (PL) spectra of bare and nanostructured GaInP 2 samples were
measured using a Raman microscope (XploRA
TM
, HORIBA Jobin Yvon Inc.) with 100X objective
lens (NA: 0.90), where a 532-nm laser was focused on the sample surface with a beam diameter
of ~1 m.
2. 4. Photoelectrochemical and photovoltaic measurement
PEC performance was measured in an aqueous solution (0.5 M, pH: ~0.33) of sulfuric acid
(H2SO4, EMD Chemicals, ACS grade, 95-98%) under simulated AM 1.5G standard solar
illumination (1000 W/m
2
) on a full-spectrum solar simulator (94042A, Oriel). The simulated light
was reflected by silver mirror (Thorlabs, PFSQ20-03-P01) to be exposed to the side of
photoelectrochemical cuvette cell (Colorimeter Cells 93-G-20, Starna). Figure 2. 17 shows a
schematic illustration of photoelectrochemical measurement setup for GaInP2 and GaAs
photocathode with a full-spectrum solar simulator in three-electrode configuration. The electrolyte
solution was purged with N2 for 15 min before each PEC measurement. Linear sweep voltammetry
data were collected by a potentiostat (Reference 600, Gamry) under a three-electrode configuration
with Ag/AgCl (3M NaCl, RE-5B, Bioanalytical Systems) and platinum (MW-1032, Bioanalytical
Systems) as reference and counter electrodes, respectively. A nonionic surfactant (Triton X-100,
SPI Supplies, ~0.1 mol/L) was incorporated in the electrolyte to release hydrogen bubbles from
the reactive interface. The potential sweep was done from negative to positive potential to avoid
37
overestimation of electrode performance. For the conversion of electrode potential from Ag/AgCl
to the reversible hydrogen electrode (RHE), a linear sweep voltammetry scan was performed using
a platinum electrode (MF-2013, Bioanalytical Systems) as a cathode to experimentally determine
the onset potential of hydrogen evolution. The photovoltaic performances were recorded using a
semiconductor parameter analyzer (4156C, Agilent) and a full spectrum solar simulator (94042A,
Oriel), under simulated AM 1.5G solar illumination (~1,000Wm
2
) calibrated by a reference silicon
solar cell (91150V, Newport). The illumination area was confined by aligning with an anodized
black aluminum aperture (1.5 × 1.5 mm
2
).
2. 5. IPCE measurement
Incident-photon-to-current efficiency (IPCE) for ultrathin GaAs photocathodes was
measured by National Renewable Energy Laboratory. IPCE measurements were performed in a
three-electrode configuration using a mercury/mercurous sulfate reference electrode (MSE) in an
Figure 2. 17. A schematic illustration of photoelectrochemical measurement setup for GaInP 2 and GaAs photocathode
with a full-spectrum solar simulator in three-electrode configuration.
38
aqueous solution (0.5 M) of sulfuric acid (5100A, Koslow Scientific) at 0 V (vs. RHE). To
accurately confine the light-incident area, titanium aperture (~600 × 600 mm
2
) was deposited on
to the backside of glass substrate by photolithography (see Figure 2. 18) and electron beam
evaporation (Ti 200 nm), and the illumination was chopped once at each wavelength at a frequency
of 0.2Hz. Each type of samples was measured by more than two separate cells and the
performances were averaged, showing a reliable reproducibility. A xenon arc lamp (67005,
Newport, 300 W) and monochromator (SP-50, Acton) generated monochromatic illumination in
Figure 2. 18. Optical microscopic images of GaAs photocathodes with titanium aperture captured from backside for
the IPCE purpose. Two separate GaAs cells were visible at the (a) out-focused and (b) in-focused depth through the
aperture and a transparent glass. (c) An optical microscopic image of GaAs photocathodes captured from frontside
under back illumination, where the bright region was the transmitted light of the back illumination. (d) Schematic
illustration of IPCE measurements for Ti-aperture-confined GaAs photocathodes.
39
10-nm increments (<10nm full-width at half-maximum)
8
. Overfill illumination was used, where
the flux density was measured at each wavelength using a calibrated Si photodiode (S1336-8BQ,
Hamamatsu). More detailed information of IPCE measurements appears elsewhere
8,40
2. 6. X-ray photoelectron spectroscopy
X-ray photoelectron spectroscopy (XPS) is a surface-analyzing technique that measures
the elemental composition, and empirical formula and oxidation states of elements by exposing an
beam of X-ray (typically Al Kα with a photon energy of 1486.6eV) to the surface of specimen and
measuring the kinetic energy of photoelectrons from their electronic quantum states (i.e. spin
orbitals). XPS was performed on a Kratos Axis Ultra DLD, where photoelectrons were generated
by monochromatic Al Kα X-ray at 1486.7 eV at a base pressure of 4 x 10
-8
torr. Binding energies
were calibrated by C 1s peak at 284.8 eV. Linear-least-squares fitting of XPS spectra was
performed using CasaXPS software (Casa Software Ltd.) using convolution of Gaussian (70%)
and Lorentzian (30%) line-shapes. The peak positions and relative intensity ratios are nearly
identical between bare and nanostructured samples under the present experimental conditions
2. 7. Faradaic efficiency measurement
Faradaic efficiency was measured in collaboration with the group of Dr. Todd Deutsch at
National Renewable Energy Laboratory. Hydrogen and oxygen gases were collected
volumetrically by a Hoffman-type apparatus from bare GaInP2 photocathodes with and without
(NH4)2S-treatment
7
. The photocathodes were operated at 0 V vs. RHE using Pt and Hg/Hg2SO4
(MSE) as counter and reference electrodes, respectively, in 0.5 M sulfuric acid illuminated by a
tungsten-halogen lamp with water filter calibrated to one Sun intensity using a GaInP2 reference
cell. Faradaic efficiency was calculated with the following equation:
40
η
H
2
=
Collected gas quantity ( mol)
Expected gas quantity ( mol)
=
(
P
H
2
V
RT
)
( Charges passed)× (
mol e
−
96485 C
) × (
1 mol H
2
2 mol e
−
)
, where 𝑃 𝐻 2
is the pressure of the evolved hydrogen gas, V is the volume, R is the gas constant
(62363 mL torr K
-1
mol
-1
) and T is temperature (292.59 K). PH2 was adjusted from atmospheric
pressure Patm by subtracting out the water vapor pressure PH2O vapor and the pressure from the
suspended solution Psuspended as
𝑃 𝐻 2
= 𝑃 𝑎𝑡 𝑚 − 𝑃 𝐻 2
𝑂 𝑣𝑎𝑝𝑜𝑟 − 𝑃 𝑠𝑢𝑠𝑝𝑒𝑛𝑑𝑒𝑑
where, Patm and PH2O vapor were 615.08 and 16.8 torr, respectively. Psuspended was calculated by
measuring the height h1 of the suspended solution above the solution level in the PEC cell.
𝑃 𝑠𝑢𝑠𝑝𝑒𝑛𝑑𝑒𝑑 = ℎ
1
×
𝐻𝑔 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑆𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 ×
1 𝑎𝑡𝑚 760 𝑚𝑚𝐻𝑔 = ℎ
1
× 0.0736 ( 𝑡𝑜𝑟𝑟 −1
𝑚𝑚
−1
)
41
Chapter 3. Monofacial Nanostructured GaInP2 Photocathodes
3. 1. Introduction
Superior materials properties including a direct bandgap, appropriate band-edge energetics,
as well as capability to form monolithically multiple junctions rendered III-V compound
semiconductors one of the most attractive candidates as photoelectrodes in solar-driven PEC water
splitting. Gallium indium phosphide (Ga0.51In0.49P, referred to as GaInP2) among III-V compound
semiconductors, a ternary alloy lattice-matched to gallium arsenide (GaAs) shown in Figure 3.
1(a), represents an enabling material for ultrahigh efficiency water splitting systems because of its
near- ideal bandgap (~1.8 eV) for the top junction in dual-junction (2J) photoelectrodes by
absorbing up to ~670 nm of wavelength of light as shown Figure 3. 1(b). Since the electrolysis of
water is a photoelectrochemical reaction at the interface between electrolyte and top junction
material, it is very important to understand and optimize the top-junction interface. Figure 3. 1(c)
shows bandgap energies and band-edge positions of various semiconductor photoelectrodes at the
interface with electrolyte (pH = 0) relative to the NHE and the vacuum level
41
. As discussed in
Chapter 1, the ideal material for photoelectrodes must satisfy several characteristics of bandgap
and band edge positions; its conduction and valence band edges must straddle the hydrogen and
oxygen redox potentials to minimize overpotentials for HER and OER, respectively. Figure 3. 1(d)
shows band-edge positions of p-type GaInP2 photocathode relative to hydrogen and oxygen redox
potentials. The position of conduction band edge is more negative than hydrogen redox potential
so that, it provides sufficient potential for photogenerated electrons to drive HER. However, the
position of its valence band edge is too negative than oxygen redox potential, meaning that it
requires external potential to drive OER reaction for unassisted water splitting. This is the reason
42
why GaInP2 has been generally used as a top junction material in multiple-junction
photoelectrochemical devices instead of being used itself as a single-junction absorber. As shown
in Figure 3. 2, GaInP2 has been widely used as a top junction material in most of monolithic III-V
multiple-junction photoelectrochemical devices because its wide bandgap allows it to absorb the
wavelength of light up to 670 nm corresponding to its bandgap energy, while the bottom layer,
such as GaAs or GaInAs absorbs the longer wavelength of light above the absorption edge of
GaInP2. For example, in 1998, Khaselev et al. at NREL first demonstrated a type of monolithic
Figure 3. 1. (a) Bandgap energy and lattice constant of various III-V semiconductors. (b) Spectral irradiance of the
AM1.5 Global spectrum and absorption spectrum of GaInP 2 (672 nm). Raw spectrum data was downloaded from PV
Education (https://www.pveducation.org) (c) Bandgap energies and band-edge positions of various semiconductor
photoelectrodes relative to the NHE and the vacuum level at pH = 0. Copyright (2017) Sustainable Energy & Fuels.
(d) Band-edge alignment of p-type GaInP 2 photocathode relative to hydrogen redox potential and oxygen redox
potential in non-equilibrium state
43
III-V PEC photoelectrodes by using GaInP2/GaAs tandem structure with 12.4% of solar-to-
hydrogen (STH) efficiency, which has been a still high number among reported values
5
. In 2017,
Young et al. from the same research group demonstrated inverted metha-morphic (IMM)
GaInP2/GaInAs tandem PEC cell and reported 16.4% of STH efficiency
7
. In 2018, Cheng et al.
from Atwater research group at Caltech showed 19.2% of STH using GaInP2/GaInAs tandem PEC
cells by incorporating a TiO2 antireflection layer
9
. In such tandem electrode systems, GaInP2, in
Figure 3. 2. (a) Schematic of the monolithic GaInP 2/GaAs tandem PEC cell with 12.4% of STH and (b) its equilibrium
energy band diagram. Copyright (1998) Science. (c) Schematic of the monolithic GaInP 2/GaInAs tandem PEC cell
with 16.2% of STH. Copyright (2017) Nature Energy. (d) Schematic of the monolithic GaInP 2/GaInAs tandem PEC
cell with 19.3% of STH. Copyright (2018) ACS Energy Letters.
44
direct contact with an acidic electrolyte, is responsible for multiple roles in not only absorbing
sunlight but also catalyzing hydrogen evolution reaction (HER) while protecting itself and beneath
layers from photocorrosion. These are the motivations of this work why studying top junction
material is important as shown in Figure 3. 3. First, controlling the morphology of the surface of
top junction layer eliminate reflection loss, which is ~25% at the interface between electrolyte and
a plane III-V material. By adopting textured top surface, reflectance can be reduced. For example,
Figure 3. 3. (a) Multiple roles of catalytic interface from monofacial photoelectrodes in light absorption,
electrocatalysis, as well as and corrosion protection. Previous approaches of (a) nanostructures to reduce reflection
loss (Copyright (2011) Energy & Environmental Science and (2012) Angewandte Chemie), (c) modulation of energetic
states to improve carrier separation (Copyright (2015) Science and (2017) Nature Energy), and (d) the addition of
protective layers to prevent photocorrosion of semiconductors (Copyright (2014) Science and (2017) Nature Energy), .
45
density-graded nanostructures can suppress reflectance below 1 %
22,24
. Current density in PEC
performance is one of most important factors to determine a high and efficient yield of hydrogen
production, and it is directly controlled by light absorption. Due to these reasons, many researchers
have tried to reduce the top junction reflectance by implementing nanostructured surface
22,24,42-
45
(see Figure 3. 3(b)). The second role of top junction material is the control of energetic states of
catalytic interface to efficiently drive HER reaction. Since the energetic position of band edges at
the catalytic interface affects the transfer of photogenerated carriers from semiconductor to
electrolyte, which governs onset potential including overpotential and photovoltage in PEC
performance. To make the reaction more favorable, researchers adjusted the energetic position of
band edges by surface modifications, such as adding a TiO2 layer to block hole-induced
recombination
46
or n-type buried junction on p-type layer
7
(see Figure 3. 3(c)). The third role is
the protection which prevent photocorrosion of semiconductor because most of electrolyte are
strong acid or base solution which easily etch III-V compound semiconductors. Due to this reason,
the top junction material determines how long the material can stably operate while performing
hydrogen production, which is one of most important requirements for their practical applications.
Many studies have focused on protecting the top junction absorber by covering it with stable metal
oxides, such as amorphous TiO2
47
or MoSx/TiOx bilayers
21
as shown in Figure 3. 3(d), which often
involves complicated and/or high-temperature processes (e.g. atomic layer deposition or
electrochemical deposition) and might reduce charge transfer rates or light absorption.
Considering these three roles of top junction material, engineering GaInP2 will bring many
advantages in terms of light absorption, the transfer rate of photogenerated carriers, and
photoelectrochemical stability. Maximizing the light absorption while promoting catalytic
efficiency and electrochemical stability of GaInP2 is therefore of critical importance to realize the
46
full potential of III-V tandem electrode system. However, there has been no study that achieved
all-balanced improvements of three functionalities by solely controlling the top junction material.
In this study, we therefore explored a simple approach to improve light absorption, charge transfer,
and stability via synergistically tailored structure and stoichiometry at the catalytic interface.
3. 2. Fabrication
As a light absorbing material for solar water splitting, p-type GaInP2 (Zn-doped, 2 × 10
17
cm
-3
) with a thickness of 2.5 m was grown on a (100) GaAs substrate by an atmospheric pressure
metal organic vapor phase epitaxy (MOVPE)
7,10,18,21,27,48
. As discussed in Chapter 2, monofacial
nanostructured GaInP2 photocathodes were prepared by a simple fabrication process as
schematically illustrated in Figure 3. 4. The process begins with the removal of native oxide on
as-grown p-type GaInP2 using a dilute NH4OH and with the chemical activation of GaInP2 surface
using a perchloric acid (CR-7). Silver nanoparticles were grown on the treated surface by dipping
into an aqueous solution of silver nitrate (AgNO3) and hydrofluoric acid (HF). During the
Figure 3. 4. Schematic illustration, morphological and optical properties of surface-tailored GaInP 2 photocathodes. a
Schematic illustration of fabrication procedures for surface-tailored ‘black’ GaInP 2 photocathodes.
47
electroless deposition, the silver cations (Ag
+
) in the solution are reduced into silver nanoparticles
that can serve as a hard mask in successive plasma-induced dry etching of GaInP2, where the
density and size of silver nanoparticles can be readily controlled by adjusting activation time, the
concentration of precursors, and/or plating time
49
. Afterwards, inductively coupled plasma
reactive ion etching (ICP RIE) was conducted using a gas mixture of BCl3/N2 to form cone-shaped
nanopillars of GaInP2, followed by the removal of residual hard mask (i.e. silver) by a wet chemical
etchant. Figure 3. 5(a) and (b) show SEM images of electrolessely deposited silver nanoparticles
on plane surface of GaInP2 and nanostructured GaInP2 surface after dry-etching, respectively. As
depicted in the SEM image, the diameter of nanopillars gradually decreased from the bottom to
the tip, thereby naturally constructing a graded index of refraction to suppress reflection
50,51
. It is
evident that the ‘black’ surface of nanostructured GaInP2 is distinguished from the shiny surface
of bare GaInP2 (see Figure 3. 5(c)). After the formation of nanoporous morphology, chemical
passivation of dry-etched surface was performed through two-step process consisting of (NH4)2S-
Figure 3. 5. (a) A top-view SEM image of electrolessly deposited silver nanoparticles on GaInP 2 as a mask for dry
etching (scale bar: 1 m). (b) Tilt-view scanning electron microscope (SEM) image of nanostructured p-type GaInP 2
photocathodes after the dry etching and before the (NH 4) 2S-treatment (scale bar: 500 nm). (c) The photographic
images of fully-functional bare (labeled as “Bare”) and nanostructured (labeled as “NS”) GaInP 2 photocathodes
mounted on a slide glass with epoxy encapsulation (scale bar: 5 mm), where the ‘black’ surface of nanostructured
GaInP 2 is evidently shown in contrast to the shiny surface of bare GaInP 2.
48
passivation and thermal annealing (see Figure 3. 6). An aqueous solution of ammonium sulfide
((NH4)2S) was cast onto the surface of GaInP2 and dried at 85 C in air, followed by thermal
Figure 3. 6. (a) Schematic illustration of two-step (NH 4) 2S treatment. (b) XPS spectra of S 2p and Ga 3s before
(NH 4) 2S-treatment (left), after drying at 85 C (center), and after heating at 250 C in air (right), supporting the
formation of sulfide and sulfate species at each step. The second XPS spectra of S 2p was measured after washing the
sample with DI water to remove a thick polysulfide that shown on (c). (c) Scanning electron microscope (SEM) image
and results of energy dispersive spectroscopy (EDS) from the sample right after the casting at 85 C without washing
in DI water. Thick islands of (NH 4) 2S-based polysulfides are clearly observed. EDS spectra indicate the formation of
oxides and carbon-containing species after this first step of (NH 4) 2S-treatment. Because of the penetration depth (2~3
mm) of highly-energized electron beam (10 kV), arsenic was detected from GaAs substrate, and the atomic percentage
of gallium was slightly higher than indium.
49
annealing at 250C in air to passivate plasma-induced surface defects and to integrate corrosion-
resistant surface stoichiometry (i.e. sulfide and sulfate bonding). In the following, this two-step
process consisting of (NH4)2S-passivation and thermal annealing is referred to as (NH4)2S-
treatment unless additional specifications are given. The (NH4)2S-treated, nanostructured GaInP2
was then electrically connected with a copper wire on the backside-metal contact (Ti/Au : 10/300
nm) and encapsulated by thermally cured epoxy (Loctite 9460/9462) to form fully-functional
photocathodes ready for driving the hydrogen evolution half-reaction in solar water splitting.
3. 3. Results and discussion
Efficient optical coupling between the semiconductor photoelectrode and sunlight is one
of the key advantages for the reported ‘black’ GaInP2. Figure 3. 7 shows the average heights of
GaInP2 nanostructures measured from cross-sectional SEM images, showing approximately ~60,
~320, ~450, and ~550 nm for etching times of 1, 2, 3, and 4 min, respectively. Overall, the height
Figure 3. 7. (a) Cross-sectional SEM images of the nanostructured GaInP2 (yet without (NH4)2S-treatment) at etching
times of 1, 2, 3, and 4 min (scale bar: 300 nm) and (b) corresponding etching depth of measured from SEM images.
Error bars represent the range of values obtained from three separate measurements (n = 3).
50
of nanostructures increased with the time of ICP-RIE process, while the etch rate (i.e. a slop of
black line) diminished after 2 min due to the gradual etching of an entire tip of nanopillars upon
the current experiment conditions. Figure 3. 8 shows the corresponding total (i.e. sum of diffuse
and specular) reflectance spectra of nanostructured GaInP2 measured at near-normal incidence (
= 8°) in air. The front-surface reflectance was strongly suppressed over a broad range of
Figure 3. 8. (a) Corresponding total (i.e. specular and diffuse) reflectance spectra of nanostructured GaInP 2 measured
on spectrophotometer equipped with an integrating sphere at an incidence angle of 8° and calculated (dotted line)
reflectance spectra obtained from FDTD-based numerical optical modeling matched well with the experimental (solid
line) spectra. (b) Zoomed-in, total reflectance spectra below 5% shown in a. (c) Calculated absorption spectra and (d)
integrated solar flux absorption (S_abs) of nanostructured GaInP 2 in water calculated using the numerical model
established in a.
51
wavelength arising from the improved wave-impedance matching owing to a gradually varying
refractive index
50,52,53
. The reflectance at 500 nm decreased from ~33% for bare GaInP2 to less
than ~1% for nanostructured (3- and 4-min-etched) samples over broadband wavelength (see
Figure 3. 8(b)). The measured reflectance spectra (solid line) quantitatively well-matched with the
calculated spectra (dotted line) simulated from 3D full-wave numerical optical modeling based on
finite-difference time-domain method (FDTD, Lumerical
TM
) as described in Chapter 2. Using the
proven numerical model, we also estimated the absorption enhancement of nanostructured GaInP2
in the electrolyte (i.e. water) (see Figure 3. 8(c)). Figure 3. 8(d) shows integrated solar flux
absorption (Sabs) weighted over a simulated AM1.5G solar illumination calculated by
𝑆 𝑎𝑏𝑠 ( %) =
∫
𝜆 ℎ𝑐 𝐴 ( 𝜆 ) 𝐼 1.5𝐺 ( 𝜆 ) 𝑑𝜆 678 𝑛𝑚
400 𝑛𝑚
∫
𝜆 ℎ𝑐 678 𝑛𝑚
400 𝑛𝑚
𝐼 1.5𝐺 ( 𝜆 ) 𝑑𝜆 × 100
where h, c, A( ), and I1.5G ( ) are Planck’s constant, the speed of light, calculated absorption, and
the standard solar irradiance (AM 1.5G; ASTM G-173), respectively
8,37
. The Sabs of
nanostructured GaInP2 in water is ~93% (for 4-min-etched sample), which is considerably higher
than the maximum absorption (~72%) of bare (i.e. planar surface without nanopillars) GaInP2 in
water and can be directly translated to the enhanced electrode efficiency.
The photoelectrochemical (PEC) measurement of nanostructured GaInP2 photocathodes
for the hydrogen evolution reaction (HER) were conducted in a three-electrode configuration
under simulated AM1.5G solar illumination (1000 W m
-2
), where Pt and Ag/AgCl were used as
counter and reference electrodes, respectively, with aqueous sulfuric acid (0.5M H 2SO4) as an
electrolyte (see Figure 2. 17). Figure 3. 9 shows current density (J)- potential (E) curves of the
nanostructured GaInP2 photocathodes prepared at various etching times, obtained from a linear
52
sweep voltammetry from -0.5 to 0.4 V (vs. reversible hydrogen electrode (RHE)), where the data
from the first scan were plotted. All samples in this study were prepared still without (NH4)2S-
treatment. The half-reaction diagnostic efficiency ( cathode) of GaInP2 photocathodes for the HER
was calculated by
Figure 3. 9. Photoelectrochemical performance of nanostructured GaInP 2 photocathodes for the HER. Representative
J-E curves of bare and nanostructured GaInP2 photocathodes driving the hydrogen evolution reaction (HER),
measured under simulated AM1.5G solar illumination (1000 W/m
2
). All samples were measured without (NH 4) 2S-
treatment.
Table 3. 1. A summary of photoelectrochemical (PEC) performance characteristics of nanostructured GaInP 2
photocathodes performing the HER, extracted from Figure 3. 9.
53
𝜂 𝑐 𝑎 𝑡 ℎ𝑜𝑑𝑒 ( %) =
𝐽 𝑚𝑎𝑥
∙ (𝐸 𝑚𝑎𝑥
− 𝐸 (
H
+
H
2
))
𝑃 𝑖𝑛
× 100
, where Jmax and Emax are the current density and electrode potential at a maximum power point,
E(H
+
/H2) is the thermodynamic HER potential, and Pin is the power density of simulated AM1.5G
solar illumination
8,29,54
. While this diagnostic efficiency obtained from three-electrode
configuration does not fully reflect the solar-to-hydrogen (STH) efficiency of overall water
splitting reactions, it is used here as a simple metric to compare the photoelectrochemical
performance quantitatively
29,54
. As summarized in Table 3. 1, both the onset potential (Vonset) and
fill factor (FF) of all nanostructured GaInP2 photocathodes anodically (i.e. positively along the x-
axis) shifted compared to the bare (i.e. planar surface) GaInP2 because of the reduction of local
current density related with the enlarged surface area and corresponding decrease of over-
potential
22,24
. Consequently, the overall efficiency greatly improved in nanostructured samples.
Figure 3. 10. (a) Corresponding steady-state photoluminescence (PL) spectra of bare and nanostructured GaInP 2
photocathodes without (NH4)2S passivation. (b) Steady-state PL spectra of as-received bare GaInP 2 before and after
the dry-etching (4 min) without silver nanoparticles (i.e. still plane surface but dry-damaged) measured at room
temperature. The etching was performed without silver nanoparticles using the same condition (BCl 3/N 2 (1.5/9.0
sccm), 100W/500W , 5 mTorr, 100°C) as for nanostructured GaInP 2
54
On the other hand, the saturated current density (Jsat) slightly increased for 1-min-etched sample
compared to the bare GaInP2 owing to the suppressed reflection loss but became smaller at longer
etching times, which is in contrast to the estimation from optical calculations. This conflicting
observation implies a large degree of surface-non-radiative recombination of photogenerated
carriers caused by plasma-induced crystalline defects and oxidation at the nanostructured surface,
Figure 3. 11. (a) XPS spectra of Ga 2p 3/2 for bare and nanostructured GaInP 2. The measured spectra (black line)
matched quantitatively with fitted spectra (green dotted line) composed of deconvoluted Ga-O (blue line) and Ga-P
(red line) peaks. (b) XPS spectra of P 2p 1/2 and 2p 3/2 for bare and nanostructured GaInP 2. The fitted spectra were
deconvoluted to resolve P-Ga (red line) and P-O (blue line) peaks. (c) XPS spectra of In 3d 5/2 for bare and
nanostructured GaInP 2. The measured spectra (black line) matched quantitatively with the fitted spectra (green dotted
line) composed of deconvoluted In-O (blue line) and In-P (red line) peaks. All samples were yet without (NH 4) 2S-
treatment. (d) Inter-band defect states of oxygen- and/or crystalline-defect sites.
55
as supported by the severe attenuation of steady-state photoluminescence (PL) with the dry-etched
GaInP2 (see Figure 3. 10). Regardless of having nanostructured interface, dry etching significantly
lowered the peak intensity at bandgap edge (i.e. ~670 nm), which might be related with plasma-
induced surface defects. To elucidate what kind of defects hinders the coupling between the
saturation current density and optical enhancement, X-ray photoelectron spectroscopy (XPS) was
performed on the surface of GaInP2 before and after the dry etching. In Figure 3. 11(a), the Ga
2p5/2 peaks with binding energies of 1118.0 and 1116.7 eV observed from bare GaInP2 (i.e. GaInP2
without dry etching) correspond to 3+ oxidation states for Ga2O3 and GaInP2, respectively.
Notably, the integrated area of Ga-O peak, reflecting the relative amount of Ga-O bonding,
substantially increased after the dry etching (i.e. NS, nanostructured GaInP2), suggesting the
incorporation of oxygen atoms at the etched surface of GaInP2. With similar origins, the relative
amount of P-O and In-O peaks increased over P-Ga and In-P peaks after the formation of
nanopillars by dry etching (see Figure 3. 11(b) and (c)). Figure 3. 11(d) shows a schematic inter-
band defect states of oxygen- and/or crystalline-defect sites within conduction and valence band
edges of GaInP2 or III-V compound semiconductors. It is well-known that the generation of
oxygen defect states within the bandgap can act as centers for non-radiative carrier
recombination
29-32
.
To transfer a full advantage of ‘black’ GaInP2 into photo-generated current, it is therefore
essential to address the issue of increased carrier recombination associated with etching-induced
surface oxidation but also to protect the nanoporous morphology from corrosion. To this end, we
soaked the nanostructured GaInP2 in a dilute solution of ammonium sulfide ((NH4)2S), followed
by thermal annealing in air (250 C for 1 h, Figure 3. 6)
30,55
. Figure 3. 12 shows that the nanoporous
morphologies and reflectance spectra of dry-etched GaInP2 were still preserved after the (NH4)2S-
56
treatment. Figure 3. 13 shows J-E curves of nanostructured GaInP2 (4-min etched) photocathodes
before and after the (NH4)2S-treatment, plotted together with the data for bare GaInP2 as a
reference. As expected, nanostructured GaInP2 before the sulfur passivation exhibited the smaller
saturated current density (~10.9 mA cm
-2
) than that of bare GaInP2 (~13.3 mA cm
-2
) owing to the
above-described non-radiative carrier recombination. By contrast, the Jsat of the nanostructured
Figure 3. 12. Top-view and tilted-view SEM images of (a) nanostructured (4-min etching) GaInP 2 without (NH 4) 2S-
treatment nor MoS 2-deposition, (b) nanostructured GaInP 2 after (NH 4) 2S-treatment (15 min, yet without MoS 2-
deposition) and (c) nanostructured GaInP 2 after MoS 2-deposition (yet without (NH 4) 2S-treatment) All samples used
in (a), (b), and (c) were prepared together using one piece of wafer up to the process of dry-etching, and they were
cleaved into three pieces for each case. (d) Corresponding total reflectance spectra of nanostructured (4-min etching)
GaInP 2 without (NH 4) 2S-treatment nor MoS 2-deposition, nanostructured GaInP 2 after (NH 4) 2S-treatment (15 min, yet
without MoS 2-deposition), and nanostructured GaInP 2 after MoS 2-deposition (yet without (NH 4) 2S-treatment)
57
GaInP2, after the (NH4)2S-treatment for 3 and 5 min, increased to ~14.4 and ~15.2 mA cm
-2
,
respectively, resulting in a large enhancement of half-reaction efficiency ( cathode) by over 100%
(relative) compared to the untreated samples (see Table 3. 2). Such large improvement of Jsat was
further studied by PL and XPS analysis. Figure 3. 14 shows the partial recovery of PL intensities
(blue line) as well as the reduced areas of Ga-O-, P-O-, and In-O-related peaks in XPS spectra
(blue line), suggesting that the substitution of oxygen atoms by sulfur at the damaged surface of
Table 3. 2. A summary of photoelectrochemical (PEC) performance characteristics of bare and nanostructured GaInP 2
Photocathodes with and without (NH 4) 2S treatment, extracted from Figure 3. 13.
Figure 3. 13. Representative J-E curves of bare and nanostructured GaInP 2 photocathodes for the HER after the
(NH 4) 2S-treatment of 3 and 5 min, measured under simulated AM1.5G solar illumination (1000 W m
-2
).
58
nanostructured GaInP2 and corresponding decrease of oxide-related defect states effectively
alleviated the extent of non-radiative surface recombination and thus improved the efficiency of
charge transfer at the catalytic interface
30,33
.
The photoelectrochemical efficiency of (NH4)2S-treated, nanostructured GaInP2 can be
further boosted by additionally depositing materials of highly active catalyst (i.e. co-catalyst) on
the surface of GaInP2 photocathodes. Typically, noble metals such as Pt have been widely used as
effective co-catalyst, which provides a minimum over potential. Alternatively, earth-abundant co-
Figure 3. 14. (a) Photoluminescence intensities of nanostructured (4-min-etched) GaInP 2 before and after (NH 4) 2S-
treatment (5 min). XPS spectra of (b) Ga 2p 3/2 and (c) P 2p 1/2 and 2p 3/2, and (d) In 3d 5/2 for nanostructured GaInP 2
before and after the (NH 4) 2S-treatment (15 min).
59
catalysts such as amorphous MoS2 or MoSx have been intensively studied in the last decade to
replace such noble-metal co-catalysts
56-60
. In the present study, a thin layer of amorphous
molybdenum disulfide (MoS2) was photochemically deposited on the nanostructured and (NH4)2S-
treated GaInP2 as a HER co-catalyst (see Figure 3. 15). As expected, the MoS2 co-catalyst
considerably improved the photoelectrochemical efficiency of (NH4)2S-treated, nanostructured
GaInP2, with a large enhancement in both onset potential and fill factor, resulting in the significant
Figure 3. 15. (a) Representative J-E curves of bare GaInP 2 (black line), nanostructured GaInP 2 after (NH 4) 2S-treatment
(red line), and nanostructured GaInP 2 after (NH 4) 2S-treatment and MoS 2 deposition (blue line), measured in an acidic
electrolyte (0.5M H 2SO 4) under simulated AM 1.5G illumination. (b) XPS spectra of Mo 3d and S 2p of amorphous
molybdenum disulfide deposited on GaInP 2 as a HER co-catalyst.
Table 3. 3. A summary of photoelectrochemical (PEC) performance characteristics of nanostructured GaInP 2
photocathodes deposited with MoS 2 performing the HER, extracted from Figure 3. 15.
60
increase of half-reaction efficiency (cathode) by ~20 times compared to the bare GaInP2 without
co-catalysts (see Table 3. 3). Due to the ultrathin layer of MoS2 (~10-30 nm, see Figure 3. 16)
confirmed by atomic-force microscopy (AFM), saturation current density was maintained without
any optical loss, which was also confirmed by reflectance of MoS2-deposited, nanostructured
GaInP2 photocathodes (see Figure 3. 12).
Long-term lifetime and catalytic performance of III-V semiconductor photoelectrodes is
one of the most critical requirements for their practical application in solar water splitting.
Figure 3. 16. Tapping-mode AFM images of (a) bare GaInP 2 and (b) MoS 2-deposited bare GaInP 2. (c) Zoomed-in
image of MoS 2-deposited bare GaInP 2. The inset shows the distribution of height on the image. (d) The height (z)
profiles corresponding to the scan lines in (c).
61
Nevertheless, III-V compound semiconductors including GaInP2 suffer from intrinsic
thermodynamic instability and corrode rapidly in a wide range of pH under the potentials of water
splitting reactions, thereby leading to the fast degradation of photoelectrode functionality with an
impractically short lifetime
47,61-63
. In this regard, the reported surface-tailoring strategy (i.e.
tailoring both surface stoichiometry and morphology at the same time) provides a potential route
to strongly improve the durability of III-V photoelectrodes. Figure 3. 17 shows the current density
of GaInP2 photocathodes as a function of time (J-t) for various materials configurations including
bare GaInP2 (i.e. with unetched and untreated surface), bare GaInP2 deposited with MoS2 co-
catalyst, nanostructured GaInP2 with (NH4)2S treatment, and nanostructured GaInP2 with MoS2-
deposition, measured 0 V (vs. RHE) in an acidic electrolyte (0.5 M H2SO4) under simulated
AM1.5G solar illumination. In this figure, all MoS2-deposited samples were not treated by
(NH4)2S and were used for references as protected GaInP2 photocathodes because MoS2 have been
Figure 3. 17. Electrochemical durability of surface-tailored GaInP 2 photocathodes performing the HER under bias.
Current density–time (J–t) plots of GaInP 2 photocathodes in an acidic electrolyte (0.5M H 2SO 4) for (a) short- and
(b) long-term measurements, at various materials configurations including bare GaInP 2 (black data), bare GaInP 2
deposited with MoS 2 (yet without (NH 4) 2S-treatment, red data), nanostructured GaInP 2 with (NH 4) 2S-treatment
(yet without MoS 2-deposition, blue data), and nanostructured GaInP 2 with MoS 2 (yet without (NH 4) 2S treatment,
green data), measured at an electrode potential of 0 V (vs. RHE) under simulated AM1.5G solar illumination. Dry
etching and (NH 4) 2S-treatment were performed for 4 min and 15 min, respectively.
62
widely used as a protective layer as well as HER co-catalyst
10,11,21,64-66
. For all samples presented
here, processes of dry etching and (NH4)2S-treatment were performed for 4 min and 15 min,
respectively. In short-term J-t measurements (i.e. up to ~60 min, Figure 3. 17(a)), all tested samples
exhibited nearly constant current densities except the bare GaInP2, where the rapid degradation of
current density (J) with bare GaInP2 (black data) is attributed to the cathodic shift of the J-E curve
arising from the formation of surface oxide in water, which gradually plateaued due to the self-
limiting nature of wet oxidation
67
. The bare and nanostructured GaInP2 deposited with MoS2 (red
and green data) showed comparatively stable performance because of the temporary prevention of
surface oxidation and photocorrosion of GaInP2 by the MoS2 layer. It is noteworthy that the
nanostructured GaInP2 after the (NH4)2S-treatment (blue data) remained stable even without
relying on additional protective materials. The extraordinary PEC durability of nanostructured
GaInP2 after the (NH4)2S-treatment was verified more evidently in long-term measurements as
depicted in Figure 3. 17(b). The current density of bare and nanostructured GaInP2 with MoS2
was maintained for ~6 and ~20 hours, respectively, which is attributed to the electrochemical
dissolution of amorphous MoS2
64
and is still comparable to the reported lifetime by other research
groups
21
. It is noteworthy that nanostructured GaInP2 with MoS2 showed longer lifetime, resulting
from the enlarged surface area which could retard the electrochemical dissolution of MoS2 by
accelerating the charge transfer rates for HER which is competing with dissolution rate of MoS 2.
This effect of enlarged surface area is also very important to interpret one of the reasons why
(NH4)2S-treated, nanostructured GaInP2 maintained current density without an aid of additional
protective layers as follows. The current density of nanostructured GaInP2 with (NH4)2S-treatment
(blue data) was maintained nearly undiminished ( J < ~2%) for over ~124 hours, where the
63
measurement was terminated without observing the degradation of photoelectrode performance.
For this sample, it is also notable that the onset potential continuously improved during the J-t
stability test (see Figure 3. 18(a)), which might be attributed to several factors including the
activation of catalytic sites of sulfurized GaInP2 as well as enhanced photovoltage and charge
transfer efficiency, all occurring with the removal of oxides (see Figure 3. 18(b), (c), and (d))
and/or carbon-containing species that are unstable in the HER not by the redeposition of Pt from
counter electrode (see Figure 3. 19). By contrast, the bare GaInP2 both with (red data) and without
MoS2 (black data) rapidly degraded fast in the early stage of measurement. As mentioned above,
Figure 3. 18. (a) J-E curves of nanostructured and (NH 4) 2S-treated GaInP 2 measured during the stability test (Figure
3. 17(b)). XPS spectra of (b) Ga 2p, (c) In 3d, and (d) P 2p obtained from (NH 4) 2S-treatd bare GaInP 2 photocathode
before and after the stability test performed for 124 h shown in Figure 3. 17(b).
64
in case of the nanostructured GaInP2 with MoS2 but without (NH4)2S-treatment (green data),
significant degradation was still noted owing to the delamination or dissolution of MoS2 during
the HER. Consistent with these observations, the density-graded surface morphology of
nanostructured and (NH4)2S-treated GaInP2 remained nearly intact after the chronoamperometry
study (Figure 3. 17), as evidenced by the SEM images (Figure 3. 20(a)) as well as the preservation
of ‘black’ appearance and low reflectance (Figure 3. 20(b)). For bare GaInP2, however, large
degrees of photocorrosion already occurred just after the ~2.5-hour measurement, where In-rich
particles with sizes ranging from a few microns to tens of nanometers appeared on the surface due
to the selective dissolution of Ga as reported in the previous literature
68,69
. The change of surface
Figure 3. 19. Pt-related XPS survey spectra of (NH 4) 2S-treated bare GaInP 2 photocathode before and after the stability
test (Fig. 4f).
65
morphology observed after the stability test with (NH4)2S-treated, nanostructured GaInP2 might
be attributed to the desorption and re-adsorption of Ga, In, P, or S atoms at the catalytic interface.
To further elucidate the synergistic contributions of (NH4)2S-treatment and nanoporous
surface morphology to the exceptional improvement of corrosion-resistance, we further explored
the evolution of surface atomic composition in bare GaInP2 by XPS before and after the (NH4)2S-
treatment (Figure 3. 21(a)). It is notable that the (NH4)2S-treated bare GaInP2 has peaks
corresponding to sulfate (SO4
2-
) group (Figure 3. 21(a)), while showing the extended lifetime. As
previously mentioned, the sulfate group was introduced during the thermal annealing in the
(NH4)2S-treatment (see Figure 3. 6 and Figure 3. 22). Such sulfate (SO4
2-
)- and sulfide (S
2-
)-related
signals are, however, completely missing in untreated bare GaInP2 (the upper XPS spectra in
Figure 3. 21(a)) that degraded rapidly. Therefore, it is concluded that the sulfate group, rather than
sulfide group, on the sulfurized surface of GaInP2 played a crucial role in the strongly enhanced
durability (see Figure 3. 22). On the other hand, the (NH4)2S-treated bare GaInP2 (i.e. without
surface nanostructure, cyan data in Figure 3. 21(b)) did not exhibit a long-term durability
Figure 3. 20. (a) Top-view SEM images of bare and nanostructured/(NH 4) 2S-treated GaInP 2 (scale bar: 500 nm) before
and after the stability test in b (i.e. ~1 h for bare, ~124 h for NS). (b) Reflectance spectra of nanostructured and
(NH 4) 2S-treated GaInP 2 photocathodes before and after the stability test in b. Insets show corresponding photographic
images of samples (scale bar: 5 mm).
66
comparable to the (NH4)2S-treated and nanostructured GaInP2 (blue data in Figure 3. 21(b)), while
it was still more stable than the untreated bare GaInP2 (black data in Figure 3. 21(b)). Figure 3.
21(c) shows J-E curves of (NH4)2S-treated bare GaInP2, showing the onset potential of bare GaInP2
after the (NH4)2S treatment anodically shifted, which might also result from the activation of
catalytic sites of sulfurized GaInP2 as well as the removal of oxides as shown from (NH4)2S-treated,
nanostructured GaInP2 (see Figure 3. 18(a)). Despite the protective property of sulfurized surface,
Figure 3. 21. (a) XPS spectra of bare GaInP 2 before and after the (NH 4) 2S-treatment. (b) J-t plots of bare and
nanostructured GaInP 2 photocathodes with and without (NH 4) 2S-treatment, obtained under the same measurement
condition as in (a). (c) Representative J-E curves of (NH 4) 2S-treated bare GaInP 2 before (red) and after (blue) the 1-h
stability test (b), measured in an acidic electrolyte (0.5M H 2SO 4) under simulated AM 1.5G illumination. The J-E
curve from the bare GaInP 2 (i.e. without (NH 4) 2S-treatment) is also shown for comparison. (d) XPS spectra of S 2p
and Ga 3s from bare GaInP 2 photocathodes with (NH 4) 2S-treatment before and after the stability test (J-t) for 1 h
shown in (b) (cyan data). After the stability test, the relative area of sulfate-related peaks decreased.
67
it is noteworthy that the sulfate group keep corroding during HER, and Figure 3. 21(d) shows that
the peak intensity of sulfate from (NH4)2S-treated bare GaInP2 diminished after 1 hour of PEC
stability test, suggesting that the synergistic interplay between the nanostructured morphology to
enlarge the surface area and limit the progression of corroded region, and the incorporation of
sulfide/sulfate group to confer the improved corrosion resistance is collectively responsible for
suppressing the kinetics of corrosion and enabling significantly longer lifetime.
Faradaic efficiencies of H2 measured from bare GaInP2 without and with (NH4)2S-
treatment are ~80 and ~91% (see Table 3. 4), respectively, possibly due to the contribution of
corrosion to the photocurrent as observed in Figure 3. 21(b). As expected, the faradaic efficiency
of (NH4)2-treated GaInP2 is higher than the untreated GaInP2 owing to combined effects of surface
passivation and suppressed corrosion on the sulfurized reactive interface. Nevertheless, the
nanoporous morphology was maintained in ways that produce similar levels of reflectance and
Figure 3. 22. (a) XPS spectra of S 2p and Ga 3s measured from bare GaInP 2 photocathodes with and without thermal
annealing (250°C, 1 h, in air) during the (NH 4) 2S-treatment. (b) Representative J-E curves of bare GaInP 2
photocathodes with and without thermal annealing during the (NH 4) 2S-treatment, measured in an acidic electrolyte
(0.5M H 2SO 4) under simulated AM 1.5G illumination.
68
light absorption, and thus the photocurrent to those obtained before the stability test. Given that
the degradation of (NH4)2S-treated bare GaInP2 has also accompanied the gradual decrease of
sulfate group on the electrode surface as evidenced by XPS spectra (Figure 3. 21(d)), we postulate
that the rate of electrochemical dissolution of sulfate group is substantially lowered by the
nanoporous morphology owing to the reduced rate of local charge transfer owing to the enlarged
surface area as similarly observed from the PEC lifetime (J-t) of MoS2-deposited bare GaInP2 (see
Figure 3. 17(b)). While the nanostructured GaInP2 without (NH4)2S-treatment (orange data in
Figure 3. 21(b)) exhibited improved durability compared to the untreated bare electrode, it also
steadily degraded at the rate much higher than the nanostructured and (NH4)2S-treated GaInP2,
reiterating the importance of collaborative contributions from nanoporous surface morphology and
corrosion-resistant surface stoichiometry. It is preeminent that photoelectrochemical durability
(~124 hours) observed in this study without the usage of protective layers is comparably
outstanding to previous III-V photoelectrodes protected by the state-of-art protective layers
Table 3. 4. Values for determining faradaic efficiency of bare GaInP 2 photocathodes with and without (NH 4) 2S-
treatment.
69
including graded MoSx/TiO2 (~20 hours), Mo/MoS2, TiO2/Pt, TiO2/Co-TiO2, and a-TiO2/Ni (see
Table 3. 5)
7,10,18,21,47,70,71
.
3. 4. Conclusion
In summary, we demonstrated an unique approach that can simultaneously enhance the
light absorption, catalytic efficiency, and durability of monofacial GaInP2 photocathodes in the
Table 3. 5. Summary of published stability data of III-V compound semiconductor photoelectrodes in solar water
splitting.
70
HER of solar water splitting by collaboratively exploiting corrosion-resistant surface
stoichiometry and structurally tailored reactive interface. Firstly, a simple fabrication method was
newly developed in this study by employing electroless deposition of silver nanoparticles and dry
etching of GaInP2, followed by (NH4)2S treatment. Secondly, the two-step (NH4)2S treatment
exhibited the effective passivation of the surface states and suppressed the surface recombination ,
which would be further applied in a wide range of semiconductor materials
33,69,72
. Finally, the
synergistic interplay between the nanostructured morphology and sulfurized surface significantly
enhanced photoelectrochemical durability. We therefore expect our approach capitalizing the
synergistic effect of surface nanostructure and corrosion-resistant surface stoichiometry would be
broadly applicable to various semiconductor photoelectrodes (e.g. III-V, III-N) and even to other
electrochemical reactions (e.g. oxygen evolution reaction (OER), CO2 reduction) that can benefit
from simultaneously enhanced light absorption, catalytic efficiency, and corrosion resistance, all
without solely relying on the development of new protective materials, thereby offering practical
pathways towards high efficiency, high durability PEC solar water splitting. This work has been
accepted in Nature Communication and will be open to public. Therefore, Nature Communication
reserves the copyright for the all copied contents in this chapter.
71
Chapter 4. Bifacially Optimized, Nanostructured Ultrathin GaAs
Photocathodes
4. 1. Introduction
III-V compound semiconductors are the most promising class of photovoltaic materials
that have recorded high efficiencies among diverse photovoltaic technologies. Figure 4. 1 shows
the highest confirmed conversion efficiencies for research cells for a range of photovoltaic
technologies, plotted from 1976 to the present
73
, where cell efficiencies of III-V compound
semiconductors (i.e. purple line) including various configurations (i.e. single-junction and multi-
junction cells) exceeded that of other materials such as Si, CIGS, organic, or perovskite solar cells.
Figure 4. 2 (a) shows the quantitative fraction of Shockley-Queisser detailed-balance limit for
voltage and current achieved by record cells, where GaAs solar cells provided the highest degree
Figure 4. 1. A chart of the highest confirmed conversion efficiencies for research cells for a range of photovoltaic
technologies, plotted from 1976 to the present. Copyright (2019) National Renewable Energy Laboratory, Golden,
CO.
72
of both light management (i.e. the ratio of voltage × fill factor = VOC × FF / (VSQ × FFSQ in x axis)
and charge carrier management (i.e. the ratio of current density = JSC / JSQ in y axis)
4
. Because of
the high optical bandgap absorption co-efficiency, the cell thickness of GaAs can be kept relatively
small (< 2 m) to harvest the solar spectrum up to the bandgap, while achieving the record
efficiency (28.8%) among single junction solar-cells under one-sun illumination. Because of
superior photovoltaic properties, III-V compound semiconductors can realize ultrahigh efficiency
(>15%) photoelectrochemical (PEC) solar water splitting
74,75
, where they offer a number of
desirable materials attributes for photocatalytic electrodes including excellent photophysical
properties
4,76
. In addition, III-V compound semiconductors have another advantage to incorporate
multiple solid-state junctions
7,9
. Figure 4. 2(b) shows comparison chart of realized limiting solar-
to-hydrogen (STH) efficiencies and historic development in dual-junction limiting efficiency
9
,
where the record-high STH efficiencies have been reported from III-V multi-junction
photoelectrodes. For the record-high STH efficiency, it is noteworthy that there is still a room to
be improve further from ~19.3% to ~25%. One of main limitations in current technologies in III-
Figure 4. 2. (a) Fraction of Shockley-Queisser detailed-balance limit for voltage and current achieved by record cells.
Copyright (2016) Science. (b) Comparison of realized limiting STH efficiencies and historic development in dual-
junction limiting efficiency. Copyright (2018) ACS Energy Letters.
73
V photoelectrodes is a constraint of integration rule to incorporate efficient light management and
the state-of-art electrocatalyst as well as protective layers into monofacial photoelectrode
configuration. Kang et al. in our research group successfully first demonstrated a bifacial approach,
which separately optimize light absorption and catalytic efficiency, by employing transfer-printed
assembly design
8
(see Figure 4. 3) and reported 13.1% STH efficiency by coupling two series-
connected GaAs photoelectrodes. Nonetheless, the high cost for preparing device-quality III-V
materials still limits their practical deployment in PEC solar hydrogen generation
77-79
. For example,
Figure 4. 3. (a) Cross-sectional illustration of an integrated GaAs photocathode fabricated by printing-based materials
assemblies. (b) Representative current density–potential (J–E) curves of integrated GaAs photocathodes. (c) Epitaxial
design of ‘p-on-n’ GaAs photocathodes. Copyright (2017) Nature Energy. (d) Cost estimation (~2.6 $/Watt) for
growing p-GaAs base layer (2.5 um) using MOVPE. Copyright (2013) National Renewable Energy Laboratory.
74
the cost for growing p-GaAs base layer with 2.5 m thickness is ~2.6 $/Watt, which is already
over the cost for Si photovoltaic modules (see Figure 4. 3(d)). In our previous work, the thickness
of active layer was ~ 4 m, which can be further minimized to reduce the material cost (see Figure
4. 3(c)). However, a large fraction of the incident light in long wavelength is not fully absorbed in
such optically thin absorbers (see Figure 4. 4(a))
80
. For sub-microscale absorbers, the light
Figure 4. 4. (a) AM 1.5 solar spectrum and the absorption of solar energy in a 2-μm-thick crystalline Si film (assuming
single-pass absorption and no reflection). Copyright (2010) Nature Materials. (b) Short-circuit current density (J sc)
vs. thickness of light absorbers (GaAs and Si) with planar surface or Lambertian light trapping. Copyright (2012)
Journal of Applied Physics. (c) Epitaxial design of triple-stack ultrathin GaAs solar cells grown by MBE. (d)
Representative current density (J)−voltage (V) curves of ultrathin GaAs solar cells with (red) and without (orange)
TiO 2 NPs after the printing and on the growth wafer, respectively. The insets show an optical image of printed GaAs
solar cells with TiO 2 NPs and a SEM image of TiO 2 NPs, respectively. Copyright (2017) ACS Nano.
75
absorption which is directly related with short-circuit current density (Jsc) becomes more limited
for GaAs or Si absorbers (see Figure 4. 4(b))
81
. To overcome the limitation of thin photovoltaic
materials, Lambertian light trapping can boost the light absorption by utilizing nanophotonic
structures and effectively coupling light into thin absorbers. In addition to the cost of growing
materials (i.e. absorber thickness), a single-crystalline growth wafer (e.g. > $11,000/m
2
for epi-
ready GaAs) consists of a major portion of materials cost in III-V epitaxy. Overall, lowering their
cost contribution is an essential step to achieve economically viable PV or PEC solar water
splitting
82,83
. To the end, one straightforward route to overcome this challenge is to reuse the
growth substrate for epitaxy multiple times after detaching epitaxially-grown active materials with
Figure 4. 5. Relationship between power conversion efficiency, module areal costs and cost per peak watt (in $/Wp).
Each of dashed black lines indicates a given $/Wp, and the light blue lines represents the current record-high efficiency
obtained from bulk crystal Si photovoltaic module, while the blue horizontal line is the Shockley-Queisser limit for
single-junction devices. For next-generation technologies the goal is to reach 0.03-0.05 $/kWh, to replace fossil fuels
such as gasoline (i.e. 0.03 $/kWh). Copyright (2014) Nature Nanotechnology.
76
the epitaxial liftoff (ELO) technology
37,84-86
. Moreover, ELO for III-V photoelectrodes provides
crucial advantages in terms of overall device performance and cost effectiveness. First, the
thickness of semiconductor absorber can be reduced to lower the cost of epitaxial materials without
compromising the performance by exploiting nanophotonic optical management. Gai et al. in our
research group evidently demonstrated epitaxially-lifted-off ultrathin (~300 nm) GaAs
photovoltaic cells and effectively improved short-circuit current density (~22.3 mA/cm
2
) and
solar-to-electric power conversion efficiency (~17.2%) by combining with TiO2 nanoposts for
light trapping (Figure 4. 4(c) and (d)). Although actively pursued in photovoltaics
36,37,83,87
, similar
efforts in this strategy have been rarely reported in the field of III-V PEC solar water splitting.
Figure 4. 5 shows the relationship between power conversion efficiency, module areal costs and
cost per peak watt (in $/Wp), where “current-day technologies” such as Si photovoltaics placed in
the range of relatively high module cost. It is noteworthy that “next-generation technologies” must
reach 0.03-0.05 $/kWh (blue-dashed region), which is comparable to fossil fuels such as gasoline
(i.e. 0.03 $/kWh). Second, the ELO-enabled bifacial photoelectrodes can independently and
simultaneously optimize each of material interfaces for optical absorption and electrocatalysis
8
.
This advantage can effectively address the limitation of conventional monofacial photoelectrodes
by isolating a catalytic interface from light absorbing interface and by optimizing for both charge
transfer and electrochemical durability
16,21,43,47,88,89
. Figure 4. 6 and Figure 4. 7 show the state-of-
art of electrocatalysts and protective materials, respectively
90,91
, and the reported bifacial
photoelectrodes greatly expanded choices of materials that can be readily employed without
optical constraints to enhance the corrosion resistance and overall photoelectrochemical efficiency
beyond conventional monofacial photoelectrodes
8,92
.
77
In this chapter, we present a unique approach for ultrathin GaAs photocathodes for high-
performance and cost-effective solar water splitting that can accommodate all of aforementioned
Figure 4. 7. Benchmarking of reported stabilities of photocathodes and protective materials for the HER, versus tested
pH condition, with resulting photocurrent and degradation rate indicated. Copyright (2017) Chemical Society Reviews.
Figure 4. 6. Benchmarking of reported electrocatalysts for hydrogen evolving reaction and oxygen evolving reaction
for solar water splitting devices. Copyright (2015) Journal of the American Chemical Society.
78
advantages in cost, light management, charge transfer, and electrochemical durability. 270-nm-
thick ultrathin GaAs photocathodes (i.e. ~10 thinner than conventional photoelectrodes) were
bifacially optimized to incorporate TiO2 diffractive nanostructure at the light-absorbing interface
(top side) to boost optical absorption and thus obtain comparable or even superior performance to
the optically thick devices (e.g. ~3 m GaAs) (see Figure 4. 8. A schematic illustration of bifacial,
nanostructured, ultrathin GaAs photocathodes with ohmic contact, TiO2, and HER co-catalyst.).
Multicomponent protective and electrocatalytic layers implemented at the catalytic interface
(bottom side) strongly improved the charge transfer to yield photovoltaic-level fill-factor and onset
potential but also enhanced the photoelectrochemical durability of ultrathin GaAs photocathodes.
In the following, we describe systematic studies of optical, morphological, electrical, and
photoelectrochemical properties of nanostructured, ultrathin GaAs photocathodes in experiments
as well as optical modeling to elucidate underlying materials science and photoelectronic
mechanism, as well as design rules for multifaceted optimization of integrated solar water splitting
devices.
Figure 4. 8. A schematic illustration of bifacial, nanostructured, ultrathin GaAs photocathodes with ohmic contact,
TiO 2, and HER co-catalyst.
79
4. 2. Fabrication
Figure 4. 9(a) shows the epitaxial design of ultrathin GaAs photocathodes, composing of
p
+
-GaAs top contact (200 nm, Be-doped, 1 × 10
19
cm
-3
), p-Al0.40Ga0.60As window (40 nm, Be-
doped, 2 × 10
18
cm
-3
), p-GaAs emitter (50 nm, Be-doped, 2 × 10
18
cm
-3
), n-GaAs base (220 nm,
Si-doped, 3 × 10
17
cm
-3
), n-Al0.40Ga0.60As back-surface field (50 nm, Si-doped, 3 × 10
18
cm
-3
), n
+
-
GaAs bottom contact (50 nm, Si-doped, 5 × 10
18
cm
-3
), and Al0.90Ga0.1As sacrificial layer (400 nm,
undoped), grown on a semi-insulating (100) GaAs wafer. As shown in Figure 4. 9(b), the epitaxial
stacks for ultrathin GaAs photocathodes corresponding to the detailed epitaxial design were grown
by molecular beam epitaxy (MBE) without any visible defects on GaAs substrate. As designed,
the thickness of active layers was ~270 nm, including emitter, base, and back-surface field (BSF)
layers which were responsible for light absorption and the source of generated photocurrents.
Figure 4. 10 illustrates fabrication procedures of ultrathin, nanostructured GaAs
photocathodes with a bifacial electrode configuration. The fabrication of bifacial GaAs
photocathodes began with the deposition of a p-type ohmic contact (Cr/Au = 5 nm/70 nm) and wet
Figure 4. 9. (a) Epitaxial design of ‘p-on-n’ ultrathin GaAs photocathodes. (b) Cross-sectional view SEM image of
the ultrathin GaAs photocathode grown by MBE using Si and Be as n- and p-type dopants, respectively.
80
chemical etching of the p
+
-GaAs contact layer (i.e. step i), followed by the formation of TiO2
nanoposts on Al0.4GaAs window layer through radio-frequency (RF) magnetron sputtering of TiO2,
softimprint lithography, oxygen reactive ion etching (O2 RIE), electron beam evaporation, reactive
ion etching (RIE), and wet-chemical etching to produce hexagonally periodic TiO2 nanoposts (NPs)
within the electrode boundary (i.e. step ii)
93
. For this step, more detailed information of IPCE
measurements is available in chapter 2. Subsequently, isolated arrays of microscale (~580 580
m
2
) photoelectrodes were delineated by photolithography and wet chemical etching, followed by
the undercut etching of sacrificial layer (Al0.9GaAs) in dilute hydrochloric acid (i.e. step iii). A
fully isolated set of GaAs photoelectrodes were then picked up from the growth substrate by an
elastomeric stamp made of polydimethylsiloxane (PDMS) (i.e. step iv) and transferred to another
elastomeric stamp (i.e. step v). Using photocurable polyurethane (NOA 61), GaAs photocathodes
were printed over a transparent glass substrate, while the side of Al0.40Ga0.60As window
Figure 4. 10. Schematic illustration of fabrication processes of bifacial, nanostructured GaAs photocathodes printed
on a glass substrate for solar-driven photoelectrochemical water splitting.
81
implemented with TiO2 NPs was faced down and embedded in NOA 61 (i.e. step vi). After
forming the n-type ohmic metal contact (Pd/Ge/Au/Ti = 5 nm/35 nm/100 nm/30 nm or
AuGe/Ni/Au/Ti = 100 nm/30 nm/100 nm/30 nm), the p-type metal contact was exposed, while
forming polymeric ‘via’ by wet chemical etching (i.e. step vii). After the incorporation of metal
interconnects (Cr/Ag/Au = 15 nm/1500 nm/100 nm) and encapsulation of photoelectrodes by
polyimide, ultrathin GaAs photocathodes were ready for driving the hydrogen evolution half-
reaction in solar water splitting. In particular configurations, TiO2 protective layer and/or HER
Figure 4. 11. Optical microscopic images of a nanostructured GaAs photocathodes integrated on a glass substrate on
(a) a catalytic interface and (b) light-absorbing interface. (c) Cross-sectional and (d) tilt-view scanning electron
microscope (SEM) images of TiO 2 NPs implemented onto the light-absorbing interface (i.e. Al 0.4Ga 0.6As) of
nanostructured GaAs photocathode.
82
co-catalyst (i.e. Pt or MoSx) were additionally incorporated, and complete details of fabrication
processes are also available in Chapter 2. Notably, the resulting integrated GaAs photoelectrode
has a bifacial configuration that decouples the catalytic and light-absorbing materials interfaces to
allow their independent optimization (see Figure 4. 8. A schematic illustration of bifacial,
nanostructured, ultrathin GaAs photocathodes with ohmic contact, TiO2, and HER co-catalyst.).
The TiO2 NPs at the light-absorbing interface served as a diffractive optical element to reduce the
reflection and increase the optical path-length of the incident solar light, leading to the significantly
enhanced absorption of ultrathin GaAs photocathodes. Figure 4. 11 (a) and (b) show optical
micrographs of such isolated catalytic (left) and light-absorbing (right) interfaces of a completed
nanostructured GaAs photocathode integrated on a glass substrate, where the dark color seen
through the glass substrate supports the effectiveness of TiO2 NPs to produce broadband
antireflection. Figure 4. 11(c) and (d) show cross-sectional and tilt-view, respectively, scanning
electron microscope (SEM) images of TiO2 NPs implemented onto the light-absorbing interface
(i.e. Al0.4Ga0.6As) of nanostructured GaAs photocathode. Figure 4. 12 shows a photograph of a
completed ultrathin GaAs photoelectrode cells integrated on a transparent glass with the
encapsulation of polyimide, where the transparent-yellow color indicated the color of polyimide
Figure 4. 12. A photograph of a completed ultrathin GaAs photoelectrode cells integrated on a transparent glass,
encapsulated by polyimide.
83
and the four metal pads were the metal interconnection for n-type and p-type ohmic contact per
two microcells printed on a glass.
4. 3. Results and Discussion
In the reported electrode design with an isolated light-absorbing interface, the photocurrent
of optically thin (i.e. emitter + base = 270 nm) GaAs (t-GaAs) photoelectrodes can be strongly
Figure 4. 13. (a) Schematic illustration of antireflection and light trapping, resulting from periodic nanostructures with
1D gratings (p ≤ λ). (b) Schematic of a 2D grating structure with p. (c) Allowed diffraction modes in k-space or
reciprocal space. Copyright (2012) Journal of Applied Physics. (d) Distribution of the forward diffraction efficiencies
(DET mn) of a 2D-grating. Copyright (2007) Progress in Photovoltaics: Research and Applications.
84
enhanced by employing an optimized scheme of bifacial photon management employing TiO2 NPs
and a back-side metallic reflector (BSR). Figure 4. 13(a) shows a schematic illustration of
antireflection and light trapping, arising from periodic nanostructures with 1D gratings (p ≤ λ). For
1D grating, the propagation angle for a diffraction order (m) can be calculated using grating
equation by
𝑛 sin 𝜃 𝑚 =
𝑚𝜆
𝑝
where n, θm, m, λ, and p are refractive index of medium, diffraction angle, diffraction mode,
wavelength of incident light, periodicity of grating, respectively. Incident light can couple to a
larger number of propagating diffraction modes (i.e. blue arrows) inside the bottom medium with
a high refractive index (e.g. TiO2 = 2.5), on the other hand, couple to zeroth mode (red arrow)
inside the upper medium with a lower index (e.g. air = 1), causing the antireflection effect by
suppressing the coupling of outward light in upper medium. Figure 4. 13 (b) and (c) shows 2D
dielectric gratings with periodicity (p) smaller than the wavelength ( λ) of incident light and allowed
diffraction modes (m, n), respectively
81
. Figure 4. 13(d) shows the distribution of the forward
diffraction efficiencies (DETmn) of a 2D-grating, confirmed by a rigorous coupled-wave analysis
(RCWA)
94
. It is noted that the transmitted light mostly distributed at (0, ±1) and (±1, 0) orders,
and the transmitted light propagates at such oblique diffraction angles, resulting in the increase of
optical path lengths. Consequently, there is an increase in the light absorption, which is the light
trapping effect.
To identify optimal materials configurations, optical properties of GaAs photocathodes
were studied numerically using finite-difference time-domain (FDTD) methods. Figure 4. 14(a)
schematically depicts a cross-sectional layout of bifacial GaAs photocathodes integrated on a glass
substrate for numerical modeling, where the incident light is illuminated on the side of TiO 2 NPs
85
through the glass substrate, while an ohmic metal contact (OM) at the catalytic interface serves as
a BSR to bounce back unabsorbed long-wavelength photons. Figure 4. 14(b), (c), and (d) show
schematic illustration of cross-sectional layout of a bifacial-planar GaAs photocathode printed on
a glass substrate, and a monofacial-planar GaAs photocathode on GaAs wafer with and without Pt
co-catalyst, respectively. For the quantitative evaluation of photoelectrode performance under
sunlight, calculated absorption spectra were integrated over the simulated AM1.5G solar
illumination to obtain integrated solar flux absorption (Sabs) given by,
Figure 4. 14. Schematic illustration of (a) a bifacial, nanostructured GaAs photocathodes printed on a glass substrate,
(b) a bifacial, planar GaAs photocathodes printed on a glass substrate, and (c), (d) a monofacial, planar GaAs
photocathode on GaAs wafer with and without Pt co-catalyst, considered for FDTD optical modeling.
86
𝑆 𝑎𝑏𝑠 =
∫
𝜆 ℎ𝑐 𝐴 ( 𝜆 ) 𝐼 1.5𝐺 ( 𝜆 ) 𝑑𝜆 870 𝑛𝑚
400 𝑛𝑚
∫
𝜆 ℎ𝑐 𝐼 1.5𝐺 ( 𝜆 ) 𝑑𝜆 870 𝑛𝑚
400 𝑛𝑚
, where h, c, A( ), and I1.5G are Planck’s constant, speed of light, calculated absorbance, and the
standard solar irradiance (AM 1.5G; ASTM G-173), respectively. Figure 4. 15(a) shows A
schematic illustration of optical model for periodic TiO2 NPs on ultrathin GaAs photocathodes for
FDTD optical calculations, and Figure 4. 15(b) presents a contour plot of calculated Sabs of
nanostructured GaAs photocathode as a function of diameter (D) and height (h) of TiO2 NPs at a
DS fixed NP period (p = 500 nm), where a n-type ohmic metal contact (i.e. AuGe/Ni/Au/Ti)
was incorporated on a BSF layer (Al0.40Ga0.60As) as a BSR (i.e. without a GaAs bottom contact
(BC) layer). For this baseline materials configuration, the optimum D and h of TiO2 NPs that
maximize the Sabs are 300 and 400 nm, respectively (the increment of D and h in the simulation
was 50 nm). Similar sets of calculations performed over a range of TiO2 periods (p = 100 - 900
nm) were performed to find optimal D and h at each p for other configurations (see Figure 4. 16,
Figure 4. 15. (a) A schematic illustration of optical model for periodic TiO 2 NPs on ultrathin GaAs photocathodes for
FDTD optical calculations. (b) A contour plot of the integrated solar flux absorption (S_abs) of a bifacial,
nanostructured GaAs photocathode in (a) as a function of diameter (D) and height (h) of TiO 2 NPs at the period (p) of
500 nm, where optical absorption in window, emitter, base, and back-surface field (BSF) layers were considered.
87
Figure 4. 16. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes without bottom contact layer incorporating TiO 2 NPs and gold as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm.
Table 4. 1. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, obtained
from Figure 4. 16. The thickness of TiO 2 base was fixed at 50 nm.
88
Figure 4. 17. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes without bottom contact layer incorporating TiO 2 NPs and ohmic metal as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm.
Table 4. 2. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, exported
from Figure 4. 17. The thickness of TiO 2 base was fixed at 50 nm.
89
Figure 4. 18. Contour plot of integrated solar flux absorption (S_abs) in window, emitter, base, and back-surface field
layers of ultrathin GaAs photocathodes with bottom contact layer incorporating TiO 2 NPs and ohmic metal as BSR
calculated by a FDTD method at an NP period from 100 nm to 900 nm. Optimum dimensions corresponding to the
maximum S_abs per each period are available in the table. The thickness of TiO 2 base was fixed at 50 nm.
Table 4. 3. A summary of optimum dimensions corresponding to the maximum S_abs per each periodicity, exported
from Figure 4. 18. The thickness of TiO 2 base was fixed at 50 nm.
90
Figure 4. 17, and Figure 4. 18). Figure 4. 19 shows the summarized maximum values of calculated
S_abs of bifacial, nanostructured GaAs photocathode as a function of NP period at various
materials configurations. It is noteworthy that the maximum integrated absorption (~70.3%, red
diamond) of GaAs photoelectrode implemented with TiO2 NPs and an ohmic metal contact
(formed on the BSR) is significantly higher than the case without TiO2 NPs (~58.9%, cyan line)
and can be further increased by employing an ideal (e.g. gold) BSR (~79.8%, black square). It is
notable that the maximum Sabs of nanostructured GaAs photoelectrodes with an ideal BSR is even
higher than that (~74.5%) of optically thick (~3 m), monofacial GaAs photoelectrodes (green
dotted line) that receive the sunlight directly through the catalytic surface. While HER co-catalysts
such as Pt are an essential component to obtain a higher (e.g. >10%) efficiency, they inevitably
degrade the light absorption of monofacial photoelectrodes due to associated reflection and
scattering losses
8,47
. Even with 1-nm-thick Pt, for example, the Sabs of optically thick GaAs is
Figure 4. 19. Maximum values of calculated S_abs of bifacial, nanostructured GaAs photocathode as a function of NP
period at various materials configurations including (i) ultrathin GaAs with TiO2 NPs and ohmic metal contact after
the removal of bottom contact layer (red diamond), (ii) ultrathin GaAs with TiO 2 NPs and silver reflector after the
removal of bottom contact layer (black square), (iii) ultrathin GaAs with TiO 2 NPs and ohmic metal contact without
the removal of bottom contact layer (blue triangle), (iv) ultrathin GaAs with SLARC and ohmic metal contact after
the removal of bottom contact layer (orange dotted line), (v) ultrathin GaAs with ohmic metal contact after the removal
of bottom contact layer (cyan dotted line), and (vi) thick GaAs directly immersed in water (green dotted line).
91
further decreased to ~65.3% (pink dotted line), which is an unavoidable dilemma in all monofacial
photocathodes. Another important consideration required to maximize the light trapping in
optically thin GaAs photocathodes is their epitaxial design. In particular, a heavily-doped bottom
contact layer (i.e. n
+
-GaAs) that is typically incorporated in optically thick GaAs photoelectrode
can significantly deteriorate the absorption of long-wavelength photons reflected by the BSR due
to the recombination of photogenerated carriers (blue triangle) and therefore must be thinned down
or, if possible, removed. Figure 4. 20 (b) shows contour plots of normalized absorbed power
density of bifacial ultrathin GaAs photocathodes at three different configurations including bare
GaAs with a bottom contact layer and metal contact (top, BC/OM), nanostructured GaAs (i.e. with
TiO2 NPs) with a bottom contact layer and metal contact (middle, NP/BC/OM), and nanostructured
GaAs with metal contact yet without a bottom contact (bottom NP/OM), where the augmented
absorption resulting from the combined effects of antireflection, diffraction, and light trapping by
Figure 4. 20. (a) A schematic illustration of optical model for periodic TiO 2 NPs on ultrathin GaAs photocathodes
with the plane of incidence for FDTD optical calculations. (b) Contour plots of normalized absorbed power density
of bifacial ultrathin GaAs photocathodes at three different configurations including bare GaAs with a bottom contact
layer and metal contact (top, BC/OM), nanostructured GaAs (i.e. with TiO 2 NPs) with a bottom contact layer and
metal contact (middle, NP/BC/OM), and nanostructured GaAs with metal contact yet without a bottom contact
(bottom NP/OM).
92
TiO2 NPs and metallic BSR (i.e. middle and bottom) as well as the effect of bottom contact layer
are clearly shown. As expected, the removal of the bottom contact substantially enhanced the
absorption of long wavelength photons by eliminating the carrier recombination. Based on these
computational studies, our approach to combine ultrathin GaAs absorber with nanoscale photon
Figure 4. 22. Maximum integrated solar flux from various thickness of GaAs photoelectrodes in different
configurations including optically thick monofacial GaAs photoelectrode without (green) and with co-catalyst (pink),
ultrathin, nanostructured GaAs photoelectrode with Au and without bottom contact (BC) (black), with ohmic metal
and without BC (red), and with ohmic metal and with BC (blue). Optical models shown in Figure S3 were used by
varying the thickness of GaAs emitter/base while maintaining the thickness of Al 0.4GaAs window and BSF layer.
Figure 4. 21. Corresponding calculated reflectance and absorption spectra in water of bifacial, nanostructured GaAs
photocathodes, where optimum geometries of TiO 2 NPs (i.e. D: 300 nm, H: 400 nm, p: 500 nm) (green) were assumed.
93
management using TiO2 NPs and ohmic metal contact as a BSR is quantitatively justified, where
it is shown that comparable photocurrent or efficiency to optically thick GaAs is readily achievable
with ~10X lower consumption of expensive epitaxial materials (see Figure 4. 22). Figure 4. 21
shows corresponding calculated reflectance and absorption spectra at optimum geometries of TiO2
NPs in water. Consistent with the calculated S_abs, the absorption of ultrathin, bifacial GaAs with
TiO2 NPs and gold reflector (black line) is even higher than optically thick, ‘monofacial’ GaAs
(green line) over nearly entire wavelengths.
For the proof-of-concept demonstration of ultrathin GaAs photocathodes, we employed
50-nm-thick n
+
-GaAs bottom contact layer and ohmic metal contact (i.e. AuGe/Ni/Au/Ti or
Pd/Ge/Au/Ti) as a BSR were employed. Figure 4. 23(a) shows the reflectance spectra of GaAs
photocathodes measured in air (solid line) with three different materials configurations, including
Figure 4. 23. (a) Measured reflectance spectra of ultrathin GaAs photocathodes in air at various materials
configurations at the light absorbing interface, including bare GaAs on wafer (black), bare GaAs printed on glass
(red), and GaAs with TiO 2 NP printed on glass (blue), where the light was illuminated through the glass substrate and
ohmic metal contact was incorporated for all printed samples. The dotted lines depict calculated reflectance spectra
from numerical optical modeling, while dashed lines show corresponding calculated absorption spectra in water. (b)
A schematic illustration of tapered TiO 2 NPs for FDTD optical calculations.
94
bare (red data) and nanostructured (i.e. with TiO2 NPs (p = 500 nm, Dtop/Dbottom = 380/420 nm
(Figure 4. 23(b)), h = 360 nm, blue data) GaAs printed on glass as well as bare GaAs on wafer
(black data). In these measurements, a silver mirror was used as a calibration standard for 100%
reflectance. Figure 4. 24 shows optical micrographs from light-absorbing interface of (a) bare
GaAs on wafer, (b) bare GaAs on glass, and (c) NS GaAs on a glass substrate after printing,
corresponding to Figure 4. 23(a). Fringe color from Figure 4. 24(b) was originated from the effects
of back-surface reflector (ohmic metal contact) and the antireflection effect of printing medium
(NOA 61). As the light was illuminated through the glass substrate for printed samples, the
reflectance of bare GaAs in a bifacial design considerably decreased compared to the monofacial
design due to the glass and polymeric printing medium that lowered the refractive index contrast.
The incorporation of TiO2 NPs as diffractive optical elements further promoted the suppression of
reflection losses. Given that the calculated reflectance spectra (dotted line) from the FDTD
modeling matched well with the experimental data (solid line), we utilized this optical model to
evaluate absorption spectra (dashed line) of GaAs photocathodes in water, where the calculated
Figure 4. 24. Optical micrographs from light-absorbing interface of (a) bare GaAs on wafer, (b) bare GaAs on glass,
and (c) NS GaAs on a glass substrate after printing, measured under the same illumination condition. Fringe color
from (b) was originated from the effects of back-surface reflector (ohmic metal contact) and the antireflection effect
of printing medium (NOA 61).
95
values of integrated solar flux absorption (S_abs) are 47.4% (bare on wafer), 53.5% (bare on glass),
and 61.9% (NP on glass), respectively (see Table 4. 4). The enhanced optical absorption directly
translates to the boost of photocurrent in both photovoltaic and photoelectrochemical devices.
Figure 4. 25 shows representative current density (J)-voltage (V) curves of GaAs photocathodes,
measured in a photovoltaic mode under simulated AM1.5G illumination (1000 W/m
2
).
Corresponding photovoltaic characteristics including short-circuit current density (Jsc), open-
circuit voltage (Voc) and fill factor (FF) are summarized in Table 4. 5. The Jsc of bare and
nanostructured GaAs photocathodes printed on a glass substrate were 14.1 and 16.5 mA/cm
2
,
which correspond to the enhancement by ~21 and ~41% compared to the bare GaAs on wafer
Table 4. 4. Comparisons of relative light absorption enhancement between measured short-circuit current densities
(J SC) from photovoltaic performance shown in Figure 3a and calculated integrated solar flux absorption (S_abs) in air
with different configurations including bare GaAs on wafer, bare GaAs on glass, and nanostructured GaAs on glass
with bottom contact layer and ohmic metal as back-side reflector.
96
(11.7 mA/cm
2
), respectively, and is quantitatively consistent with the calculated improvement of
solar flux absorption.
Photoelectrochemical performance of ultrathin GaAs photocathodes with and without TiO2
NPs, using electron-beam evaporated Pt (~30 nm) and Ti (5 nm) as a HER co-catalyst and adhesion
layer, was measured in a three-electrode configuration with an acidic electrolyte (0.5M H2SO4)
under the AM1.5G solar illumination, where Pt and Ag/AgCl were employed as counter and
reference electrodes, respectively (see Figure 4. 26). The bifacial photoelectrode was illuminated
through the glass substrate onto the nanostructured surface of photocathode, while the opposite
side deposited with a n-type ohmic contact and Pt co-catalyst was responsible for the hydrogen
Figure 4. 25. Representative current density (J)-voltage (V) curves of ultrathin GaAs photocathodes at various
configurations including bare GaAs on wafer (black), bare GaAs printed on glass (red), and GaAs with TiO 2 NP
printed on glass (blue), under simulated AM1.5G solar illumination (1000 W/m
2
).
Table 4. 5. Summarized photovoltaic performance shown in Figure 4. 25.
97
evolution reaction. Similar to the photovoltaic characteristics, the saturation current density (Jsat)
of ultrathin GaAs photocathodes was increased due to the effects of nanoscale photon management,
resulting in strong enhancement of half-reaction diagnostic efficiency ( cathode) as defined by the
following equation from ~8.6% to ~11.3%,
𝜂 𝑐𝑎𝑡 ℎ𝑜𝑑𝑒 ( %) =
𝐽 𝑚𝑎𝑥
∙ ( 𝐸 𝑚𝑎𝑥
− 𝐸 ( 𝐻 +
/𝐻 2
) )
𝑃 𝑖𝑛
× 100
, where Jmax and Emax are the current density and electrode potential at a maximum power point,
E(H
+
/H2) is the thermodynamic HER potential, and Pin is the power density of simulated AM1.5G
solar illumination
8
. Notably, the Jsat of ultrathin (>10X thinner) GaAs photocathodes with TiO2
NPs were comparable to that of previously reported optically thick GaAs photoelectrodes with
metallic co-catalysts
8,47
. Furthermore, the recorded onset potential (i.e. Eonset ~0.89 V vs. RHE)
was almost equal to the open-circuit voltage (Voc, ~0.94 V) in photovoltaic measurements (see
Figure 4. 25), where the small discrepancy is explained by the overpotential (~50 mV) of Pt for
the HER (see Figure 4. 27). The fill-factor ( 𝐹𝐹 = ( 𝐽 𝑚𝑎𝑥
∙ 𝐸 𝑚𝑎𝑥
) /( 𝐽 𝑠𝑎𝑡
∙ 𝐸 𝑜𝑛𝑠𝑒𝑡 ) ) in PEC
Figure 4. 26. Representative current density (J)-potential (E) curves of ultrathin GaAs photocathodes printed on glass
in various configurations including bare GaAs (red) and GaAs with TiO 2 NP (blue), measured in a three-electrode
configuration with an acidic electrolyte (0.5M H 2SO 4) under simulated AM1.5G solar illumination (1000 W/m
2
). Pt
and Ag/AgCl were employed as counter and reference electrodes, respectively.
98
decreased only slightly from that observed in the photovoltaic measurement, which is attributed to
the Tafel slope (~30 mV/dec) of Pt. Critically, such PV-level photovoltages and fill-factors have
Figure 4. 27. (a) Cyclic voltammetric curves of platinum with various thickness of TiO 2 protective layer as a dark
HER electrode measured in 0.5M H 2SO 4. Onset potential from platinum catalyst deposited by ebeam evaporator
(black) showed a slightly higher overpotential than commercially purchased platinum rod (pink). (b) HER onset
potentials at -0.5 mA/cm
2
(black, triangle) and -10 mA/cm
2
(red, rectangular) captured from (a). It is note that onset
potential after adding TiO 2 protective layer still maintained onset potential.
Figure 4. 28. Comparisons between photovoltaic and photoelectrochemical performance of nanostructured, ultrathin
GaAs photocathodes with Pt co-catalyst without TiO 2 protective layer captured from Figure 3b and 3c. It is note that
all the properties from PV performance were maintained in PEC performance by the effect of bifacial design ; adopting
n-type ohmic contact metal and high performance HER catalyst. The small decrease of V oc to V onset would be originated
from the overpotential (~40 mV) and Tafel slope (~30 mV/dec)of ebeam-deposited Pt as HER catalyst. The higher
saturation current density from PEC might be due to a small fraction of light waveguide effect by the glass substrate
as well as PEC cell.
99
never been achieved in conventional monofacial photoelectrodes, which is enabled by the unique
capabilities of the reported design to incorporate an ohmic metal contact and a thick layer of Pt co-
catalyst that effectively removed the Schottky barrier and fully optimized charge transfer at the
catalytic interface (see Figure 4. 28). Figure 4. 29 shows corresponding incident photon-to-current
efficiency (IPCE) spectra of ultrathin GaAs photocathodes with and without TiO2 NPs measured
at 0 V (vs. RHE) under simulated AM 1.5G illumination. The enhancement in light absorption
enabled with TiO2 NPs was clearly demonstrated in the long-wavelength region as projected in
Figure 4. 20 and Figure 4. 21, showing an overall increase of photocurrent by ~17.7% from the
integration of IPCE, which closely matched with the experimentally measured increase of Jsc from
PV and Jsat from PEC data.
With the removal of optical constraints, we can also employ unconventional materials
configurations at the catalytic interface of bifacial photoelectrodes to augment their
electrochemical durability. To this end, optically thick metal layers that can serve as both an ohmic
contact and a cathodic protection layer have been successfully demonstrated for bifacial GaAs
Figure 4. 29. Incident-photon-to-current efficiency (IPCE) spectra for the ultrathin GaAs photocathodes including
bare GaAs and GaAs with TiO 2 NP measured at 0 V (vs. RHE) under AM 1.5G illumination.
100
photocathodes in the HER. However, metallic protection layers can work only in a narrow range
of pH or electrode potential and may not be ideal for long-term protection in diurnal cycles. In the
present study, we therefore explored the possibility to incorporate a layer of sputter-deposited TiO2
between the ohmic metal contact and HER co-catalyst as TiO2 is thermodynamically stable in a
wide range of pH and potential but also, as an n-type semiconductor, conduct electrons efficiently.
Figure 4. 30 shows current density (J)-potential (E) curves of bifacial, nanostructured GaAs
Table 4. 6. Summarized photoelectrochemical performance shown in Figure 4. 30.
Figure 4. 30. Representative current density (J)-potential (E) curves of bifacial, nanostructured GaAs photocathodes
at various materials configurations at the catalytic interface, including GaAs with Pt, GaAs with n-type ohmic contact
and Pt, GaAs with n-type ohmic contact, Ti, TiO 2, and Pt, measured in a three-electrode configuration with an acidic
electrolyte (0.5M H 2SO 4) under simulated AM1.5G solar illumination (1000 W/m
2
).
101
photocathodes under AM1.5G illumination at various materials configurations at the reactive
interface, including (i) ohmic metals (OM; Pd/Ge/Au/Ti = 5 nm/35 nm/100 nm/30 nm), black data),
(ii) ohmic metals with TiO2 (OM/TiO2 (20 nm), red data), (iii) ohmic metals with Pt (OM/Ti (3
nm)/Pt (30 nm), blue data), (iv) ohmic metals with TiO2 and Pt (OM/TiO2 (20 nm)/Ti (3 nm)/Pt
(30 nm), green data), (v) ohmic metals with amorphous MoSx (OM/MoSx (40 nm), x=~2.5, orange
data), and (vi) ohmic metals with TiO2 and MoSx (OM/TiO2 (50 nm)/MoSx (40 nm), cyan data),
respectively. The incorporation of TiO2 with ohmic metals (red data) cathodically shifted the onset
potential from ~0.42 to ~0.37 V (vs. RHE) owing to its comparatively lower catalytic activity for
the HER than Ti in ohmic metals
21,95
. As expected, Pt co-catalysts deposited over ohmic metals
(blue data) strongly improved the onset potential to ~0.89 V (vs. RHE). It is noteworthy that the
TiO2 incorporated between the ohmic metals and Pt co-catalyst (green data) did not make any
detrimental effect on the charge transfer and catalytic performance so that the onset potential and
fill factor was maintained in comparable levels to the case without TiO2 (i.e. OM/Pt). Motivated
by the necessity to replace the precious metal catalysts, we also employed cathodically-deposited
MoSx as an earth-abundant HER co-catalyst in conjunction with the reported bifacial electrode
configuration. At a fixed potential (-0.5 V vs. RHE), the thickness of MoSx films deposited on a
ITO glass changed in proportion to deposition time (see Figure 4. 31(a)), where transmittance
spectra of corresponding MoSx films gradually decreased as the time increased (see Figure 4.
31(b)). XPS analysis confirmed that the ratio (x) of Mo/S was ~2.5, where the observed
overpotential (~230 mV) for HER is comparable to previously reported values (see Figure 4. 31(c)
and (d))
60,64,96
. In Figure 4. 30, the MoSx deposited over the ohmic metals (orange data) anodically
shifted the onset potential to ~0.71 V (vs. RHE). Similar to the OM/TiO2/Pt case, the incorporation
of TiO2 between ohmic metals and MoSx rarely affected the onset potential and fill factor, thereby
102
validating the use of TiO2 as an intermediate layer between ohmic metals and high performance
HER co-catalysts for efficient charge transfer and enhanced protection against corrosion. To
evaluate the multicomponent protection layer comprised of OM/TiO2/HER co-catalyst,
photoelectrochemical durability tests were conducted at three-electrode configuration (0 V vs.
RHE) under AM1.5G illumination in an acidic electrolyte (0.5 M H2SO4) as summarized in Figure
4. 32. Bifacial GaAs photocathodes without TiO2 protective layer (i.e. OM/Pt, blue data) exhibited
Figure 4. 31. (a) Thickness (nm) vs. deposition time (s) of MoS x film deposited on an ITO glass using cathodic
electrodeposition in an aqueous solution of (NH 4) 2MoS 4 (2 mM), 0.5 M of Na 2SO 4 (pH 6.4), and (NH 4) 2S ((NH 4) 2S:DI
water = 1: 750 by volume, Macron, 20.0-24.0 wt% in H 2O) at -0.5 V vs. RHE. (b) Corresponding transmittance
spectra of MoS x films measured by UV-Vis spectroscopy. (c) XPS spectra of Mo 3d and S 2p of MoS x films. (d)
Linear voltammetric curves of MoS x deposited on a ITO glass as a dark HER electrode measured in 0.5M H 2SO 4
103
~14 hours of lifetime with ~20% reduction of current density. As previously reported, the presence
of thick ohmic metals can act as a barrier against the diffusion of protons in the acidic electrolyte
to retard the reductive corrosion of GaAs photocathodes as
8
GaAs + 3e
-
+ 3H
+
→ Ga + AsH3.
Figure 4. 32. Corresponding current density (J)-time (t) plots of bifacial, nanostructured GaAs photocathodes,
measured under simulated AM1.5G illumination at an applied potential of 0 V (vs. RHE).
Figure 4. 33. Degradation of bare ultrathin GaAs with OM/Ti/Pt without TiO 2 protective layer (a) (b) before and (c)
(d) after photoelectrochemical stability test (~14 hours) in 0.5M H 2SO 4 under AM 1.5G illumination shown on. Left
images were microscopic images from the light-absorbing interface (i.e. TiO 2 nanoposts) and right images were from
the reactive interface (i.e. Pt co-catalyst.). The failure was caused by a crack formation of underlying ultrathin GaAs
photocathodes, which was led by the stress corrosion cracking of Ti.
104
The failure of electrode here is attributed to the oxidation of titanium during the HER (see Figure
4. 33) and associated accumulation of residual stresses, leading to the crack formation in
underlying ultrathin GaAs photocathodes
97,98
. By contrast, the incorporation of TiO2 layer
between ohmic metals and Pt dramatically improved the stability with lifetime of ~175 h (green
data), similar to other monofacial photoelectrodes with a TiO2 protection layer
17,20
. Such strongly
enhanced electrochemical stability is attributed to several factors as illustrated in Figure 4. 34; (1)
TiO2 is thermodynamically stable in the HER condition
61
, (2) TiO2 retards the reductive corrosion
of GaAs by blocking proton diffusion from electrolyte
8
, (3) TiO2 suppresses the oxidation rate of
titanium (in ohmic metals) and therefore inhibits related crack formation in GaAs photocathodes
99
,
(4) titanium in ohmic metals can serve a ‘sacrificial metal’ layer to consume any protons passing
through the TiO2 by self-oxidation. The stability of bifacial GaAs photocathodes was also
improved by replacing Pt with a-MoS2, which was stable for ~25 hours with 20% reduction of
current density as shown in Figure 4. 32 (OM/MoS2, cyan) due to a relatively good stability of
Figure 4. 34. energy band diagrams of bifacial, nanostructured, ultrathin GaAs photocathodes with ohmic contact,
TiO 2, and HER co-catalyst.
105
MoSx in HER condition
64
. The main reason for the decreased current density was the
electrochemical dissolution or delamination of MoSx layer from ohmic metals possibly due to its
non-crystalline nature
64
, where GaAs photocathode still remained nearly intact without corrosion
(see Figure 4. 35). Similar to the case with Pt, the stability of bifacial GaAs photocathodes was
further enhanced by adding TiO2 layer between MoSx and ohmic metals, showing ~85 h of lifetime
with 20% reduction of current density, which is attributed to the improved adhesion of MoSx with
TiO2 by forming graded MoSx/TiO2 at the interface
21
and the enhanced durability of the
multicomponent protection layer that effectively prevented the corrosion of underlying GaAs
photocathodes.
4. 4. Conclusion
In summary, we reported a type of GaAs photocathodes that can provide simultaneous
advantages in materials cost, corrosion resistance, light management, and charge transfer.
Ultrathin GaAs photocathodes were bifacially configured to incorporate optically transparent
diffractive nanostructure at the light-absorbing interface to boost optical absorption and thus obtain
Figure 4. 35. The electrochemical dissolution/delamination of MoS x layer from ohmic metals during stability test in
0.5 M H 2SO 4. Microscopic images (a) before depositing a-MoSx on ohmic metals, (b) after depositing MoS x on ohmic
metals, (c) and (d) after 25 hours of photoelectrochemical stability test shown in Figure 4(b) (cyan), where GaAs
photocathode still remained nearly intact without corrosion.
106
comparable or even superior performance to the optically thick devices. Multicomponent
protection layers implemented at the catalytic interface, composed of ohmic metals, titanium
dioxide (TiO2), and HER co-catalysts, strongly improved the charge transfer to yield photovoltaic-
level fill-factor and current density, but also augmented the lifetime of GaAs photocathodes. This
work will be submitted in a journal for a publication. Therefore, the future journal may reserve the
copyright for the all copied contents in this chapter
107
Chapter 5. Future works
5. 1. Nanostructured, pn-junction GaInP2 photocathode
As discussed in previous chapters, an interface between semiconductor and electrolyte
plays many critical roles including light absorption, protection, charge transfer, and photovoltage
in photoelectrolysis. Among them, photovoltage can be also optimized by applying co-catalysts,
conjugation molecules, and/or buried junction to the interface, while it can be quantitatively
calculated by onset potential (Vonset) referenced to hydrogen redox potential (EHER, i.e. 0 V vs. RHE)
subtracted by overpotential ( η), which were obtainable from photocurrent density vs. potential (J-
E) in HER. To understand photovoltage, it is important to understand the energy levels of
semiconductor bands and energy distribution of the occupied and unoccupied states of the redox
acceptor
100
. Figure 5. 1(a) denotes the energy bands of semiconductors and the redox states versus
electron energy, respectively. The distribution functions of states for occupied and unoccupied
states (Dox and Drex, respectively) were given by
Figure 5. 1. (a) Energy levels of semiconductor bands and energy distribution of the occupied and unoccupied states
of the redox acceptor. (b) Band bending in n-type (left side) and p-type (right side) semiconductor electrodes upon
equilibration of the Fermi levels of the semiconductor with the redox species. Copyright (1996) The Journal of
Physical Chemistry.
108
𝐷 𝑜𝑥
= exp [−
( 𝐸 − 𝐸 𝐹 ,𝑟𝑒𝑑𝑜𝑥 − 𝜆 )
2
4𝑘𝑇𝜆 ]
𝐷 𝑟𝑒𝑑
= exp [−
( 𝐸 − 𝐸 𝐹 ,𝑟𝑒𝑑𝑜𝑥 + 𝜆 )
2
4𝑘𝑇𝜆 ]
, where EF,redox, k, T, λ were the electrochemical potential of electrons in a redox system, Boltzmann
constant, temperature, the reorganization energy of electron transfer theory, respectively.
Generally, λ ranges 0.5 ~ 2.0 eV depending on the interaction of the redox molecule with the
electrolyte. Figure 5. 1(b) shows band bending in n-type (left side) and p-type (right side)
semiconductor electrodes upon equilibration of the Fermi levels of the semiconductor with the
redox species. It is well known that bands of p-type semiconductor were bent downward, on the
contrary, bands of n-type semiconductor were bent upward, where potential varied depending on
Fermi levels of semiconductors and doping concentrations. Figure 5. 2 shows the energy positions
Figure 5. 2. Positions of energy bands of various semiconductors in the dark (d) and in light (l) with respect to the
electrochemical scale. Copyright (1996) The Journal of Physical Chemistry.
109
at the surface for various semiconductors in contact with aqueous solutions. Such band positions
and band gap of semiconductors contacting with electrolyte are crucial values to determine
photovoltage (i.e. onset potential) in PEC devices. It is noteworthy that there is a free energy loss
(Vloss), which corresponds to the energy gap between conduction band edge and hydrogen redox
potential in HER. As discussed in Chapter 1, the ideal photoelectrode must satisfy that conduction
and valence band edges must straddle the hydrogen and oxygen redox potentials to minimize such
Vloss. Typically, n-type semiconductors have a small portion of Vloss contrary to p-type
Figure 5. 3. Effects of buried junction in photoelectrodes for solar water splitting. (a) Conceptual band alignments of
p-GaInP 2 and np-GaInP 2 photocathodes and (b) photo-current density vs. potential (J-E) of IMM (Inverted
metamorphic multi-junction) without and with ultrathin n-GaInP 2 buried junction (IMM-p and IMM-pn). Copyright
(2017) Nature Energy. Similarly, (c) Conceptual band alignments and (d) photo-current density vs. potential (J-E) of
p-Si and np-Si photocathodes, respectively. Copyright (2011) Journal of the American Chemical Society.
110
semiconductors (see Figure 5. 1(b)), which can provide higher photovoltage for the more efficient
hydrogen evolution reaction. Many researchers have demonstrated ultrathin n-type buried junction
at the interface of p-type semiconductor photocathodes
7,42,101
. Figure 5. 3 shows the effect of n-
type buried junction in III-V and Si photoelectrodes for solar water splitting. By inserting a
ultrathin n-doped layer at the interface, conduction band edge lowered and almost straddled to
hydrogen redox potential, while improving onset potential by ~0.6 V and ~0.3V, respectively (see
Figure 5. 3 (b) and (d)).
Figure 5. 4(a) shows a cross-sectional SEM image of as-grown pn-GaInP2, where n-type
GaInP2 (~400 nm) is grown on p-type GaInP2 (~2 um) by a group of Dr. John Geisz in National
Renewable Energy Laboratory (NREL) using metal organic vapor phase epitaxy (MOVPE). Using
electrolessely deposited Ag nanoparticles as hard mask, n-type GaInP2 was dry-etched by
inductively coupled plasma reactive ion etching (ICP RIE) (see Figure 5. 4(b)). In this study, the
electroless Ag deposition which was described in Chapter 2 has been further modified to enable
reduction of Ag
+
by illuminating white LED light to pn-GaInP2 immersed in an aqueous solution
(see Figure 5. 5(a)). This is because the lowered valence-band edge is even lower than Ag
+
/Ag
Figure 5. 4. Cross-sectional SEM images of (a) as-grown pn-GaInP 2 and (b) nanostructured np-GaInP 2 by electroless
Ag deposition and dry-etching process. Inset image indicates zoom-in cross-sectional SEM image of nanostructured
surface.
111
redox potential (e.g. -5.2 eV in respect to vacuum level), which hinders spontaneous reduction of
Ag
+
. Also, the Schottky barrier might act as a barrier for electron transfer between n-type GaInP2
and AgNO3 solution because of the upward band bending by n-type GaInP2
102
. Therefore,
illumination was a key factor for the deposition of Ag nanoparticles on n-type semiconductor or
otherwise nothing will be created. Figure 5. 6 (a) shows reflectance spectra of bare p-GaInP2 and
nanostructured pn-GaInP2 measured by UV-Vis spectroscopy, showing a broadband antireflection.
Efficient light coupling was observed from the current density-potential curves (J-E) of
nanostructured pn-GaInP2 measured under 1-sun illumination (see Figure 5. 6 (b)), providing a
higher saturation current density (Jsat) from nanostructured pn-GaInP2. It is also noteworthy that
Vonset was remarkably improved from ~0.14 V to ~0.81 because of the presence of n-type buried
junction by tuning band position ,which are comparable to reported values (see Table 5. 1)
7
.
Overall photoelectrochemical performance of nanostructured pn-GaInP2 was further improved
Figure 5. 5. (a) A schematic illustration of photo-induced electroless deposition of Ag nanoparticles for n-type GaInP 2,
where the electrolyte-junction layer is n-type GaInP 2. (b) A photograph of pn-GaInP 2 immersed in an aqueous AgNO 3
solution during photo-induced electroless deposition under white LED illuminated.
112
after applying earth-abundant a-MoS2 co-catalyst, where overall half-reaction diagnostic
efficiency ( cathode) was significantly improved from ~0.2% to 7.7% while noticeably increasing
both FF and Jsat. In this work, there are still rooms and problems to be solved and it requires
further development of fabrication process, chemical or physical passivation, and/or the
optimization of epistructure. Nevertheless, materials design and fabrication concepts that will be
done in this study are readily applicable to other materials systems (e.g., group IV, III-V, III-N)
and electrochemical reactions (e.g., CO2 reduction) that can benefit from structurally and
compositionally tailored materials interfaces for photocatalysis, opening up new paradigm in the
efficiency and durability of unassisted solar water splitting.
Figure 5. 6. (a) reflectance spectra of bare p-GaInP 2 and nanostructured pn-GaInP 2 measured by UV-Vis spectroscopy.
(b) the current density-potential curves (J-E) of bare, NS pn-GaInP 2, and a-MoS 2 deposited NS pn-GaInP 2
photocathodes measured under 1-sun illumination.
113
5. 2. Multilayer-grown nanostructured, ultrathin GaAs photoanodes
Nanostructured, ultrathin GaAs photocathodes were successfully demonstrated in the
previous chapter, showing a potential for high-performance and cost-effective III-V photocathodes.
As a reminder, bifacial strategy significantly expanded choices of materials and design rules and
improved photoelectrochemical efficiency beyond conventional monofacial photoelectrodes,
while reducing production cost for III-V by minimizing the thickness of absorber and by reusing
GaAs growth substrate via epitaxial liftoff (ELO). However, the reported approach still cannot
eliminate “Depreciation” comprising a large portion in production cost (see Figure 5. 7(a)).
Figure 5. 7. (a) Illustration of multi-epitaxial liftoff for multi-layer ultrathin GaAs photoelectrodes. (b) (b) Cost
estimation (~2.6 $/Watt) for growing p-GaAs base layer (2.5 um) using MOVPE. Copyright (2013) National
Renewable Energy Laboratory. (c) A cross-sectional SEM image of multilayer-grown 3-stack ultrathin GaAs
photoanodes. (b) A cross-sectional SEM image of top layer consisting of n
+
-GaAs contact, AlInP window, n-GaAs
emitter, p-GaAs base, AlGaAs BSF, p
+
-GaAs contact, and sacrificial layer.
114
Depreciation is the time-consuming load-unload procedure per each device growth, which can be
effectively reduced by growing multiple layers. Figure 5. 7(b) shows a schematic illustration of
multi-epitaxial liftoff (multi ELO) for multi-layer ultrathin GaAs photoelectrodes. Figure 5. 7(c)
and (d) show cross-sectional SEM images of multilayer-grown 3-stack ultrathin GaAs
photoanodes and top layer, respectively without any visible defects, where ultrathin pn-GaAs
epilayers were grown by Professor Minjoo L. Lee’s group at the University of Illinois Urbana-
Champaign. These epilayers were intentionally grown desirably for oxygen evolution reaction (i.e.
photoanodes) with n-on-p structure. As a preliminary result, we have obtained very uniform
photovoltaic performances from 3 stacks of ultrathin GaAs photoanodes. Figure 5. 8 shows
representative photovoltaic current density (J)-voltage (V) curves of GaAs photocathodes on wafer,
where isolated ultrathin GaAs photoelectrodes from top, middle, and bottom layers showed very
similar photovoltaic characteristics including short-circuit current density, open-circuit voltage,
and power conversion efficiency. To further optimize the light-absorbing interface by employing
TiO2 nanoposts (NPs), optical properties and absorption within ultrathin GaAs photoanodes must
Figure 5. 8. representative photovoltaic current density (J)-voltage (V) curves of GaAs photocathodes on wafer, where
isolated ultrathin GaAs photoelectrodes from top, middle, and bottom layers
115
be numerically modelled and calculated by FDTD Lumerical
TM
as demonstrated in Chapter 2 and
Chapter 4. To functionalize this bifacial nanostructured, ultrathin photoanodes, there are a lot of
things to be resolved in development of fabrication processes, material selections of OER co-
catalysts and protective layers, and optical optimization.
5. 3. Conclusion
In summary, we proposed novel strategies to further improve monofacial and bifacial
approaches, respectively, for realizing high-performance, durable, and cost-competitive III-V
photoelectrodes by incorporating nanostructured interface. For monofacial approach, by
employing nanostructured, ultrathin n-type buried layer on p-type GaInP2 monofacial
photocathodes, overall half-reaction diagnostic efficiency increased from ~0.2% to ~7.7% with an
aid of amorphous molybdenum disulfide HER co-catalyst. For bifacial strategy, multi-layer grown
epitaxial liftoff have been utilized to further reduce the production cost for competitive III-V
photoelectrode in solar water splitting. We anticipate that the proposed strategies suggest feasible
routes to realize the practical application of III-V photoelectrodes for high performance and high
durable solar water splitting.
116
References
1 Roger, I., Shipman, M. A. & Symes, M. D. Earth-abundant catalysts for electrochemical
and photoelectrochemical water splitting. Nature Reviews Chemistry 1, 0003, (2017).
2 Seh, Z. W. et al. Combining theory and experiment in electrocatalysis: Insights into
materials design. Science 355, eaad4998, (2017).
3 McCrory, C. C. L., Jung, S., Peters, J. C. & Jaramillo, T. F. Benchmarking Heterogeneous
Electrocatalysts for the Oxygen Evolution Reaction. Journal of the American Chemical
Society 135, 16977-16987, (2013).
4 Polman, A., Knight, M., Garnett, E. C., Ehrler, B. & Sinke, W. C. Photovoltaic materials:
Present efficiencies and future challenges. Science 352, aad4424, (2016).
5 Khaselev, O. & Turner, J. A. A Monolithic Photovoltaic-Photoelectrochemical Device for
Hydrogen Production via Water Splitting. Science 280, 425-427, (1998).
6 Verlage, E. et al. A monolithically integrated, intrinsically safe, 10% efficient, solar-driven
water-splitting system based on active, stable earth-abundant electrocatalysts in
conjunction with tandem III-V light absorbers protected by amorphous TiO2 films. Energy
& Environmental Science 8, 3166-3172, (2015).
7 Young, J. L. et al. Direct solar-to-hydrogen conversion via inverted metamorphic multi-
junction semiconductor architectures. Nature Energy 2, 17028, (2017).
8 Kang, D. et al. Printed assemblies of GaAs photoelectrodes with decoupled optical and
reactive interfaces for unassisted solar water splitting. Nature Energy 2, 17043, (2017).
9 Cheng, W.-H. et al. Monolithic Photoelectrochemical Device for Direct Water Splitting
with 19% Efficiency. ACS Energy Letters 3, 1795-1800, (2018).
117
10 Britto, R. J. et al. Molybdenum Disulfide as a Protection Layer and Catalyst for Gallium
Indium Phosphide Solar Water Splitting Photocathodes. The Journal of Physical Chemistry
Letters 7, 2044-2049, (2016).
11 King, L. A., Hellstern, T. R., Park, J., Sinclair, R. & Jaramillo, T. F. Highly Stable
Molybdenum Disulfide Protected Silicon Photocathodes for Photoelectrochemical Water
Splitting. ACS Applied Materials & Interfaces 9, 36792-36798, (2017).
12 Vanka, S. et al. High Efficiency Si Photocathode Protected by Multifunctional GaN
Nanostructures. Nano Letters 18, 6530-6537, (2018).
13 Kibria, M. G. et al. Visible light-driven efficient overall water splitting using p-type metal-
nitride nanowire arrays. Nature Communications 6, 6797, (2015).
14 Fan, S. et al. High Efficiency Solar-to-Hydrogen Conversion on a Monolithically
Integrated InGaN/GaN/Si Adaptive Tunnel Junction Photocathode. Nano Letters 15, 2721-
2726, (2015).
15 Zheng, J. et al. Crystalline TiO2 protective layer with graded oxygen defects for efficient
and stable silicon-based photocathode. Nature Communications 9, 3572, (2018).
16 Yu, Y. et al. Enhanced photoelectrochemical efficiency and stability using a conformal
TiO2 film on a black silicon photoanode. Nature Energy 2, 17045, (2017).
17 Mei, B. et al. Crystalline TiO2: A Generic and Effective Electron-Conducting Protection
Layer for Photoanodes and -cathodes. The Journal of Physical Chemistry C 119, 15019-
15027, (2015).
18 Gu, J. et al. Water reduction by a p-GaInP2 photoelectrode stabilized by an amorphous
TiO2 coating and a molecular cobalt catalyst. Nature Materials 15, 456, (2015).
118
19 Kast, M. G., Enman, L. J., Gurnon, N. J., Nadarajah, A. & Boettcher, S. W. Solution-
Deposited F:SnO2/TiO2 as a Base-Stable Protective Layer and Antireflective Coating for
Microtextured Buried-Junction H2-evolving Si Photocathodes. ACS Applied Materials &
Interfaces 6, 22830-22837, (2014).
20 Seger, B. et al. Using TiO2 as a Conductive Protective Layer for Photocathodic H2
Evolution. Journal of the American Chemical Society 135, 1057-1064, (2013).
21 Gu, J. et al. A graded catalytic–protective layer for an efficient and stable water-splitting
photocathode. Nature Energy 2, 16192, (2017).
22 Oh, J., Deutsch, T. G., Yuan, H.-C. & Branz, H. M. Nanoporous black silicon photocathode
for H2 production by photoelectrochemical water splitting. Energy & Environmental
Science 4, 1690-1694, (2011).
23 Sun, K. et al. Metal on metal oxide nanowire Co-catalyzed Si photocathode for solar water
splitting. Nanotechnology 23, 194013, (2012).
24 Lee, M. H. et al. p-Type InP Nanopillar Photocathodes for Efficient Solar-Driven
Hydrogen Production. Angewandte Chemie 124, 10918-10922, (2012).
25 Van de Walle, C. G. & Neugebauer, J. Universal alignment of hydrogen levels in
semiconductors, insulators and solutions. Nature 423, 626-628, (2003).
26 Kocha, S. S., Turner, J. A. & Nozik, A. J. Study of the Schottky barrier and determination
of the energetic positions of band edges at the n- and p-type gallium indium phosphide
electrode | electrolyte interface. Journal of Electroanalytical Chemistry 367, 27-30, (1994).
27 MacLeod, B. A. et al. Phosphonic Acid Modification of GaInP2 Photocathodes Toward
Unbiased Photoelectrochemical Water Splitting. ACS Applied Materials & Interfaces 7,
11346-11350, (2015).
119
28 Foyt, A. G. The electro-optic applications of InP. Journal of Crystal Growth 54, 1-8,
(1981).
29 Dean, P. J., Henry, C. H. & Frosch, C. J. Infrared Donor-Acceptor Pair Spectra Involving
the Deep Oxygen Donor in Gallium Phosphide. Physical Review 168, 812-816, (1968).
30 Tsai, C.-D. & Lee, C.-T. Passivation mechanism analysis of sulfur-passivated InGaP
surfaces using x-ray photoelectron spectroscopy. Journal of Applied Physics 87, 4230-
4233, (2000).
31 Xiang, N., Tukiainen, A., Dekker, J., Likonen, J. & Pessa, M. Oxygen-related deep level
defects in solid-source MBE grown GaInP. Journal of Crystal Growth 227-228, 244-248,
(2001).
32 Schubert, E. F. Doping in III-V Semiconductors. (Cambridge University Press, 1993).
33 Bessolov, V. N. & Lebedev, M. V. Chalcogenide passivation of III–V semiconductor
surfaces. Semiconductors 32, 1141-1156, (1998).
34 Seo, J. M., Kim, Y. K., Lee, H. G., Chung, Y. S. & Kim, S. Reduction of gap states of
ternary III–V semiconductor surfaces by sulfur passivation: Comparative studies of
AlGaAs and InGaP. Journal of Vacuum Science & Technology A 14, 941-945, (1996).
35 Kwon, S. D., Kwon, H. K., Choe, B. D., Lim, H. & Lee, J. Y. Investigation of electrical
properties and stability of Schottky contacts on (NH4)2Sx‐treated n‐ and p‐type
In0.5Ga0.5P. Journal of Applied Physics 78, 2482-2488, (1995).
36 Gai, B. et al. Multilayer-Grown Ultrathin Nanostructured GaAs Solar Cells as a Cost-
Competitive Materials Platform for III–V Photovoltaics. ACS Nano 11, 992-999, (2017).
120
37 Lee, S.-M. et al. High Performance Ultrathin GaAs Solar Cells Enabled with
Heterogeneously Integrated Dielectric Periodic Nanostructures. ACS Nano 9, 10356-10365,
(2015).
38 Tompkins, H. G. & Hilfiker, J. N. Spectroscopic ellipsometry : practical application to
thin film characterization. (2016).
39 Schubert, M. et al. Optical constants of GaxIn1−xP lattice matched to GaAs. Journal of
Applied Physics 77, 3416-3419, (1995).
40 Döscher, H., Young, J. L., Geisz, J. F., Turner, J. A. & Deutsch, T. G. Solar-to-hydrogen
efficiency: shining light on photoelectrochemical device performance. Energy &
Environmental Science 9, 74-80, (2016).
41 Wang, M., Yang, Y., Shen, J., Jiang, J. & Sun, L. Visible-light-absorbing
semiconductor/molecular catalyst hybrid photoelectrodes for H2 or O2 evolution: recent
advances and challenges. Sustainable Energy & Fuels 1, 1641-1663, (2017).
42 Boettcher, S. W. et al. Photoelectrochemical Hydrogen Evolution Using Si Microwire
Arrays. Journal of the American Chemical Society 133, 1216-1219, (2011).
43 Chu, S. et al. Solar Water Oxidation by an InGaN Nanowire Photoanode with a Bandgap
of 1.7 eV. ACS Energy Letters 3, 307-314, (2018).
44 Ling, Y., Wang, G., Wheeler, D. A., Zhang, J. Z. & Li, Y. Sn-Doped Hematite
Nanostructures for Photoelectrochemical Water Splitting. Nano Letters 11, 2119-2125,
(2011).
45 Tilley, S. D., Cornuz, M., Sivula, K. & Grätzel, M. Light-Induced Water Splitting with
Hematite: Improved Nanostructure and Iridium Oxide Catalysis. Angewandte Chemie 122,
6549-6552, (2010).
121
46 Yang, Y. et al. Semiconductor interfacial carrier dynamics via photoinduced electric fields.
Science 350, 1061-1065, (2015).
47 Hu, S. et al. Amorphous TiO2 coatings stabilize Si, GaAs, and GaP photoanodes for
efficient water oxidation. Science 344, 1005-1009, (2014).
48 Essig, S. et al. Raising the one-sun conversion efficiency of III–V/Si solar cells to 32.8%
for two junctions and 35.9% for three junctions. Nature Energy 2, 17144, (2017).
49 Sayed, S. Y., Daly, B. & Buriak, J. M. Characterization of the Interface of Gold and Silver
Nanostructures on InP and GaAs Synthesized via Galvanic Displacement. The Journal of
Physical Chemistry C 112, 12291-12298, (2008).
50 Huang, Y.-F. et al. Improved broadband and quasi-omnidirectional anti-reflection
properties with biomimetic silicon nanostructures. Nature Nanotechnology 2, 770, (2007).
51 Chan, L. et al. Broadband antireflection and absorption enhancement of ultrathin silicon
solar microcells enabled with density-graded surface nanostructures. Applied Physics
Letters 104, 223905, (2014).
52 Raut, H. K., Ganesh, V. A., Nair, A. S. & Ramakrishna, S. Anti-reflective coatings: A
critical, in-depth review. Energy & Environmental Science 4, 3779-3804, (2011).
53 Xi, J. Q. et al. Optical thin-film materials with low refractive index for broadband
elimination of Fresnel reflection. Nature Photonics 1, 176, (2007).
54 Chen, Z. et al. Accelerating materials development for photoelectrochemical hydrogen
production: Standards for methods, definitions, and reporting protocols. Journal of
Materials Research 25, 3-16, (2010).
55 Tajik, N., Haapamaki, C. M. & LaPierre, R. R. Photoluminescence model of sulfur
passivated p-InP nanowires. Nanotechnology 23, 315703, (2012).
122
56 Jaramillo, T. F. et al. Identification of Active Edge Sites for Electrochemical
H
2
Evolution from MoS
2
Nanocatalysts. Science 317, 100-102,
(2007).
57 Kibsgaard, J., Chen, Z., Reinecke, B. N. & Jaramillo, T. F. Engineering the surface
structure of MoS2 to preferentially expose active edge sites for electrocatalysis. Nature
Materials 11, 963, (2012).
58 Li, H. et al. Activating and optimizing MoS2 basal planes for hydrogen evolution through
the formation of strained sulphur vacancies. Nature Materials 15, 48, (2015).
59 Merki, D. & Hu, X. Recent developments of molybdenum and tungsten sulfides as
hydrogen evolution catalysts. Energy & Environmental Science 4, 3878-3888, (2011).
60 Merki, D., Fierro, S., Vrubel, H. & Hu, X. Amorphous molybdenum sulfide films as
catalysts for electrochemical hydrogen production in water. Chemical Science 2, 1262-
1267, (2011).
61 Pourbaix, M. Atlas of electrochemical equilibria in aqueous solutions. 2nd edn, (National
Association of Corrosion Engineers, 1974).
62 Park, S.-M. & Barber, M. E. Thermodynamic stabilities of semiconductor electrodes.
Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 99, 67-75, (1979).
63 Young, J. L., Steirer, K. X., Dzara, M. J., Turner, J. A. & Deutsch, T. G. Remarkable
stability of unmodified GaAs photocathodes during hydrogen evolution in acidic
electrolyte. Journal of Materials Chemistry A 4, 2831-2836, (2016).
64 Benck, J. D., Hellstern, T. R., Kibsgaard, J., Chakthranont, P. & Jaramillo, T. F. Catalyzing
the Hydrogen Evolution Reaction (HER) with Molybdenum Sulfide Nanomaterials. ACS
Catalysis 4, 3957-3971, (2014).
123
65 Hellstern, T. R. et al. Molybdenum Disulfide Catalytic Coatings via Atomic Layer
Deposition for Solar Hydrogen Production from Copper Gallium Diselenide
Photocathodes. ACS Applied Energy Materials 2, 1060-1066, (2019).
66 Hellstern, T. R. et al. Nanostructuring Strategies To Increase the Photoelectrochemical
Water Splitting Activity of Silicon Photocathodes. ACS Applied Nano Materials 2, 6-11,
(2019).
67 Shen, L. et al. Nanostructured Silicon Photocathodes for Solar Water Splitting Patterned
by the Self-Assembly of Lamellar Block Copolymers. ACS Applied Materials & Interfaces
7, 26043-26049, (2015).
68 Wang, H., Deutsch, T., Welch, A. & Turner, J. A. The stability of illuminated p-GaInP2
semiconductor photoelectrode. International Journal of Hydrogen Energy 37, 14009-
14014, (2012).
69 Lee, C.-T., Lan, M.-H. & Tsai, C.-D. Improved performances of InGaP Schottky contact
with Ti/Pt/Au metals and MSM photodetectors by (NH4)2Sx treatment. Solid-State
Electronics 41, 1715-1719, (1997).
70 Gao, L. et al. High-Efficiency InP-Based Photocathode for Hydrogen Production by
Interface Energetics Design and Photon Management. Advanced Functional Materials 26,
679-686, (2016).
71 Malizia, M., Seger, B., Chorkendorff, I. & Vesborg, P. C. K. Formation of a p-n
heterojunction on GaP photocathodes for H2 production providing an open-circuit voltage
of 710 mV. Journal of Materials Chemistry A 2, 6847-6853, (2014).
72 Varadhan, P. et al. Surface Passivation of GaN Nanowires for Enhanced
Photoelectrochemical Water-Splitting. Nano Letters 17, 1520-1528, (2017).
124
73 National Renewable Energy Laboratory. Golden, CO, Best Research-Cell Efficiency Chart,
(2019). (https://www.nrel.gov/pv/cell-efficiency.html).
74 Hu, S., Xiang, C., Haussener, S., Berger, A. D. & Lewis, N. S. An analysis of the optimal
band gaps of light absorbers in integrated tandem photoelectrochemical water-splitting
systems. Energy & Environmental Science 6, 2984-2993, (2013).
75 Seitz, L. C. et al. Modeling Practical Performance Limits of Photoelectrochemical Water
Splitting Based on the Current State of Materials Research. ChemSusChem 7, 1372-1385,
(2014).
76 Green, M. A. et al. Solar cell efficiency tables (Version 53). Progress in Photovoltaics:
Research and Applications 27, 3-12, (2019).
77 Yamaguchi, M. III–V compound multi-junction solar cells: present and future. Solar
Energy Materials and Solar Cells 75, 261-269, (2003).
78 Bosi, M. & Pelosi, C. The potential of III-V semiconductors as terrestrial photovoltaic
devices. Progress in Photovoltaics: Research and Applications 15, 51-68, (2007).
79 Pinaud, B. A. et al. Technical and economic feasibility of centralized facilities for solar
hydrogen production via photocatalysis and photoelectrochemistry. Energy &
Environmental Science 6, 1983-2002, (2013).
80 Atwater, H. A. & Polman, A. Plasmonics for improved photovoltaic devices. Nature
Materials 9, 205, (2010).
81 Mokkapati, S. & Catchpole, K. R. Nanophotonic light trapping in solar cells. Journal of
Applied Physics 112, 101101, (2012).
125
82 Woodhouse, M. & Goodich, A. A Manufacturing Cost Analysis Relevant to Single- and
Dual-Junction Photovoltaic Cells Fabricated with III-Vs and III-Vs Grown on Czochralski
Silicon. 1-92 (NREL, 2013).
83 Beard, M. C., Luther, J. M. & Nozik, A. J. The promise and challenge of nanostructured
solar cells. Nature Nanotechnology 9, 951, (2014).
84 Konagai, M., Sugimoto, M. & Takahashi, K. High efficiency GaAs thin film solar cells by
peeled film technology. Journal of Crystal Growth 45, 277-280, (1978).
85 Yablonovitch, E., Gmitter, T., Harbison, J. P. & Bhat, R. Extreme selectivity in the lift‐off
of epitaxial GaAs films. Applied Physics Letters 51, 2222-2224, (1987).
86 Cheng, C.-W. et al. Epitaxial lift-off process for gallium arsenide substrate reuse and
flexible electronics. Nature Communications 4, 1577, (2013).
87 Gai, B., Sun, Y., Chen, H., Lee, M. L. & Yoon, J. 10-Fold-Stack Multilayer-Grown
Nanomembrane GaAs Solar Cells. ACS Photonics 5, 2786-2790, (2018).
88 Lee, D. K. & Choi, K.-S. Enhancing long-term photostability of BiVO4 photoanodes for
solar water splitting by tuning electrolyte composition. Nature Energy 3, 53-60, (2018).
89 Jiang, C., Moniz, S. J. A., Wang, A., Zhang, T. & Tang, J. Photoelectrochemical devices
for solar water splitting - materials and challenges. Chemical Society Reviews 46, 4645-
4660, (2017).
90 McCrory, C. C. L. et al. Benchmarking Hydrogen Evolving Reaction and Oxygen Evolving
Reaction Electrocatalysts for Solar Water Splitting Devices. Journal of the American
Chemical Society 137, 4347-4357, (2015).
126
91 Bae, D., Seger, B., Vesborg, P. C. K., Hansen, O. & Chorkendorff, I. Strategies for stable
water splitting via protected photoelectrodes. Chemical Society Reviews 46, 1933-1954,
(2017).
92 Bae, D. et al. Back-illuminated Si photocathode: a combined experimental and theoretical
study for photocatalytic hydrogen evolution. Energy & Environmental Science 8, 650-660,
(2015).
93 Lee, S. M. et al. High Performance Ultrathin GaAs Solar Cells Enabled with
Heterogeneously Integrated Dielectric Periodic Nanostructures. Acs Nano 9, 10356-10365,
(2015).
94 Sai, H., Kanamori, Y., Arafune, K., Ohshita, Y. & Yamaguchi, M. Light trapping effect of
submicron surface textures in crystalline Si solar cells. Progress in Photovoltaics:
Research and Applications 15, 415-423, (2007).
95 Thomas, N. T. & Nobe, K. Kinetics of the Hydrogen Evolution Reaction on Titanium.
Journal of The Electrochemical Society 117, 622-626, (1970).
96 Ting, L. R. L. et al. Catalytic Activities of Sulfur Atoms in Amorphous Molybdenum
Sulfide for the Electrochemical Hydrogen Evolution Reaction. ACS Catalysis 6, 861-867,
(2016).
97 Woodtli, J. & Kieselbach, R. Damage due to hydrogen embrittlement and stress corrosion
cracking. Engineering Failure Analysis 7, 427-450, (2000).
98 Seger, B. et al. Hydrogen Production Using a Molybdenum Sulfide Catalyst on a Titanium-
Protected n+p-Silicon Photocathode. Angewandte Chemie International Edition 51, 9128-
9131, (2012).
127
99 Nelson, J. G. & Murray, G. T. Prevention of hydrogen embrittlement by a TiO2 surface
layer. Metallurgical Transactions A 15, 597-600, (1984).
100 Nozik, A. J. & Memming, R. Physical Chemistry of Semiconductor−Liquid Interfaces. The
Journal of Physical Chemistry 100, 13061-13078, (1996).
101 Bae, D. et al. Carrier-selective p- and n-contacts for efficient and stable photocatalytic
water reduction. Catalysis Today 290, 59-64, (2017).
102 Chen, C.-Y. & Wong, C.-P. Unveiling the shape-diversified silicon nanowires made by
HF/HNO3 isotropic etching with the assistance of silver. Nanoscale 7, 1216-1223, (2015).
Abstract (if available)
Abstract
High performance and high durability of photoelectrodes are the most critical factors to realize practical utilization of III-V compound semiconductors in solar water splitting. Here we analytically departmentalized such several factors and proposed different approaches to develop those factors by categorizing the form factors of photoelectrodes into two configurations
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Multilayer grown ultrathin nanostructured GaAs solar cells towards high-efficiency, cost-competitive III-V photovoltaics
PDF
Nanomaterials for energy storage devices and electronic/optoelectronic devices
PDF
Enhanced photocatalysis on titanium oxide passivated III-V semiconductors
PDF
Synthesis and characterization of novel donor-acceptor systems for solar water splitting
PDF
Light management in nanostructures: nanowire solar cells and optical epitaxial growth
PDF
Cathode and anode materials for sodium ion batteries
PDF
Nanostructure design of sulfur cathodes and lithium metal anodes for lithium-ion batteries
PDF
Optical and electrical characterization of one-dimensional (1D) and two-dimensional (2D) nanostructures
PDF
GaN nanostructures grown by selective area growth for solid-state lighting
PDF
Nanostructured silicon for lithium-ion battery anode
PDF
Development of novel optical materials for whispering gallery mode resonator for nonlinear optics
PDF
Molecular dynamics simulations of nanostructures
PDF
Printed electronics based on carbon nanotubes and two-dimensional transition metal dichalcogenides
PDF
2D layered materials: fundamental properties and device applications
PDF
Hot carriers in bare metals and photocatalytically active defect sites in dielectric/metal structures
PDF
Synthesis, characterization, and device application of two-dimensional materials beyond graphene
PDF
Novel nanomaterials for electronics, optoelectronics and sensing applications
PDF
Fabrication and application of plasmonic nanostructures: a story of nano-fingers
PDF
Catalysis for solar fuels production: molecular and materials approaches
PDF
Plasmonic enhancement of catalysis and solar energy conversion
Asset Metadata
Creator
Lim, Haneol
(author)
Core Title
Nanostructured III-V photoelectrodes for high performance, high durability solar water splitting
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Materials Science
Publication Date
07/25/2021
Defense Date
04/30/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
GaAs,GaInP,III-V,nanostructures,nanotechnology,OAI-PMH Harvest,photoelectrodes,solar fuel,solar water splitting
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Yoon, Jongseung (
committee chair
), Wu, Wei (
committee member
), Zhou, Chongwu (
committee member
)
Creator Email
haneol.lim@gmail.com,haneolli@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-198011
Unique identifier
UC11662789
Identifier
etd-LimHaneol-7667.pdf (filename),usctheses-c89-198011 (legacy record id)
Legacy Identifier
etd-LimHaneol-7667.pdf
Dmrecord
198011
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Lim, Haneol
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
GaAs
GaInP
III-V
nanostructures
nanotechnology
photoelectrodes
solar fuel
solar water splitting