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
/
Artificial photosynthesis on titanium oxide passivated III-V semiconductors
(USC Thesis Other)
Artificial photosynthesis on titanium oxide passivated III-V semiconductors
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
ARTIFICIAL PHOTOSYNTHESIS ON TITANIUM OXIDE PASSIV ATED III-V
SEMICONDUCTORS
By
Guangtong Zeng
_________________________________________________________________
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
(CHEMISTRY)
May 2016
Copyright 2015 GUANGTONG ZENG
ii
Dedication
Dedicated to my beloved parents, brothers and my girlfriend Miss Beibei Zhu.
iii
Acknowledgments
First of all, I would like to express my sincere gratitude to my academic advisor Prof.
Stephen B. Cronin, who has always been a very supportive mentor full of creative ideas.
During my years as a graduate student, Professor Cronin has provided me with invaluable
suggestions and guidance to enable the fruitfulness and success of my projects. His
comprehensive knowledge, innovative ideas, and in-depth insight into research have
always been of great help to me. It is extremely joyful to work in such a free and inspiring
research environment cultivated by Professor Cronin.
My sincere thanks also go to my dissertation committees Prof. Alexandar Bernderskii
and Prof. Wei Wu for their invaluable suggestions, as well as Prof. Surya Prakash, Prof.
Curt Wittig and Prof. Ralf Haiges for serving on my qualifying exam committee.
I extend my gratitude to our collaborators Dr. Yonjing Lin and Mr. Mark Hettick from
University of California, Berkeley, and Dr. Robert Peterson from UCLA for help on NMR.
Also, Ms. Kim Reid and Ms. Dea Mitchell were extremely helpful in their administrative
support and management of funding.
I am extremely thankful Dr. Jing Qiu for her continuous support during my PhD study.
I am exceedingly grateful to her for everything she taught me and the collaboration projects
would not be possible without her. Thanks to Jing for being such a good mentor, colleague
and friend.
iv
I am very grateful to spend my graduate student life in such a great research group and
I have to acknowledge my friends and labmates who have maintained a highly professional
and friendly environment to work. Among these people are Miss Bingya Hou, Mr. Haotian
Shi, Mr. Lang Shen, Mr. Rohan Dhall, Mr. Zhen Li and Mr. Nirakar Poudel, Dr. Chia-Chi
Chang, Dr. Chun-Chung Chen, Dr. Shermin Arab, Dr. Shun-Wen Chang, Dr. Wenbo Hou,
Dr. Mehmet Aykol, Dr. Mohammed Amer, Dr. Jesse Theiss, Dr. Ioannis Chatzakis, thank
you all for making PhD journey fun. I really enjoyed working with you.
Finally, I would like to express my whole-hearted gratefulness to my parents, my
siblings who have given me endless love and encouragement. I would like to thank my
beloved girlfriend Miss Beibei Zhu, who is always considerate and supportive, and brought
sunshine and happiness to my graduate life.
5
Contents
Dedication ii
Acknowledgments iii
List of Tables 7
List of Figures 8
Abstract 15
Chapter 1: Introduction and Background 19
1.1 Fundamentals ...................................................................................................... 19
1.2 General Introduction of CO2 ............................................................................... 21
1.3 The Difficulty of CO2 Reduction ........................................................................ 23
1.4 Photochemical and Photoelectrochemical Reduction of CO2 ............................. 25
1.5 Factors Limiting the Photocatalytic Efficiency .................................................. 29
1.6 Strategies to Enhance Photoelectrochemical Performance ................................. 32
1.6.1 Advantages of III-V Semiconductor Photocatalysis ................................ 33
1.6.2 Advantages of Thin Layer TiO2 Passivation ............................................ 36
Chapter 2: Photocatalytic Conversion of CO2 to Hydrocarbon Fuels on GaP 38
2.1 Introduction ......................................................................................................... 38
2.2 Experimental Procedure ...................................................................................... 39
2.3 Results and Discussion ....................................................................................... 41
2.4 Conclusion .......................................................................................................... 50
6
Chapter 3: Artificial Photosynthesis on TiO2-passivated InP Nanopillars 52
3.1 Introducton .......................................................................................................... 52
3.2 Experimental Procedure ...................................................................................... 52
3.3 Results and Discussion ....................................................................................... 56
3.4 Conclusion .......................................................................................................... 65
Chapter 4: Enhanced Photocatalytic Reduction of CO2 to CO through
TiO2 Passivation of InP in Ionic Liquids 67
4.1 Introduction ......................................................................................................... 67
4.2 Experimental Procedure ...................................................................................... 69
4.3 Results and Discussion ....................................................................................... 71
4.4 Conclusion .......................................................................................................... 81
Chapter 5: Plasmon-enhanced water splitting on TiO2-passivated GaP photocatalysts
82
5.1 Introducton .......................................................................................................... 82
5.2 Experimental procedure ...................................................................................... 84
5.3 Results and Discussion ....................................................................................... 85
5.4 Conclusion .......................................................................................................... 94
Chapter 6: Outlook and Future Work 96
References: 102
7
List of Tables
Table 2.1 Shift of onset overpotential for samples with different thicknesses of TiO2
Table 3.1 Faraday efficiencies of methanol of 12-hour reaction on InP/TiO2/Cu
sample under 532nm illumination in a CO2 saturated 0.5M KCl solution
Table 3.2 Faraday efficiencies of hydrogen and methanol of 12-hour reaction on
InP/TiO2/Cu sample under 532nm illumination in a CO2 saturated 0.5M KCl solution
.....42
…………...….63
....65
8
List of Figures
Figure 1.1 Energy band diagram of various semiconductors, plotted together with the
redox potentials of some chemical reaction
Figure 1.2 Schematic diagram of photoelectrochemical system
Figure 1.3 Energy band diagram of various semiconductors, plotted together with the
redox potentials of CO2 reduction
Figure 1.4 Schematic diagrams of four different schemes for light-assisted CO2
reduction on a semiconducting photocathode: (a) heterogeneous catalysis on a
semiconductor electrode, (b) heterogeneous catalysis on a metal-decorated
semiconductor electrode, (c) homogeneous catalysis through a
semiconductor/molecular catalyst junction, and (d) heterogeneous catalysis through
a molecular catalyst–decorated semiconductor electrode.
Figure 1.5 Overall Proposed Mechanism for the Pyridinium-Catalyzed Reduction of
CO2 to the Various Products of Formic Acid, Formaldehyde, and Methanol.
Figure 1.6 Solar spectrum and absorption spectrum of TiO2 and α-Fe2O3
Figure 1.7 Theoretical photocurrent densities for semiconductors under one-sun
illumination. The blue curve is the integrated photon flux at different cutoff energies,
based on the AM1.5G solar spectrum (ASTM G173-03 reference). The theoretical
photocurrent is calculated assuming that all incident photons above the
semiconductor band gap contribute to the photocurrent. The arrows at the bottom
...……………………………………......….20
…………………………………………………..…...24
……………………………….27
………………28
……………...31
…………...31
...……………………………......….22
9
indicate the regions of ultraviolet (UV) (below 400nm), visible (400−750 nm), and
infrared (IR) spectra
Figure 1.8 Solar spectrum and absorption spectrum of TiO2 and III-V semiconductors
Figure 1.9 Shockley-Quesisser Limit of TiO2 and III-V semiconductors
Figure 2.1 Schematic diagram of sample geometry
Figure 2.2 (a) Optical microscope image, (b) atomic force microscope image, and (c)
surface topography of bare GaP surface after 8h reaction at -0.5V overpotential. (d)
Optical microscope image, (e) atomic force microscope image, and (f) surface
topography of 5nm TiO2 on GaP surface after 8h reaction
Figure 2.3 (a) Photocatalytic current-potential curves of GaP photocatalysts with
different TiO2 thicknesses in a 0.5M NaCl, 10mM pyridine solution under 532nm
wavelength laser illumination. (b) Decrease of overpotential plotted as a function of
TiO2 thickness on GaP. (c) Calculated built-in voltage plotted as a function of TiO2
thickness. (d) NMR spectra showing methanol production using bare GaP and 5nm
TiO2-passiavated GaP photocatalysts at an overpotential of -0.50V
Figure 2.4 Gas chromatograph (GC) data taken after 8 hour illumination (532nm) of
bare and 5nm TiO2 passivated-GaP in 0.5M NaCl, 10mM pyridine solution at an
overpotential of -0.5V (vs NHE). The GC data is plotted together with a calibration
standard consisting of 10
-4
M methanol in aqueous solution. Based on this data, the
……………………………………………………………………..32
………………………………….....41
……………………………..41
………………….....43
……………………………………………………………………………………….36
………………36
10
5nm TiO2 passivated-GaP was found to have a Faradaic efficiency of 55%
Figure 2.5
13
C-NMR spectrum showing methanol peak taken after 8 hour
illumination (532nm) of 5nm TiO2 passivated-GaP at an overpotential of -0.5V (vs
NHE) in 0.5M NaCl, with 10mM pyridine solution
Figure 2.6
1
H-NMR spectra showing methanol peaks taken after 8 hour illumination
(532nm) of 5nm TiO2 passivated-GaP at an overpotential of -0.5V (vs NHE) in 0.5M
NaCl, with 10mM pyridine solution and without 10mM pyridine
Figure 2.7
(a) Optical microscope image, (b) atomic force microscope image, and (c)
surface topography of TiO2 passivated GaP surface after 8h reaction in 0.5M NaCl
without pyridine at -0.5V overpotential
Figure 2.8 Energy band alignment of GaP and TiO2 together with the relevant redox
potentials of CO2
Figure 3.1 (a) Schematic diagram of TiO2-passivated InP nanopillars with Cu
cocatalyst nanoparticles. (b-e) SEM and TEM images of InP nanopillar array with
TiO2 deposition layer and Cu nanoparticles. The high resolution TEM image in (e)
resolves the crystal lattice of the Cu nanoparticles
Figure 3.2 (a) Cu 2p for CuO and Cu, (b) O 1s core level XPS spectra of air-oxide
Cu nanoparticles
Figure 3.3 Time dependence of the photocurrent density of InP with 3nm TiO2 and
Cu nanoparticles illuminated with 532nm light at an applied overpotential of -0.6V
……………46
……………………………………..47
……………………....47
………………………………………………....48
………………………………………………………………………....49
……………………………………...54
………………………………………………………………………....55
11
vs. NHE
Figure 3.4 (a) Photocatalytic current-potential curves where the dashed line indicates
the potential applied during the methanol test. (b) Log plot of photocatalytic current-
potential curves. (c) Methanol peak in NMR spectra and (d) Faraday efficiencies of
methanol production for InP nanopillars with and without 3nm TiO2 under 532nm
illumination in a CO2 saturated 0.5M KCl solution
Figure 3.5 Ti 2p level XPS spectra of TiO2 on InP, which shows the presence of Ti
3+
states
Figure 3.6 PW-DFT calculated structure for anatase TiO2 with O vacancies (a) before
CO2 adsorption and (b) after CO2 adsorption and relaxation
Figure 3.7 PW-DFT result of neutral CO2 adsorbed to the stoichiometric anatase 101
surface with a binding energy of -0.48 eV
Figure 3.8 (a) Photocatalytic current-potential curves and (b) Faraday efficiencies of
methanol production for samples of bare InP, InP/Cu, and InP/TiO2/Cu under 532nm
illumination in a CO2 saturated 0.5M KCl solution for a12-hour reaction
Figure 3.9 Gas chromatograph (GC) data taken after 12-hour reaction on
InP/TiO2/Cu sample under 532nm illumination in a CO2 saturated 0.5M KCl solution.
Figure 4.1 (a) Schematic of the electrochemical cell and (b) sample geometry of
TiO2-passivated p-InP photocathode.
……………………………………………………………………………….....56
…………………………………......58
…………………………………………………………………………………......58
…………………………....60
…………………………………………….....60
……………...63
....65
……………………………...………………….....70
12
Figure 4.2 Photocatalytic current-potential curves of InP photocatalysts with
different TiO2 thicknesses in a 0.02M [EMIM]BF4 non-aqueous electrolyte, under
532nm wavelength illumination.
Figure 4.3 (a) Photocatalytic current-potential curves of InP photocatalysts with Pt
nanoparticles with various thicknesses of TiO2 in a 0.02M [EMIM]BF4 non-aqueous
electrolyte under 532nm wavelength laser illumination. (b) Photocurrent-potential
curves of 1nm TiO2–passivated InP compared to 1nm TiO2–passivated InP with Pt
nanoparticles.
Figure 4.4 Photoluminescence spectra of (a) InP passivated with various thickness
of TiO2, (b) Pt nanoparticles deposited on TiO2-passviated InP
Figure 4.5 Cyclic voltammograms of 3nm TiO2-passivated InP in electrolytes
saturated with N2 and CO2 as indicated in 0.02 M [EMIM]BF4/AcN and 0.02M
[EMIM]BF4/AcN with 0.1M LiClO4/AcN
Figure 4.6 (a)
1
H NMR spectrum (in DMSO, reference to TMS) of 0.02M
[EMIM]BF4 electrolyte solution after 4h of CO2 reduction. (b)
1
H NMR spetrum of
pure [EMIM]BF4. (c) Schematic representation of the intermediate complex that is
formed between the [EMIM] ion and the CO2
-
.
Figure 4.7 Quadruple mass spectrometer data for molecular mass of 29 (
13
CO).
Figure 4.8 CO Faradaic efficiency for different thickness TiO2 passivated InP in
different molarity of [EMIM]BF4 in AcN electrolyte, under 532nm wavelength laser
…………………………………...………………….....72
……………………………………………………...………………….....73
…………………………………….....78
……79
………………………………………….....76
……………………….....75
13
illumination after 4 hours irradiation and -1.37V applied voltage vs. NHE.
Figure 5.1 Schematic diagram of sample geometry for GaP photocatalysts with TiO2
and Au nanoparticles
Figure 5.2 (a) Photocatalytic current-potential curves measured for GaP
photocatalysts with various thicknesses of TiO2 under 1W/cm
2
532nm illumination
in a 0.5M Na2SO4 pH=7 solution. (b) Photocatalytic current plotted as a function of
potential for thicker TiO2 layers. (c) Relative decrease of the overpotential required
to initiate this reaction and (d) calculated built-in voltage plotted as a function of TiO2
thickness.Energy band diagrams for (e) thin and (f) thick TiO2 layers
Figure 5.3 Photocurrent plotted as a function of voltage for GaP photocatalysts with
various thicknesses of TiO2 with Au nanoparticles under 1W/cm
2
532nm illumination
in a 0.5M Na2SO4 solution
Figure 5.4 Electric field distributions calculated using the finite difference time
domain method (a) in the plane of the Au nanoparticles and (c,d) in the perpendicular
direction across the GaP/TiO2/Au/electrolyte interface. (b) Calculated electric field
enhancement factor plotted as a function of TiO2 thickness using Eq. 1
Figure 5.5 Two terminal photocurrent density (absolute value) plotted as a function
of the applied overpotential for various GaP/TiO 2/Au nanoparticle photocatalysts
measured in a pH=0, 0.5M H2SO4 solution under 1W/cm
2
532nm illumination
Figure 5.6 (a) Time dependence of the photocurrent density of bare GaP illuminated
……………………………………………………………….....85
……………….....87
………………………………………………………………..89
……………......89
……......92
…………...80
14
with 1W/cm
2
532nm light in a 0.5M Na2SO4 solution at an applied overpotential of -
0.7V . (b) Optical microscope image, (c) atomic force microscope image, and (d)
surface topography of the GaP surface after the 5 hour reaction
Figure 5.7 (a) Time dependence of the photocurrent density of TiO2 passivated GaP
illuminated with 1W/cm
2
532nm light in a 0.5M Na2SO4 solution at an applied
overpotential of -0.7V . (b) Optical microscope image, (c) atomic force microscope
image, and (d) surface topography of the GaP/TiO2 surface after the 12 hour reaction.
Figure 6.1 (a,b) Schematic diagrams illustrating the thin layer electrochemical cell
with in situ SFG spectroscopy, where light is irradiated through the top CaF2 window.
(c) Photograph showing previous micro-channel reactor with glass cover and metal
eyelets. ……….………………………………………………….………………………99
....94
……………………......93
15
Abstract
Artificial photosynthesis has become a hot research area since first demonstration of
photocatalytic water splitting using TiO2 in 1972. For photocatalytic materials, III-V
semiconductors such as GaP, InP, and GaAs are promising candidates with theoretical
rational band gap energies and high maximum photocurrent densities of 9mA/cm
2
,
35mA/cm
2
, and 32mA/cm
2
, respectively. However, photocatlytic corrosion of III-V
semiconductors prevents them from being utilized as reliable photocatalysts. During my
PhD program, our research group has successfully developed a strategy of using the atomic
layer deposition (ALD) passivates the surface of III-V semiconductors with a thin layer of
TiO2 (less than 10nm), which protects them from corrosion. What’s more, this thin layer of
TiO2 enhance overall photoconversion efficiency substantially.
This dissertation will begin with an introduction of the fundamentals of
photocatalytical semiconductors, followed by the application of CO2 reduction. After that,
we also discuss about the factors limiting the photocatalytic conversion efficiency. And we
discuss the advantages of III-V semiconductors and TiO2 passivation. In the following
chapters, we will report the achievements we have made on the enhanced photocatalytic
CO2 reduction processes.
In Chapter 2, we report photocatalytic CO2 reduction with water to produce methanol
using TiO2-passivated GaP photocathodes under 532nm wavelength illumination. The TiO2
16
layer prevents corrosion of the GaP, as evidenced by atomic force microscopy and
photoelectrochemical measurements. Here, the GaP surface is passivated using a thin film
of TiO2 deposited by atomic layer deposition (ALD), which provides a viable, stable
photocatalyst without sacrificing photocatalytic efficiency. In addition to providing a stable
photocatalytic surface, the TiO2-passivation provides substantial enhancement in the
photoconversion efficiency through passivation of surface states, which cause non-
radiative carrier recombination. In addition to passivation effects, the TiO 2 deposited by
ALD is n-type due to oxygen vacancies, and forms a pn-junction with the underlying p-
type GaP photocathode. This creates a built-in field that assists in the separation of
photogenerated electron-hole pairs, further reducing recombination. This reduction in the
surface recombination velocity (SRV) corresponds to a shift in the overpotential of almost
0.5V . No enhancement is observed for TiO2 thicknesses above 10nm, due to the insulating
nature of the TiO2, which eventually outweighs the benefits of passivation.
In Chapter 3, photocatalytic CO2 reduction with water to produce methanol is
demonstrated using TiO2-passivated InP nanopillar photocathodes under 532nm
wavelength illumination. In addition to providing a stable photocatalytic surface, the TiO2-
passivation layer provides substantial enhancement in the photoconversion efficiency
through the introduction of O vacancies associated with the non-stoichiometric growth of
TiO2 by atomic layer deposition. Plane wave-density functional theory (PW-DFT)
calculations confirm the role of oxygen vacancies in the TiO2 surface, which serve as
17
catalytically active sites in the CO2 reduction process. PW-DFT shows that CO2 binds
stably to these oxygen vacancies and CO2 gains an electron (-0.897e) spontaneously from
the TiO2 support. This calculation indicates that the O vacancies provide active sites for
CO2 absorption, and no overpotential is required to form the CO2
-
intermediate.
In Chapter 4, we present a robust and reliable method for improving the photocatalytic
performance of InP, which is one of the best known materials for solar photoconversion
(i.e., solar cells). In this article, we report substantial improvements (up to 18X) in the
photocatalytic yields for CO2 reduction to CO through the surface passivation of InP with
TiO2 deposited by atomic layer deposition (ALD). Here, the main mechanisms of
enhancement are the introduction of catalytically active sites and the formation of a pn-
junction. Photoelectrochemical reactions were carried out in a non-aqueous solution
consisting of ionic liquid (1-ethyl-3-methylimidazolium tetrafluoroborate ([EMIM]BF4))
dissolved in acetontrile, which enables CO2 reduction with a Faradaic efficiency of 99% at
an underpotential of +0.78V . While the photocatalytic yield increases with the addition of
the TiO2 layer, a corresponding drop in the photoluminescence intensity indicates the
presence of catalytically active sites, which cause in increase in the electron-hole pair
recombination rate. NMR spectra show that the [EMIM]
+
ions in solution form an
intermediate complex with CO2
-
, thus lowering the energy barrier of this reaction.
In Chapter 5, we demonstrate that a thin layer of n-type TiO2 using atomic layer
deposition (ALD) prevents corrosion of p-type GaP, as evidenced by atomic force
18
microscopy and photoelectrochemical measurements. In addition, the TiO2 passivation
layer provides an enhancement in photoconversion efficiency through the formation of a
charge separating pn-region. Plasmonic Au nanoparticles deposited on top of the TiO2-
passivated GaP further increases the photoconversion efficiency through local field
enhancement. Finite difference time domain (FDTD) simulations of the electric field
profiles in this photocatalytic heterostructure corroborate the experimental results.
In Chapter 6, in order to separate the various mechanisms of the catalysis and
enhancement, we will use the use vibrational sum frequency generation (vSFG)
spectroscopy to identify the reactant and intermediate species adsorbed at the active surface
sites on the photocatalytic substrate. Also, we will use the reflection and total internal
reflection Fourier transform infrared spectroscopy (FTIR) measurements to analyze the
vibrational frequency range of the OH, CH and CO stretches. Those two measurements
will provide a more complete understanding of the surface bound intermediates in this
photocatalytic reaction system.
19
Chapter 1: Introduction and Background
1.1 Fundamentals
Solar energy is a clean, renewable, carbon neutral and abundant energy alternative to
fossil fuels. In addition to direct solar-to-electric conversion (as in solar cell),
photocatalysis provides an alternative method for storing the solar energy in chemical
bonds that can be released later without producing harmful byproducts. Many
semiconductors serve as good photocatalysts due to their electronic band structures
1-4
.
From energy-band theory of single-crystal materials, bands of allowed energy states
that the electrons may occupy separated by bands of forbidden energies. At absolute zero
degree, electrons are in the lowest energy state. For some single-crystal materials, all states
in the lower bands (normally we call the valence band) will be full, and all states in the
upper bands (normally we call it the conduction band) will be empty. And the gap width of
the forbidden energy bands between the top-most of the valence bands and the bottom-
most of the conduction bands is the Bandgap Energy (Eg).
From wave-particle duality principle, light waves can be treated as particles, which are
referred to as photons. The energy of one photon is E=hν, where h is Planck’s constant and
ν is the light frequency. We can relate the wavelength and energy by λ=1240/E (nm). When
a photon collides with an electron at the valance band in a solid (usually a semiconductor),
if the incident light with energy matching or exceeding the bandgap energy(Eg) of the
20
semiconductor, creating electron-hole pairs, the photo-generated electrons enter the
conduction band, leaving the holes in the valence band. These generated electron-hole pairs
in the semiconductor can transfer to the reactants in an electrolyte solution, and cause the
photoassisted decomposition or synthesis of chemical compounds. More generally thus can
drive redox reactions.
-8
-7
-6
-5
-4
-3
InP
H
2
O/H
2
0V
H
2
O/O
2
0.82 V
CO
2
/CO
2
-
-1.90 V
1.7 eV
anatase rutile
2.8 eV
3.2 eV
1.34 eV
2.25 eV
1.1 eV
2.8 eV
2.3 eV
2.4 eV
1.6 eV
3.2 eV
TiO
2
TiO
2
CuO
Cu
2
O
Fe
2
O
3
WO
3
Si
GaP
CdSe
SrTiO
3
PbO
3.0 eV
Vacuum (eV)
NHE (Volts)
CO
2
/HAc
-0.61 V
CO
2
/CH
3
OH -0.38 V
CO
2
/CH
4
-0.244 V
3
2
1
0
-1
Figure 1.1. Energy band diagram of various metal oxide semiconductors, together with the
redox potentials of some chemical reaction.
In figure 1.1, we list the band gap and band edge position respect to the potentials
versus NHE for some semiconductors. This data is derived from flat band potential
determinations via capacity measurements. The standard potentials for these redox couples
indicate the thermodynamic limitations for the photoreactions that can be carried out with
the charge carriers (at pH=7). For example, if a reduction of the species in the electrolyte
is to be performed, the conduction band position of the semiconductor must be positioned
21
above the relevant redox level.
It has been suggested that at semiconductor electrodes, the charge transfer rates
between photogenerated carriers in semiconductors and the solution species depend on the
correlation of energy levels between the semiconductor and the redox agents in the solution.
That is, the solution species of the redox potential is more positive than the conduction
band energy is better reduced at semiconductors.
If the conduction band position of the semiconductor is not positioned above the
relevant redox level, the photo-excited electrons do not have the ability to drive the
reduction half-reactions. In this situation, an external bias voltage must be applied.
Normally referred to as applied an overpotential.
1.2 General Introduction of CO2
The whole world faces two major energy related problems in the next 50 years. Firstly,
the increasing worldwide competition for the gradually depleting fossil fuels. Secondly,
atmospheric CO2 level is at the highest level since records began, which is the main source
of the green-house effect. As it is known to all, carbon dioxide is the basic source of all the
fossil fuels used in our daily lives. The process that drives the carbon dioxide into these
fuels is photosynthesis, combined the sunlight, water and carbon dioxide into the organic
materials
5
. Research in the field of photochemical and photoelectrochemical reduction of
CO2 has grown rapidly in the last few decades. This growing research effort is a response
by physical scientists and engineers to the increasing amount of CO2 in the atmosphere and
22
the steady growth in global fuel demand. The catalytic conversion of CO2 to liquid fuels is
a critical goal that would positively impact the global carbon balance by recycling CO2 into
usable fuels. The challenges presented here are great, but the potential rewards are
enormous. CO2 is an extremely stable molecule generally produced by fossil fuel
combustion and respiration. Returning CO2 to a useful state by activation/reduction is a
scientifically challenging problem, requiring appropriate catalysts and energy input. This
poses several fundamental challenges in chemical catalysis, electrochemistry,
photochemistry, and semiconductor physics and engineering
6-14
.
One method to achieve artificial photosynthesis employs photoactive semiconductors
in a photoelectrochemical cell, which has four components: a cathode, an anode, a
reference electrode and an electrolyte, like the figure 1.2 shows. Carbon dioxide gets
reduced at the cathode to form other organic products or carbon monoxide. In such a system,
the anode needs to be a photoactive semiconductor that absorbs light to produce electrons
and holes, which also generates the photovoltage to reduce the CO2.
23
Figure 1.2. Schematic diagram of photoelectrochemical system.
1.3 The Difficulty of CO2 Reduction
Carbon dioxide is a linear molecule, it has a C-O bond distance of 1.16 Å. CO2 is a
nonpolar molecule as a whole, however, it contains polar bonds between C atom and O
atom, because of the electronegativity, C atom shows partial positive and O atom shows
partial negative, indicating its possibilities to nucleophilic attack at carbon and electrophilic
attack at oxygen
15,16
. With a carbon-localized LUMO, carbon dioxide is possible to attack
by nucleophiles and to reduce. The first step to reduce the carbon dioxide is “activation of
the CO2”, normally, CO2 get one electron to be a CO2
-
(as eq.(5) shows) is such kind of
activation. Few semiconductors have its conduction band edges lie above this potential.
Thus an applied potential is usually needed to enable a CO2 reduction reaction occurring.
Where the standard redox potentials are given in V olts versus NHE at pH 7.0, and (g) and
(aq) denote the gaseous state and aqueous solution, respectively. The thermodynamic
potentials for various CO2 reduction products can be seen in Equations 1–5 (pH 7 in
aqueous solution versus a normal hydrogen electrode (NHE), 25
◦
C
CO2+ 2H
+
+ 2e
−
→ CO + H2O E
o
= −0.53V (1)
CO2+ 2H
+
+ 2e
−
→ HCO2H E
o
= −0.61V (2)
CO2+ 6H
+
+ 6e
−
→ CH3OH + H2O E
o
= −0.38V (3)
CO2+ 8H
+
+ 8e
−
→ CH4+ 2H2O E
o
= −0.24V (4)
CO2+ e
−
→ CO2
•−
E
o
= −1.90V (5)
24
Figure 1.3. Energy band diagram of various metal oxide semiconductors, together with the
redox potentials of CO2 reduction.
Although CO2 has been shown to be reduced directly on metal surfaces, the
overpotentials are either exceedingly high or the metal surfaces become poisoned and
deactivated by the reduction products. In addition to thermodynamic considerations, there
are also considerable kinetic challenges to the conversion of CO2 to more complex products.
Typically, multiple proton coupled electron transfer (PCET) steps must be orchestrated
with their own associated activation energies presenting kinetic barriers to the forward
reaction. A great deal of success has been achieved in the reduction of CO 2 to CO and
formate. However, the multiple electron and proton transfers necessary to produce more
useful products such as methane or methanol have only been demonstrated with low
efficiency. To achieve success at efficient production of a CO2 reduction product that can
serve as a liquid fuel directly (i.e., methanol) would be a considerable milestone for
renewable energy and energy storage research.
-8
-7
-6
-5
-4
-3
InP
H
2
O/O
2
0.82 V
CO
2
/CO
2
-
-1.90 V
1.7 eV
anatase rutile
2.8 eV
3.2 eV
1.34 eV
2.25 eV
1.1 eV
2.8 eV
2.3 eV
2.4 eV
1.6 eV
3.2 eV
TiO
2
TiO
2
CuO
Cu
2
O
Fe
2
O
3
WO
3
Si
GaP
CdSe
SrTiO
3
PbO
3.0 eV
Vacuum (eV)
NHE (Volts)
CO
2
/HAc
-0.61 V
CO
2
/CH
3
OH -0.38 V
CO
2
/CH
4
-0.244 V
3
2
1
0
-1
25
Another difficult associated with carbon dioxide reduction in aqueous condition (i.e.
water solution) is that the standard reduction potential of H2O to form H2 is considerably
lower (E
o
= 0V) than the standard reduction potential of CO2 to CO2
•−
form E
o
= −1.90V (as
Figure 1.3 shows). Indicating it is a competitive reaction in aqueous condition, how to
improve the CO2 reduction products selectivity other than water splitting is very
important.
17,18
1.4 Photochemical and Photoelectrochemical Reduction of CO2
There are several ways to reduce CO2 with the assistance of renewable solar energy,
and these methods can be divided into three major categories: homogeneous photo
reduction by a molecular catalyst, photoelectrochemical reduction by a semiconducting
photocathode, and electrochemical reduction by an electrolyzer powered by commercial
photovoltaic (PV) devices. A homogeneous CO2 photo reduction system consists of a
molecular catalyst, light absorber, sacrificial electron donor, and/or electron relay. When
looking at these types of systems, the main figure of merit is the photochemical quantum
yield, defined as
Photochemical quantum yield (Φ%)=(moles products/absorbed photons) (number of
electrons needed for conversion).
In a heterogeneous system, p-type semiconductor/liquid junctions are extensively
studied as PV devices. The p-type semiconducting electrodes can act as photocathodes for
photoassisted CO2 reduction. Figure 1.4 shows four different schemes of photoassisted
26
reduction of CO2 using a semiconducting photocathode: (a) direct heterogeneous CO2
reduction by a biased semiconductor photocathode
12,19-21
(b) heterogeneous CO2 reduction
by metal particles on a biased semiconductor photocathode
22-24
, (c) homogeneous CO2
reduction by a molecular catalyst through a semiconductor/molecular catalyst junction
25,26
,
and (d) heterogeneous CO2 reduction by a molecular catalyst attached to the semiconductor
photocathode surface
26
.
There are several examples where PV-powered commercial electrolyzers have been
used for hydrogen generation
27
, but very few of these setups exist for CO2 reduction to an
energy dense product. Recently, Delacourt et al
28
reported a PV-powered electrolyzer that
forms syngas (CO and H2) from CO2 and water.
For these different systems, the expressions for solar to chemical energy conversion
efficiency are complicated because multiple products can form at the cathode and anode.
In most cases, CO2 photoelectrochemical reduction on photocathodes happens at high
overpotentials, which further complicates this calculation.
27
Figure 1.4. Schematic diagrams of four different schemes for light-assisted CO2 reduction
on a semiconducting photocathode: (a) heterogeneous catalysis on a semiconductor
electrode, (b) heterogeneous catalysis on a metal-decorated semiconductor electrode, (c)
homogeneous catalysis through a semiconductor/molecular catalyst junction, and (d)
heterogeneous catalysis through a molecular catalyst–decorated semiconductor electrode.
Despite the promise of such technology, there is only one known selective
electrocatalyst for CO2 to methanol, which is the surprisingly simple pyridine molecule.
Bocarsly and co-workers have shown that pyridine, at pH=5, will selectively reduce CO2
to methanol in a series of one-electron steps and at low overpotentials. Other deep reduction
products such as formaldehyde or higher alcohols are only produced in trace quantity
25
.
The overall proposed mechanism for the reduction of CO2 to the various products of
formic acid, formaldehyde, and methanol is represented in Scheme 1
26
. The light gray area
28
represents alternate possible routes to formic acid and formaldehyde that could potentially
occur but could not be unambiguously identified from the available data at a Pt interface.
The main mechanistic route correlates to the observed data presented in the previous
sections. Importantly, it seems that the formation of either formic acid or formaldehyde
(which is further reduced to methanol) goes through a key intermediate, the hydroxyformyl
radical. The direct reduction of this species yields either formic acid or the formyl radical,
as seen in figure 1.5. The formyl radical is then further reduced to methanol through
reduction by the pyridinium radical. It should be noted that, at Pt, any formic acid generated
can also dissociatively adsorb at the electrode surface, regenerating the hydroxyformyl
radical and an adsorbed hydrogen atom that could then be reduced to the formyl radical.
This would reduce formic acid concentrations and ultimately increase methanol yields.
Figure 1.5. Overall Proposed Mechanism for the Pyridinium-Catalyzed Reduction of CO2
to the Various Products of Formic Acid, Formaldehyde, and Methanol.
29
1.5 Factors Limiting the Photocatalytic Efficiency
While photocatalysis has a great potential for environmental and solar energy
conversion applications, low photocatalytic efficiencies in the solar spectral range have
made most applications unfeasible. And the major factor limiting the photocatalyst
efficiencies is the inherent mismatch between the absorption spectra of semiconductors and
the solar spectrum.
For example, TiO2, one of the most preferable photo-catalysts for its self-cleaning
property and anti-corrosion property in photovoltaic applications. However, it seldom
absorbs light in the visible region of the electromagnetic spectrum. Figure 1.6 shows this
problem graphically with the absorption spectrum of TiO2 superimposed over the solar
spectrum (AM1.5). Because TiO2’s has such a short wavelength cutoff, few solar photons
(~4%) that can be used to drive this photocatalyst catalyze chemistry reactions. Some
research groups tried to extend the cutoff wavelength of this catalyst by doping
29,30
, and
defecting creation
31
. While these efforts have not done any big improvements in the
absorption in the visible range, still leaving a majority of the solar spectrum unable to drive
this photocatalyst. Hematite (α-Fe2O3) is another well-known photocatalyst for water
splitting and methane production. With a smaller band gap (2–2.3 eV), Fe2O3 can capture
approximately 40% of the incident sunlight, as illustrated in Figure 1.5. While Fe2O3 is
highly stable over a wide range of pH environments, it typically requires the application of
an external bias to initiate the hydrogen evolution reaction
30-32
.
30
In order to make most use of the incident sunlight as illustrated in Figure 1.7. Small
energy bandgap semiconductor is needed. According to the equation E=1240/ λ(eV), the
energy bandgap should be at least smaller than 3.1eV with which to start to absorb the
incident light in the visible range. And with the bandgap about 1.5eV , the semiconductor
can capture most of the incident sunlight.
From theoretical calculations, considering the overpotentials and other losses, the most
desirable band gaps for photoelectrodes are between 1.1eV and 1.7eV for optimized
efficiency.
33,34
What’s more, the choice of available materials with band gaps ranging
between 1.5 and 2.0eV is quite limited.
35
Figure 1.6 shows the theoretical photocurrent
density under one-sun illumination, where Si (44mA/cm
2
) and III-V semicondutors, such
as GaP (9mA/cm
2
), InP (35mA/cm
2
) and GaAs (32mA/cm
2
), are all better than that of TiO2
(1~2mA/cm
2
). However, most semiconductors in such range between 1.5 and 2.0eV
corrode easily when immersed in an aqueous photoelectrochemical cell.
36,37
Such corrosion
is not expected in practical use for photocatalytic reaction.
Small bandgap semiconductor can make use of the incident sunlight, but it also meets
another problem of drive the reaction, in the other words, the short bandgap energy maybe
not sufficient for the photocatalytic CO2 reduction reaction and water splitting. As
discussed in above, in order to have the generated electrons and holes energetically
favorable transfer from the semiconductor to the reagents, the conduction band should lie
above the reduction potential and the valence band should lie below the oxidation potential.
31
But most of the small bandgap materials has the band position mismatching the redox
potential. This mismatch typically requires the application of an external potential to
initiate the photocatalytic chemical reaction.
Figure 1.6. Solar spectrum and absorption spectrum of TiO2 and α-Fe2O3
32
Figure 1.7 Theoretical photocurrent densities for semiconductors under one-sun
illumination. The blue curve is the integrated photon flux at different cutoff energies, based
on the AM1.5G solar spectrum (ASTM G173-03 reference). The theoretical photocurrent
is calculated assuming that all incident photons above the semiconductor band gap
contribute to the photocurrent. The arrows at the bottom indicate the regions of ultraviolet
(UV) (below 400nm), visible (400−750 nm), and infrared (IR) spectra
1.6 Strategies to Enhance Photoelectrochemical Performance
The primary of focus of researchers in this field is to engineer materials for more
efficient solar energy to convert into chemical fuels. And some strategies has been raised
33
to enhance the performance include
(1) To improve the charge-carrier transfer and collection efficiency by nano structure
the semiconductors
38-41
;
(2) To lower the reaction overpotential by attaching the cayatalysts to the surface to
the anode
42,43
;
(3) To improve charge separation by use of a surface layer to create a buried p–n
junction to achieve a higher voltage
44
;
(4) To design surface passivation layers that chemically or physically protect the
semiconductor from corrosion
45
;
(5) To reduce the rate of electron–hole recombination by surface state passivation
46
.
1.6.1 Advantages of III-V Semiconductor Photocatalysis
III-V compound semiconductors, such as GaP, GaAs, and InP, have smaller band gaps
than TiO2, and can therefore make more efficient use of the solar spectrum like Figure 1.8
and Figure 1.9 show. Gallium phosphide is a photocatalytic semiconductor with an indirect
band gap of 2.25eV . While it absorbs a more extended range of the solar spectrum than
TiO2, GaP’s indirect band gap results in a particularly low absorption coefficient, which is
disadvantageous for photocatalysis. Its main advantage as a photocatalyst is its relatively
high conduction band energy, which exceeds that of TiO2 by 0.5eV , as illustrated in Figure
1.8. In addition, GaP is more expensive than TiO2 and is not oxidatively stable over a wide
range of pH. As a result, a vast majority of photocatalytic studies have been carried out on
34
TiO2. In the proposed work, we present a strategy for mitigating the degradation of GaP
through surface passivation. While neither GaP nor TiO2 possesses all the properties of an
ideal photocatalyst, this proposed study is aimed at developing an understanding of the
photocatalytic mechanisms that occur in these systems. InP, a direct band gap
semiconductor, is an excellent candidate for solar photocatalysis. The record high solar-to-
hydrogen efficiency of 13.3% was demonstrated in 1982 using a system that employed p-
type InP photocathodes covered with Pt catalysts.
47
More recently, Ali Javey’s group
reported 14% efficient water splitting using InP nanopillars with a Ru catalyst.
48
While the conduction band of GaP lies well above the redox potential of the CO2
reduction half-reactions to form CH4 and CH3OH (as shown in Figure 1.3), methanol
formation is a process involving 6 electrons and methane requires 8 electrons. The first
step in these multi-electron processes is likely the reduction of CO2 to form CO2
-
(Eq. 5
listed above), which is an uphill reaction with a reduction potential of -1.9V vs. NHE. As
such, photocatalytic reduction of CO2 on GaP (conduction band = -0.7V vs. NHE) typically
requires a significant overpotential to induce the charge transfer mechanism.
49,50
Although
CO2 has been shown to be reduced directly on metal surfaces, the overpotentials are either
exceedingly high or the metal surfaces become poisoned and deactivated by the reduction
products.
51
In addition to thermodynamic considerations, there are also considerable
kinetic challenges to the conversion of CO2 to more complex products. It is important to
note, however, that this reduction potential of -1.9eV vs. NHE is, in fact, calculated from
35
simple thermodynamic considerations for an isolated CO2
-
species and does not include the
effects of the solution or catalytic surface. As a result, the energetics of the actual CO2
-
species can be quite different due to the presence of the aqueous solution and/or the
catalytic surface. In a mechanism proposed by Anpo et al., the CO2
-
intermediate is strongly
bound to a proposed Ti
3+
active site (oxygen vacancy) in the TiO2 surface, thus lowering
its energy
52
. Another strategy for lowering the reaction barrier is stabilizing the (CO2)
-
intermediate, which was recently demonstrated using an ionic liquid electrolyte co-catalyst
where the cation forms a complex with the anionic intermediate.
53
Two-electron processes
have also been proposed by Tananka et al., which would circumvent this first intermediate
step altogether
54
. The existing controversy in the literature perhaps illustrates both the
importance of the problem and the need for a rigorous in situ mechanistic study to
understand the chemical mechanisms and adsorption-desorption of the reactants, products,
and intermediates involved. While there are multiple mechanisms and intermediates
proposed in the literature, direct spectroscopic studies are few.
36
Figure 1.8. Solar spectrum and absorption spectrum of TiO2 and III-V semiconductors.
Figure 1.9. Shockley-Quesisser Limit of TiO2 and III-V semiconductors
GaAs, InP
37
1.6.2 Advantages of Thin Layer TiO2 Passivation
Passivation layers like TiO2 were originally applied to semiconductor electrodes to
minimize the corrode and partially improve the chmical or photoelectrochemical stability
when immersed into the electrolyte. What’s more, such surface layers can also be used to
prevent formation of band gap states that promote the electron-hole pairs recombination,
which help the CO2 reduction at the semiconductor-electrolyte interface, also it contributes
the band position shift. In our previous publications, we facricate the TiO2 layer by atomic
layer depostion (ALD), such very thin layer of TiO2 (less than 10nm), can prevent the
parastic light absorption or charge-transfer inibition, also such thin layer of TiO2 on III-V
semiconductors increases the overall performance of the CO2 reduction, including both the
Faradaic efficiency and photovoltage.
55-57
38
Chapter 2: Photocatalytic Conversion of CO2 to Hydrocarbon Fuels
on GaP
2.1 Introduction
The photoelectrochemical reduction of CO2 is an exciting reaction system with the
ability to convert an abundant greenhouse gas to combustible hydrocarbon fuels using
sunlight. The direct conversion of solar-to-chemical energy has several advantages over
solar-to-electric energy conversion, most notably, the ability to store large amounts of
energy (~GW) in chemical bonds that can later be released in a carbon neutral cycle.
6-
11,14,19,58-60
Many attempts have been made to reduce CO2 by 2e
-
to various species such as
CO and formic acid, as reported in previous literature.
13,61-67
Few researchers have achieved
further reduction to CH3OH or CH4.
25,68,69
Methanol is an attractive product with a
relatively high energy density, which can be easily integrated into the existing liquid fuel
technologies.
14,70
However, the photocatalytic reduction of CO2 with H2O to methanol
requires 6 electrons and many intermediate species, some of which have extremely high
energy barriers.
71
The most likely first step in this multi-electron reaction is the one electron
reduction to the CO2
-
intermediate,
72
which lies 1.7eV above the conduction band of TiO2
and 1.2eV above GaP. The mechanism for electrochemical CO2 reduction was first
proposed by Bockris et al.
73-75
The high overpotential required for this reaction was
attributed to the formation of the CO2
-
intermediate, which consequently converts to CO
39
via the general process CO2 + e
-
CO2
-
, CO2
-
+2H
+
+ e
-
CO + H2O.
8,25,58,76
In 1978,
Hallman’s group first reported CO2 reduction on p-GaP under 365nm illumination with an
applied overpotential of -1.4V (vs SCE).
61
Fujishima and Honda demonstrated
photoelectrocatalytic reduction of CO2 to formaldehyde and methanol by irradiating TiO2
and GaP with the UV light at an overpotential of -1.5V (vs SCE).
13,77
Canfield later reported
CO2 reduction to methanol on p-InP with an overpotential of -1.3V (vs SCE).
67
More
recently, Bocarsly’s group demonstrated pyridinium-catalyzed CO2 reduction on GaP
photocathodes with overpotentials between -0.7V and -0.2V (vs SCE) under UV light.
25
Despite these interesting prior results, the stability of these materials against
photocorrosion has not been addressed.
In the work presented here, we investigate the photocatalytic performance and stability
of TiO2-passivated p-GaP using atomic force microscopy (AFM), photoelectrochemistry,
and optical microscopy. The photocatalytic efficiency is studied systematically as a
function of TiO2 layer thickness using a three-terminal potentiostat. The products are
detected using NMR spectroscopy and gas chromatography, systematically as a function
of applied overpotential.
2.2 Experimental Procedure
Zn doped p-type (100) oriented GaP with a dopant concentration of 2x10
18
cm
-3
was
used as the photocatalyst for CO2 reduction with an active area of 0.5cm×1cm. Atomic
layer deposition (ALD) of anatase TiO2 was performed at 250ºC on the p-GaP wafers with
40
TiCl4 as the titanium source and water vapor as the oxygen source. The antase crystal phase
was verified by Raman spectroscopy. Using ellipsometry, we established that 100 cycles
of ALD produces a 4nm thick TiO2 film and 1000 cycles produces a 40nm film. A Ga-In
eutectic film was painted on the back of the p-GaP to form an Ohmic contact. The Ga-In
contact was then connected to the external circuitry with a copper wire and coated with
epoxy cement to insulate it from the electrolytic solution, as illustrated in Figure 2.1. While
this planar geometry is not ideal for high efficiency photoconversion, it enables us to study
surface stability. A three-terminal potentiostat was used with the prepared semiconductor
samples as the working electrode, a Ag/AgCl electrode as the reference electrode, and a Pt
electrode functioning as a counter electrode. The photocatalytic reaction rates of two sets
of samples were measured in a 2ml solution of 0.5M NaCl, with and without 10mM
pyridine, while continuously bubbling CO2 through the solution. 0.5M NaCl was chosen
as the electrolyte solution because of its high conductance, and its ability to stabilize the
intermediate states involved in the CO2 reduction.
78
In this setup, we analyze the products
evolving at the working electrode, instead of at the counter electrode. It is likely that
oxygen is also produced in the reaction.
41
Figure 2.1 Schematic diagram of sample geometry.
2.3 Results and Discussion
Figure 2.2 (a) Optical microscope image, (b) atomic force microscope image, and (c)
surface topography of bare GaP surface after 8h reaction at -0.5V overpotential. (d) Optical
microscope image, (e) atomic force microscope image, and (f) surface topography of 5nm
TiO2 on GaP surface after 8h reaction.
While photocatalysis on GaP (and other III-V compound semiconductors) has been
demonstrated previously,
25,67,79
this material corrodes rapidly under photo-electrochemical
conditions and is significantly degraded after just 30 minutes of illumination. In order to
(b)
(d)
(c)
(e)
0 1 2 3 4 5
-60
-40
-20
0
20
40
60
Height (nm)
Scan Range ( m)
(a)
0 1 2 3 4 5
-10
-5
0
5
10
Height (nm)
Scan Range ( m)
(f)
10µm
10µm
1µm
±1nm
1µm
±54nm
42
make GaP photochemically stable, we passivated the surface using a thin film of TiO2
deposited by ALD. Figures 2.2a and 2.2b show optical microscope and atomic force
microscope images of the bare GaP surface after 8 hours of illumination. Figure 2.2c
shows a plot of the surface topography obtained along the dashed white line in Figure 2.2b,
showing an RMS roughness of ±54nm, which indicates that substantial photocorrosion has
taken place and that this will not serve as a viable photocatalyst. In contrast, the
photocurrent density of TiO2-passivated GaP is stable for 8 hours. The optical microscope
image (Figure 2.2d) and atomic force microscope image (Figure 2.2e) exhibit no evidence
of surface corrosion or damage after 8 hours, with an RMS roughness of ±1nm (Figure
2.2f), indicating that this is a long term, stable photocatalyst. Here, the TiO2 significantly
improves the photo-stablilty of the GaP surface, however, more extensive time-dependent
studies are needed in order to establish the extent of this long term stability.
Table 2.1 Shift of onset overpotential for samples with different thicknesses of TiO2.
Catalysts Shift of onset overpotential
compared to Bare GaP (V)
Bare GaP 0
1nm TiO
2
@GaP 0.13
3nm TiO
2
@GaP 0.31
5nm TiO
2
@GaP 0.40
10nm TiO
2
@GaP 0.52
43
Figure 2.3 (a) Photocatalytic current-potential curves of GaP photocatalysts with different
TiO2 thicknesses in a 0.5M NaCl, 10mM pyridine solution under 532nm wavelength laser
illumination. (b) Decrease of overpotential plotted as a function of TiO2 thickness on GaP.
(c) Calculated built-in voltage plotted as a function of TiO2 thickness. (d) NMR spectra
showing methanol production using bare GaP and 5nm TiO2-passiavated GaP
photocatalysts at an overpotential of -0.50V.
In addition to providing a stable photocatalytic surface, the TiO2 passivation layer
results in an increase in the photoconversion efficiency. Figure 2.3a shows the
photocurrent-voltage curves for GaP passivated with various thicknesses of TiO2 measured
in a 0.5M NaCl, 10mM pyridine solution under 532nm illumination. During these
measurements, CO2 is continuously bubbled through the solution. Bare GaP (green curve)
has an onset of photocurrent at a potential of approximately -0.15V (vs NHE). For TiO2-
(d) (c)
3.5 3.4 3.3 3.2 3.1 3.0
Bare GaP
5nm TiO
2
passivated GaP
ppm
Methanol
-1.0 -0.5 0.0
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
Dark
Bare GaP
GaP w/1nm TiO
2
GaP w/3nm TiO
2
GaP w/5nm TiO
2
GaP w/10nm TiO
2
Potential vs. NHE (V)
Current Density (mA/cm
2
)
(a) (b)
0 1 2 3 4 5 6 7 8 9 10
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Thickness of TiO
2
(nm)
Decrease of Overpotential (V)
0.5V
1 2 3 4 5 6 7 8 9 10
0.0
0.1
0.2
0.3
0.4
0.5
Built-in Voltage (eV)
TiO
2
Thickness (nm)
44
passivated GaP, we see a clear shift in the overpotential required to drive this reaction with
increasing thickness of the TiO2, as plotted in Figure 2.3b. Table 2.1 lists the shift of onset
overpotential of samples with different thicknesses of TiO2. For example, the onset
potential for 10nm TiO2 (red curve) is shifted by 0.5V with the respect to bare GaP. This
shift is attributed to the passivation of surface states that cause non-radiative recombination
and the formation of a pn-junction, which is created because the ALD-deposited TiO2 tends
to be n-type doped due to oxygen vacancies
80,81
. Figure 2.3c shows the built-in potential
for the junction calculated using 𝑉 𝑏𝑖
=
𝑊 𝐷 2
𝑞 2𝜖 0
𝜖 𝑎 𝜖 𝑑 𝑁 𝑎 𝑁 𝑑 (𝑁 𝑎 𝜖 𝑎 +𝑁 𝑑 𝜖 𝑑 )
(𝑁 𝑎 +𝑁 𝑑 )
2
V
bi
=
W
D
2
q
2ϵ
0
ϵ
a
ϵ
d
N
a
N
d
(N
a
ϵ
a
+N
d
ϵ
d
)
(N
a
+N
d
)
2
, assuming a doping concentration of Na=5x10
18
cm
3
.
82
Here, 𝑊 𝐷 is
the depletion width of the GaP-TiO2 junction, which is a function of the TiO2 layer
thickness. This simple calculation predicts values similar to the experimentally observed
shift in the overpotential plotted in Figure 2.3b. Beyond 10nm, however, the photocurrent
decreases rapidly with increasing TiO2 thickness due to band bending at the n-type
TiO2/electrolyte interface, which blocks electrons. No enhancement is observed for TiO2
thicknesses above 10nm, due to the insulating nature of the TiO2, which eventually
outweighs the benefits of passivation. While, TiO2 does not absorb light at 532nm, the pn-
junction formed with the GaP enables separation of the photo-generated charge in the
actively absorbing GaP. Figure 2.3d shows the NMR spectra taken after 8 hours of
illumination with an overpotential of -0.50V vs NHE for GaP with and without TiO2
passivation. This data shows a clear peak corresponding to methanol, as reported
45
previously by Barton et al.
25
Gas chromatography FID data has also been used to verify the
production of methanol, as shown in Figure 2.4. Based on this GC FID data for the 5nm
thick TiO2 sample, we calculated that 4.9 (±0.02) µmol of CH3OH are produced during an
8 hour reaction consuming 5.2 Coulombs of charge. Dividing by this ratio by the
stoichiometric factor of 6, yields a Faradaic efficiency of 55%. Also, according to the GC
TCD data of the same experiment, H2 is produced with a Faradaic efficiency of 30%. The
photo-conversion efficiency, however, can be significantly less than this due to non-
radiative recombination, which is unknown, particularly for this planar sample geometry.
As a control experiment, the same reaction was run under the same electrochemical
conditions of -0.50V vs NHE without laser illumination, which resulted in no measureable
current and no detectable methanol in the NMR spectra. In order to rule out other sources
of carbon in this reaction, we used isotopically labeled
13
CO2 as the carbon source in this
reaction and observed
13
CH3OH in the
13
C-NMR spectrum shown in Figure 2.5
83
. In
addition, we repeated the experiment, purging with Ar instead of CO2, and found no
production of hydrocarbons. Therefore, we are confident that CO2 is the only carbon source
in this reaction. Previously, it was reported that a pyridine catalyst is required to drive this
reaction on GaP. The pyridinium radical serves as a one-electron charge-transfer mediator,
which is capable of efficiently transferring all six electrons to reduce CO2 to methanol,
thereby circumventing the high energy barrier of the one-electron reduction of CO2
mentioned above.
79
However, we observe the same methanol peak in our NMR spectra
46
without pyridine in solution (see Figure 2.6), indicating that this catalyst is not, in fact,
required to drive this reaction at low overpotentials. While CH3OH products are observed
without pyridine, the yield is one third that of the system with pyridine, indicating that the
pyridine, in fact, helps lower the energy barriers of the reaction by forming an inner-sphere-
type electron transfer system.
79,84
. Atomic force microscopy shows that the GaP/TiO2 is
photochemically stable without pyridine, as shown in Figure 2.7.
Figure 2.4 Gas chromatograph (GC) data taken after 8 hour illumination (532nm) of bare
and 5nm TiO2 passivated-GaP in 0.5M NaCl, 10mM pyridine solution at an overpotential
of -0.5V (vs NHE). The GC data is plotted together with a calibration standard consisting
of 10
-4
M methanol in aqueous solution. Based on this data, the 5nm TiO2 passivated-GaP
was found to have a Faradaic efficiency of 55%.
4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8 6.0
Bare-gap
10
-4
methanol
TiO
2
passivated p-GaP
GC Signal (a.u.)
Retention Time (min)
methanol peak
47
Figure 2.5
13
C-NMR spectrum showing methanol peak taken after 8 hour illumination
(532nm) of 5nm TiO2 passivated-GaP at an overpotential of -0.5V (vs NHE) in 0.5M NaCl,
with 10mM pyridine solution.
Figure 2.6
1
H-NMR spectra showing methanol peaks taken after 8 hour illumination
(532nm) of 5nm TiO2 passivated-GaP at an overpotential of -0.5V (vs NHE) in 0.5M NaCl,
with 10mM pyridine solution and without 10mM pyridine.
52 51 50 49 48
ppm
5nm TiO
2
w/ GaP
13
CH
3
OH, 13-C peak
3.5 3.4 3.3 3.2 3.1 3.0
With pyridine
Without pyridine
ppm
Methanol
48
Figure 2.7
(a) Optical microscope image, (b) atomic force microscope image, and (c)
surface topography of TiO2 passivated GaP surface after 8h reaction in 0.5M NaCl without
pyridine at -0.5V overpotential.
In order to understand the mechanism of this reaction, we must consider the energetics
of the electrons in this photocatalytic structure. The conduction bands of GaP and TiO2 lie
slightly above NHE at -0.7V and -0.2V vs NHE, respectively, as plotted in Figure 2.8. This
leaves an energy barrier of at least 1.2V for the electrons to overcome in the reduction of
CO2. The -0.5V externally applied overpotential (Vext) accounts for part of this required
energy, and the photovoltage produced at the internal pn-junction and/or the liquid-
semiconductor junction (VPV) can easily account for the remaining -0.7V . From the flat
band voltage, we can obtain the open circuit voltage, as follows: (Voc)max=|Vfb-Vredox|,
where Vfb is flat-band potential and Vredox is the potential of the redox couple
85
. From Mott-
Schottky measurements, we obtained a flat-band potential of 0.4 V versus NHE, which is
consistent with previous values from literature
25,86
. Using Vredox
(Ferricyanide/Ferroyanide)=0.36V , we obtain an open circuit voltage of Voc=0.76V , which
is large enough to cover the remaining -0.7V depicted in Figure 2.8. This photovoltage is
reasonable considering GaP’s relatively large band gap of 2.25eV .
0 1 2 3 4 5
-10
-5
0
5
10
Height (nm)
Scan Range ( m)
1µm
10µm
±4nm
(a) (b) (c)
49
Figure 2.8 Energy band alignment of GaP and TiO2 together with the relevant redox
potentials of CO2.
These estimations assume that the electrons traverse the TiO2 layer ballistically, and
do not equilibrate to the TiO2 conduction band edge. In the diffusive case, an additional
0.5V would be required. We believe this is one of the reasons why the TiO2 layer must be
made very thin. Several aspects of these results and their underlying mechanism are quite
surprising. First, we had initially thought that the TiO2 layer, which is insulating and has a
low conduction band energy, would lower the overall photocatalytic efficiency (i.e.,
photocurrent), but would, at least, provide a stable, viable catalyst. Much to our surprise,
the TiO2 layer actually improved the overall photo-conversion efficiency. The reasons for
this are three-fold: 1.) The TiO2 reduces non-radiative recombination of the photo-excited
e
-
-1.9V CO
2
-
/CO
2
p-GaP n-TiO
2
-0.7V
0V (NHE)
GaP
Conduction
Band
E
g
=2.25eV
e
+
V
ext
=-0.5V
V
pv
=-0.7V
Electrolyte
50
electron-hole pairs. 2.) The electrons traverse the TiO2 ballistically and, therefore, do not
relax to the conduction band edge. 3.) The formation of a pn-junction provides an
additional photovoltage required to drive the reaction.
It is important to note, however, that the CO2
-
reduction potential of -1.9V vs NHE is
calculated from simple thermodynamic considerations for isolated CO2
-
species, and does
not include the effects of the solution or catalytic surface. As a result, the energetics of the
actual CO2
-
intermediates can be quite different due to the presence of the aqueous solution
and/or the catalytic surface. In a mechanism proposed by Anpo et al., the CO2
-
intermediate
is strongly bound to a proposed Ti
3+
active site (oxygen vacancy) on the TiO2 surface, thus
lowering its energy
52
. Another strategy for lowering the reaction barrier is stabilizing the
CO2
-
intermediate, which was recently demonstrated using an ionic liquid electrolyte co-
catalyst where the cation forms a complex with the anionic intermediate.
53
Two-electron
processes have also been proposed by Tananka et al., which would circumvent this first
intermediate step altogether
54
. While several mechanisms have been proposed in the
literature, further spectroscopic studies are needed in order to verify the catalytic reaction
pathway.
2.4 Conclusion
In conclusion, we report photocatalytic CO2 reduction on TiO2-passivated GaP. The
TiO2 passivation layer successfully stabilizes the GaP surface in solution, preventing it
from photocorrosion. In addition, the TiO2 passivation layer provides enhancement in the
51
photoconversion efficiency through the passivation of surface states and the formation of
a charge separating pn-region, which reduces carrier recombination and lowers the
overpotential required to initiate this reaction by approximately 0.5V . This general
approach of passivating narrower band gap semiconductors with TiO2 will enable more
efficient photocatalysts to be developed and a broad range of materials to be considered for
photocatalysis that make more efficient use of the solar spectrum. We also observe CH3OH
evolution with and without pyridine catalyst, indicating that this catalysts is not, in fact,
required to drive this reaction at low overpotentials.
52
Chapter 3: Artificial Photosynthesis on TiO2-passivated InP
Nanopillars
3.1 Introduction
Here, we utilize TiO2-passivated nanotextured InP photo-cathodes to explore aqueous
CO2 reduction to methanol under 532nm illumination as a function of applied potential.
The selectivity of methanol production is compared for InP samples prepared with and
without Cu nanoparticles and TiO2-passivation layers. The photocatalytic surface is
characterized by high resolution transmission electron microscopy (HRTEM), in order to
provide a detailed picture of the nanoparticle/TiO2/InP interface. Plane wave density
functional theory (PW-DFT) calculations are carried out to explore the role of surface
binding of reactants and intermediate species to the TiO2 surface.
3.2 Experimental procedure
A schematic diagram of the sample geometry is illustrated in Figure 3.1a. Zn doped p-
type (100) oriented InP nanopillars (NPLs) with a dopant concentration of 5x10
17
cm
-3
were
used as the photocatalyst for CO2 reduction. The InP NPLs are around 80nm in diameter
and approximately 400 to 600 nm tall, and have an average pitch of approximately 50 nm
(Figure 3.1b). Briefly, InP bulk wafers are treated in a reactive ion O2 plasma treatment
followed by a two-minute wet-etching step in HCl/H3PO4 (3:1) to remove the surface-
53
damaged layers and contaminants.
87
Atomic layer deposition (ALD) of TiO2 was
performed at 250
o
C on the p-InP wafers with TiCl4 as the titanium source and water vapor
as the oxygen source. The average rate of deposition is approximately 0.44Å per cycle, as
calibrated by ellipsometry. We then evaporate copper with a nominal thickness of 0.5nm
on the top surface of the TiO2. A Zn-Au film was evaporated on the back of the p-InP to
form an Ohmic contact. The Zn-Au contact was then connected to the external circuitry
with a copper wire and coated with epoxy cement to insulate it from the electrolytic
solution. Figures 3.1c–3.1e show transmission electron microscope (TEM) images of the
0.5nm Cu on the top of the InP NPLs. These figures indicate that, instead of forming a
uniform continuous film, the evaporated 0.5nm Cu forms crystal nanoparticles with
diameters around 20nm. In Figure 3.1e, a thin amorphous layer of CuO can be seen on the
surface of the Cu nanoparticles, which is formed in air. The XPS spectra shown in Figure
3.2. confirms the existence of Cu 2p and O 1s peaks corresponding to CuO. The thickness
of the deposited TiO2 is around 3nm, as shown in Figure 3.1d.
54
Figure 3.1 (a) Schematic diagram of TiO2-passivated InP nanopillars with Cu cocatalyst
nanoparticles. (b-e) SEM and TEM images of InP nanopillar array with TiO2 deposition
layer and Cu nanoparticles. The high resolution TEM image in (e) resolves the crystal
lattice of the Cu nanoparticles.
500nm
100nm
InP
Cu
Nanoparticles
20nm
Cu
TiO
2
InP
10nm
Cu
(c) (b)
(e) (d)
(a)
InP
TiO
2
Cu
55
Figure 3.2 (a) Cu 2p for CuO and Cu, (b) O 1s core level XPS spectra of air-oxide Cu
nanoparticles.
A three-terminal potentiostat was used with the prepared semiconductor samples as
the working electrode, a Ag/AgCl electrode as the reference electrode, and a Pt electrode
functioning as the counter electrode. The photocatalytic reaction rates of two sets of
samples were measured in a 0.5 M KCl solution, while continuously bubbling CO2 through
the solution. The products are detected using NMR spectroscopy and gas chromatography
(GC). While photocatalytic CO2 reduction on InP and other III-V compound
semiconductors has been demonstrated previously,
25,88
these materials corrode rapidly
under photo-electrochemical conditions and are significantly degraded after just 30
minutes of illumination.
81,89
Several research groups have shown that by depositing thin
films of TiO2 on these unstable semiconductors, they can be protected from
corrosion.
81,87,89-91
In order to make InP photochemically stable, we passivated the surface
using a 3nm thin film of TiO2 deposited by ALD, as illustrated schematically in Figure 3.1a.
536 534 532 530 528 526 524
Intensity (a.u.)
O
Binding Energy (eV)
980 970 960 950 940 930
Intensity (a.u.)
Experimental Data
Fitting Data
Cu Peak
CuO Peak
Binding Energy (eV)
56
Under 532nm illumination, the photocurrent of InP nanopillars with TiO2-passivation is
stable for at least 12 hours, as shown in Figure 3.3.
Figure 3.3 Time dependence of the photocurrent density of InP with 3nm TiO2 and Cu
nanoparticles illuminated with 532nm light at an applied overpotential of -0.6V vs. NHE.
3.3 Results and Discussion
Figure 3.4a shows the photocurrent-voltage curves measured in a 0.5M KCl solution
under 532nm illumination for bare InP nanopillars and InP nanopillars passivated with TiO2.
With TiO2-passivation, the photocurrent is substantially increased for all applied potentials.
The bare InP nanopillar sample (blue curve) has an onset of photocurrent at a potential of
approximately 0.03V (vs NHE), as indicated in Figure 3.4b. For TiO2-passivated InP (red
curve), we observe a clear shift in the onset potential by about 0.1V in these photo-I-V
characteristics. We attribute this, in part, to a pn-junction formed between the TiO2, which
is n-type due to oxygen vacancies, and the p-type InP.
80,81
This pn-junction creates a built-
in potential that assists in the separation of photogenerated electron-hole pairs and results
0 2 4 6 8 10 12
0
1
2
3
4
5
6
7
8
Time (hours)
Current Density (mA/cm
2
)
57
in a shift of the onset potential of this reaction. Figure 3.4c shows the methanol peaks
observed in the NMR spectra of bare InP nanopillars and TiO2-passivated InP nanopillars
after 12-hour reactions at an applied potential of -0.6V vs. NHE under 532nm illumination.
If we consider the one electron reduction of CO2 to CO2
-
as the first step on this reaction,
this is 1.3V below the E
o
(CO2/CO2
-
)=-1.9eV standard redox potential. Figure 3.4d shows
the Faraday efficiencies of methanol (i.e., the selectivity of methanol) for these two types
of samples. This figure indicates that the TiO2-passivation layer not only enhances the
overall photon-conversion efficiency, but also increases the selectivity of methanol from
H2 and other hydrocarbons from 0.85% to 4.8%. We attribute the increased methanol
selectivity to oxygen vacancies that are inherent to these thin TiO2 films, which provide
the catalytically active sites for CO2 reduction.
92-94
Peaks corresponding to Ti
3+
states (i.e.,
O-vacancies) are observed in the XPS spectra taken from TiO2 deposited on InP, as shown
in Figure 3.5.
58
Figure 3.4 (a) Photocatalytic current-potential curves where the dashed line indicates the
potential applied during the methanol test. (b) Log plot of photocatalytic current-potential
curves. (c) Methanol peak in NMR spectra and (d) Faraday efficiencies of methanol
production for InP nanopillars with and without 3nm TiO2 under 532nm illumination in a
CO2 saturated 0.5M KCl solution.
Figure 3.5 Ti 2p level XPS spectra of TiO2 on InP, which shows the presence of Ti
3+
states.
-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15
0.1
1
Bare InP
InP w/TiO
2
Potential vs. NHE (V)
Absolute Current Density (mA/cm
2
)
(b)
0
1
2
3
4
5
6
4.79%
InP with TiO
2
Faraday Efficency of Methanol (%)
Bare InP
0.846%
(d)
3.50 3.45 3.40 3.35 3.30
Bare InP
InP w/TiO
2
ppm
Methanol
(c)
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4
-30
-25
-20
-15
-10
-5
0
Dark
Bare InP
InP w/TiO
2
Potential vs. NHE(V)
Current Density (mA/cm
2
)
(a)
468 466 464 462 460 458 456 454
Intensity (a.u.)
Binding Energy (eV)
Ti
3+
Experimatal data
Fitting data
59
In order to verify the role of these oxygen vacancies in the photocatalytic reduction of
CO2, all plane wave density functional theory (PW-DFT) calculations were performed with
the Quantum Espresso package using the most recently available ultrasoft pseudopotentials
with scalar relativistic corrections,
95-98
and spin-unrestricted calculations were done
employing the Perdew-Burke-Ernzerhof (PBE) functional.
99
Clean, stoichiometric anatase
and defective anatase with an oxygen vacancy were both investigated in this study.
In all plane wave density functional theory (PW-DFT) calculations, large kinetic
energy cutoffs of 435.2 eV and 4.352 eV were applied to the wave functions and charge
density, respectively. A 1x1x1 Monkhorst-Pack k-point grid was used for all calculations.
The anatase slab was modeled as a 3x3 cell with lattice constants of a = 3.7845 Å, c =
9.5143 Å from experimental crystallographic data.
100
The appropriate cuts were made to
construct the most stable and dominant facet (>94%) of the anatase crystal, the 101
surface,
101
as a 16 TiO2 cell, the optimal size for molecular deposition of small molecules
such as water.
102
A vacuum gap of about 9 Å (a factor of 2.5 greater than the height of the
cell) was used. A vacuum gap of about 11 Å (a factor of 2.5 greater than the height of the
cell) was used. The effect of nonlocal forces such as dispersion forces was accounted for
using Grimme’s method,
4
which correctly predicted the thermodynamic stability of the
phases of TiO2 (rutile > brookite > anatase).
5
Differences in adsorption energies and Bader
charges between our calculations and Sorescu, et al.’s study may be accounted for by our
different methods (their dispersion method was developed by Tkatchenko and Scheffler),
60
cell sizes (doubled in the x and y-axes), and U (their U = 3.5 eV, our U = 3.6 eV). We
attributed our higher energies to interaction across the periodic cells.
6
Figure 3.6 PW-DFT calculated structure for anatase TiO2 with O vacancies (a) before CO2
adsorption and (b) after CO2 adsorption and relaxation.
Figure 3.7 PW-DFT result of neutral CO2 adsorbed to the stoichiometric anatase 101
surface with a binding energy of -0.48 eV .
The DFT+U approach was adopted in order to recover these highly localized states
61
with a self-consistently computed U term of 3.6 eV applied to Ti atoms, well within Finazzi
et. al.’s suggested range of 3-4 eV for anatase.
103,104
Although one study found the oxygen
vacancy to be present deep in slabs of varying size,
105
others were unable to determine the
most stable vacancy structure.
80
Our DFT+U calculations identified the oxygen vacancy to
be at the surface, specifically, the two-fold coordinated bridging oxygen (see Figure 3.6),
as energetically favorable compared to other sites. Adsorption energies were calculated by
subtracting the two components (molecule and surface) from the adsorbed system:
Eads=E[surf+molecule]-E[surf]-E[molecule]
where the molecule was CO2, or CO2
-
and the surface, stoichiometric anatase or defective
anatase with a surface oxygen vacancy. A thorough, manual search was performed to
determine the global minimum of adsorbed molecules to the anatase support. Adsorption
energies were favorable between neutral CO2 adsorbed to stoichiometric (-0.48 eV) and
defective anatase (-0.94 eV). The global minima found reproduced the geometries of
Sorescu, et al.’s PW-DFT study of the adsorption of CO2 on anatase.
34
The difference in
adsorption energies to Sorescu, et al.’s results may be attributed to our smaller cell and
computational parameter. In our study, we investigated alternative roles of the anatase
support such as stabilization of CO
2-
intermediate. However, the adsorption of CO2
-
to
stoichiometric and defective anatase resulted in repulsive, unstable systems, requiring
thermodynamically unfavorable energies of 4.39 eV and 2.57 eV , respectively, to form. The
chemical bonding analysis obtained using the Bader charge localization scheme is shown
62
in Figure 3.7.
106-108
In Figure 5.6b, the linear CO2 molecule becomes bent upon adsorption
to the defective anatase support, its C effectively filling the bridging oxygen vacancy.
Moreover, CO2 gains an electron (-0.897e) spontaneously from the TiO2 support. This
calculation indicates that the O vacancies provide active sites for CO2 absorption, and no
overpotential is required to form the CO2
-
intermediate. In fact, 4 of the 8 minima found by
Sorescu, et. al. formed the CO
2-
intermediate.
34
These CO2
-
intermediates then react with
H2O to form methanol. Therefore, this thin TiO2 film can not only protect InP from
photocorrosion, but also increase the methanol yields without applying high overpotentials.
In order to further improve the selectivity of methanol with respect to H2 evolution in
aqueous solution, we deposited copper nanoparticles on the TiO2-passivated InP
nanopillars. Cu and its oxide are known catalysts for CO2 reduction by lowering the energy
barriers of intermediate states in the reaction
109,110
and have been intensively studied in the
electrochemical reduction of CO2.
111,112
Figure 3.8a shows the photo-I-V characteristics of
InP nanopillars with copper nanoparticles (green curve) and TiO2-passivated InP
nanopillars with copper (purple line). From this Figure, we find that the addition of Cu
nanoparticles to the bare InP NPLs (green curve) does not change the photo-I-V
characteristics. However, it does increase the Faraday efficiency of methanol from 0.85%
to 2.8%, indicating that copper is an active catalyst for CO2 reduction to methanol. By
deposition of copper nanoparticles on the TiO2-passivated InP nanopillars, we successfully
increase the selectivity to 8.7%, as shown in Figure 3.8b. This improvement in the
63
selectivity likely results from the interface of the copper nanoparticles and the TiO 2-
passivation layer. For Cu as a metal catalyst (and likely CuO), there have been several
previous works reporting that it has a moderate hydrogen overvoltage and weak CO 2
adsorption characteristics, and that it can facilitate the reaction of CO with H 2 to generate
hydrocarbons, aldehydes, and alcohols as major products.
111,113
We believe it is the oxygen
vacancies in the TiO2 and Cu together, which promote the selectivity of methanol over H2
production, however a detailed reaction mechanism is still under investigation.
Figure 3.8 (a) Photocatalytic current-potential curves and (b) Faraday efficiencies of
methanol production for samples of bare InP, InP/Cu, and InP/TiO2/Cu under 532nm
illumination in a CO2 saturated 0.5M KCl solution for a12-hour reaction.
Table 3.1. Faraday efficiencies of methanol of 12-hour reaction on InP/TiO2/Cu sample
under 532nm illumination in a CO2 saturated 0.5M KCl solution.
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4
-35
-30
-25
-20
-15
-10
-5
0
Dark
Bare InP
InP w/Cu
InP w/ TiO
2
&Cu
Potential vs. NHE (V)
Current density (mA/cm
2
)
0
2
4
6
8
InP with Cu InP with TiO
2
&Cu
Faraday Efficency of Methanol (%)
Bare InP
-0.6V vs.NHE
12hr Reaction
Methanol (mol) Q(C) Faraday Efficiency
Planar 2.76E-8 1.414 1.13%
Nanopillar 5.0E-8 3.39 0.864%
64
The previous work of Lee et al. showed that these nanotextured InP photo-cathodes
exhibit about a 40% enhancement in the photoconversion efficiency compared to planar
surfaces in photocatalytic water splitting.
87
In the work presented here, we also observed
an enhancement by a factor of 2 in total conversion efficiency when comparing InP
nanopillars to planar InP. The Faraday efficiencies of methanol production, however, are
around 1% for both planar and nanopillar InP, and, thus, the selectivity of methanol with
respect to H2 is almost the same, as listed in Table 3.1. By depositing Cu nanoparticles on
TiO2-passivated InP nanopillars, the Faraday efficiency increases to 8.7%. The H2 yields
produced by these samples were measured by GC after 12 hours of illumination with
532nm wavelength light under an applied potential of -0.6V vs. NHE, as shown in Figure
3.9. Table 3.2 lists the H2 yields from these GC measurements, methanol yields from NMR
spectroscopy, and the total integrated charge running through the device. For 3nm TiO2-
passivated InP nanopillars, the Faraday efficiency is 83.6% in total and 76.9% for H2
evolution. This indicates that, in aqueous solution, simultaneous H2 evolution is an
inevitable competing reaction with a lower energy barrier than CO2 reduction. In order to
suppress the H2 competing reaction, a non-aqueous solution should be used in future
studies.
65
Figure 3.9 Gas chromatograph (GC) data taken after 12-hour reaction on InP/TiO2/Cu
sample under 532nm illumination in a CO2 saturated 0.5M KCl solution.
Table 3.2. Faraday efficiencies of hydrogen and methanol of 12-hour reaction on
InP/TiO2/Cu sample under 532nm illumination in a CO2 saturated 0.5M KCl solution.
3.4 Conclusion
In summary, photocatalytic CO2 reduction with water to produce methanol is observed
using TiO2-passivated InP nanopillar photocathodes under 532nm wavelength
illumination. In addition to providing a stable photocatalytic surface, the TiO2-passivation
provides substantial enhancement in the photoconversion efficiency through the
1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
H
2
CO
2
Re tention Time (min)
N2
3nm TiO2 & Cu H2 CH3OH H2 & CH3OH
Faraday Efficiency 76.9% 8.7%
85.6%
66
introduction of O vacancies associated with the non-stoichiometric growth of TiO2 by
atomic layer deposition. The role of these oxygen vacancies as catalytically active sites in
the photocatalytic reduction of CO2 is established by PW-DFT calculations, which indicate
that CO2 binds stably to these oxygen vacancies and gains an electron (-0.897e)
spontaneously from the TiO2 support. Therefore, no externally applied overpotential is
required to form the CO2
-
intermediate, which can subsequently react with H2O to form
methanol. Copper nanoparticles deposited on the TiO2 act as a cocatalyst, which further
improves the selectivity and yield of methanol production by up to 8 fold giving a Faraday
efficiency of 8.7%.
67
Chapter 4: Enhanced Photocatalytic Reduction of CO2 to CO
through TiO2 Passivation of InP in Ionic Liquids
4.1 Introduction
Photocatalytic reduction of CO2 is an exciting reaction system with the ability to
convert an abundant greenhouse gas to combustible hydrocarbon fuels using sunlight.
13,114
The direct conversion of solar energy into chemical bonds that can later be released in a
carbon neutral cycle has several advantages over solar-to-electric energy conversion, most
notably, the ability to store large amounts of energy (~GW).
6-11,14,19,58-60
CO2 is recognized
as a naturally abundant and inexpensive carbon source for the production of organic
fuels.
115
The two electron reduction of CO2 to CO (i.e., CO2 + 2e
-
+ 2H
+
CO + H2O) is
an important reaction, since CO can react with water to generate hydrogen gas, and this
CO/H2 mixture (syngas) can subsequently be used to produce synthetic petroleum using
the Fischer−Tropsch process.
116
Many attempts have been made to reduce CO2 by 2e
-
processes to various species such as CO and formic acid, as reported in previous
literature.
13,61-67
The most likely first step in these multi-electron reactions is the one
electron reduction to the CO2
-
intermediate,
72
which lies 1.7eV above the conduction band
of TiO2 and 1.6eV above InP, as first proposed by Bockris et al.
73-75
Since then, others have
also attributed the high overpotentials required to drive this reaction to the formation of the
CO2
-
intermediate.
8,25,58,76
Recently, some researcher attempted to reduce CO2 into CO in
68
ionic liquids
117,118
. In an aqueous system, Rosen et al. reported an effective electrocatalytic
system that reduces CO2 to CO using an ionic liquid 1-ethyl-3-methylimidazolium
tetrafluoroborate ([EMIM]BF4) to lower the energy of the CO2
-
intermediate.
18
In a non-
aqueous solution, Sun et al. used a similar ionic liquid 1-ethyl-3-methylimidazolium
bis(trifluoromethylsulfonyl)imide [EMIM][Tf2N] dissolved in acetonitrile for the
electrochemical reduction of CO2, which resulted in a 0.18V reduction in the overpotential
required to drive this reaction.
119
Also, Rosenthal’s group reported a Bi-based
electrocatalyst demonstrating highly selective conversion of CO2 to CO with a 0.195V
overpotential.
42,43,120
Metal co-catalysts, such as Pt are known to play an important role in
the semiconductor-based photocatalytic systems
121,122
. S. Xie’s group reported a Pt-
promoted TiO2 photocatalyst under UV illumination with enhanced selectivity for CO2
reduction.
123
Recently, we have observed the photoreduction of CO2 to methanol in an aqueous
solution on bare GaP and TiO2–passivated GaP.
115
In addition to providing a stable
photocatalytic surface, the TiO2–passivation layer improves the photocatalytic efficiency
and lowers the overpotential required to drive this reaction.
45,90,115,124,125
However, GaP’s
indirect band gap results in a particularly low absorption coefficient, which is
disadvantageous for photocatalysis and results in a low overall photoconversion efficiency.
Also, a particularly low Faradaic efficiency was observed due to the formation of H 2. InP,
on the other hand, is a direct band gap semiconductor and an excellent candidate for solar
69
photocatalysis. The record high solar-to-hydrogen efficiency of 13.3% was demonstrated
in 1982 using a system that employed p-type InP photocathodes covered with Pt catalysts.
47
More recently, Lee et al. reported 14% efficient water splitting using InP nanopillars with
a Ru catalyst.
48
However, photocatalytic reduction of CO2 on InP has not been reported
with high efficiency.
4.2 Experimental Procedure
The non-aqueous solution was prepared using acetonitrile (AcN, 99.99%), dimethyl
sulfoxide (DMSO, 99.96% D, Sigma Aldrich), and 1-Ethyl-3-methylimidazolium
tetrafluoroborate ([EMIM]BF4, 99.0%, HPLC, Sigma Aldrich). An anion exchange
membrane (Selemion AMV , Anion Exchange Membrane, AGC Inc.) was used to separate
the working and counter electrodes to prevent the oxidation of reduced CO2 products. This
anion exchange membrane only allows negative ions to transfer through the membrane,
which prevents the oxylate intermediate from participating in the oxygen evolution half-
reaction at the counter electrode. p-type (100)-oriented InP with a Zn dopant concentration
of 5×10
17
cm
-3
was used as the photocatalyst for CO2 reduction with an active area of
0.5cm×1cm. Atomic layer deposition (ALD) of anatase TiO2 was performed at 250ºC on
the p-GaP wafers with TiCl4 as the titanium source and water vapor as the oxygen source.
Using ellipsometry, we established that 100 cycles of ALD produces a 4nm thick TiO2 film
and 1000 cycles produces a 40nm film. Pt nanoparticles were deposited by electron beam
evaporation with a nominal thickness of 0.8nm. This is not thick enough to form a
70
continuous film, and instead forms small islands or nanoparticles of Pt. A Zn-Au film was
evaporated on the back of the p-InP to form an Ohmic contact. The Zn-Au contact was then
connected to the external circuitry with a copper wire and coated with epoxy cement to
insulate it from the electrolytic solution, as shown in Figure 4.1b.
Figure 4.1. (a) Schematic of the electrochemical cell and (b) sample geometry of TiO2-
passivated p-InP photocathode.
A three-terminal potentiostat was used with the prepared semiconductor samples as the
working electrode, a Ag/AgNO3 reference electrode, and a Pt wire as the counter electrode,
as shown in Figure 4.1a. The Ag/AgNO3 electrode was made of a silver wire immersed in
0.01 M silver nitrate dissolved in 0.1M TEAP/AcN. Also, the reference electrode was
calibrated against a ferrocene/ferrocenium (Fc
+
/Fc) redox couple to confirm that it gave
the right potential with respect to NHE. The potential applied between the reference and
working electrodes was maintained by the potentionstat (Gamry, Inc.). Before each
measurement, CO2 was purged through the solution on the working electrode side of the
71
electrochemical cell for 30 minutes to ensure a CO2 saturated solution. In this setup, we
analyze the products evolving at the working electrode, instead of at the counter electrode.
It is likely that oxygen is produced at the counter electrode in the reaction 4OH
-
+ 4e
+
O2 + 2H2O. The gaseous products from the CO2 reduction reaction were analyzed with a
gas chromatograph (GC) (Bruker GC-450) equipped with a thermal conductivity detector
(TCD) and a carbon-plot column (Agilent). A standard calibration curve was developed to
quantify the CO yields from the photocatalytic surfaces. NMR spectroscopy was performed
using a Varian spectrometer operating at 600 MHz. The samples were dried on a rotary
evaporator at 40 ºC under vacuum and then dissolved in DMSO. We observed only CO
product on the cathode side.
4.3 Results and Discussion
In the work presented here, we investigate CO2 reduction to CO by illuminating TiO2–
passivated InP with and without Pt nanoparticles in a non-aqueous ionic liquid solution.
We investigate the role that ionic liquid [EMIM]BF4 plays as a homogeneous catalyst for
CO2. The effects of the TiO2 are explored by studying the photocurrent vs. reference
potential characteristics systematically as a function of TiO2 layer thickness. To our
knowledge, this is the first report of photoelectrochemical reduction of CO2 to CO using a
III-V compound semiconductor photocatalyst in an ionic liquid solution. The high Faradaic
efficiencies achieved at substantial underpotentials demonstrates the ability to drive CO2
reduction efficiently using visible light.
72
Figure 4.2. Photocatalytic current-potential curves of InP photocatalysts with different
TiO2 thicknesses in a 0.02M [EMIM]BF4 non-aqueous electrolyte, under 532nm
wavelength illumination.
Figure 4.2 shows the photocurrent-voltage curves for InP passivated with various
thicknesses of TiO2 measured in an AcN electrolyte with 0.02M [EMIM]BF4 under 532nm
illumination. At any given voltage, the TiO2–passivated InP samples have a higher
photocurrent than the bare InP (black curve). Under an applied voltage of -1.37V vs. NHE,
we can see a clear improvement in the photocurrent of samples with TiO2 passivation. From
this data, we find that 3nm thick TiO2 produces the best conditions for photocatalysis. The
strong dependence on TiO2 thickness indicates there is a tradeoff between the benefits of
the TiO2 layer and its insulating nature, which eventually outweighs its benefits in thicker
TiO2 films (20nm curve). Figure 1 also shows the current-potential characteristics of 3nm
TiO2 passivated InP with and without laser illumination. This improvement observed with
TiO2 is attributed to an increased photovoltage produced at the pn-junction formed between
-2.0 -1.5 -1.0 -0.5 0.0 0.5
-15
-10
-5
0
bare InP
InP w/1nm TiO
2
InP w/3nm TiO
2
InP w/5nm TiO
2
InP w/20nm TiO
2
dark InP w/3nm TiO
2
Photocurrent (mA/cm
2
)
Applied Voltage (vs. NHE)
73
the TiO2 and InP. Here, the TiO2 deposited by ALD is n-type due to oxygen vacancies and
forms a pn-junction with the underlying p-type III-V compound photocathode.
80,81
This
junction creates a built-in potential as well as band offsets between TiO2 and InP that assist
in the separation of photogenerated electron-hole pairs, and enables electrons to overcome
the high barriers associated with this reaction through the associated increase in
photovoltage produced at the junction.
Figure 4.3. (a) Photocatalytic current-potential curves of InP photocatalysts with Pt
nanoparticles with various thicknesses of TiO2 in a 0.02M [EMIM]BF4 non-aqueous
electrolyte under 532nm wavelength laser illumination. (b) Photocurrent-potential curves
of 1nm TiO2–passivated InP compared to 1nm TiO2–passivated InP with Pt nanoparticles.
Figure 4.3a shows the photocurrent-voltage curves for Pt nanoparticles deposited on
various thicknesses of TiO2–passivated InP measured in an AcN electrolyte with 0.02M
[EMIM]BF4 under 532nm illumination The 1nm TiO2 film gives the highest photocurrent.
However, the photocatalytic performance does not depend strongly on the TiO2 thickness
with the Pt co-catalyst. Figure 4.3b shows a comparison of the 1nm TiO2–passivated InP
(a)
(b)
-2.0 -1.5 -1.0 -0.5 0.0 0.5
-25
-20
-15
-10
-5
0
Pt 1nm TiO
2
InP
1nm TiO
2
InP
Photocurrent (mA/cm
2
)
Applied Voltage (vs. NHE)
-2.0 -1.5 -1.0 -0.5 0.0 0.5
-25
-20
-15
-10
-5
0
Pt 1nm TiO
2
InP
Pt 3nm TiO
2
InP
Pt 5nm TiO
2
InP
dark Pt 1nm TiO
2
InP
dark Pt 3nm TiO
2
InP
dark Pt 5nm TiO
2
InP
Photocurrent (mA/cm
2
)
Applied Voltage (vs. NHE)
74
photocatalysts with and without Pt nanoparticles from Figures 4.2 and 4.3a, respectively.
Here, the Pt co-catalyst increases the photocurrent by more than 25-fold at 0V vs. NHE.
There appears to be a synergistic effect between TiO2 and Pt perhaps by the introduction
of new active sites at the tri-phase boundary between the TiO2, Pt, and electrolyte. The Pt
co-catalyst provides enhancement in the photocatalytic reduction of CO2 by extracting
electrons from the semiconductor (TiO2/InP), thus decreasing the probability of
recombination with photogenerated holes, and enhancing charge transfer and surface
binding of reactants in solution.
123
Photoluminescence (PL) spectroscopy gives a relative measure of the non-radiative
recombination in a material. Figure 4.4 shows the PL spectra of InP with various
thicknesses of TiO2 and with/without Pt nanoparticles. (For each type of sample measured,
several photoluminescence spectra were taken from different regions of the sample. These
spectra were consistent within <1% variation.) Here, the TiO2 film reduces the PL intensity
of InP by a factor of 2.5 (Figure 4.4a), and the Pt nanoparticles further reduce this by an
additional factor of 2 (Figure 4.4b). This is likely due to the high density of surface states
in the TiO2, which act as non-radiative recombination centers, thus lowering the PL
efficiency. So, contrary to what one might expect, the best photocatalyst has the lowest PL
efficiency and vice versa. From this data, it is apparent that the TiO2 films and their surface
states cause strong electron-hole recombination, however, the benefit that they provide by
lowering the potential barriers of the reaction and promoting charge transfer outweighs
75
their detriment associated with charge recombination.
Figure 4.4. Photoluminescence spectra of (a) InP passivated with various thickness of TiO2,
(b) Pt nanoparticles deposited on TiO2-passivated InP.
The role of the active surface states can be further confirmed by running the same
reaction in LiClO4 electrolyte, as shown previously by Ramesha et al.
126
Here, the small
Li
+
cations bind strongly to the charged surface states essentially poisoning these sites,
making them catalytically inactive.
127-129
Figure 4.5 shows the photo-I-V curves taken with
and without the LiClO4 electrolyte. From this data, it is clear that the CO2 reduction
reaction is blocked by the presence of the Li
+
ions, demonstrating the importance of the
catalytically active surface states, as previously shown by Ramesha et al.
130
As a
comparison, the photo-I-V characteristics were obtained using EMIM
+
solution with N2 gas
instead of CO2, as shown in Figure 4.6. The results obtained with CO2 in the Li
+
solution
closely resemble those taken with N2, where no photoreduction of CO2 is possible. By
contrast, when running the reaction without Li
+
ions in ionic liquid solution, the surface
(a) (b)
750 800 850 900 950 1000
0
20
40
60
Bare
1nm TiO
2
InP
3nm TiO
2
InP
5nm TiO
2
InP
20nm TiO
2
InP
Wavelength(nm)
PL Intensity (10
3
Counts)
750 800 850 900 950 1000
0
20
40
60
Bare
Pt 1nm TiO
2
InP
Pt 3nm TiO
2
InP
Pt 5nm TiO
2
InP
Wavelength(nm)
PL Intensity (10
3
Counts)
With Pt cocatalyst
76
sites remain active and can catalyze the reduction of CO2. Here, the [EMIM]
+
ions in the
ionic liquid solution do not have the same poisoning effect since this is a larger molecule,
unable to provide sufficient Coloumbic screening of these active surface states.
Figure 4.5. Cyclic voltammograms of 3nm TiO2-passivated InP in electrolytes saturated
with N2 and CO2 as indicated in 0.02 M [EMIM]BF4/AcN and 0.02M [EMIM]BF4/AcN
with 0.1M LiClO4/AcN.
The non-aqueous ionic liquid electrolyte is important for suppressing hydrogen
evolution (i.e., water splitting), which has a much lower energy barrier. When illuminated,
TiO2–passivated InP produces CO as quantified by gas chromatography described in the
Experimental Methods section below. Of all the samples without Pt nanoparticles measured
in this study, the 3nm TiO2–passivated InP sample exhibits the highest CO Faradaic
efficiency of 89% at -1.57V (vs. NHE). This corresponds to an overpotential of just 0.02V
compared to the standard redox potential E°(CO2/CO)=-1.55V (vs. NHE) in IL/AcN. The
results of the gas analysis were consistent with the photo-I-V data, which also showed the
3nm TiO2–passivated sample to have the highest performance. We also observe CO 2
reduction at underpotentials of 0.18V and 0.68V with Faradaic efficiencies of 50% and
-1.5 -1.0 -0.5 0.0 0.5
-15
-10
-5
0
EMIM.BF
4
saturated N
2
EMIM.BF
4
saturated CO
2
LiClO
4
EMIM.BF
4
saturated N
2
LiClO
4
EMIM.BF
4
saturated CO
2
Photocurrent (mA/cm
2
)
Applied Voltage (NHE)
77
39%, respectively, as listed in Table 4.1. For the samples with Pt co-catalyst, 1nm TiO2–
passivated InP gives the best performance with 99% Faradaic efficiency at an
underpotential of 0.78V (-0.77V vs. NHE), as listed in Table 1. Table S1 in the
supplemental Information lists the Faradaic efficiencies of CO for various thicknesses of
TiO2–passivated InP in 0.02M [EMIM]BF4 ionic liquid electrolyte taken after 4 hours of
illumination with 532nm wavelength light at an underpotential of 0.18V .
Figure 4.6a shows the
1
H NMR spectrum of the reaction products after 4h illumination
of 3nm TiO2–passivated InP in [EMIM]BF4, Figure 4.6b shows the
1
H NMR spectrum of
pure [EMIM]BF4. In Figure 4a, the main resonance lines in the spectrum correspond to the
protons of the pure IL. The [EMIM]BF4 ionic liquid was chosen for these experiments
since CO2 is known to form weak complexes with BF4 anions, as compared with other
ionic liquids. As reported by Rosen et al., the intermediate complex formed can be
expressed as EMIM-CO2*
…
BF4, where the intermediate EMIM-CO2* is bonded weakly
with BF4
-
ions. Ideally, this CO2* complex is bound strongly enough to facilitate CO2
reduction, but not too strongly as to render the CO2* unreactive.
18
The formation of the
bound EMIM-CO2* intermediate compound, as previous established by Rosen and Sun, is
evidenced by the low intensity set of peaks in the NMR spectrum of Figure 4.6a, which are
labeled with prime symbols (e.g., 4’, 5’).
18,119
No resonant H peaks appear on C2, which
indicates that the CO2 molecule is binding to the C2 of the imidazolium ring to form the
intermediate species, as illustrated in Figure 4.6c.
78
Figure 4.6. (a)
1
H NMR spectrum (in DMSO, reference to TMS) of 0.02M [EMIM]BF4
electrolyte solution after 4h of CO2 reduction. (b)
1
H NMR spetrum of pure [EMIM]BF4.
(c) Schematic representation of the intermediate complex that is formed between the
[EMIM] ion and the CO2
-
.
As a control experiment, we compared photocatalysis performed in the 0.02M
[EMIM]BF4 in AcN electrolyte with that performed in a CO2 saturated 0.5M KCl (pH=7)
aqueous electrolyte at -1.37V (vs. NHE) under 532nm irradiation. Here, we found that only
hydrogen was produced in the aqueous solution, further establishing the importance of the
non-aqueous electrolyte for CO2 reduction reactions. Also, we ran the experiment at -1.37V
(vs. NHE) on 3nm TiO2–passivated InP without light irradiation and detected no CO. In
(b)
(c)
(a)
9 8 5 4 3 2
7
7'
7
8'
8
8
6
6
6'
5 4
2
2
5 4
5' 4'
AcN-H
79
order to rule out other possible sources of carbon in this reaction, we repeated the
experiment bubbling argon as the control gas instead of CO2, and observed no CO2
reduction products. We also ran the experiment using isotopically labeled
13
CO2, and
observed
13
CO (molecular weight 29), as shown in Figure 4.7, verifying that CO2 is indeed
the main carbon source in this reaction. Figure 4.8 plots the Faradaic efficiencies of the CO
and carboxylate products as a function of [EMIM]BF4 concentration in an AcN electrolyte
under 532nm wavelength illumination. From this data, it is clear that the molarity has
little effect on the Faradaic efficiency of the products, and 0.02M is sufficient to run the
reaction.
Figure 4.7. Quadruple mass spectrometer data for molecular mass of 29 (
13
CO).
100 150 200 250
Molecular mass 29
13
CO
Ion Current
Time (S)
Light on
80
Figure 4.8. CO Faradaic efficiency for different thickness TiO2 passivated InP in different
molarity of [EMIM]BF4 in AcN electrolyte, under 532nm wavelength laser illumination
after 4 hours irradiation and -1.37V applied voltage vs. NHE.
These photocatalysts represent complex photoelectrochemical systems, and it is
difficult to separate the physical (e.g., photovoltaic) and chemical (e.g., surface
intermediates) processes in the overall efficiency. The main advantage of this system,
compared to bulk TiO2, are physical in nature. Namely, the band gap of InP is close to the
optimum band gap derived in the Shockley-Queisser limit for solar utilization. Also, InP’s
carrier transport and lifetimes far exceed those of TiO2. One non-trivial aspect of these
photocatalysts is the low resistance Ohmic contact formatted on the back of the InP wafer,
without which efficiencies would be considerably lower. On the chemical side,
overpotential losses are reduced by the introduction of catalytically active sites and the
formation of intermediate complexes.
0.00 0.02 0.04 0.06 0.08 0.10
0
20
40
60
CO FE of InP w/ 3nm TiO
2
CO FE of InP w/ 5nm TiO
2
Different concentration of [emim]BF
4
in AcN
Faradaic Efficiency (%)
81
4.4 Conclusion
In conclusion, we report photocatalytic CO2 reduction to CO at high underpotentials
on TiO2–passivated InP with a Pt co-catalyst in a non-aqueous [EMIM]BF4 ionic liquid
electrolyte. The non-aqueous solution prohibits hydrogen formation and enables CO2
reduction with a Faradaic efficiency of 99% at an underpotential of 0.78V . The TiO2
passivation layer provides enhancement in the photoconversion efficiency through the
formation of a charge separating pn-region and active surface states, which reduces carrier
recombination and lowers the external potential required to initiate this reaction. The TiO2
layer reduces the photoluminescence intensity indicating that these active surface states
cause electron-hole recombination, however, the benefits of the TiO2 film outweigh its
detriment associated with charge recombination. The Pt co-catalyst provides substantial
enhancement in the Faradaic efficiency of this reaction system. NMR spectroscopy shows
the formation of an intermediate complex [EMIM-CO2]
*
, which further lowers the energy
of the reaction pathway.
82
Chapter 5: Plasmon-enhanced water splitting on TiO2-passivated
GaP Photocatalysts
5.1 Introductions
In photocatalysis, the energy of photons can be utilized to drive many important
chemical reactions, including H2 production
131
, CO2 reduction
49,50,77
, and water
purification
132,133
. Unfortunately, the efficiencies of most photocatalytic processes are far
too low for practical large scale applications. The direct conversion of solar-to-chemical
energy has several advantages over solar-to-electric energy conversion, most notably, the
ability to store large amounts of energy (~GW) in chemical bonds that can later be released
without producing harmful byproducts. As the cost of direct solar-to-electrical energy
becomes competitive with fossil fuels, there will be a need to store large amounts of the
solar energy for use during nights, cloudy days, and winter months. Therefore, our energy
infrastructure has much to gain by finding more efficient ways to enhance these
photocatalytic processes. Over the past few years, several research groups have
demonstrated a new method for improving the efficiency of photocatalytic processes by
exploiting the strong plasmon resonance of small metal nanoparticles
4,134-136
. While these
studies have clearly shown proof-of-principle of this enhancement mechanism, the overall
photoconversion efficiencies are still very low
134
. Torimoto et al. also demonstrated
enhanced photocatalytic water splitting by depositing CdS nanoparticles on SiO2-coated
83
Au nanoparticles
137
. However, these plasmonic-photocatalytic complexes demonstrated
enhancement factors less than 2. Liu et al. and Ingram et al. observed enhanced
photocatalytic water splitting under visible illumination by depositing plasmonic metal
nanoparticles on top of anatase TiO2
137-142
. While these previous works reported
enhancement factors of approximately 10-fold with the incorporation of plasmonic
nanoparticles, the overall photoconversion efficiencies were still quite low in the visible
wavelength range because of TiO2’s large band gap (Eg=3.2eV). As a result, these proof-
of-principle studies relied on short-lived, sub-band gap defect states for optical absorption
in the visible wavelength range.
GaP (Eg=2.25eV) has a substantially smaller band gap than TiO2, can absorb more than
18% of the solar spectrum, and is better matched to the plasmon resonance energy of Au
nanoparticles. While GaP absorbs a more extended range of the solar spectrum than TiO 2,
its main advantage as a photocatalyst is its relatively high conduction band energy, which
exceeds that of TiO2 by more than 1eV . However, GaP is more expensive than TiO2 and is
not oxidatively stable over a wide range of pH. As a result, a vast majority of previous
photocatalytic studies have been carried out on TiO2. In the work presented here, we
investigate the photocatalytic stability of TiO2-passivated GaP using atomic force
microscopy (AFM), photoelectrochemistry, and optical microscopy. The photocatalytic
efficiency is studied systematically as a function of TiO2 layer thickness. Further catalytic
enhancement is explored with the addition of plasmonic gold nanoparticles, and is studied
84
as a function of TiO2 layer thickness. The effects of plasmonic enhancementare
distinguished from the natural catalytic properties of Au by evaluating similar
photocatalytic TiO2/GaP structures with catalytic, non-plasmonic metals (i.e., Pt) instead
of Au.
5.2 Experimental Procedure
In the work presented here, Zn doped p-type (100) oriented GaP with a dopant
concentration of 2x10
18
cm
-3
was used as a photocatalyst for water splitting. Atomic layer
deposition (ALD) of TiO2 was performed at 250
o
C on the p-GaP wafers with TiCl4 as the
titanium source and water vapor as the oxygen source. The carrier gas during the deposition
was argon with a flow rate of 20sccm. The rate of deposition was about 0.4Å per cycle. A
500nm thick aluminum film was evaporated on the back of the p-GaP to form an Ohmic
contact. We then evaporated a gold film with a nominal thickness of 5nm on the top surface
of the TiO2. This thin gold film is known to form island-like growth that is strongly
plasmonic and serves as a good substrate for surface enhanced Raman spectroscopy
(SERS)
143,144
and photocatalytic enhancement
138,145
. Samples were prepared with and
without the plasmonic gold nanoparticles. The schematic diagram of sample geometry is
shown in the schematic diagrams of Figure 5.1. The aluminum contact was then connected
to the external circuitry with a copper wire and coated with epoxy cement to insulate it
from the electrolytic solution. The photocatalytic reaction rates of two sets of samples were
measured in 0.5 M Na2SO4 and 0.5M H2SO4 solutions using a three-terminal potentiostat
-1.5 -1.0 -0.5 0.0
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
Dark
Bare GaP
GaP w/1nm TiO
2
GaP w/3nm TiO
2
GaP w/5nm TiO
2
GaP w/10nm TiO
2
Potential vs. Ag/AgCl (V)
Current Density (mA/cm
2
)
RHE=0
85
with the prepared samples, a Ag/AgCl electrode, and a graphite electrode functioning as
the working, reference, and counter electrodes, respectively.
Figure 5.1 Schematic diagram of sample geometry for GaP photocatalysts with TiO2 and
Au nanoparticles.
Finite difference time domain (FDTD) simulations were carried out using a cell of size
1000 nm x 800 nm x 500 nm. A grid spacing of 2Å is used in the volume of 500 nm x 500
nm x 40 nm around the film and 10 nm elsewhere. A 0.002 fs temporal grid is used with a
total of 100 000 time steps. In the simulation, the samples are irradiated with a planewave
source with a Gaussian pulse containing a spectrum of wavelengths ranging from 300 nm
to 800 nm. Perfectly matched layers (PML) boundary conditions are used with 25 layers.
The dielectric function of Au is based on the optical constants given by Palik and Ghosh.
139
5.3 Results and Discussion
Figure 5.2a shows the photocurrent-voltage curves for GaP passivated with various
thicknesses of TiO2 measured in a pH=7 solution of 0.5M Na2SO4 under 532nm
illumination. Bare GaP (blue curve) has an onset of photocurrent at a potential of
approximately -0.66V . For TiO2 passivated GaP, we see a clear shift in the overpotential
Green Laser
(532nm)
Au
TiO
2
Al
Epoxy
p-GaP
86
that scales linearly with the thickness of the TiO2, as shown in Figure 5.2c. For 10nm TiO2
(pink curve), this overpotential/onset potential is shifted by approximately 0.46V . This shift
is attributed to the formation of a pn-junction, since the TiO2 is n-type doped due to oxygen
vacancies
808080363531314343
. While TiO2 does not absorb light at 532nm, the pn-junction
formed with the GaP enables separation of photogenerated charge in the actively absorbing
GaP. Figure 5.2d shows the built-in potential for the junction calculated using the
relation 𝑉 𝑏𝑖
=
𝑊 𝐷 2
𝑞 2𝜖 0
𝜖 𝑎 𝜖 𝑑 𝑁 𝑎 𝑁 𝑑 (𝑁 𝑎 𝜖 𝑎 +𝑁 𝑑 𝜖 𝑑 )
(𝑁 𝑎 +𝑁 𝑑 )
2
, with a doping concentration of Na=5x10
18
cm
3
. Here,
𝑊 𝐷 is the depletion width of the GaP-TiO2 junction, which is a function of the TiO2
thickness layer. This calculation shows a similar trend to the experimentally observed shift
in the overpotential. Typically, p-type GaP acts as a photocathode due to the direction of
band bending at its interface with water, promoting photoelectron flow to the water. For n-
type semiconductors, the bending is usually upward, resulting in photohole flow to the
water. It is, therefore, somewhat surprising that when n-type TiO2 up to 10 nm thick is
coated on top of p-GaP, the electrode still functions as a photocathode. This is most likely
due to the finite thickness of the TiO2 layer, which is fully depleted and does not provide
enough
87
Figure 5.2 Photocatalytic current-potential curves measured for GaP photocatalysts with
various thicknesses of TiO2 under 1W/cm
2
532nm illumination in a 0.5M Na2SO4 pH=7
solution. (c) Relative decrease of the overpotential required to initiate this reaction and (d)
calculated built-in voltage plotted as a function of TiO2 thickness.Energy band diagrams
for (e) thin and (f) thick TiO2 layers. (g) Photocatalytic current plotted as a function of
potential for thicker TiO2 layers.
(e)
-1.5 -1.0 -0.5 0.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
Dark
Bare GaP
GaP w/1nm TiO
2
GaP w/3nm TiO
2
GaP w/5nm TiO
2
GaP w/10nm TiO
2
Potential vs. Ag/AgCl (V)
Current Density (mA/cm
2
)
RHE=0
-1.5 -1.0 -0.5 0.0
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
Dark
Bare GaP
GaP w/15nm TiO2
GaP w/20nm TiO2
GaP w/25nm TiO2
Potential vs. Ag/AgCl (V)
Current density (mA/cm
2
)
0 1 2 3 4 5 6 7 8 9 10
0.0
0.1
0.2
0.3
0.4
0.5
Thickness of TiO
2
(nm)
Decrease of Overpotential (V)
1 2 3 4 5 6 7 8 9 10
0.0
0.1
0.2
0.3
0.4
0.5
Built-in Voltage (eV)
TiO
2
Thickness (nm)
(a) (b)
(c)
(d)
(f)
88
surface charge to invert the material from p- to n-type, as illustrated in Figure 5.2e. When
the TiO2 thickness is further increased, enough donor impurities to bend the band upward,
hindering the electron flow to the water, as shown in Figure 5.2f. This result was
corroborated experimentally, when the thickness of TiO2 is increased beyond 15nm, we
observe no photocurrent, as shown in Figure 5.2b.
Figure 5.3 shows the photocatalytic I-V characteristics for TiO2 passivated GaP with
and without gold nanoparticles. Au nanoparticles deposited on GaP without TiO2
passivation (purple curve) improve the photocatalytic reaction rate by a slight shift of the
I-V curve to a lower overpotential with respect to bare GaP by approximately 0.2V .
Depositing Au nanoparticles on top of TiO2 passivated GaP results in a further
improvement in the I-V characteristics. For 0.5nm thick TiO2, however, significant
enhancement is observed in the photocatalytic reaction rate. For this dataset, we observe
both a downshift of the overpotential (by approximately -0.58V) and an increase in the
photocurrent (i.e., increased photoinduced charge) due to plasmonic field enhancement. At
V=-0.7V , the photocurrent is enhanced by a factor of 4X with respect to bare GaP, after
accounting for the shift in the overpotential. The photocatalytic properties of samples with
0.5nm TiO2 are substantially better than those with 1nm TiO2. This anomalous behavior of
0.5nm thick TiO2 was observed in several other samples consistently, and is the result of a
tradeoff between pn-junction formation and coupling of the localized plasmonic fields of
the Au nanoparticles to the actively absorbing GaP layer, as discussed below.
89
Figure 5.3 Photocurrent plotted as a function of voltage for GaP photocatalysts with
various thicknesses of TiO2 with Au nanoparticles under 1W/cm
2
532nm illumination in a
0.5M Na2SO4 solution.
Figure 5.4 Electric field distributions calculated using the finite difference time domain
method (a) in the plane of the Au nanoparticles and (c,d) in the perpendicular direction
across the GaP/TiO2/Au/electrolyte interface. (b) Calculated electric field enhancement
factor plotted as a function of TiO2 thickness using Eq. 5.1.
-1.5 -1.0 -0.5 0.0
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
Dark
Bare GaP
GaP w/ 5nm Au
GaP w/ 0.5nmTiO
2
and 5nm Au
GaP w/ 1nmTiO
2
and 5nm Au
GaP w/ 3nmTiO
2
and 5nm Au
Potential vs. Ag/AgCl (V)
Current Density (mA/cm
2
)
0 1 2 3 4 5
1.0
1.2
1.4
Enhancement Factor
TiO
2
Thickness (nm)
Au Au
GaP
TiO
2
(a)
(d)
(c)
(b)
50 nm
Au
Au
GaP
TiO
2
(d)
90
Figure 5.4 shows the results of a finite difference time domain (FDTD) simulation
performed on a 5nm gold nanoparticle (nano island) film. The electric field
distribution in the plane of the gold island film is shown in Figure 5.4a. Here,
localized hot spots can be seen in regions between nearly touching
nanoparticles/nanoislands separated by approximately 2-3nm. This phenomenon has
been studied in detail previously
138,145-147
. Figure 5.4c shows the electric field
distribution in the perpendicular direction, across the GaP/TiO2/Au/electrolyte
interface at one of the hot spots between two nearly touching Au nanoparticles with a
TiO2 thickness of 0.5nm. Here, the electric field intensity can reach 1000X the incident
electric field. These plasmonic nanoparticles couple light very effectively from the far
field to the near field at the GaP/TiO2 pn-junction. This is advantageous for
photocatalysis for two reasons. First, there is an increased electron-hole pair generation
rate in close proximity to the electrolyte interface and charge separating region,
enabling a larger fraction of the photoinduced charge to diffuse to the catalytic surface
and contribute to catalysis. We can calculate this fraction by integrating E
2
over the
volume of the catalyst from the GaP surface to one minority carrier diffusion length
(100nm) below this surface, as described by Eq 5.1.
(5.1)
91
In the denominator, the electric field intensity without the TiO2 and Au nanoparticles,
Eo, is integrated over the same volume as in the numerator. The second enhancement
mechanism arises from the increased light intensity at the pn-junction, which produces a
larger open circuit voltage 𝑉 𝑜𝑐
=
𝑛𝑘𝑇 𝑞 ln (
𝐼 𝐿 𝐼 0
+ 1), and this further reduces the overpotential.
Here, the plasmon-enhanced electric fields create a larger photocurrent IL, which in turn
produces a larger Voc. In addition to electric field enhancement, these plasmonic
nanoparticles may produce hot electrons that can drive catalytic processes at a lower
applied overpotential
135
. Figure 5.4d shows the electric field distribution for a
GaP/TiO2/Au nanoparticle structure with a 3nm thick TiO2 film. Here, the localized
plasmonic fields do not extend into the actively absorbing GaP layer. Figure 5.4b shows
the integrated electric field enhancement factor calculated using Eq. 5.1 plotted as a
function of TiO2 thickness. In this Figure, a maximum enhancement of 42% can be seen at
0.5nm TiO2 thickness. The sharp drop off of the EF as the TiO2 thickness increases is a
direct result of the highly localized nature of the plasmon-enhanced fields. The 42%
electric field enhancement factor is significantly smaller than those obtained in our
previous studies of plasmon-enhanced TiO2 (without GaP)
138
. This is a direct result of
GaP’s relatively long minority carrier diffusion length (100nm), which results in a larger
integrated volume in Eq 5.1.
Figure 5.5 shows the two terminal photocurrent densities plotted as a function of the
applied overpotential for various GaP/TiO2/Au nanoparticle photocatalysts, measured with
92
a Pt counter electrode in a pH=0, 0.5M H2SO4 solution illuminated at 532nm. A large
reduction in the overpotential from approximately -0.8 to -0.2V can be seen for the
GaP/TiO2/Au structure with 0.5nm TiO2 thickness. Again, the plasmon-enhanced
photocatalyst with 0.5nm TiO2 is substantially better than those with thicker TiO2 films and
TiO2 without gold nanoparticles.
Figure 5.5 Two terminal photocurrent density (absolute value) plotted as a function of the
applied overpotential for various GaP/TiO2/Au nanoparticle photocatalysts measured in a
pH=0, 0.5M H2SO4 solution under 1W/cm
2
532nm illumination.
Perhaps the most important aspect of these TiO2 passivated GaP photocatalysts is their
photochemical stability in the electrolytic solution. Figures 5.6 and 5.7 show the time
dependence of the photocatalytic current and surface roughness without and with TiO 2
passivation, respectively. In Figure 5.6a, the photocurrent density is plotted as a function
of time for bare GaP illuminated at 532nm in a 0.5M Na2SO4 solution with an applied
overpotential of -0.7V for 5 hours. An exponential decay can be seen with a time constant
of 0.45 hours indicating significant corrosion of the surface. An abrupt drop in the
-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0
0.01
0.1
Dark
Bare GaP
GaP w/0.5nmTiO
2
GaP w/0.5nmTiO
2
&Au
GaP w/10nmTiO
2
GaP w/10nmTiO
2
&Au
Applied Overpotential (V)
Current Density (mA/cm
2
)
93
photocurrent occurs at 4.8 hours when the device failed. Figures 5.6b and 5.6c show optical
microscope and atomic force microscope images of the GaP surface after 5 hours of
illumination. Figure 5.6d shows a plot of the surface topography obtained along the dashed
white line indicated in Figure 5.6c, showing an RMS roughness of ±143nm. In contrast,
the photocurrent density of TiO2 passivated GaP is stable for 12 hours, as plotted in Figure
5.7a. The optical microscope image (Figure 5.7b) and atomic force microscope image
(Figure 5.7c) exhibit no evidence of surface corrosion or damage after 12 hours, with an
RMS surface roughness of ±1.0nm (Figure 5.7d).
Figure 5.6 (a) Time dependence of the photocurrent density of bare GaP illuminated with
1W/cm
2
532nm light in a 0.5M Na2SO4 solution at an applied overpotential of -0.7V . (b)
Optical microscope image, (c) atomic force microscope image, and (d) surface topography
of the GaP surface after the 5 hour reaction.
3 4 5 6 7
-200
-100
0
100
200
300
Scan Range ( m)
Height (nm)
0 1 2 3 4 5
0.0
0.1
0.2
0.3
0.4
Bare GaP
Time (Hours)
Current Density (mA/cm
2
)
(a)
(b)
10 m
(c)
(d)
2 m
±143nm
94
Figure 5.7 (a) Time dependence of the photocurrent density of TiO2 passivated GaP
illuminated with 1W/cm
2
532nm light in a 0.5M Na2SO4 solution at an applied
overpotential of -0.7V . (b) Optical microscope image, (c) atomic force microscope image,
and (d) surface topography of the GaP/TiO2 surface after the 12 hour reaction.
5.4 Conclusion
In conclusion, plasmon-enhanced photocatalytic water splitting is observed on TiO2
passivated GaP. The TiO2 passivation layer prevents corrosion of the GaP surface, making
it stable in pH=0 solution. In addition to preventing corrosion, the TiO2 passivation layer
provides enhancement in the photoconversion efficiency through the formation of a charge
separating pn-region, which decreases carrier recombination and lowers the overpotential
required to initiate this reaction. Plasmonic Au nanoparticles deposited on top of TiO2
0 2 4 6 8 10 12
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
GaP w/ 3nm TiO
2
Time (Hours)
Current Density (mA/cm
2
)
10 m
0 1 2 3 4 5
-4
-3
-2
-1
0
1
2
Scan Range (um)
Height (nm)
(a)
(b)
(c)
(d)
1 m
±1nm
95
passivated GaP further improve the photocatalytic process through plasmonic field
enhancement. These two enhancement mechanisms result in an optimum thickness of the
TiO2 layer of 0.5nm. Electromagnetic simulations performed using the finite difference
time domain (FDTD) method indicate that a 0.5nm film of TiO2 enables significant
coupling of the localized plasmonic fields of the Au nanoparticles with the actively
absorbing GaP layer. This general approach of passivating narrower band gap
semiconductors with TiO2will enable more efficient photocatalysts to be developed.
96
Chapter 6: Outlook and Future Work
We have developed photocatalysts for CO2 reduction and water splitting that are based
on TiO2-passivated III-V compounds (i.e., GaAs, GaP, and InP). These III-V compound
semiconductors have smaller band gaps than TiO2 and, therefore, make more efficient use
of the solar spectrum. The TiO2 makes these surfaces photochemically stable and improves
their catalytic efficiency significantly. Here, the TiO2 surface layer (1-5nm thick) is
deposited by atomic layer deposition (ALD). While the TiO2 layer provides substantial
enhancement, the underlying mechanisms of enhancement are not understood. In the future
work, we will elucidate the mechanisms contributing to the photocatalytic enhancement
using a systematic approach. Our preliminary findings were obtained on planar surfaces
without any co-catalyst and are far from optimized. The future work will develop a
fundamental understanding of the reaction mechanism and a detailed reaction pathway that
can later be engineered to produce high efficiency photocatalysts.
We will systematically vary several aspects of the sample preparation to explore and
separate various mechanisms of catalysis and enhancement. We will use vibrational sum
frequency generation (vSFG) spectroscopy, which is a surface selective technique, to
identify reaction intermediate species and catalytically active sites on these carefully
engineered photocatalytic surfaces. The primary reaction of interest will be the
photocatalytic reduction of CO2 to CO and to various hydrocarbons, which is a complex
reaction system requiring up to 8 electrons and many intermediate species, some of which
97
have extremely high energy barriers. Comparing the intermediates observed on various
sample configurations (i.e., TiO2 thickness, ALD precursors, doping, etc.) over a range of
electrochemical conditions will elucidate key enabling factors in the reaction pathway and
establish the roles of defects and the metal-semiconductor interface. We will vary the
surface coverage, size, and composition of metal co-catalyst nanoparticles, and
controllably introduce defects in the TiO2 surface to determine their effect on the reaction
surface-bound intermediates. By varying the sample morphology, we will ascertain the
extent to which hot electrons give rise to new reaction pathways and intermediates, identify
specific adsorption sites and catalytically active sites, and ultimately constrain the reaction
mechanism to provide a detailed reaction pathway.
Vibrational sum frequency generation (vSFG) will be used to identify the reactant and
intermediate species adsorbed at the active surface sites on the photocatalytic substrate. In
particular, we will focus on carbon monoxide, a key reaction intermediate and common
catalyst poison, and the adsorbed CO2 species, the necessary reaction precursor. The
adsorbed CO stretch frequency is sensitive to the adsorption site, ranging between 1850
and 2150 cm
-1
,
148-157
and was tentatively identified in our preliminary studies of random
Au nano-islands to be at 2120 cm
-1
. The physisorbed CO2 is expected to have the
asymmetric stretch frequency within 50 cm
-1
of the gas-phase value (2396 cm
-1
),
158
while
the chemisorbed species at the reactive sites are likely partially charged and bent, with
frequencies reported at 1640-1670 cm
-1
and 1240-1245 cm
-1
.
159
We will also scan the
98
2800-3000 cm
-1
range of C-H stretch modes to detect the nascent hydrocarbon products,
and possible intermediates such as aldehyde and methanol will be detected using their O-
H, C=O, and C-O stretches. Although the primary focus of this proposal is on the reaction
intermediates, we realize that interfacial water likely plays an important role in both the
energetics and dynamics of the electron transfer processes. Following our earlier work,
160-
162
and many other spectroscopic studies of interfacial water,
163-170
we will monitor water
hydrogen bonding near the electrode surfaces by using the OH-stretch mode (3150-3500
cm
-1
H-bonded) and H/D isotopic substitution.
161
Figure 6.1 illustrates the experimental setup for taking vSFG spectra of the catalyst
surface while immersed in solution in an electrochemical cell. The design is similar to that
used previously by the groups of G. Somorjai and D. Dlott,
154,155
and enables the in situ
application of an overpotential to initiate electron transfer in the reduction of CO2. In order
to avoid absorption of the IR beam by the electrolyte solution (water or ionic liquid), these
experiments will be conducted in the thin layer cell with a 25-50 μm space separating the
top window (CaF2) and the bottom electrode. Figure 6.1c shows an image of a similar
micro-channel reactor used in our previous work.
171
The surface mounted eyelets (metal or
Teflon) enable easy interface with an external micro-fluidic system and the reference
electrode. This design represents a significant improvement over the electrochemical cells
used by Dlott and Somorjai, in which their counter and reference electrodes were
substantially separated from the working electrode, making it difficult to pass current
99
through. Our microfluidic electrochemical cell eliminates this problem by the close
proximity of the working, counter, and reference electrodes. In our microfluidic
electrochemical cell, a 25-50µm channel is first etched in the III-V material (e.g., GaP),
followed by TiO2 deposition by ALD. A Pt strip serving as the counter electrode will be
deposited on the bottom surface of the CaF2 window. A dielectric oxide film (e.g., Al2O3)
will be deposited on the TiO2/GaP substrate in regions away from the photocatalytically
active channel region in order to electrically insulate it from the Pt counter electrode. Prior
to bonding the two wafers together, Teflon eyelets will be installed in the CaF2 window as
illustrated in Figure 6.1.
While SFG will serve as our primary tool for studying these reactions, additional
complementary techniques will also be used, including reflection and total internal
reflection Fourier transform infrared spectroscopy (FTIR) measurements, which is
particularly sensitive in the vibrational frequency range of the OH, CH and CO stretches.
This instrument, in conjunction with SFG spectroscopy, will provide a more complete
understanding of the surface bound intermediates in this photocatalytic reaction system.
Figure 6.1. (a,b) Schematic diagrams illustrating the thin layer electrochemical cell with
CaF
2
TiO
2
on GaP
Reactants Products
Reference
25um
(a) (b)
5mm 5mm
(b) (a)
(a) (b)
5mm 5mm
(a) (b)
5mm 5mm
(a) (b)
5mm 5mm
(a) (b)
5mm 5mm
(a) (b)
5mm 5mm
(c)
counter
electrode
100
in situ SFG spectroscopy, where light is irradiated through the top CaF2 window. (c)
Photograph showing previous micro-channel reactor with glass cover and metal eyelets.
FTIR spectroscopic studies of photoinduced activation of CO2 on neat TiO2 powders
have revealed a number of surface-bound intermediates such as bent and partially
negatively charged CO2 (bands at 1640 cm
-1
and 1219 cm
-1
) and CO.
159
Carbon monoxide
adsorbed on various metal and metal oxide surfaces has been actively studied for more than
two decades. CO adsorbs readily to a number of surfaces, and the consequence for
heterogeneous catalysis is that CO tends to block active sites by blocking adsorption of
reactants and, thus, is well known as a catalyst poison. On the other hand, CO is a common
reaction intermediate in conversions between CO2 and hydrocarbons. Due to its strong
interactions with the substrate, the frequency of the CO stretch vibration is a sensitive
indicator of the nature of the adsorption site; frequency shifts of more than 100 cm
-1
for
CO adsorbed at different sites on the same surface are commonly observed. For example,
FTIR studies of CO adsorption on Au(deposited from solution)/TiO2(powder) observed
several spectral features: 2184 cm
-1
CO on TiO2 powder, 2112 cm
-1
CO on metallic gold,
and 2151 cm
-1
CO on oxidized gold.
172
UHV IRAS studies of CO adsorption on the
Au(110)-(1 × 2) surface performed as a function of CO pressure and sample temperature
showed a single CO peak whose frequency varied from 2108 cm
-1
to 2118 cm
-1
depending
on coverage and temperature.
173
UHV SFG studies of the interaction of CO with Au(111)
revealed a single peak at 2100 cm
-1
for CO atop a Au atom for surfaces modified by ion
bombardment.
174
More dramatic frequency shifts are observed for CO adsorbed on Pd and
101
Pt. On Pd(111), two spectral signatures were seen at 1890 cm
-1
(bridging site) and 2109
cm
-1
(atop site), at 190K.
175
On Pt(111), the bridging site was measured at 1855 cm
-1
, while
the atop site peak was reported between 2083-2100 cm
-1
depending on temperature and
surface coverage.
176,177
Dissolved CO adsorbed on Pt electrodes was also noted to have
similar bridge (1850 cm
-1
) and a top (2100 cm
-1
) site frequencies.
155
102
References:
(1) Bhatkhande, D. S.; Pangarkar, V . G.; Beenackers, A. A. Journal of Chemical
Technology and Biotechnology 2002, 77, 102.
(2) Roy, S. C.; Varghese, O. K.; Paulose, M.; Grimes, C. A. Acs Nano 2010, 4, 1259.
(3) Liu, Z.; Hou, W.; Pavaskar, P.; Aykol, M.; Cronin, S. B. Nano Letters 2011, 11,
1111.
(4) Linic, S.; Christopher, P.; Ingram, D. B. Nature materials 2011, 10, 911.
(5) Kondratenko, E. V .; Mul, G.; Baltrusaitis, J.; Larrazábal, G. O.; Pérez-Ramírez, J.
Energy & environmental science 2013, 6, 3112.
(6) Benson, E. E.; Kubiak, C. P.; Sathrum, A. J.; Smieja, J. M. Chemical Society
Reviews 2009, 38, 89.
(7) Kumar, B.; Llorente, M.; Froehlich, J.; Dang, T.; Sathrum, A.; Kubiak, C. P.
Annual Review of Physical Chemistry 2012, 63, 541.
(8) Amatore, C.; Saveant, J. M. Journal of the American Chemical Society 1981, 103,
5021.
(9) Cokoja, M.; Bruckmeier, C.; Rieger, B.; Herrmann, W. A.; Kühn, F. E.
Angewandte Chemie International Edition 2011, 50, 8510.
(10) Finn, C.; Schnittger, S.; Yellowlees, L. J.; Love, J. B. Chemical Communications
2012, 48, 1392.
(11) Goeppert, A.; Czaun, M.; May, R. B.; Prakash, G. K. S.; Olah, G. A.; Narayanan,
S. R. Journal of the American Chemical Society 2011, 133, 20164.
(12) Halmann, M. 1978.
(13) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979, 277, 637.
(14) Olah, G. A.; Prakash, G. K. S.; Goeppert, A. Journal of the American Chemical
Society 2011, 133, 12881.
(15) D'Alessandro, D. M.; Smit, B.; Long, J. R. Angewandte Chemie International
Edition 2010, 49, 6058.
(16) Appel, A. M.; Bercaw, J. E.; Bocarsly, A. B.; Dobbek, H.; DuBois, D. L.; Dupuis,
M.; Ferry, J. G.; Fujita, E.; Hille, R.; Kenis, P. J. Chemical reviews 2013, 113, 6621.
(17) Rosen, B. A.; Haan, J. L.; Mukherjee, P.; Braunschweig, B.; Zhu, W.; Salehi-
Khojin, A.; Dlott, D. D.; Masel, R. I. The Journal of Physical Chemistry C 2012, 116,
15307.
(18) Rosen, B. A.; Salehi-Khojin, A.; Thorson, M. R.; Zhu, W.; Whipple, D. T.; Kenis,
P. J. A.; Masel, R. I. Science 2011, 334, 643.
(19) Le, M.; Ren, M.; Zhang, Z.; Sprunger, P. T.; Kurtz, R. L.; Flake, J. C. Journal of
The Electrochemical Society 2011, 158, E45.
(20) Hirota, K.; Tryk, D. A.; Yamamoto, T.; Hashimoto, K.; Okawa, M.; Fujishima, A.
The Journal of Physical Chemistry B 1998, 102, 9834.
103
(21) Hirota, K.; Tryk, D. A.; Hashimoto, K.; Okawa, M.; Fujishima, A. Journal of The
Electrochemical Society 1998, 145, L82.
(22) Junfu, L.; Baozhu, C. Journal of Electroanalytical Chemistry 1992, 324, 191.
(23) Kaneco, S.; Katsumata, H.; Suzuki, T.; Ohta, K. Applied Catalysis B:
Environmental 2006, 64, 139.
(24) Kaneco, S.; Ueno, Y .; Katsumata, H.; Suzuki, T.; Ohta, K. Chemical Engineering
Journal 2009, 148, 57.
(25) Barton, E. E.; Rampulla, D. M.; Bocarsly, A. B. Journal of the American Chemical
Society 2008, 130, 6342.
(26) Cole, E. B.; Bocarsly, A. B. Carbon Dioxide as Chemical Feedstock 2010, 291.
(27) Atlam, O.; Barbir, F.; Bezmalinovic, D. International journal of hydrogen energy
2011, 36, 7012.
(28) Delacourt, C.; Ridgway, P. L.; Kerr, J. B.; Newman, J. Journal of The
Electrochemical Society 2008, 155, B42.
(29) Sajjad, A. K. L.; Shamaila, S.; Tian, B.; Chen, F.; Zhang, J. Journal of Hazardous
materials 2010, 177, 781.
(30) Wang, X.; Li, J.-G.; Kamiyama, H.; Moriyoshi, Y .; Ishigaki, T. The Journal of
Physical Chemistry B 2006, 110, 6804.
(31) Takeuchi, M.; Yamashita, H.; Matsuoka, M.; Anpo, M.; Hirao, T.; Itoh, N.;
Iwamoto, N. Catalysis Letters 2000, 67, 135.
(32) Xie, Y .; Zhao, X. Journal of Molecular Catalysis A: Chemical 2008, 285, 142.
(33) Pinaud, B. A.; Benck, J. D.; Seitz, L. C.; Forman, A. J.; Chen, Z.; Deutsch, T. G.;
James, B. D.; Baum, K. N.; Baum, G. N.; Ardo, S. Energy & Environmental Science 2013,
6, 1983.
(34) Hu, S.; Xiang, C.; Haussener, S.; Berger, A. D.; Lewis, N. S. Energy &
Environmental Science 2013, 6, 2984.
(35) Liu, C.; Dasgupta, N. P.; Yang, P. Chemistry of Materials 2013, 26, 415.
(36) Harris, L. A.; Wilson, R. Annual review of materials science 1978, 8, 99.
(37) Frese Jr, K.; Madou, M.; Morrison, S. The Journal of Physical Chemistry 1980,
84, 3172.
(38) Zeng, G.; Qiu, J.; Hou, B.; Shi, H.; Lin, Y .; Hettick, M.; Javey, A.; Cronin, S. B.
Chemistry-A European Journal 2015, 21, 13502.
(39) Qiu, J.; Zeng, G.; Ha, M.-A.; Ge, M.; Lin, Y .; Hettick, M.; Hou, B.; Alexandrova,
A. N.; Javey, A.; Cronin, S. B. Nano letters 2015, 15, 6177.
(40) Zeng, G.; Qiu, J.; Li, Z.; Pavaskar, P.; Cronin, S. B. ACS Catalysis 2014, 4, 3512.
(41) Qiu, J.; Zeng, G.; Pavaskar, P.; Li, Z.; Cronin, S. B. Physical Chemistry Chemical
Physics 2014, 16, 3115.
(42) Medina-Ramos, J.; Pupillo, R. C.; Keane, T. P.; DiMeglio, J. L.; Rosenthal, J.
Journal of the American Chemical Society 2015, 137, 5021.
(43) Medina-Ramos, J.; DiMeglio, J. L.; Rosenthal, J. Journal of the American
Chemical Society 2014, 136, 8361.
104
(44) Lin, Y .; Xu, Y .; Mayer, M. T.; Simpson, Z. I.; McMahon, G.; Zhou, S.; Wang, D.
Journal of the American Chemical Society 2012, 134, 5508.
(45) Chen, Y . W.; Prange, J. D.; Dühnen, S.; Park, Y .; Gunji, M.; Chidsey, C. E. D.;
McIntyre, P. C. Nat Mater 2011, 10, 539.
(46) Le Formal, F.; Tetreault, N.; Cornuz, M.; Moehl, T.; Gratzel, M.; Sivula, K.
Chemical Science 2011, 2, 737.
(47) Aharonshalom, E.; Heller, A. J Electrochem Soc 1982, 129, 2865.
(48) Lee, M. H.; Takei, K.; Zhang, J. J.; Kapadia, R.; Zheng, M.; Chen, Y . Z.; Nah, J.;
Matthews, T. S.; Chueh, Y . L.; Ager, J. W.; Javey, A. Angewandte Chemie-International
Edition 2012, 51, 10760.
(49) Barton, E. E.; Rampulla, D. M.; Bocarsly, A. B. J. Am. Chem. Soc. 2008, 130,
6342.
(50) Cole, E. B.; Lakkaraju, P. S.; Rampulla, D. M.; Morris, A. J.; Abelev, E.; Bocarsly,
A. B. Journal of the American Chemical Society 2010, 132, 11539.
(51) Kedzierzawski, P.; Augustynski, J. J Electrochem Soc 1994, 141, L58.
(52) Anpo, M.; Yamashita, H.; Ichihashi, Y .; Ehara, S. J Electroanal Chem 1995, 396,
21.
(53) Rosen, B. A.; Salehi-Khojin, A.; Thorson, M. R.; Zhu, W.; Whipple, D. T.; Kenis,
P. J. A.; Masel, R. I. Science 2011, 334, 643.
(54) Tanaka, K.; Miyahara, K.; Toyoshima, I. J Phys Chem-Us 1984, 88, 3504.
(55) Le Formal, F.; Pendlebury, S. R.; Cornuz, M.; Tilley, S. D.; Gra ̈ tzel, M.; Durrant,
J. R. Journal of the American Chemical Society 2014, 136, 2564.
(56) Sivula, K. J. Phys. Chem. Lett 2013, 4, 1624.
(57) Brennan, T. P.; Tanskanen, J. T.; Roelofs, K. E.; To, J. W.; Nguyen, W. H.; Bakke,
J. R.; Ding, I.-K.; Hardin, B. E.; Sellinger, A.; McGehee, M. D. The Journal of Physical
Chemistry C 2013, 117, 24138.
(58) Bard, A. J. F., L.R. 2001, 2nd edition.
(59) Izumi, Y . Coordination Chemistry Reviews 2013, 257, 171.
(60) Varghese, O. K.; Paulose, M.; LaTempa, T. J.; Grimes, C. A. Nano Letters 2009,
9, 731.
(61) Halmann, M. Nature 1978, 275, 115.
(62) Chen, Z.; Chen, C.; Weinberg, D. R.; Kang, P.; Concepcion, J. J.; Harrison, D. P.;
Brookhart, M. S.; Meyer, T. J. Chemical Communications 2011, 47, 12607.
(63) Morris, A. J.; Meyer, G. J.; Fujita, E. Accounts of chemical research 2009, 42,
1983.
(64) Jacquet, O.; Frogneux, X.; Das Neves Gomes, C.; Cantat, T. Chemical Science
2013, 4, 2127.
(65) Costentin, C.; Drouet, S.; Robert, M.; Savéant, J.-M. Science 2012, 338, 90.
(66) Costentin, C.; Canales, J. C.; Haddou, B.; Savéant, J.-M. Journal of the American
Chemical Society 2013, 135, 17671.
(67)Canfield, D.; Frese, K. W. Journal of The Electrochemical Society 1983, 130, 1772.
105
(68) Doherty, M. D.; Grills, D. C.; Muckerman, J. T.; Polyansky, D. E.; Fujita, E.
Coordination Chemistry Reviews 2010, 254, 2472.
(69) Savéant, J.-M. Chemical Reviews 2008, 108, 2348.
(70) Olah, G. A.; Goeppert, A.; Prakash, G. K. S. The Journal of Organic Chemistry
2008, 74, 487.
(71) Kumar, B.; Llorente, M.; Froehlich, J.; Dang, T.; Sathrum, A.; Kubiak, C. P. In
Annual Review of Physical Chemistry, Vol 63; Johnson, M. A., Martinez, T. J., Eds. 2012;
V ol. 63, p 541.
(72) Taniguchi, I.; Aurian-Blajeni, B.; Bockris, J. O. M. Electrochimica Acta 1984, 29,
923.
(73) Chandrasekaran, K.; Bockris, J. O. Surf Sci 1987, 185, 495.
(74) Bockris, J. O.; Wass, J. C. J Electrochem Soc 1989, 136, 2521.
(75) Bockris, J. O.; Wass, J. C. Mater Chem Phys 1989, 22, 249.
(76)Seshadri, G.; Lin, C.; Bocarsly, A. B. Journal of Electroanalytical Chemistry 1994,
372, 145.
(77) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979, 277, 637.
(78) Boston, D. J.; Xu, C.; Armstrong, D. W.; MacDonnell, F. M. Journal of the
American Chemical Society 2013, 135, 16252.
(79) Barton Cole, E.; Lakkaraju, P. S.; Rampulla, D. M.; Morris, A. J.; Abelev, E.;
Bocarsly, A. B. Journal of the American Chemical Society 2010, 132, 11539.
(80) Morgan, B. J.; Watson, G. W. The Journal of Physical Chemistry C 2010, 114,
2321.
(81) Qiu, J.; Zeng, G.; Pavaskar, P.; Li, Z.; Cronin, S. B. Physical Chemistry Chemical
Physics 2014, 16, 3115.
(82) Neamen, D. A. McGraw Hill press 2010, 241.
(83) Fulmer, G. R.; Miller, A. J. M.; Sherden, N. H.; Gottlieb, H. E.; Nudelman, A.;
Stoltz, B. M.; Bercaw, J. E.; Goldberg, K. I. Organometallics 2010, 29, 2176.
(84) Yan, Y .; Zeitler, E. L.; Gu, J.; Hu, Y .; Bocarsly, A. B. Journal of the American
Chemical Society 2013, 135, 14020.
(85) Bard, A. J.; Bocarsly, A. B.; Fan, F. R. F.; Walton, E. G.; Wrighton, M. S. Journal
of the American Chemical Society 1980, 102, 3671.
(86) Grätzel, M. Nature 2001, 414, 338.
(87) Lee, M. H.; Takei, K.; Zhang, J.; Kapadia, R.; Zheng, M.; Chen, Y . Z.; Nah, J.;
Matthews, T. S.; Chueh, Y . L.; Ager, J. W. Angewandte Chemie 2012, 124, 10918.
(88) Canfield, D.; Frese Jr, K. Journal of the Electrochemical Society 1983, 130, 1772.
(89) Zeng, G.; Qiu, J.; Li, Z.; Pavaskar, P.; Cronin, S. B. ACS Catalysis 2014.
(90) Hu, S.; Shaner, M. R.; Beardslee, J. A.; Lichterman, M.; Brunschwig, B. S.; Lewis,
N. S. Science 2014, 344, 1005.
(91) Chen, Y . W.; Prange, J. D.; Dühnen, S.; Park, Y .; Gunji, M.; Chidsey, C. E.;
McIntyre, P. C. Nature materials 2011, 10, 539.
(92) Lindan, P.; Harrison, N.; Gillan, M.; White, J. Physical Review B 1997, 55, 15919.
106
(93) Krüger, P.; Bourgeois, S.; Domenichini, B.; Magnan, H.; Chandesris, D.; Le Fevre,
P.; Flank, A.; Jupille, J.; Floreano, L.; Cossaro, A. Physical review letters 2008, 100,
055501.
(94) V on Oertzen, G.; Gerson, A. International journal of quantum chemistry 2006,
106, 2054.
(95) Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.;
Ceresoli, D.; Chiarotti, G. L.; Cococcioni, M.; Dabo, I. Journal of Physics: Condensed
Matter 2009, 21, 395502.
(96) Kohn, W.; Sham, L. J. Physical Review 1965, 140, A1133.
(97) Lee, C.; Yang, W.; Parr, R. G. Physical review B 1988, 37, 785.
(98) Burke, K.; Werschnik, J.; Gross, E. The Journal of chemical physics 2005, 123,
062206.
(99) Perdew, J. P.; Burke, K.; Ernzerhof, M. Physical review letters 1996, 77, 3865.
(100) Howard, C.; Sabine, T.; Dickson, F. Acta Crystallographica Section B:
Structural Science 1991, 47, 462.
(101) Lazzeri, M.; Vittadini, A.; Selloni, A. Physical Review B 2001, 63, 155409.
(102) Vittadini, A.; Selloni, A.; Rotzinger, F.; Grätzel, M. Physical Review Letters
1998, 81, 2954.
(103) Cococcioni, M.; De Gironcoli, S. Physical Review B 2005, 71, 035105.
(104) Finazzi, E.; Di Valentin, C.; Pacchioni, G.; Selloni, A. The Journal of
chemical physics 2008, 129, 154113.
(105) Cheng, H.; Selloni, A. Physical Review B 2009, 79, 092101.
(106) Tang, W.; Sanville, E.; Henkelman, G. Journal of Physics: Condensed Matter
2009, 21, 084204.
(107) Sanville, E.; Kenny, S. D.; Smith, R.; Henkelman, G. Journal of
computational chemistry 2007, 28, 899.
(108) Henkelman, G.; Arnaldsson, A.; Jónsson, H. Computational Materials
Science 2006, 36, 354.
(109) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. Energy &
Environmental Science 2012, 5, 7050.
(110) Li, Y .; Wang, W.-N.; Zhan, Z.; Woo, M.-H.; Wu, C.-Y .; Biswas, P. Applied
Catalysis B: Environmental 2010, 100, 386.
(111) Gattrell, M.; Gupta, N.; Co, A. Journal of Electroanalytical Chemistry 2006,
594, 1.
(112) Chinchen, G. C.; Spencer, M. S.; Waugh, K. C.; Whan, D. A. Journal of the
Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases 1987,
83, 2193.
(113) Hori, Y .; Murata, A.; Takahashi, R. Journal of the Chemical Society, Faraday
Transactions 1: Physical Chemistry in Condensed Phases 1989, 85, 2309.
(114) Halmann, M. Nature 1978, 275, 115.
(115) Zeng, G.; Qiu, J.; Li, Z.; Pavaskar, P.; Cronin, S. B. ACS Catalysis 2014, 3512.
107
(116) Vennestrøm, P. N. R.; Osmundsen, C. M.; Christensen, C. H.; Taarning, E.
Angewandte Chemie International Edition 2011, 50, 10502.
(117) Koh, J. H.; Jeon, H. S.; Jee, M. S.; Nursanto, E. B.; Lee, H.; Hwang, Y . J.;
Min, B. K. The Journal of Physical Chemistry C 2015, 119, 883.
(118) Snuffin, L. L.; Whaley, L. W.; Yu, L. Journal of The Electrochemical Society
2011, 158, F155.
(119) Sun, L.; Ramesha, G. K.; Kamat, P. V .; Brennecke, J. F. Langmuir 2014, 30,
6302.
(120) DiMeglio, J. L.; Rosenthal, J. Journal of the American Chemical Society 2013,
135, 8798.
(121) Yan, H.; Yang, J.; Ma, G.; Wu, G.; Zong, X.; Lei, Z.; Shi, J.; Li, C. Journal of
Catalysis 2009, 266, 165.
(122) Zhai, Q.; Xie, S.; Fan, W.; Zhang, Q.; Wang, Y .; Deng, W.; Wang, Y .
Angewandte Chemie International Edition 2013, 52, 5776.
(123) Xie, S.; Wang, Y .; Zhang, Q.; Deng, W.; Wang, Y . ACS Catalysis 2014, 4,
3644.
(124) Lin, Y .; Kapadia, R.; Yang, J.; Zheng, M.; Chen, K.; Hettick, M.; Yin, X.;
Battaglia, C.; Sharp, I. D.; Ager, J. W.; Javey, A. The Journal of Physical Chemistry C 2015,
119, 2308.
(125) Lichterman, M. F.; Carim, A. I.; McDowell, M. T.; Hu, S.; Gray, H. B.;
Brunschwig, B. S.; Lewis, N. S. Energy & Environmental Science 2014, 7, 3334.
(126) Ramesha, G. K.; Brennecke, J. F.; Kamat, P. V . ACS Catalysis 2014, 4, 3249.
(127) Anpo, M.; Yamashita, H.; Ichihashi, Y .; Ehara, S. Journal of Electroanalytical
Chemistry 1995, 396, 21.
(128) Pan, J. M.; Maschhoff, B.; Diebold, U.; Madey, T. Journal of Vacuum Science
& Technology A 1992, 10, 2470.
(129) Lu, G.; Linsebigler, A.; Yates Jr, J. T. The Journal of Physical Chemistry 1994,
98, 11733.
(130) Meekins, B. H.; Kamat, P. V . ACS Nano 2009, 3, 3437.
(131) Fujishima. A; Honda, K. Nature 1972, 238, 37.
(132) Hou, W.; Liu, Z.; Pavaskar, P.; Hung, W. H.; Cronin, S. B. J. Catal. 2011, 277,
149.
(133) Mills, A.; Davies, R. H.; Worsley, D. Chemical Society Reviews 1993, 22, 417.
(134) Hou, W.; Cronin, S. Advanced Functinoal Materials 2012, 23, 1612.
(135) Shaunak Mukherjee, F. L., Nicolas Large, Oara Neumann, Lisa V . Brown, Jin
Cheng, J. Britt Lassiter, Emily A. Carter, Peter Nordlander, Naomi J. Halas Nano Lett. 2013,
Online ASAP.
(136) Thomann, I.; Pinaud, B. A.; Chen, Z.; Clemens, B. M.; Jaramillo, T. F.;
Brongersma, M. L. Nano Lett. 2011, 11, 3440.
(137) Torimoto, T.; Horibe, H.; Kameyama, T.; Okazaki, K.; Ikeda, S.; Matsumura,
M.; Ishikawa, A.; Ishihara, H. J. Phys. Chem. Lett. 2011, 2, 2057.
108
(138) Liu, Z.; Hou, W.; Pavaskar, P.; Aykol, M.; Cronin, S. Nano Lett. 2011, 11,
1111.
(139) Duan, H. L.; Xuan, Y . M. Physica E-Low-Dimensional Systems &
Nanostructures 2011, 43, 1475.
(140) Liu, Z. W.; Hou, W. B.; Pavaskar, P.; Aykol, M.; Cronin, S. B. Nano Lett. 2011,
11, 1111.
(141) Rosseler, O.; Shankar, M. V .; Du, M. K. L.; Schmidlin, L.; Keller, N.; Keller,
V . J. Catal. 2010, 269, 179.
(142) Silva, C. G.; Juarez, R.; Marino, T.; Molinari, R.; Garcia, H. J. Am. Chem.
Soc. 2011, 133, 595.
(143) Kumar, R.; Zhou, H.; Cronin, S. B. Applied Physics Letters 2007, 91, 223105.
(144) Van Duyne, R. P.; Hulteen, J. C.; Treichel, D. A. Journal of Chemical Physics
1993, 99, 2101.
(145) Hou, W.; Hung, W. H.; Pavaskar, P.; Goeppert, A.; Aykol, M.; Cronin, S. B.
ACS Catalysis 2011, 1, 929.
(146) Pavaskar, P.; Theiss, J.; Cronin, S. B. Opt. Express 2012, 20, 14656.
(147) Pavaskar, P.; Hsu, I. K.; Theiss, J.; Hung, W. H.; Cronin, S. B. J. Appl. Phys.
2013, 113.
(148) Wang, J.; Masel, R. I. Journal of Catalysis 1990, 126, 519.
(149) Chang, S. C.; Jiang, X.; Roth, J. D.; Weaver, M. J. J Phys Chem-Us 1991, 95,
5378.
(150) Jiang, X. D.; Chang, S. C.; Weaver, M. J. J Phys Chem-Us 1991, 95, 7453.
(151) Weaver, M. J.; Chang, S. C.; Leung, L. W. H.; Jiang, X.; Rubel, M.;
Szklarczyk, M.; Zurawski, D.; Wieckowski, A. J Electroanal Chem 1992, 327, 247.
(152) Jiang, X. D.; Weaver, M. J. Surf Sci 1992, 275, 237.
(153) Kizhakevariam, N.; Jiang, X. D.; Weaver, M. J. J Chem Phys 1994, 100, 6750.
(154) Baldelli, S.; Markovic, N.; Ross, P.; Shen, Y . R.; Somorjai, G. J Phys Chem B
1999, 103, 8920.
(155) Lagutchev, A.; Lu, G. Q.; Takeshita, T.; Dlott, D. D.; Wieckowski, A. J Chem
Phys 2006, 125.
(156) Hoffer, S.; Baldelli, S.; Chou, K.; Ross, P.; Somorjai, G. A. J Phys Chem B
2002, 106, 6473.
(157) Chou, K. C.; Kim, J.; Baldelli, S.; Somorjai, G. A. J Electroanal Chem 2003,
554, 253.
(158) Rosen, B. A.; Haan, J. L.; Mukherjee, P.; Braunschweig, B.; Zhu, W.; Salehi-
Khojin, A.; Dlott, D. D.; Masel, R. I. J Phys Chem C 2012, 116, 15307.
(159) Rasko, J.; Solymosi, F. J Phys Chem-Us 1994, 98, 7147.
(160) Bordenyuk, A. N.; Benderskii, A. V . J Chem Phys 2005, 122, 134713.
(161) Stiopkin, I. V .; Weeraman, C.; Pieniazek, P. A.; Shalhout, F. Y .; Skinner, J. L.;
Benderskii, A. V . Nature 2011, 474, 192.
(162) Vinaykin, M.; Benderskii, A. V . Journal of Physical Chemistry Letters 2012,
109
3, 3348.
(163) Du, Q.; Superfine, R.; Freysz, E.; Shen, Y . R. Physical Review Letters 1993,
70, 2313.
(164) Kim, J.; Kim, G.; Cremer, P. S. Langmuir 2001, 17, 7255.
(165) Ostroverkhov, V .; Waychunas, G. A.; Shen, Y . R. Physical Review Letters
2005, 94, 046102.
(166) Richmond, G. L. Chemical Reviews 2002, 102, 2693.
(167) Sovago, M.; Campen, R. K.; Bakker, H. J.; Bonn, M. Chem Phys Lett 2009,
470, 7.
(168) McGuire, J. A.; Shen, Y . R. Science 2006, 313, 1945.
(169) Eftekhari-Bafrooei, A.; Borguet, E. Journal of the American Chemical
Society 2010, 132, 3756.
(170) Eftekhari-Bafrooei, A.; Borguet, E. Journal of the American Chemical
Society 2009, 131, 12034.
(171) Hung, W. H.; Aykol, M.; Valley, D.; Hou, W. B.; Cronin, S. B. Nano Letters
2010, 10, 1314.
(172) Dekkers, M. A. P.; Lippits, M. J.; Nieuwenhuys, B. E. Catal Lett 1998, 56,
195.
(173) Meier, D. C.; Bukhtiyarov, V .; Goodman, A. W. J Phys Chem B 2003, 107,
12668.
(174) Paszti, Z.; Hakkel, O.; Keszthelyi, T.; Berko, A.; Balazs, N.; Bako, I.; Guczi,
L. Langmuir 2010, 26, 16312.
(175) Morkel, M.; Unterhalt, H.; Kluner, T.; Rupprechter, G.; Freund, H. J. Surf Sci
2005, 586, 146.
(176) Schweizer, E.; Persson, B. N. J.; Tushaus, M.; Hoge, D.; Bradshaw, A. M.
Surf Sci 1989, 213, 49.
(177) Roeterdink, W. G.; Berg, O.; Bonn, M. J Chem Phys 2004, 121, 10174.
Abstract (if available)
Abstract
Artificial photosynthesis has become a hot research area since first demonstration of photocatalytic water splitting using TiO₂ in 1972. For photocatalytic materials, III-V semiconductors such as GaP, InP, and GaAs are promising candidates with theoretical rational band gap energies and high maximum photocurrent densities of 9mA/cm², 35mA/cm², and 32mA/cm², respectively. However, photocatlytic corrosion of III-V semiconductors prevents them from being utilized as reliable photocatalysts. During my PhD program, our research group has successfully developed a strategy of using the atomic layer deposition (ALD) passivates the surface of III-V semiconductors with a thin layer of TiO₂ (less than 10nm), which protects them from corrosion. What’s more, this thin layer of TiO₂ enhance overall photoconversion efficiency substantially. ❧ This dissertation will begin with an introduction of the fundamentals of photocatalytical semiconductors, followed by the application of CO₂ reduction. After that, we also discuss about the factors limiting the photocatalytic conversion efficiency. And we discuss the advantages of III-V semiconductors and TiO₂ passivation. In the following chapters, we will report the achievements we have made on the enhanced photocatalytic CO₂ reduction processes. ❧ In Chapter 2, we report photocatalytic CO₂ reduction with water to produce methanol using TiO₂-passivated GaP photocathodes under 532nm wavelength illumination. The TiO₂ layer prevents corrosion of the GaP, as evidenced by atomic force microscopy and photoelectrochemical measurements. Here, the GaP surface is passivated using a thin film of TiO₂ deposited by atomic layer deposition (ALD), which provides a viable, stable photocatalyst without sacrificing photocatalytic efficiency. In addition to providing a stable photocatalytic surface, the TiO₂‐passivation provides substantial enhancement in the photoconversion efficiency through passivation of surface states, which cause non‐radiative carrier recombination. In addition to passivation effects, the TiO₂ deposited by ALD is n‐type due to oxygen vacancies, and forms a pn‐junction with the underlying p‐type GaP photocathode. This creates a built‐in field that assists in the separation of photogenerated electron‐hole pairs, further reducing recombination. This reduction in the surface recombination velocity (SRV) corresponds to a shift in the overpotential of almost 0.5V. No enhancement is observed for TiO₂ thicknesses above 10nm, due to the insulating nature of the TiO₂, which eventually outweighs the benefits of passivation. ❧ In Chapter 3, photocatalytic CO₂ reduction with water to produce methanol is demonstrated using TiO₂‐passivated InP nanopillar photocathodes under 532nm wavelength illumination. In addition to providing a stable photocatalytic surface, the TiO₂‐passivation layer provides substantial enhancement in the photoconversion efficiency through the introduction of O vacancies associated with the non‐stoichiometric growth of TiO₂ by atomic layer deposition. Plane wave-density functional theory (PW‐DFT) calculations confirm the role of oxygen vacancies in the TiO₂ surface, which serve as catalytically active sites in the CO₂ reduction process. PW‐DFT shows that CO₂ binds stably to these oxygen vacancies and CO₂ gains an electron (-0.897e) spontaneously from the TiO₂ support. This calculation indicates that the O vacancies provide active sites for CO₂ absorption, and no overpotential is required to form the CO₂⁻ intermediate. ❧ In Chapter 4, we present a robust and reliable method for improving the photocatalytic performance of InP, which is one of the best known materials for solar photoconversion (i.e., solar cells). In this article, we report substantial improvements (up to 18X) in the photocatalytic yields for CO₂ reduction to CO through the surface passivation of InP with TiO₂ deposited by atomic layer deposition (ALD). Here, the main mechanisms of enhancement are the introduction of catalytically active sites and the formation of a pn‐junction. Photoelectrochemical reactions were carried out in a non‐aqueous solution consisting of ionic liquid (1‐ethyl‐3‐methylimidazolium tetrafluoroborate ([EMIM]BF₄)) dissolved in acetontrile, which enables CO₂ reduction with a Faradaic efficiency of 99% at an underpotential of +0.78V. While the photocatalytic yield increases with the addition of the TiO₂ layer, a corresponding drop in the photoluminescence intensity indicates the presence of catalytically active sites, which cause in increase in the electron‐hole pair recombination rate. NMR spectra show that the [EMIM]⁺ ions in solution form an intermediate complex with CO₂⁻, thus lowering the energy barrier of this reaction. ❧ In Chapter 5, we demonstrate that a thin layer of n‐type TiO₂ using atomic layer deposition (ALD) prevents corrosion of p‐type GaP, as evidenced by atomic force microscopy and photoelectrochemical measurements. In addition, the TiO₂ passivation layer provides an enhancement in photoconversion efficiency through the formation of a charge separating pn‐region. Plasmonic Au nanoparticles deposited on top of the TiO₂‐passivated GaP further increases the photoconversion efficiency through local field enhancement. Finite difference time domain (FDTD) simulations of the electric field profiles in this photocatalytic heterostructure corroborate the experimental results. ❧ In Chapter 6, in order to separate the various mechanisms of the catalysis and enhancement, we will use the use vibrational sum frequency generation (vSFG) spectroscopy to identify the reactant and intermediate species adsorbed at the active surface sites on the photocatalytic substrate. Also, we will use the reflection and total internal reflection Fourier transform infrared spectroscopy (FTIR) measurements to analyze the vibrational frequency range of the OH, CH and CO stretches. Those two measurements will provide a more complete understanding of the surface bound intermediates in this photocatalytic reaction system.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Enhanced photocatalysis on titanium oxide passivated III-V semiconductors
PDF
Surface plasmon resonant enhancement of photocatalysis
PDF
Hot carriers in bare metals and photocatalytically active defect sites in dielectric/metal structures
PDF
Electromagnetic modeling of plasmonic nanostructures
PDF
Plasma emission across metal nanoparticle surfaces and semiconductor -liquid interface characterizations
PDF
Plasmonic enhancement of catalysis and solar energy conversion
PDF
Monitoring local electric fields, charge densities at electrode surfaces using in-situ graphene/surface enhanced Raman spectroscopy (GERS/SERS) based Stark-shifts
PDF
Understanding organic and inorganic semiconductor surfaces by coherent nonlinear spectroscopy
PDF
Fabrication and application of plasmonic nanostructures: a story of nano-fingers
PDF
Semiconductor inks for solution processed electronic thin‐films
PDF
Expanding the library of surface ligands for semiconductor nanocrystal synthesis and photovoltaic applications
PDF
Optoelectronic, thermoelectric, and photocatalytic properties of low dimensional materials
PDF
Optical, mechanical, and electrical properties of nano-structured materials
PDF
Design and modification of electrocatalysts for use in fuel cells and CO₂ reduction
PDF
In-situ characterization of nanoscale opto-electronic devices through optical spectroscopy and electron microscopy
PDF
Hot electron driven photocatalysis and plasmonic sensing using metallic gratings
PDF
The modification of catalysts and their supports for use in various fuel cells
PDF
Nanomaterials for macroelectronics and energy storage device
PDF
Multilayer grown ultrathin nanostructured GaAs solar cells towards high-efficiency, cost-competitive III-V photovoltaics
PDF
Design and characterization of metal and semiconducting nanostructures and nanodevices
Asset Metadata
Creator
Zeng, Guangtong
(author)
Core Title
Artificial photosynthesis on titanium oxide passivated III-V semiconductors
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Chemistry
Publication Date
02/09/2016
Defense Date
12/07/2015
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
artificial photosynthesis,carbon dioxide,III-V semiconductors,OAI-PMH Harvest,solar fuel,titanium oxide
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Cronin, Stephen B. (
committee chair
), Benderskii, Alexander (
committee member
), Wu, Wei (
committee member
)
Creator Email
guangtoz@usc.edu,zengguangtong@outlook.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c40-208148
Unique identifier
UC11277207
Identifier
etd-ZengGuangt-4095.pdf (filename),usctheses-c40-208148 (legacy record id)
Legacy Identifier
etd-ZengGuangt-4095.pdf
Dmrecord
208148
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Zeng, Guangtong
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
artificial photosynthesis
carbon dioxide
III-V semiconductors
solar fuel
titanium oxide