Post on 22-Apr-2022
The Pennsylvania State University
The Graduate School
OLIGOMERIC RUTHENIUM LIGHT ABSORBERS AND
NANOSTRUCTURED LENSES FOR IMPROVED HARVESTING OF SOLAR
ENERGY
A Dissertation in
Chemistry
by
Christopher L. Gray
© 2019 Christopher Gray
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Doctor of Philosophy
December 2019
ii
The dissertation of Christopher L. Gray was reviewed and approved* by the following:
Thomas E. Mallouk
Evan Pugh University Professor of Chemistry
Professor of Biochemistry and Molecular Biology
Professor of Physics, and Engineering Science and Mechanics EDIT
Eberly Family Distinguished Chair in Chemistry
Dissertation Advisor
Chair of Committee
John B. Asbury
Professor of Chemistry
Raymond Schaak
DuPont Professor of Materials Chemistry
Professor of Chemistry
Akhlesh Lakhtakia
Evan Pugh University Professor and Charles Godfrey Binder Professor of
Engineering Science and Mechanics
Philip C. Bevilacqua
Head of the Department or Chair of the Graduate Program
Professor of Chemistry
Professor of Biochemistry and Molecular Biology
*Signatures are on file in the Graduate School
iii
Abstract
Nanomaterials and organic synthesis can be used to harvest light energy from the
sun and replace fossil fuels. Water-splitting dye-sensitized photoelectrochemical cells
(WS-DSPECs) can be used to convert light energy into renewable fuels. Current water-
splitting solar cells are limited by the stability of the light-absorbing dye molecules. Water-
splitting dye-sensitized photoelectrochemical cells utilize high surface area semiconductor
electrodes sensitized with a molecular light absorber and catalyst in order to drive
photoelectrochemical water oxidation. Electron transfer occurs from the photoexcited dye
to the electrode material while holes are laterally transferred across the surface until
reaching a water oxidation catalyst. A novel oligomeric ruthenium bipyridine dye has been
sensitized that is attached to the electrode through phosphonate groups. The oligomer was
designed to mitigate desorption of the dye and increases lateral hole transport. These two
effects are demonstrated by electrochemical hole diffusion studies. All of these studies are
compared to electrodes sensitized by the traditionally-used molecular ruthenium bipyridine
complex, which is essentially a monomeric analogue of the oligomer dye. Nanostructured
lenses are explored as a means to separate light spatially to optimize the absorption in
tandem solar cells. In chapter 4, strategies for the synthesis of ruthenium-based dyes and
catalysts are explored.
iv
TABLE OF CONTENTS
LIST OF FIGURES ..................................................................................................... vi
LIST OF TABLES ....................................................................................................... x
ACKNOWLEDGEMENTS ......................................................................................... xi
Chapter 1 Introduction ................................................................................................. 1
Global Warming and Current Energy Usage ........................................................ 1
Natural Photosynthesis ......................................................................................... 2
Artificial Photosynthesis for the Generation of Solar Fuels ................................. 3
Integrated Water-Splitting Dye-Sensitized Photoelectrochemical cells .............. 4
Chapter 2 Oligomeric Ruthenium Dye for Improved Stability of Water-Splitting
Dye-Sensitized Photoelectrochemical Cells ......................................................... 7
Abstract ................................................................................................................. 7
Background on Water-Splitting Dye-Sensitized Photoelectrochemical Cells ...... 8
Experimental ......................................................................................................... 10
Results and Discussion ......................................................................................... 15
Conclusions........................................................................................................... 29
Acknowledgements ............................................................................................... 29
Chapter 3 Nano-structured Lenses for Solar Spectrum Splitting................................ 30
Abstract ................................................................................................................. 30
Background ........................................................................................................... 30
Tandem Solar Cells ....................................................................................... 30
Mirror Spectrum Splitting Tandem Cell ........................................................ 33
Optical Measurements of Spectrum Splitting Lenses ................................... 34
Early Lens Fabrication .......................................................................................... 40
Hemisphere PMMA Lenses .......................................................................... 40
Spin Coating Sphere Layers for a Planar Lens Prototype ............................. 42
Experimental .......................................................................................... 42
Preliminary Sol-Gel Back-Filled Lenses and Multi-Layered Lenses ............ 43
Experimental .......................................................................................... 44
Results .................................................................................................... 45
Collaboration on Spectrum Splitting Grating Simulations ................................... 49
Advanced Planar Lens Fabrication ....................................................................... 52
Improved Sphere Monolayer via Langmuir Blodgett Deposition ................. 55
Experimental .......................................................................................... 56
Liquid Phase Deposition of TiO2 Matrix ....................................................... 58
v
Experimental .......................................................................................... 58
Single Layer Planar Prototype Lens with Rough Overlayer ......................... 62
Alumina Back-Filled Lenses ......................................................................... 65
Final Optical Results ..................................................................................... 67
Future Directions .................................................................................................. 68
Chemical Vapor Deposition for Spectrum Splitting Applications ................ 68 Future Simulations ................................................................................................... 69
Conclusions ...................................................................................................................... 71
Chapter 4 Synthesis of Ruthenium Polypyridyl Compounds for WS-DSPECs ................... 74
Abstract ................................................................................................................. 74
Molecular Ruthenium Polypyridyl OER Catalyst ................................................ 75 Background .............................................................................................................. 75 Results .......................................................................................................... 76
Future Directions ........................................................................................... 79
Hydroxamate-Anchored Ruthenium Polypyridyl Light Absorber ....................... 79
Previous Attempts to Synthesize Hydroxamate-Anchored Sensitizers ......... 79
Synthesis of Hydroxamate Anchored Sensitizer ........................................... 81
Results ........................................................................................................... 83
Future Directions: Hydroxamate-Anchored Porphyrin Sensitizers ............... 84
Conclusions........................................................................................................... 85
Chapter 5 Conclusions ................................................................................................. 86
References ................................................................................................................... 90
Appendix A Intensity Modulated Photovoltage Details ............................................. 104
Appendix B Additional Characterization Data for the Oligomeric Dye ..................... 108
Appendix C Time-Resolved Emission of Oligomer Electrodes ................................. 112
Appendix D 1H NMR Spectra for Molecular Catalyst .............................................. 117
Appendix E 1H NMR Spectra for Hydroxamate Dye ................................................. 119
vi
LIST OF FIGURES
Figure 1-1: Simplified electron transfer scheme of a WS-DSPEC .............................. 4
Figure 2-1: Structure of oligomeric sensitizer ([RuP]n). ............................................. 9
Figure 2-2: Synthesis of oligomeric [RuP]n.. .............................................................. 11
Figure 2-3: Cyclic voltammetry of TiO2 films sensitized with monomer and
oligomer ................................................................................................................ 17
Figure 2-4: Absorbance (dashed lines) of dyes in methanol and steady-state
luminescence (solid lines) of monomer (black) and oligomer (red) on ZrO2
powder in aqueous 0.1 M HClO4 solution.. .......................................................... 20
Figure 2-5: Steady state emission of monomeric RuP (black) and oligomeric
[RuP]n (red) sensitizers adsorbed to ZrO2 powder and to TiO2 electrodes. ......... 21
Figure 2-6: Changes in the absorption spectrum versus time of [RuP]n (red) and
RuP (black) adsorbed on TiO2 electrodes. ........................................................... 23
Figure 2-7: A representative set of spectroelectrochemical absorbance-time data
used to calculate Dapp values for [RuP]n (red) and RuP (black). ......................... 25
Figure 2-8: Plots at pH 4.8 of recombination rate (kIMVS) of the Monomer and
Oligomer as a function of open-circuit photovoltage (Voc).. ................................ 27
Figure 2-9: Combined plot of recombination rate (kIMVS) as a function of open-
circuit photovoltage (Voc) for oligomer samples at various pH ranging from
4.8 to 6.8. .............................................................................................................. 28
Figure 3-1: A) A) Traditional Stacked Tandem Solar Cell. B) Mirrored Spectrum
Splitting Tandem Solar Cell.1. .............................................................................. 32
Figure 3-2: A) Target final hemispherical Structure. B) Planar spectrum splitting
prototype lens. ....................................................................................................... 34
Figure 3-3: Two methods of evaluating scattered light with the integrating sphere
attachment of a UV-Vis Spectrometer.. ................................................................ 35
Figure 3-4: Photograph (A) and diagram (B) of Lakhtakia Lab Spectroscope with
rotating CCD detector fixed at 7 inches. .............................................................. 38
vii
Figure 3-5: Photograph (A) and diagram (B) of Mallouk Lab spectroscope with
rotating detector. Note the adjustable detector distance that can range from 1-
4 inches. ................................................................................................................ 39
Figure 3-6: Photograph of hemispherical lenses created from 500 nm (left) and
1000nm (right) SiO2 spheres suspended in poly(methyl-methacyrlate) ............... 41
Figure 3-7: Light microscope images of spin-coated sphere layers spun at 2000
RPM (A) and 3500 RPM (B). ............................................................................... 43
Figure 3-8: Cross-section SEM Micrograph of spin-coated SiO2 spheres
surrounded by 15% TiO2/SiO2 spin-coated host matrix ....................................... 45
Figure 3-9: Plots of Transmission and Back-Scattering after deposition of
successive sphere layers. ...................................................................................... 46
Figure 3-10: Transmission spectra at detector angles of 0°, 5°, and 10°– as a
function of sphere layers ....................................................................................... 48
Figure 3-11: Structures simulated by Fan and coworkers.2 ........................................ 49
Figure 3-12: A) ALD back-filled samples with reduced titanium centers. B)
Oxidized films after annealing in a box furnace. .................................................. 54
Figure 3-13: SEM cross-section analysis of lens back-filled by ALD. ....................... 55
Figure 3-14: Sphere layers deposited by LB trough method. ...................................... 57
Figure 3-15: SEM Micrographs of LPD back-filled spheres at 40 °C. ........................ 60
Figure 3-16: SEM Micrographs of LPD back-filled spheres at 50 °C. ........................ 61
Figure 3-17: SEM Micrographs of LPD back-filled spheres at 70 °C. ........................ 62
Figure 3-18: SEM micrographs of rough surface overlayer of planar lenses back-
filled with extended ALD. .................................................................................... 64
Figure 3-19: SEM micrograph of milled cross section of planar lenses back-filled
with extended ALD. .............................................................................................. 65
Figure 3-20: LB deposited SiO2 spheres back-filled with Al2O3 via ALD. ................. 66
Figure 3-21: Transmission of sphere lenses embedded in TiO2 (A) and in Al2O3
(B). ........................................................................................................................ 68
viii
Figure 4-1: A) New class of Ru(bda)L2 catalysts. B) Structure of 2,2′‐bipyridine‐
6,6′‐dicarboxylate (bda) ligand.. ........................................................................... 76
Figure 4-2: Synthetic Scheme for silanol anchor ligand. ............................................. 78
Figure 4-3: Synthetic Scheme for Ru(bda) catalyst. .................................................... 78
Figure 4-4: Desired synthesis of hydroxamate anchored Ru(bpy)3 derivative ............ 80
Figure 4-5: General scheme for the activation of carboxylic acids for nucleophilic
attack by forming an anhydride intermediate ....................................................... 82
Figure 4-6: Synthetic scheme for the synthesis of a hydroxamate ligand first, then
the coordination reaction to form a hydroxamate anchored dye .......................... 82
Figure 4-7: Scheme for synthesizing a hydroxamate dye from the analogous
carboxylic dye. ...................................................................................................... 83
Figure A-1: IMVS Nyquist plots of the measured oligomeric sample at pH 4.8. ....... 106
Figure A-2: IMVS Nyquist plots of the measured sample for oligomeric samples
at various pHs. ...................................................................................................... 107
Figure B-1: MALDI-TOF Mass spectrum of oligomer ............................................... 108
Figure B-2: 1H NMR Spectrum of oligomeric [RuP]n. ............................................... 110
Figure B-3: 1H NMR Spectrum of linker ligand .......................................................... 111
Figure C-1: Photograph of non-injecting TaO2 core-shell control electrodes. ............ 114
Figure C-2: Emission decays fit to a decaying exponential ........................................ 115
Figure C-3: Extracted excited state lifetimes (A) and average transient peak
values (B) from the single exponential fits of the emission data .......................... 116
Figure D-1: 1H NMR Spectrum for the blue dimer decomposition product ................ 117
Figure D-2: 1H NMR Spectrum for the pure asymmetric pre-catalyst Ru(bda)(Me-
pyr)(DMSO) ......................................................................................................... 118
Figure E-1: 1H NMR Spectrum for the attempted synthesis of a hydroxamate
bipyridine .............................................................................................................. 119
ix
Figure E-2: 1H NMR Spectrum for the attempted isolation of the anhydride
intermediate .......................................................................................................... 120
x
LIST OF TABLES
Table 2-1: Estimated ground and excited-state redox properties of RuP and
[RuP]n on electrodes ............................................................................................ 18
Table 3-1: Material refractive indices used in simulations by Fan and coworkers.2 ... 50
Table 3-2: Optimized structure parameters from simulations of spheres interacting
with 6° and 15° incident light.2 ............................................................................. 52
Table 3-3: Various high refractive index metal oxides ................................................ 71
Table B-1: Absorbances of samples from sequential oligomer and monomer
adsorption ............................................................................................................. 109
Table C-1: ALD Deposition Parameters and Resulting Shell Thickness .................... 113
xi
ACKNOWLEDGEMENTS
I must first thank my advisor, Tom Mallouk, for his guidance and for allowing me
the opportunity to explore my research interests. There isn’t enough space here to list the
things I’ve learned from Tom, but I’m eternally grateful for the scientific knowledge and
general wisdom he has passed along. I’d like to thank my committee, Akhlesh Lakhtakia,
John Asbury, and Ray Schaak. They have always been willing to help when I needed it.
I’d also like to thank Greg Barber. He showed me the ropes when I joined Tom’s lab and
he’s been a great mentor over the years.
I must also thank Trevor Clark and Bangzhi Liu, who work in the Millennium
Science Center and have taught me everything I know about SEM and TEM. Bangzhi
also went to great lengths to accommodate the unusual ALD work I’ve done and to repair
the instruments when my samples have clogged the lines. I’m also indebted to Dr. Bo
Wang for synthetic advice throughout my PhD career.
I would like to thank all of my colleagues in the Mallouk research group for all of
the small things they’ve done on a day-to-day basis to make coming to work an enjoyable
experience, especially John Swierk, Pengtao Xu, Nella Vargas-Barbosa, Megan Strayer,
and Nick McCool. I’m excited to see what future scientific work you all accomplish. I’ll
also greatly miss the group hot pot celebrations.
I’m grateful for the friends I’ve made while at Penn State who helped me
maintain a balanced life – particularly Brian Conway, Matt Bauerle, Jimmy Morse, and
Juan Callejas.
xii
I must thank my parents for teaching me to value education and for understanding
as I pursue a career in science so far from home.
Last, but not least, I would like to thank Erin McCarthy for supporting me over
the years and standing by my side at the best and worst of times. Thank you for the many
things have done to support me in this long journey. I look forward to the next stage in
our lives.
1
Chapter 1
Introduction
Global Warming and Current Energy Usage
Over the next several decades, energy requirements for the human population are
expected to double as a result of population and economic growth.3,4 Global warming as a
result of human combustion of fossil fuels is a global problem that must be addressed in
the next century.3–5 In 2018, the United states energy usage consisted primarily of fossil
fuels. Renewable sources of energy were mostly limited to the generation of electricity and
the generation of electricity only accounts for approximately 40% of total energy
consumption by the United States.6 The majority of energy consumption is in the form of
fuels. Solar energy is a particularly attractive source for renewable energy because of its
vast abundance. Despite this, renewable solar energy (photovoltaic and thermal)
contributed less than 1% of the total electricity produced in the United States in 2018.7 Due
to the intermittent nature of these renewable energy sources they have only displaced fossil
fuels used to generate electricity. Conversion and storage of renewable sources remains a
significant challenge to becoming completely independent of fossil fuels. Even if we can
replace CO2 emitting sources of energy with renewable energy sources, there will still be
a need for energy-dense fuels. This fact is particularly true in the transportation industries
where energy applications are sensitive to weight. Gasoline is far more energy dense than
the best batteries available. On average batteries store 200 watt-hours per kilogram of
2
material. In contrast, gasoline’s chemical bonds store 12,000 watt-hours per kilogram. As
a result of energy density, it is estimated that approximately 40% of the current global
transportation cannot be replaced by electric vehicles.3 For example, an electric passenger
airplane is unlikely, and it is probable that they will always rely on an energy dense fuel.
This is a significant issue because airplanes are responsible for generating 2.5% of global
CO2 emissions.8 There are many approaches to storing renewable energy in chemical bonds
that can be used as fuels. The reduction of protons to hydrogen gas is often studied because
of the mechanistic ease of this two electron transfer reaction. Reduction of CO2 to various
carbon-based fuels involves the transfer of more electrons and is far more complicated
mechanistically.9 This process of absorbing light and storing the energy in chemical bonds
is broadly called artificial photosynthesis. The study of artificial photosynthesis is broadly
interdisciplinary, consisting of branches of chemistry, biochemistry, physics, materials
science, engineering and others. Common research areas include light absorption, charge
separation, and catalysis.4
Natural Photosynthesis
For millennia, plants have been successfully absorbing light and storing the energy
in the chemical bonds of carbohydrates. Human attempts to engineer devices capable of
artificial photosynthesis have drawn inspiration from the scientific discoveries of how
plants achieve natural photosynthesis. The reactions of photosynthesis occur in
photosystem I and II in the chlorophyll membranes. Photosystem II is the site of the water
splitting reaction. Chlorophyll absorbs the light energy and transfers electrons to generate
3
oxidizing equivalents to the oxygen-evolving complex. The oxygen evolving complex is a
metallo-oxo cluster comprised of four manganese atoms and a calcium atom at the center
of photosystem II. The turnover frequency (TOF) of the oxygen-evolving complex is an
impressive 100 s-1.4,10 In these biological photosynthetic systems, the proteins responsible
for photosynthesis require constant repair due to the damaging effects of light and the
reactive oxygen species involved in water splitting. In some organisms, the D1 protein of
photosystem II is synthesized and replaced as frequently as every 30 mins.5,11 Human
attempts to create more efficient or more stable devices often replace fragile biological
molecules with inorganic materials which perform the same functions but are more robust
over device lifecycles.
Artificial Photosynthesis for the Generation of Solar Fuels
There are many pathways to artificially achieve photosynthesis. The most
straightforward is to power an electrolysis cell with the renewable electricity from a
photovoltaic cell. In the past, when PV energy was more expensive, large-scale fuel
processing techniques for producing hydrogen gas were more economical than
electrolysis.12 These fuel processing techniques convert carbon-based fuels (CH3...etc.)
into hydrogen gas, but overall, they are not renewable because CO2 is a byproduct of these
reactions. The brute force approach to generating a renewable hydrogen fuel is to use four
silicon photovoltaic cells in series to generate the electricity to power a commercial water
electrolyzer. These brute-force setups usually display solar-to-chemical conversion
efficiencies of approximately 7%.13 In these setups, the function of harvesting light energy
4
is separated from the module that performs catalysis.14 Alternatively, there are integrated
approaches which combine the light absorber and catalyst within the same module on the
same electrode support. This thesis is focused on the integrated approach.
Integrated Water-Splitting Dye-Sensitized Photoelectrochemical cells
One potential method for renewable generation of hydrogen is through the solar-
driven electrochemical splitting of water. Water-splitting dye-sensitized photo-
electrochemical cells (WS-DSPECs) employ molecular sensitizers to harvest the energy
needed to split water into hydrogen and oxygen. The basic electron transfer scheme of a
WS-DSPEC is shown in Figure 1-1.
Figure 1-1: Simplified electron transfer scheme of a WS-DSPEC. Upon light
absorption by a surface-bound sensitizer, a photoexcited electron is injected into the
5
Upon light absorption by a surface-bound sensitizer, a photoexcited electron is injected
into the conduction band of an oxide semiconductor such as TiO2 or SnO2. The photo-
generated electrons are transported through an external circuit to a cathode where they can
be used to reduce water and form hydrogen gas. As electrons are transported through the
semiconductor to the electrode back contact, hole diffusion occurs along the surface
between dye molecules in order to bring oxidizing equivalents to oxygen-evolving catalyst
molecules or nanoparticles. WS-DSPECs are essentially aqueous dye-sensitized solar cells
(DSSCs) in which the anode is coupled to a water oxidation catalyst. However, while
DSSCs have been optimized extensively and can reach power conversion efficiencies
above 15%, WS-DSPEC have struggled to break 1% efficiency.4 The kinetically
demanding water oxidation reaction is slow, making electron-hole recombination the
dominant kinetic process, and the aqueous electrolyte introduces stability issues that are
much less pronounced in DSSCs. In order to generate one oxygen molecule from water,
four oxidizing equivalents (holes) must be transferred to the catalyst from excited dye
molecules. Depending on the architecture of the photoanode (porosity, dye loading,
catalyst loading, etc), only dye molecules within a certain distance of catalytic sites can
transfer oxidizing equivalents to the catalyst before they are lost to recombination. Dye-
sensitized solar cells (DSSCs) can also utilize a broad range of dyes that can be optimized
for light absorption and power generation. In contrast, WS-DSPECs are run under aqueous
conditions, limiting the choice of dyes to those that have sufficient oxidizing power to
conduction band of an oxide semiconductor. Electrons are driven through an external
circuit to a cathode where they reduce water to hydrogen.
6
generate oxygen from water, and requiring surface anchoring groups that resist hydrolysis.
Previous studies of WS-DSPECs have primarily utilized a derivative of tris(2,2’-
bipyridine)ruthenium(II) with a pair of phosphonate anchoring groups at the 4,4’ position
of one bipyridine ligand, [Ru(bpy)2(4,4-PO3H2)2bpy)]2+ (RuP),4,15,16 although some
experiments have also been carried out with porphyrin17 and perylene diimide18 sensitizers.
Despite the superior stability of the phosphonate anchoring group, dye desorption from the
electrode surface is widely recognized as factor that compromises the stability of WS-
DSPECs.4,15,19
7
Chapter 2
An Oligomeric Ruthenium Polypyridyl Dye for Improved Stability of Aqueous
Photoelectrochemical Cells
Abstract
Water-splitting dye-sensitized photoelectrochemical cells (WS-DSPECs) rely on
molecular sensitizers to harvest light energy and drive the catalytic reactions necessary to
generate hydrogen and oxygen from water. The desorption of sensitizer molecules from
the semiconductor-aqueous electrolyte interface is a significant barrier to the practical
implementation of these cells. To address this problem, we synthesized a novel oligomeric
ruthenium dye ([RuP]n) which has dramatically improved stability as a photosensitizer for
TiO2 electrodes over the pH range of interest (4 to 7.8) for DSPECs. Additionally, the
efficiency of photoelectrochemical charge separation is known to depend on the rate of
cross-surface hole diffusion between dye molecules.20,21 The oligomeric dye ([RuP]n) also
shows an order of magnitude faster cross-surface hole diffusion than the commonly used
monomeric [Ru(bpy)2(4,4-PO3H2)2bpy)]2+ (RuP) sensitizer. The enhanced stability of the
polymeric dye also enables the use of intensity-modulated photovoltage spectroscopy
(IMVS) to measure the recombination rate of photogenerated electrons and holes as a
function of electrolyte pH.
8
Background on Water-Splitting Dye-Sensitized Photoelectrochemical Cells
Over the next several decades, energy requirements for the human population are
expected to double as a result of population and economic growth.3,4 The storage of
renewable energy in the form of hydrogen and liquid fuels will be an important factor in
preventing the rise of atmospheric carbon dioxide emissions. Much of the literature
discusses these cells within the simplification of one-dimensional electron transfer without
the nuances caused by three-dimensional surface electron transfer effects. Ruthenium
centers that are distant from catalytic sites can contribute oxidizing equivalents to the
catalyst through a cross surface hole hopping mechanism between ruthenium centers.
Recent work has identified this cross-surface hole diffusion as one of the limiting factors
in catalytic turnover of these cells.20–22 In particular, it has been shown that each catalytic
site effectively draws holes from the dye on the same 20 nm nanoparticle, while most of
the dye coated surface does not contribute to the catalytic cycle.20 To improve the surface
electro-activity, it is necessary to create dyes that have higher hole diffusion constants.
Many groups have attempted to exploit the increased stability of polymers to
improve DSSCs23 and WS-DSPECs. Research groups have attempted deposition of
polymeric dyes24 or polymerized dyes after deposition25–27 or encased surface adsorbed
dyes in a polymer overlayer28,29 or metal oxide shell.16,30–32 Most of these attempts increase
stability with some kind of detrimental trade off. The hole diffusion constant has been
shown to be largely determined by the distance between ruthenium centers. With our
oligomeric dye, the light-absorbing metal centers are covalently linked before deposition
and thereby guaranteed to be a certain distance from each other on the surface, allowing
9
for holes to be transferred quicker between individual dye complexes. This should result
in an improvement in the hole diffusion constant. The phosphonate anchoring group trades
increased stability for decreased efficiency of electron injection into the TiO2. Recently,
injection efficiency has been identified as a significant bottleneck in these cells.19,33 Many
groups have searched for a water stable dye that also has a high injection efficiency. The
most promising of these are hydroxamate anchors.4,19 Although groups have demonstrated
organic absorbers with hydroxamate anchors34, they have failed to create Ru(bpy)3
analogues35,36. The search for a stable dye that strongly injects is ongoing. Currently, RuP
represents one of the best balances of injection efficiency and stability.
Figure 2-1: Structure of oligomeric sensitizer ([RuP]n)
10
A novel phosphonated oligomeric ruthenium dye ([RuP]n) has been synthesized,
which has dramatically improved electrode adhesion from pH 4 to pH 7.8. We measured
the long-term stability of the [RuP]n at higher pHs that are traditionally incompatible with
WS-DSPECs. We also determined the apparent hole diffusion constant of the oligomer
dye. These two dye properties have received recent attention as key limiting factors in the
efficiency of WS-DSPECs. The stability over a broad range of pH has allowed us to
measure the recombination rate at higher range of pHs for the first time. In addition, the
oligomer attains injection yields similar to that of the monomeric RuP sensitizer. These
results demonstrate that oligomeric dyes are an effective strategy in designing high-
stability chromophores for WS-DSPECs.
Experimental
The 4,4'-Bis(diethylphosphonate)-2,2'-bipyridine ligand, (4,4-PO3Et2)2bpy, was
purchased from Carbosynth and used without further purification. Bis(2,2-bipyridine)(4,4-
diphosphonato-2,2-bipyridine)-ruthenium bromide, [Ru(bpy)2(4,4-PO3H2)2bpy)]2+ (RuP),
was prepared according to literature methods.37
Synthesis of Oligomeric Dye ([RuP]n). Dichlorotetrakis(dimethyl
sulfoxide)ruthenium(II)38 and 1,5-bis-(4-methyl-2,2-bipyridyl-4-yl)pentane were prepared
according to literature methods39,40 and the latter was purified by recrystallization from hot
t-butyl methyl ketone. A 0.20 g (0.40 mmol) portion of Ru(DMSO)4Cl2 and 0.17 g (0.40
mmol) of 1,5-bis(4-methyl-2,2-bipyridyl-4-yl)pentane were combined with 50 mL of
chloroform and refluxed under argon for 1.5 h. The solvent was removed under reduced
11
pressure to yield a dark brown oil. The oil was dissolved in a mixture of 10 mL of H2O/15
mL of ethanol. A 0.10 g (0.64 mmol) portion of (4,4-PO3Et2)2bpy was added, and the
solution was refluxed for 2.5 h. A clear, dark-red solution resulted. The solvent was
reduced to about 10 mL under reduced pressure, and the product was precipitated by
addition of aqueous ammonium hexafluorophosphate. The precipitate was filtered and
washed with H2O and diethyl ether to yield a dark red powder (0.29 g, 78%).
Figure 2-2: Synthesis of oligomeric [RuP]n
12
In the same manner as the monomer dye (RuP), the ethyl ester groups were hydrolyzed by
reacting with excess TMS-Br.37 As has been done for similar oligomer complexes40, the
degree of polymerization was determined from 1H NMR experiments by assuming that the
end groups contained two phosphonated-2,2-bipyridine units and the interior Ru2+ centers
had one. The degree of polymerization can then be calculated from the ratio between the
average peak areas for the 1,5-bis(4-methyl-2,2-bipyridyl-4-yl)pentane units and
phosphonated-2,2-bipyridine units. The average degree of polymerization was determined
by this method to be ca. 7 ruthenium centers. MALDI-TOF and ESI mass spectrometry
were run on oligomer samples but chains longer than three ruthenium units did not result
in a detectable signal. The details and results are given in the Appendix B. Gel permeation
chromatography (GPC) was run on the oligomeric dye in a chloroform system against a
polystyrene standard. The results of the GPC experiments showed that the oligomer chain
lengths had a dispersity (ĐM) of 4.3. These results indicate a broad range of chain lengths.
Photoanode Preparation. All electrodes were prepared on 18 Ω/cm2 fluorine-
doped tin oxide-coated glass (FTO-glass, Hartford Glass Company). A colloidal
suspension of TiO2, prepared as previously described,20,41 was deposited onto the FTO-
glass via the doctor blade method to form a 1 cm2 area. Layers of transparent tape were
used to control the thickness of the suspension that was applied to the electrodes. The
resulting films were sintered at 300 °C for 20 min, 350 °C for 10 min, and 500 °C for 30
min. For one and two pieces of tape the thickness of the films was measured to be
approximately 3 µm and 6 µm, respectively, by profilometry.
13
All electrodes were sensitized by soaking in 2 mL of 100 μM of sensitizer (RuP or
[RuP]n) in ethanol overnight at room temperature in the dark, then rinsed thoroughly with
ethanol and dried under a stream of N2. Following dye deposition, electrodes were kept in
the dark until use.
Insulated silver-plated copper wire was attached to the electrode using silver paste
(DuPont CP4922N-100), and contacts were protected using white epoxy (Loctite 1C
Hysol). Before sensitization with dye, TiO2 film electrodes used in intensity-modulated
photovoltage spectroscopy (IMVS) experiments were treated with TiCl4 as described
previously.41
Electrochemical Measurements. All photoelectrochemical measurements were
carried out using an Autolab potentiostat (PGSTAT128N) in a three-electrode
electrochemical cell with a Pt wire as the counter electrode and an Ag/AgCl (3 M NaCl)
electrode as the reference electrode. Cyclic voltammetry (CV) was conducted on dye-
sensitized TiO2/FTO working electrodes with a platinum wire counter electrode, and a
Ag/AgCl reference electrode (BASi, MF-2079). CV was performed in 0.1 M HClO4
aqueous solutions at a scan rate of 1 mV/s. The slow scan rate was used to minimize the
peak to peak separation which was typically 100 mV as a result of the series resistance of
the TiO2 layers.
Photophysical Measurements. Absorption spectra of the monomer and oligomer
were collected in ethanol solutions. Steady-state emission data were collected at room
temperature on a custom-built fluorimeter. For steady-state experiments, samples were
excited using a light output from a housed 450 W Xe lamp. Sensitizer solutions were
14
freshly prepared in ethanol at 100 μM concentration, added to quartz cuvettes, and bubbled
with nitrogen for 20 min. Emission studies conducted on sensitized electrodes were
performed in argon-purged acetate buffer (pH = 4.8).
Intensity Modulated Photovoltage Spectroscopy (IMVS). A 470 nm LED light
(LDC470, Metrohm) provided the illumination. The light intensity modulation was
realized through an Autolab LED driver controlled by the potentiostat with the AC
amplitude set to 10% of the DC level. The DC light intensity was 4.1 mW/cm2 (470 nm).
The applied frequency ranged from 2000 to 1 Hz. The electrolyte was aqueous 0.1 M
sodium acetate/acetic acid (pH 4.8) or 0.1 M sodium phosphate buffer (pH 5.8 or 6.8). Four
electrodes were run at each pH. Each sample was repeated three times at each intensity and
the results averaged together.
Cross-Surface Electron Diffusion Constant (Dapp). Following the method of
Hanson et. al.,15,20 Dapp was measured in 0.1 M HClO4(aq) by applying a potential of 2.0 V
vs Ag/AgCl (3.5 M NaCl) for 5 min, then 0.0 V for an additional 5 min. Dapp was calculated
by monitoring the change in absorbance at 450 nm and then analyzing the data according
to eq 1:
ΔA = 2𝐴𝑚𝑎𝑥 𝐷𝑎𝑝𝑝
1/2 t1/2
𝑑π1/2 (1)
Here ΔA is the change in absorbance (Amax - A(t)), Amax is the initial absorbance of the
fully reduced film, t is the time in seconds, and d is the thickness of the film (6 μm). The
15
film thickness was acquired from profilometry. Dapp was determined by fitting the change
in absorbance to eq 1 over the linear portion of the experimental data.
Photostability Measurements. The long-term adhesion of the sensitizer molecules
was evaluated using a technique developed by Meyer and coworkers.15 Briefly, the
electrode was subjected to constant irradiation under open-circuit conditions. The light
from a high-power blue LED (455 nm, fwhm 30 nm, 475 mW/cm2, Thorlabs, Inc.,
M455L2) powered by a T-Cube LED driver (Thorlabs, Inc., LEDD1B) was directed onto
the sensitized electrodes placed at 45° in a standard 10 mm path length cuvette containing
5 mL of the solutions of interest. The incident light intensity was measured using a
thermopile detector (Newport Corp 1918-C meter and 818P-020-12 detector). The
absorbance of the electrodes were measured to determine the amount of dye still attached
every 5 minutes over the first hour and every 15 minutes throughout the remainder of the
experiment. The electrodes were rinsed with water between each measurement.
Results and Discussion
A phosphonated oligomeric ruthenium tris(bipyridyl) dye ([RuP]n) in which the
individual Ru(bpy)3 centers are linked by a five-carbon aliphatic chain, was synthesized
and studied as a sensitizer for TiO2 photoanodes. Because a relatively long saturated
hydrocarbon linker separates the Ru(bpy)3 units we should expect them to be electronically
isolated and to have similar electrochemical and spectroscopic properties to the monomeric
sensitizer RuP. Some differences between the oligomer and monomer could however
result from the phosphonate and alkyl substituents that are present on the bpy units within
16
and at the ends of the oligomer chains. We confirmed that the properties of the oligomer
are similar to those of RuP by using cyclic voltammetry and steady-state absorption and
emission spectroscopy.
Electrochemistry. The electrochemical properties of the oligomer were
investigated by cyclic voltammetry (CV) of the sensitizers adsorbed onto TiO2 electrodes.
CV was run on sensitizers on electrodes as these potentials provide the most relevant
estimate of the working conditions of a WS-DSPEC.42 When these Ru(bpy)3 derivatives
were anchored to electrodes, their potentials typically varied by no more than 50 mV.42
Solubility issues with the electrolyte salts prevented us from running CV on [RuP]n in
solution. Addition of lithium perchlorate or tetrabutylammonium hexafluorophosphate
caused agglomeration and precipitation of the oligomer from solution. The results of CV
on sensitized electrodes are shown in Figure 2-3 and summarized in Table 2-1. The quasi-
reversible (RuIII/II) potential of the oligomer was observed to be E1/2 = 1.04 V versus
Ag/AgCl (0.197 V vs NHE). It has been shown that the electron-withdrawing substituent
effect of the two -PO3H2 groups at the 4,4’-bpy positions results in an incremental positive
shift in E1/2(RuIII/II) of 0.05 V relative to unsubstituted Ru(bpy)3.42 Conversely, two
electron-donating methyl groups result in an incremental negative shift of approximately
0.04 V.42 Assuming an average oligomer chain length of 7 units, the ratio of phosphonated
ligands to alkylated ligands is approximately 4:5. This ratio results in the substituent effects
nearly balancing each other out. The resulting E1/2(RuIII/II) of the oligomer is 1.04 V
compared to the monomer value of E1/2(RuIII/II) =1.08 V.
17
Figure 2-3: Cyclic voltammetry of TiO2 films sensitized with monomeric RuP (red) and
oligomeric [RuP]n (black) dyes in 0.1 M HClO4 electrolyte versus an Ag/AgCl reference
electrode at a scan rate of 1 mV/s.
18
Steady-State Spectra. The absorption and emission spectra of the monomer and
oligomer in methanol solutions are shown in Figure 2-4. The absorbance maximum of the
oligomer and monomer in solution are both at 458 nm. Emission spectra were collected for
the monomer and oligomer anchored to 5 μm ZrO2 particles suspended in aqueous 0.1 M
HClO4 and are shown in Figure 2-4. The conduction band potential of ZrO2 is too negative
for either complex to inject an electron from the excited state,42 so strong emission can be
observed. The emission maximum for the adsorbed oligomer and monomer are 654 and
Complex
E1/2(RuIII/II) (V)a ΔGes (eV)b
E*1/2(RuIII/II)
(V)c
Monomer
(RuP)
1.08 1.88d -0.80
Oligomer
([RuP]n)
1.04 1.88d -0.84
Table 2-1: Estimated ground and excited-state redox properties of RuP and [RuP]n on
TiO2 and ZrO2 electrodes in aqueous 0.1 M HClO4. aFrom CV measurements on TiO2 vs
Ag/AgCl ref. electrode. bFrom fitting of emission spectrum on ZrO2. cCalculated from
eqn. (2). dLiterature value for monomer from Hanson, et al.42
19
656 nm, respectively. Excited state reduction potential (E*red) can be estimated from Ered
(Ru(bpy)3+/2+) and the free energy of emission (ΔGes) using the relationship given in eqn.
2, where n = 1 and F is Faraday’s constant.
E*red ≈ Ered (Ru(bpy)3+/2+) - ΔGes/nF (2)
Given that the oligomer emission maximum is blue shifted by only 2 nm, the literature
value of ΔGes for RuP was used to estimate the excited state reduction potential of the
oligomer (Table 1). The excited state reduction potential (E*red) of the oligomer was
estimated to be -0.84 V, according to Eqn. (2).42 This is about 40 mV more negative than
the corresponding monomer excited state potential (E*red = -0.88). These experiments show
that the ground and excited state reduction potentials of the oligomer are slightly more
negative than that of the monomeric dye, reflecting the greater degree of alkyl substitution
of bpy ligands in the oligomer.
20
Dye aggregation. Dye aggregation or multilayer adsorption on the electrode surface is
always a concern with dye-sensitized electrodes, especially with polymeric sensitizers.
Because some of the dye molecules in aggregates are not directly adsorbed on the oxide
surface, the injection yield and therefore the cell efficiency are lowered.43,44 To confirm the
absence of aggregation, steady-state luminescence spectra were collected from dye
samples in solution and adsorbed on TiO2 electrodes. When Ru(bpy)3 derivatives are
electronically anchored to the semiconducting TiO2 electrode, the excited state is rapidly
quenched through electron injection into the conduction band.42 We find (Figure 2-5) that
both the monomeric and oligomeric sensitizers exhibit similar weak luminescence when
adsorbed to TiO2 electrodes, indicating similar electron injection kinetics. These results
Figure 2-4: Absorbance (dashed lines) of dyes in methanol and steady-state luminescence
(solid lines) of monomer (black) and oligomer (red) on ZrO2 powder in aqueous 0.1 M
HClO4 solution.
21
imply that the majority of monomer units in the oligomer chains are adsorbed directly at
the TiO2 surface through their phosphonate anchoring groups.
Pore Penetration. A possible concern with a polymeric sensitizer is its ability to
penetrate into the pore network of a thick nanocrystalline TiO2 photoanode. The pore
diameter in dye cell TiO2 films is generally on the order of 20 nm and because the
monomeric dye has a radius of 1.1 nm,20 it can readily penetrate the pore network to
uniformly sensitize the TiO2 surface. Because the average length of the oligomer is 7
monomer units, we can estimate its persistence length to be in the range of 2-3 nm, which
Figure 3-5: Steady state emission of monomeric RuP (black) and oligomeric [RuP]n (red)
sensitizers adsorbed to ZrO2 powder and to TiO2 electrodes. Inset shows the dye-TiO2
spectra on an expanded scale.
22
is still much smaller than the average pore diameter. To confirm that the oligomer dye
penetrated fully into pore network, electrodes were sensitized with oligomer and
subsequently sensitized with monomer. The resulting electrodes showed on average less
than a 1% increase in absorbance after the second sensitization step. This result supports
the idea that after the oligomer adsorption step, there is little free area of the electrode
available for the smaller RuP to adsorb. Cross-sectional EDS was also attempted to
determine the distribution of Ru atoms in the electrode film, but the dye loading was below
the instrument’s detection limit.
Photostability in pH 7.8 phosphate buffer. Dye-sensitized electrodes were placed
in sodium phosphate buffer (pH 7.8) and illuminated with constant irradiation under open
circuit conditions for 4 hours. The results are displayed in Figure 2-6. As other groups have
observed15,45, the MLCT absorbance of monomer-sensitized electrodes decreased rapidly
as a result of dye desorption. By the end of the 3 h experiment the monomer (RuP) had
desorbed from the surface and the absorbance reached the baseline of the unfunctionalized
TiO2 layer. Under the same conditions, the oligomer-sensitized electrodes showed a ~15%
drop in absorbance over the first 30 min and were then stable for the 3 h duration of the
experiment. This result has profound implications for the incorporation of the oligomer
into WS-PECs. First, the oligomer remains absorbed for significantly longer than the
monomer at pH 4.8 where WS-DSPECs are studied in order to minimize dye desorption.
Second, the oligomer is persistently adsorbed at higher pH (7.8), where the highest
efficiency cells have been reported,46 but where the monomer desorbs very rapidly. This
dramatically broadens the pH range at which WS-DSPECs can be studied and the types of
23
experiments that can be done. In particular it opens up opportunities to study these cells as
a function of pH. We exploit this property to use transient electrochemical techniques to
investigate the kinetics of cross-surface hole transfer and charge recombination from TiO2
to the oxidized dye molecules as a function of solution pH.
Cross-Surface Hole Transfer. In WS-DSPECs, charge transfer diffusion between
oxidized and reduced sensitizer molecules is an important process in connecting the
photoinduced charge transfer and catalytic water oxidation steps. The apparent charge
transfer diffusion coefficient, Dapp was measured for TiO2 films sensitized with ethanol
Figure 2-6: Changes in the absorption spectrum versus time of [RuP]n (red) and RuP
(black) adsorbed on TiO2 electrodes. Electrodes were held at open circuit in 0.1 M sodium
phosphate buffer (pH = 7.8) with constant irradiation (475 mW/cm2). Inset A: Absorbance
changes at 450 nm as a function of time. Inset B. Photograph of electrodes after 3 hours.
24
solutions of RuP and [RuP]n. Spectroelectrochemical absorbance data for a typical
oligomer-sensitized electrode are shown in Figure 2-7. The inset shows the linear
regression. Dapp for the oligomer and monomer were found to be 28.1 ± 1.6 ∗ 10−10 cm2/s
and 1.79 ± 0.58 ∗ 10−10 cm2/s, respectively. The monomer values are in good agreement
with literature values for monomer deposited from ethanol.20 By covalently linking the
oligomer units together the hole diffusion constant is increased by an order of magnitude
relative to the monomer. It has also been shown that an order of magnitude increase in Dapp
can be obtained by depositing the monomer from aqueous 0.1 M HClO4.15,20 However, our
group has shown that acidic depositions negatively affect the overall performance of the
electrodes.20,47 The oligomer is not soluble in aqueous HClO4 solutions, so it was not
possible to compare the monomer and oligomer from acidic deposition solutions.
Nevertheless, the oligomer displays significant improvement over the monomer, while
avoiding the complications of acidic deposition solutions. In the case of the oligomer, it is
also possible that hole transfer within the chains is fast relative to Dapp, which is likely
limited by hole transfer between chains adsorbed on the electrode surface.
25
Intensity-modulated photovoltage spectroscopy (IMVS). IMVS has been used
to measure the recombination kinetics in DSSCs48,49 and more recently in WS-DSPECs,
where charge recombination is typically the dominant kinetic process.41 Dempsey and
coworkers50 have used transient absorbance techniques to study the recombination
dynamics of RuP anchored to TiO2 from pH 1-5 and saw that the recombination rate
decreased as pH increased. As we have discussed elsewhere,41 recombination rates
measured by transient absorbance – where high intensity laser excitation is used – are
typically much faster than those measured under lower fluence. This is a consequence of
the bimolecular nature of the recombination process. Thus it is of interest to compare the
kinetics at different pH values using IMVS, which operates under conditions closer to solar
fluence.
Figure 2-7: A representative set of spectroelectrochemical absorbance-time data used to
calculate Dapp values for [RuP]n (red) and RuP (black). Fits to eqn. 1 are shown in the
inset.
26
Details of the IMVS experiments are given in Appendix A, and the results are
shown in Figures 2-8 and 2-9. Nanocrystalline TiO2 electrodes were surface passivated by
reaction with TiCl4 solution to minimize the surface trap density29 and were then sensitized
with monomeric and oligomeric dyes. IMVS experiments were performed at open circuit
in acetate buffer solutions (pH = 4.8) and the DC light intensity was varied in order to span
a range of open circuit potentials, Voc. Figure 2-8 shows that the recombination rates
increase at more cathodic potentials, which, under open-circuit conditions, represent
electrons in higher energy levels at higher light intensities. At pH 4.8, the oligomer has a
lower recombination rate than the monomer, possibly because faster hole diffusion results
in a longer electron-hole separation distance. IMVS data from oligomer-sensitized
electrodes were also obtained in sodium phosphate buffer solutions at pH 5.8 and 6.8.
These phosphate buffer solutions were employed to demonstrate the higher pH that these
cells could run at if a more adhesive sensitizer were employed. Operating WS-DSPECs in
more basic solutions has three different effects, which are difficult to quantify with dyes
that adsorb at higher pH. Because the water oxidation potential shifts cathodically with
increasing pH, the WS-DSPECs has a larger driving force for oxidizing water, as
demonstrated by Gao et al.46 and also by Meyer and coworkers.51 At the same time, the
TiO2 flat-band potential shifts cathodically with increasing pH, which has two effects: the
driving force for electron transfer quenching of the sensitizer excited state decreases,
lowering the injection quantum yield, but the negative shift also results in an increase in
the open circuit photovoltage of the cell.
27
Interestingly, the recombination rates at the same light intensity are smaller as pH
increases. This is most likely a result of the decreased electron density which is caused by
the decreased efficiency of electron injection. It is well known that the TiO2 flat-band
potential experiences a cathodic shift as pH increases. As a result of the raised flat-band
potential, the injection efficiency for dyes will go down as the pH increases. Despite
expecting this trend to continue at higher pH, there have been no published recombination
rates as a function of pH up to pH 7.3, because up until now there were no dyes that were
stable up to this pH.
28
Figure 2-8: Plots at pH 4.8 of recombination rate (kIMVS) of the Monomer and Oligomer
as a function of open-circuit photovoltage (Voc).
Figure 2-9: Combined plot of recombination rate (kIMVS) as a function of open-circuit
photovoltage (Voc) for oligomer samples at various pH ranging from 4.8 to 6.8.
29
Conclusions
Two major limitations of WS-DSPECs are dye desorption and slow cross-surface
hole diffusion between oxidized and reduced dye molecules. We have synthesized and
characterized an oligomeric ruthenium dye that has significantly faster hole diffusion
properties and is persistently adsorbed to high surface area TiO2 electrode films over a
broader pH range than the commonly employed monomeric sensitizer RuP. We provide
evidence that the oligomer penetrates into the electrode pores and remains adsorbed under
constant illumination and throughout electrochemical experiments such as IMVS.
Additionally, the increased stability at higher pH opens up opportunities for incorporating
different water oxidation catalysts. IMVS experiments show that the oligomer has a slightly
slower recombination rate than the monomer. More interestingly, recombination rates
could be measured over a higher pH range for the first time. These measurements show
that recombination rates decrease as the pH is increased. It is likely that this is a result of
lower electron injection efficiency at higher pH. The synthetic method illustrated here
could enable the future creation of hybrid oligomers consisting of both phosphonate
anchors and carboxylate anchoring groups that could provide a “best of both worlds”
solution with both superior adsorption to the electrode and higher injection efficiency.
Acknowledgements
This work was supported by the Office of Basic Energy Sciences, Division of Chemical
Sciences, Geosciences, and Energy Biosciences, Department of Energy, under contracts DE-
FG02-07ER15911.
30
Chapter 3
Nanostructured Lenses for Solar Spectrum Splitting
Abstract
Photovoltaic efficiency can be increased through improved utilization of the solar
spectrum. By combining multiple absorbers, each matched to a piece of the solar spectrum,
the light energy can be more effectively harvested. This has been accomplished with
stacked tandem cells, but major improvements could be realized if the solar spectrum could
be spatially split by wavelength and the component PV cells placed side-by-side.
Nanostructured lenses were fabricated for the purpose of selectively scattering the solar
spectrum. These lenses are composed of low refractive index spheres embedded in a high
refractive index metal oxide layer. Simulations and possible routes to optimize these
structures are also discussed.
Background
Tandem Solar Cells
In 1961, Shockley and Queisser analyzed the theoretical losses in efficiency from
first principles. Their work showed that a photovoltaic (PV) cell comprised of a single
31
absorber has a maximum theoretical limit of approximately 33%.3 The main limiting factor
to the theoretical limit of power conversion efficiency in a single solar cell is the
thermalization of the incident photon’s energy to the bandgap of the absorber. Currently,
the most common PVs are silicon-based with bad gaps corresponding to near-IR absorption
(~1100 nm). As a result, much of the energy of absorbed visible photons is wasted via
thermalization. A blue photon (~2.8 eV) has almost twice as much energy as a red photon
(~1.8 eV). Upon absorption of a blue photon by a silicon cell, that additional energy is lost
through generation of heat. Despite this, the silicon band gap is placed correctly to absorb
the largest amount of energy inevitable fact of nature. This Shockley-Queisser limit of 33%
is specific to single absorber solar cells, Tandem solar cells, comprised of multiple
absorbers routinely exhibit efficiencies that exceed 33%. While they still experience the
same thermalization losses, they are able to increase efficiency by better utilization of the
available photons by incorporating multiple absorbers that are each matched to parts of the
solar spectrum. The most common architecture is the stacked tandem cell, which comprises
multiple cells stacked on top of each other. In this configuration, a PV with a wide
absorption gap (e.g. visible region) is placed in front of a second PV with a narrower
absorption gap (e.g. near-IR region) with the two modules electrically connected in series.
This architecture attains more complete spectral utilization, which results in a dramatic
increase in efficiency. One drawback of this architecture is that it requires area and current
matching of the two cells. Additionally, the upper cell may parasitically absorb light that
would be more optimally absorbed by the bottom cell. Often, a tandem cell is created by
combining an expensive efficient absorber, such as a silicon PV, with an inexpensive less
efficient complementary absorber. In the past, dye-sensitized solar cells (DSSCs)52,53 have
32
been used for this purpose and, more recently, perovskite solar cells31,54,55 are attractive
candidates for this application. Both are made from abundant materials and in principle can
be much less expensive than silicon cells when produced on a large scale3,56,57 Physically
separating the light spectrum in space would negate all of the drawbacks of a stacked
tandem cell. In particular, without the requirement for area matching, many new
opportunities for mixing and matching complementary cells will be available.
Figure 3-1: A) Traditional Stacked Tandem Solar Cell. B) Mirrored Spectrum
Splitting Tandem Solar Cell.1 In both cases the blue light is absorbed by the DSSC and
the red light is absorbed by the silicon cell. Note in the Mirrored Cell the DSSC and
silicon cell (Si) are different widths.
33
Mirror Spectrum Splitting Tandem Cell
Previous work from the Mallouk laboratory has demonstrated spectrum splitting
using hot mirrors to operate a low cost DSSC in tandem with an expensive silicon cell.1 A
significant result of this work is the decoupling of the component areas. In a stacked tandem
cell, the area of the silicon cell should match the area of the DSSC, but in mirrored spectrum
splitter cell, it is possible to use cells of different area. This results in a significant reduction
in cost because the cost of fabrication is proportional to the size each cell. One drawback
to the mirror-based spectrum splitting architecture is the additional requirement for a
mechanical system to track the sun throughout the day. Mechanical tracking systems are a
common component included to maximize the efficiencies of tandem cells or concentrator
cells. The goal of this project is to develop nanostructured lenses capable of passively
splitting the solar spectrum, thereby eliminating the need for expensive mechanical
tracking. A lens can be created to selectively forward scatter visible light, while
transmitting IR light un-scattered, by exploiting the wavelength-dependence of material’s
refractive indices. The primary goal is the creation of a planar prototype that demonstrates
the theoretically predicted scattering of light as a result of mismatched refractive indices.
Secondary goals include the optimization of scattering through material selection and the
development of materials and techniques that can be scaled up for commercialization at
low costs.
34
Optical Measurements of Spectrum Splitting Lenses
The first and most basic technique for evaluating a spectrum splitting lens is with
the use of an integrating sphere attachment to a UV-Vis spectrometer. An integrating
sphere is a closed sphere coated with a nearly 100 % reflective coating designed to capture
and direct all transmitted light to a detector. In the work presented here, a Perkin-Elmer
Lambda 950 was used with a 15 cm Spectralon®-coated integrating sphere. This model of
integrating sphere has a removable 2 cm diameter cap, that when removed, allows normal
transmitted light to exit the sphere and not be detected. As a result of the cap being attached
or detached there are two methods of detection available, depicted in Figure 3-3. In Method
1, the cap is attached and both the diffuse transmitted light (scattered) and the normal
transmitted (un-scattered) light is detected. In Method 2, the cap is removed, and the normal
Figure 3-2: A) Target final hemispherical Structure. B) Planar spectrum splitting
prototype lens. With a spectrum splitting lens, the DSSC and silicon cell (Si) are different
widths. Also note that the spectrum is split even with incident light from an off-zenith sun.
35
transmitted light is allowed to exit the integrating sphere. In Method 2, only the wide-angle
scattered light is detected. These two methods give an overall picture of the amount of light
scattered but do not give a precise determination of the amount of light scattered at specific
angles. To determine the wavelength dependence of scattering at specific angles a rotatable
detector is required that can be positioned at precise angles from the sample. These
integrating sphere methods are also useful for evaluating the amount of light not
transmitted as a result of back scattering losses.
Figure 3-3: Two methods of evaluating scattered light with the integrating sphere
attachment of a UV-Vis Spectrometer. Method 1 detects the total transmitted light. Method
2 detects only the wide-angle scattered light.
36
In theory, the minimum requirements for an optical system to measure the angular
scattering of a lens are simply a white light source and a movable detector, but there are
other practical concerns outlined herein. Optical components were added to two existing
optical detection systems in order to build the functionality needed to quantify the
transmitted light as a function of angular scattering. In both systems, the detector is fixed
at an angle and an entire spectrum is collected. The detector is then moved to another angle
and another full spectrum is collected. This process is repeated until the desired amount of
spectrum are collected as a function of angle.
The first system was built in Dr. Lakhtakia’s lab and is shown in Figure 3-4. In
brief, a broadband white light source is fed through a fiber optic and passed through a lens
sample and then detected by a photodiode that can be positioned at different angles. The
light source is a portable 24 V Ocean Optics halogen light source (model # HL2000). There
are several optics present in the photograph that are used for other measurements that do
not affect these measurements. The essential components for the measurements described
in this work are the light source, sample, and movable detector. The detector is a high-
resolution CCD spectrometer from Mightex (HRS-Series) that is connected to a personal
computer that records the output spectrum.
The second system is composed of additions made to the incident photon to current
efficiency (IPCE) setup in the Mallouk lab (Figure 3-5). The photodiode detector is
connected to the personal computer through a Keithley Multimeter, and the data is
collected with a LabView program. The monochromator is controlled by separate LabView
programs. The Kiethley Labview program reads, records, and converts the signal from the
37
photodiode into a spectrum. Similar to the first system, the essential components for the
measurements described in this work are the light source, sample, and movable detector.
The two main differences between these systems are the distance to the detector
and the power output of the light source. In the first system, the detector is fixed at an
unmovable distance of 7 inches from the sample stage. The detector in the Mallouk setup
is attached to a movable arm that can be adjusted to a distance of 1-4 inches from the
sample holder. The distance to the detector has significant ramifications for the detection
of scattering. As the scattered light passes out of the lens and travels farther to the detector
it is scattered more the farther it travels and the signal is weakened as a result. If the
scattering effect is weak enough it might not be detectable by the first spectroscope.
In the first system, the Ocean Optics halogen light source has a power output of 100
mW while the second system has a 500 W xenon lamp. As a result of focusing optics, spot
size, and other considerations, a direct comparison of power is not offered here, but it is
clear that the Mallouk setup results in a higher amount of incident photons and a larger
detectable signal.
It is important to note that the forward scattering occurs in all directions in a cone
shape. Therefore, the angular spectra measured by these systems represent a small spot size
along the radius of that cone. To obtain the total amount of scattered light, the value at each
angle must be integrated along the radius of the cone. The cone radii will be proportional
to the angle of scattering and proportional to the distance of the detector. A simple
alternative method to measure the total scattering, summed across all wide angles, is the
integrating sphere methods described above.
38
Figure 3-4: Photograph (A) and diagram (B) of Lakhtakia Lab Spectroscope with
rotating CCD detector fixed at 7 inches.
39
Figure 3-5: Photograph (A) and diagram (b) of Mallouk Lab spectroscope with
rotating detector. Note the adjustable detector distance that can range from 1-4 inches.
40
Early Lens Fabrication
Hemisphere PMMA Lenses
As previously mentioned, hemispherical lenses are the ideal final form of this
project because they should selectively scatter light independent of the direction of incident
light and, as a result, would not require expensive mechanical sun-tracking equipment. It
is expected that if spheres are successfully embedded in a host material that the material
would be mostly transparent across the light spectrum. To make hemispherical lens,
methyl-methacrylate (MMA) was polymerized in small glass vials. After the
polymerization the vials were shattered, and the glass was removed to yield a poly(methyl
methacrylate) cylinder which was subsequently cut in half with a band saw. During
polymerization, MMA is known to shrink in volume and form gas bubbles. Despite this,
transparent uniform films are easily obtained over temperature ranges of 40-100 °C.
Several attempts were made to create sphere-embedded polymer lenses at various
temperatures across this range. 10 mL of methyl-methacylate, 30 mg of benzoyl peroxide,
and 30 mg of SiO2 spheres were placed in glass vials. The vials were sonicated to disperse
the spheres in the viscous solution. Unless otherwise noted, all sphere sizes given
throughout this thesis chapter are the diameter of the spheres. Samples were made with 500
nm spheres, 1000 nm spheres, and a control without spheres. Samples containing spheres
resulted in significantly more gas bubbles trapped in the polymerized structure. It is
possible that the spheres provide a nucleation point for the bubbles. To eliminate bubbles
from being trapped in the film, it was necessary to heat the mixture at 40° for 15 min to
initiate the reaction. The vial was removed from the hot water bath and placed on ice to
41
slow the polymerization reaction. This process resulted in the hemispherical lenses show
in Figure 3-6. The transmission of the lenses was measured using a UV-Visible
spectrometer, and they were found to only transmit 10% of the incident light. 90% of the
light was lost to wide-angle scattering or back-scattering. Simulations discussed later in
this work will show that the difference in refractive indices between SiO2 spheres and the
PMMA polymer was not large enough to have the selective scattering effect that was
desired. To create lenses with larger differences in the refractive indices, a metal-oxide
host layer was required. Because metal-oxides are not as easily deposited as polymers, a
planar lens prototype structure (Figure 3-2 B) was targeted. The planar structure would
allow the use of a high refractive index metal oxide host matrix and would demonstrate a
scattering prototype.
Figure 3-6: Photograph of hemispherical lenses created from 500 nm (left) and
1000nm (right) SiO2 spheres suspended in poly(methyl-methacyrlate)
42
Spin Coating Sphere Layers for a Planar Lens Prototype
To create a planar lens, it is necessary to first deposit and organize spheres in a
single layer on a substrate. As spheres become very densely packed, they will exhibit long
range order which may lead to plasmonic effects that are detrimental to the goals of
spectrum splitting. Uncontrolled plasmonic effects might detrimentally waste energy in
other modes, such as heat generation. At the same time, it’s important to pack spheres
relatively close together, to increase the number of scattering points so that the magnitude
of scattering is detectable. Spin coating is a common method used to spread a thin film on
a substrate. The spacing between spheres can roughly be controlled through the variation
of spin rate. Mihi and coworkers developed a method for spin coating multiple layers of
spheres.58 This procedure was adapted for our purposes to create monolayers of spheres.
Experimental
A colloidal solution of 500 nm spheres was made by diluting 1 mL of a 50/50 (vol
%) mixture of ethanol and ethylene glycol with 9 mL of water and then dissolving 200 mg
of spheres. The colloidal solution was sonicated for 10 min and then 100 µL of the solution
was drop cast onto slides and spun for 30 s. The solution was used to create samples spun
at 2000 rotations per minute (RPM) and 3500 RPM. Light microscope images of the
resulting sphere layers (Figure 3-7) show that the spheres spun at 3500 RPM are more
dispersed than those spun at 2000 RPM. Unfortunately, the spheres have a tendency to
stack on top of themselves and form multiple layers. In Figure 3-7, single layers of spheres
are tan colored and multilayers are pink or white. Although the spheres predominantly form
43
single layers, the presence of multiple layers is evident. Even at the higher spin rates,
despite ample space to spread into, there are multiple layers of spheres. Increasing the speed
does not eliminate the presence of multi-layers. Despite the described limitations, spin
coating sphere layers provides a quick method to arrange spheres in a dense, disordered,
near monolayer.
Preliminary Sol-gel Back Filled Lenses and Multi-Layered lenses
In order to increase the difference between refractive indices of the lens materials,
a planar architecture was adopted. This architecture consists of a low-index SiO2 (n=1.4)
Figure 3-7: Light microscope images of spin-coated sphere layers spun at 2000
RPM (A) and 3500 RPM (B). Single layers of spheres are tan colored and multilayers are
pink or white.
44
sphere embedded in a higher index metal-oxide host matrix. Dr. Barber’s preliminary Mie
scattering simulations identified hybrid materials with low titania (TiO2) concentrations
(15%) as having the ideal refractive indices. Sol-gel literature has shown that the index of
silica can be tuned by adding small amounts of TiO2,59 and monodisperse silicon spheres
are easily synthesized or purchased. Simulations using MIE theory showed that spheres of
500 nm diameter would be ideal for this application because they primarily forward scatter
light in small angles (<15°), with minimal back-scattering losses. It is expected that if
spheres are successfully embedded in a host material, the lens would be mostly transparent
throughout the light spectrum.
Experimental
The spin-coated sphere layers were backfilled with a 0.2 M TiO2/SiO2 sol-gel
mixture (15 mol %/85 mol %). This mixture was created by mixing 15 mmols of titanium
bis(acetalacetonate) and 85 mmols of tetraethylorthosilicate in 20 mL of ethanol. 100 μL
was then drop cast onto the substrates containing spheres. The substrate was spun for 30 s
at 1000 RPM and then placed on a hotplate for 30 min at 500 °C to convert the precursors
into the metal oxide host matrix. The sol-gel back-fill process was repeated ten times to
fully cover the spheres with the host matrix. One of the samples was shattered with a
diamond scribe to allow cross sectional imaging. SEM was performed on the cross section
to yield the image shown in Figure 3-8. This confirms that the spheres were successfully
embedded in the metal oxide host matrix. Another sphere layer was then deposited via spin
coating. The back-fill process was repeated another 10 times. This resulted in two layers
45
of sphere. These two processes were repeated for a total of 5 sphere layers. The sphere
films became increasingly more opaque with each successive layer.
Results
After each back-filled sphere layer was completed, the normal transmission of the
samples were measured with UV-Visible spectroscopy in the absence of an integrating
sphere (Figure 3-9). The back-scattering was determined with the integrating sphere in
reflectance mode. The results of total transmittance as a function of numbers of sphere
layers are shown in Figure 3-9. The sample containing one sphere layer (blue) shows nearly
100 % transmittance of photons at 900 nm, but nearly all of the photons at 400 nm (90%)
are also normal transmitted (un-scattered). As more sphere layers are added the selective
Figure 3-8: Cross-section SEM Micrograph of spin-coated SiO2 spheres
surrounded by 15% TiO2/SiO2 spin-coated host matrix
46
scattering of 400 nm photons increases to the point where only 35 % are normal
transmitted, but this comes at the cost of increased back scattering across the entire
spectrum. The back scatter from a 5-sphere layer sample is shown in yellow and is
consistently at least 17% across the entire spectrum measured. The back scattering is most
likely caused by optical defects in the metal oxide sol-gelled host matrix film. These could
be a result of microstructure cracking because the precursor solution is too concentrated.
They could also be a result of the TiO2and SiO2 precursors condensing at different rates
and segregating into a non-homogenous film. Regardless, it is clear that a more precise
method for depositing defect-free metal oxides is required.
Figure 3-9: Plots of Transmission and Back-Scattering after deposition of
successive sphere layers. Note the amount of backscattered light (yellow) from the
sample with five periods of spheres.
47
This preliminary result demonstrates the fact that a weakly spectrum splitting
phenomenon can be enhanced by including multiple sphere layers each containing
scattering points. The drawback to this multilayer approach is the accumulation of optical
defects with each layer. Therefore, it is critical to use a method of depositing metal-oxides
that is free of optical defects. The basic measurements performed on the UV-Visible
Spectrometer provide an overall picture of the amount of normal transmitted, forward
scattered light, and back scattered light.
In order to investigate the angular distribution of the forward scattered light the
custom-built spectroscope from the Mallouk lab was used. The spectroscope was used to
detect forward scattered light as a function of angle (0°, 5°, or 10°) for samples with 1 or 4
sphere layers (Figure. 3-10). The first plot in Figure 3-10 is the transmission at 0° (normal
transmission), or the non-scattered light. The next two graphs show that the light scattered
at 5° and 10° increases with increasing periods. In order to obtain the total amount of
scattered light, it must be summed over the small angles (1-15°) and then integrated along
radius of the scattering cone. Some basic conclusions can be realized even without the total
amount of scattered light.
Conversely, in the ideal case, short wavelength light would be primarily forward
scattered and thus be detected at significant percentages in the 5° and 10° plots. Note that
the angled detector only measures a portion of the cone of scattered light, so the 5° and 10°
measurements are expected to be small percentages. The measured results show that the
spin-coated lenses do not perform well as spectrum splitting lenses. At a detector angle of
5°, more long wavelength light was detected (1%) than short wavelength light (0.8%). This
observation remained even with additional sphere layers. Further, at 10°, more long
48
wavelength light (~0.5%) was detected than short wavelength light (~0.25%). Overall, the
angular forward scattering at 5° and 10° was small and not selective for short wavelength
light. These angular measurements further confirm the presence of overwhelming losses
from optical defects, back scattering, or wide angle scattering (> 10°). These results
reinforce the critical need for a material deposition process that creates dense, defect free
metal-oxide host layers.
Figure 3-10: Transmission spectra at detector angles of 0°, 5°, and 10°– as a
function of sphere layers (1 or 4 layers). As the number of periods increase the light
detected at 0° decreases and the light detected at 5 and 10° increases. These
measurements were collected using the modified IPCE setup in the Mallouk Lab.
49
Collaboration on Spectrum Splitting Grating Simulations
After our preliminary work described above, a collaboration with Dr. Lakhtakia
and Dr. Monk was created to simulate 2-dimensional spectrum splitting gratings and
optimize their material parameters. Given the nature of Maxwell’s electromagnetic
equations, it is only possible to computationally solve 2-dimensional systems. Monk and
coworkers2 simulated the triangular structure (A) and cylindrical structure (B) shown in
Figure 3-11. They used a combination of the rigorous coupled-wave approach (RCWA) to
determine the reflection and transmission characteristics of the grating and a differential
evolution algorithm (DEA) for optimizing the geometry and the refractive indices of the
material constituents. It is important to note that the shapes extend infinitely in the y
direction in the form of cylinders, as optical gratings typically do. This differs from the
experimental sphere-based samples that have been fabricated herein. Despite these
limitations, these simulations have granted great insight into which materials and
dimensions are needed to fabricate a planar spectrum splitting lens.
Figure 3-11: Structures simulated by Fan and coworkers.2 A) Difficult to
manufacture structure predicted to scatter effectively without sensitivity to the direction
50
Variable Material
Refractive Index
na Air 1.0 (fixed)
n1 Low index metal oxide (ex: SiO2) 1.0-2.6
n2
High index metal oxide (ex: TiO2) 1.0-2.6
ng SF11 Glass 1.78 (fixed)
Simulations by Monk and coworkers were run on cylindrical gratings with incident
light coming from an angle (θ) of 6° and 15°. The algorithm sampled four variables over
the simulation window. The refractive indices of the cylinder and host matrix were n1 and
n2, L was the length of the simulation unit cell, and R/L was the ratio of the cylinder radius
to the simulation unit cell. The diameter (D) of the cylinders can be calculated using L and
R/L and are provided in Table 3-2. A ratio of R/L=0.5 would correspond to a cylinder
filling the entire simulation box, and thus the cylinder would be touching the neighboring
cylinders. The differential evolution algorithm was iterated until the parameters converged
on a global minimum which resulted in the optimal values displayed in Table 3-2. The
variable range is also listed in Table 3-2, for the wider parameter window that was
simulated for the grating at 15° incidence. Interestingly, the optimal values obtained at
of incident light. B) Manufacturable structure that would be sensitive to the direction of
incident light.
Table 3-1: Material refractive indices used in simulations by Fan and coworkers.2
Row colors match the structure diagram in Figure 3-11.
51
different angles of incidence vary widely. In the case of the refractive indices for the
cylinder material (n1) and host matrix (n2) the optimal values are actually inverted. For a
15° incidence, the cylinder is a low refractive index material (n=1.11) embedded in a higher
index host matrix (n=1.80). This low-index cylinder structure was the original design
proposed by Dr. Barber. On the other hand, for a 6° incidence, the cylinder is a high
refractive index material (n=2.4) embedded in a low index host matrix (n=1.11). Ideally
the scattering properties of the cylindrical structure would be independent of the degree of
incident light. The difficult to fabricate triangular structure appears to be less sensitive than
the cylindrical structure in this aspect. It is unclear if the cylindrical structure’s sensitivity
is a consequence of the planar nature of the lens. It may be possible that a hemispherical
lens composed of the same materials might not display this sensitivity. Unfortunately, a
hemispherical lens structure cannot be simulated due to the way the materials are
partitioned in a layered mesh. What is indisputable from these simulations is that a large
difference between the refractive indices of cylinder and host index (Δn ~ 0.8) is required
to generate selective scattering. This allowed us to abandon the complicated mixed
TiO2/SiO2 sol-gel back-fill process and focus on methods of depositing 100% TiO2.
Additionally, at no point during the simulations were touching cylinders (R/L=0.5), found
to be optimal for spectrum splitting. This reinforces our chemical intuition that it is
desirable to avoid highly ordered, close-packed sphere layers.
52
Table 3-2: Optimized structure parameters from simulations of cylinders
interacting with 6° and 15° incident light.2 a Simulation window values are given for the
wider range of 15° incidence.
Variable Simulation Windowa 6° Incidence
15°
Incidence
Simulation box (L) 300 - 600 560 nm 430 nm
Ratio of cylinder to box (R/L) 0.1 - 0.5 0.30 0.25
n1 1.0 - 2.6 1.80 1.11
n2 1.0 - 2.6 1.00 2.4
Cylinder Diameter (D) 60-600 336 nm 215 nm
Advanced Planar Lens Fabrication
With the insights from the simulations by Lakhtakia and coworkers,2 changes were
made to the prototype planar lens architecture. The complicated mixed Si/Ti host matrix
was replaced by a simpler 100% TiO2 back-fill. The simulations highlighted the need for a
large difference between sphere and host index (Δn > 0.8). SiO2 (n=1.4) spheres back-filled
with a 100% TiO2 matrix fulfils this requirement (n=2.6). Layers deposited through ALD
are also much less likely to contain optical defects since they are built up slowly, atom-by-
atom, in a vacuum-sealed process chamber. The higher optical quality of the host matrix
should eliminate the significant back-scattering losses observed with the preliminary
lenses.
53
Sphere layers from spin coating were back-filled with a TiO2 matrix using atomic
layer deposition (ALD). A Cambridge Savannah 200 system was used with a Ti[EtNH3]4
precursor heated at 75 °C as the titanium source and water vapor was the oxygen source.
Substrates were heated at 150 °C throughout the deposition. TiO2 films were formed by
alternately pulsing each precursor (0.03 s for Ti, 0.015 s for water) into the samples
chamber under a N2 gas flowing at 20 sccm. A silicon wafer was placed inside the sample
chamber during the deposition to monitor the growth of the TiO2 film. Over a 72 h period,
16,000 cycles of ALD were performed, and the deposited film was 1.5 µm thick according
to ellipsometry measurements on the silicon wafer control sample. The resulting sample is
shown in Figure 3-12-A. ALD is known to deposit slightly reduced TiO2, which has an
opaque blue color. In order to fully oxidize the TiO2 layer, the samples were annealed at
600 °C in a box furnace under an oxygen atmosphere for 5 hr. A photograph of the resulting
sample is shown in Figure 3-12-B. The areas covered by spheres show a white color,
whereas areas only covered by TiO2 are fully transparent. The transparent nature of the
pure TiO2 layer is one indication that the process produced a dense, optically uniform layer
of TiO2. Further evidence of a uniform layer deposition was obtained by SEM analysis of
a cross section of the sample. Samples were scored with a diamond scribe and shattered to
expose a cross sectional view of the buried sphere layer. Energy Dispersive Spectroscopy
(EDS) was used in conjunction with SEM to identify elements composing the sphere layer
along the rough, shattered edge of the sample. The SEM micrograph is shown in Figure 3-
13 with a colored EDS map overlay showing elemental titanium in red and elemental
silicon in green. Given the expensive and time-consuming nature of back-filling TiO2 with
ALD it became critical to ensure that only a single layer of spheres was deposited so that a
54
minimum amount (~1000 nm) of material was required to cover a single sphere layer, as
opposed to multiple layers (~3000 nm). A Langmuir-Blodgett trough was used to create
more precise sphere layers.
Figure 3-12: A) ALD back-filled samples with reduced titanium centers. B) The
same sample after being annealed in a box furnace
55
Improved Sphere Monolayers via Langmuir Blodgett Deposition
For the purposes of creating a planar spectrum splitting lens, it is critical to have a
monolayer of spheres that are densely packed, but not so densely packed that there is long
range order. More densely packed particles result in more scattering points for incident
photons, as well as more overall scattering. However, if the particles are too densely
packed, they will display long range order and may begin to exhibit plasmonic effects.
Langmuir-Blodgett troughs are commonly employed to deposit highly ordered monolayers
of particles.60 Silica spheres were modified with ionic surfactants so they could be floated
on a layer of water in the trough. The sphere films are then transferred to a dipped substrate,
Figure 3-13: SEM cross-section analysis of lens back-filled by ALD. EDS
overlay of an SEM Micrograph confirming the buried sphere layer is comprised of silicon
(green) and the overlayer is comprised of titanium (red)
56
while automatically compressing barriers to maintain the densely packed monolayer. In
order to determine the optimal deposition pressure required to deposit a highly dense
monolayer, it is typical to obtain an isotherm of a floated layer on the LB trough. The
resulting isotherm can be divided into three regions: a low density region (1), a densely
packed region (2), and a third non-linear region (3). The transition from region 2 to region
3 occurs when the dense monolayer collapses and becomes multiple layers.60 As such,
Szekeres and coworkers60, deposited their layers just below the collapse layer of each film.
This ensures that the monolayers of spheres are densely packed and highly ordered. In our
experimental setup, deposition pressures were chosen that were farther below the collapse
point but still in the densely packed region 2 such that the spheres are densely packed but
not highly ordered.
Experimental
Silica spheres were suspended in a 0.1 M solution of hexadecyltrimethylammonium
bromide surfactant. The silica suspensions were prepared fresh each time and sonicated for
30 min prior to deposition. 0.5 mL of the silica sphere solution was floated on a water layer
in the trough. The films were compressed at a barrier speed 20 cm2 min. For film
deposition, 25 mm x 25 mm glass microscope slides were cleaned by sonication in soap
and water for 20 min each. The films were deposited in the upstroke direction at a speed
of 4 mm/min, at pressures below the collapse pressures of the films (4-8 mPa).
Sphere layers as a result of the LB trough method are shown in Figure 3-14. The
spheres are densely packed in a disordered single layer. There is no evidence of multiple
57
layers throughout the sample. This is a significant improvement over the spin coating
method. Although the LB process is much slower than the spin-coating method, it is far
superior for the controlled deposition of spheres.
Poor contrast or streaking in the image is a result of the sample charging due to the
insulating nature of the SiO2 spheres and glass substrate. The charging can be reduced
through coating with a conducting substrate such as gold or iridium; however, the presence
of a conducting layer might affect the optical properties of the final structure so was method
is avoided.
Figure 3-14: Sphere layers deposited by LB trough method. Poor contrast or
streaking in the image is a result of the sample charging due to the insulating nature of
the SiO2 spheres and glass substrate.
58
Liquid Phase Deposition of TiO2 Matrix
Although ALD provided an excellent method to back-fill a single sphere layer,
multiple sphere layers were desired in order to increase the magnitude of the scattered light.
In an effort to reduce the time and cost required to back fill a single sphere layer, and make
multilayer structures feasible, an alternative liquid phase deposition (LPD) of TiO2was
investigated. Our group had previously exploited this LPD method to make inverse opal
structures out of TiO2.61 The method works by creating a reactive titanium precursor in-
situ through the reaction of Ammonium hexafluorotitanate ((NH4)2TiF6 (AHFT), and boric
acid (H3BO3). This process was run at 40 °C, 50 °C, and 70 °C in an attempt to deposit
thicker layers of TiO2.
Experimental
Sphere films were prepared using the LB trough method discussed earlier.
Ammonium hexafluorotitanate ((NH4)2TiF6 (AHFT), and boric acid (H3BO3) were
separately dissolved in deionized water and then mixed to form a solution and brought to
pH 2.88 with HCl. Water was further added to yield a final solution containing 0.3 M
AHFT and 0.25 M H3BO3 at pH 2.88. The solutions were heated to 40 °C, 50 °C, and 70
°C. The sphere substrates were immersed in this solution for 80 min. Substrates were held
vertically at the bottom of this solution to prevent particles formed in the solution from
accumulating on the surface. The substrate was then sonicated for 1 min, rinsed with water,
and dried at room temperature. The films were then imaged with SEM and the results are
59
shown in Figures 3-15, 3-16, and 3-17. The samples back-filled at 40 °C (Figure 3-15)
show the beginning nucleation of TiO2 crystallites.
When the reaction is carried out at 50 °C (Figure 3-16), the deposition covers the
spheres, but the texture of the material is very rough and porous in nature. Additionally,
the process forms cracks in the overall film and appears to peel the sphere layer away from
the substrate.. Films deposited at 70 °C (Figure 3-17) exhibit a similar observed peeling
effect which is more pronounced than the films deposited at 50 °C. In this case, entire
segments of the film had peeled away and flaked off of the substrate. Regardless of the
temperature, the LPD process resulted in porous, defect containing TiO2. In conclusion,
this LPD process is not adequate for the deposition of a dense, defect-free host matrix
required for our spectrum splitting application.
60
Figure 3-15: SEM Micrographs of LPD back-filled spheres at 40 °C. The reaction
has not covered the spheres with a complete TiO2 host matrix. Small crystallites can be
seen where the deposition has begun
61
Figure 3-16: SEM Micrograph of LPD back-filled spheres at 50 °C. Although the
back-fill is nearly complete, the texture of the material is very rough and porous in
nature. Additionally, the process forms cracks in the overall film and appears to peel the
sphere layer away from the substrate.
62
Single Layer Planar Prototype Lens with Rough Overlayer
Sphere films were prepared using the LB trough method and back-filled using
30,000 cycles of ALD. Given the rough nature of the ALD surface, additional material was
deposited to allow for the polishing of the surface. The surface of the back-filled lenses are
shown in Figure 3-18 and the cross section of the films are shown in Figure 3-19. The
layered appearance is an artifact of the milling process. The magnitude of roughness that
resulted from the conformal nature of ALD was unexpected. It is desirable to remove the
surface roughness so that the samples closely match the simulated structures. Abrasive
Figure 3-17: SEM Micrograph of LPD back-filled spheres at 70 °C. Note the
flakes that peeled away from the substrate.
63
polishing and ion-milling were attempted to polish the surface but yielded no results. In
particular the problem is more complicated than just the polishing of a few nanometers of
material. Planarization of these samples would require the removal of hundreds of
nanometers of the host matrix. Spin coating, liquid phase deposition, and chemical vapor
deposition all deposit material in a conformal coating. Regardless of the method used to
back-fill the host matrix, a more advanced method of planarization is required. One
possible alternative technique that may be more successful is chemical mechanical
polishing (CMP). CMP achieves planarization and material removal through chemical
etching and large abrasive pads.62–64 Unfortunately, TiO2 is not easily etched chemically.
Incorporating CMP into the fabrication process would require substituting a different host
material that is capable of being chemically etched.
64
Figure 3-18: SEM micrographs of rough surface overlayer of planar lenses back-
filled with extended ALD.
65
Alumina Back-Filled Lenses
The same method described previously for the Savannah 200 system was used to
deposit an Al2O3 matrix. The substrate was heated at 100 °C, and Al2O3 films were formed
by alternately pulsing each precursor (0.015 s for Al, 0.015 s for water) into the sample
chamber under a N2 gas flowing at 20 sccm. Over a 72 h period, 20,000 cycles of ALD
were performed, and the deposited film was 2.1 µm thick according to ellipsometry
Figure 3-19: SEM micrograph of milled cross section of planar lenses back-filled
with extended ALD. This confirms that the spheres are successfully embedded in the
TiO2deposited by ALD. This also shows the rough overlayer of conformal spheres. The
layered appearance of the cross sections was an artifact of the milling process.
66
measurements on the silicon wafer control sample. A thicker film was deposited to ensure
an overlayer that could be polished down to remove surface roughness. After ALD, the
samples were annealed in a box furnace for 6 h at 600 °C. The Al2O3 matrix did not prove
to be easier to polish than the TiO2. Interestingly, despite the closeness in refractive indices
of SiO2 (n=1.45 and Al2O3 (n~1.6), the spheres are still visible as a white film in the
photograph shown in Figure 3-20. This might indicate optical decoupling between the
sphere material and the host matrix. Regardless of the cause of the whiteness, the spheres
will show significant backscattering losses.
Figure 3-20: LB deposited SiO2 spheres back-filled with Al2O3 via ALD. Note the
unexpected visibility of the spheres as a white film. It was predicted that spheres
embedded in a bulk material of similar refractive index would be mostly transparent.
67
Final Optical Results
The total transmission of the TiO2 (Figure 3-20-A) and Al2O3 (Figure 3-20-B)
lenses were measured using the integrating sphere. Both 500 and 700 nm spheres
embedded in TiO2 only transmitted approximately 30% of the light. Back reflectance
accounted for nearly 70% of the incident light. The Al2O3 samples transmitted more of the
light, but still had significant losses (>20%) from back scattering. Of the light that was
diffusely transmitted (scattered), no selectivity by wavelength was observed for the Al2O3
lenses. Despite fabricating several structures reasonably close to the simulations, the lenses
showed no spectrum splitting properties. The possibility of creating 3-dimensional
spectrum splitting structures cannot be ruled out by this work, but it is clear that the systems
that have been fabricated here do not exhibit spectrum splitting properties. A 2-dimensional
system consisting of nano-rods might be the more realistic pathway towards experimentally
demonstrating spectrum splitting.
68
Future Directions
Chemical Vapor Deposition for Spectrum Splitting Applications
Current user facility systems in the Penn State Materials Research Institute have
been optimized for the deposition of ultra-thin films. The Savannah 200 boasts single cycle
deposition layers of 0.6 Ang/cycle (0.06 nm/cycle) for a variety of high index materials.
The average user employs these instruments to deposit 0.1-100 nm of material and a typical
run consists of a few hours. Using the ALD to backfill TiO2 for a planar spectrum splitting
lens requires the deposition of at least 800 nm which requires 16,000 cycles and 72 h for
complete backfilling. Often, even thicker films have been created with a large overlayer
Figure 3-20: Transmission of sphere lenses embedded in TiO2 (A) and in Al2O3
(B). Both total transmission and diffuse transmission are show. None of the structures show
selective scattering of short wavelength light.
69
that can be polished off to yield a planar lens. While the run length and cost are
inconvenient, the true burden comes from accumulation of material in the instrument lines.
The water vapor is supplied through the same manifold as the metal oxide precursor so
miniscule amounts come into contact and react in the supply lines. Newer ALD instruments
correct this design flaw by supplying the water vapor though a separate inlet. Under routine
operating conditions, the Savannah 200 is taken apart once a year to replace the manifold
and vacuum lines that become clogged with built up material. As a result of the
exceptionally long deposition times required for these samples, the manifold must be
replaced after every two of our extended runs. Given the difficulty of using ALD to create
single layer spectrum splitting lenses, future research might devote time towards building
a crude chemical vapor deposition system. This system would have separate inlets for the
precursor and water vapor. This would enable faster back-filling and minimize
maintenance, while still allowing for deposition of defect free metal oxide films. More
importantly this would allow for structures containing multiple sphere layers to be created.
Future Simulations
While the work by Fan and coworkers2 constrained simulations to physically
achievable shapes, the indices of the materials were allowed to range from 1.0 to 2.6. While
there is a broad continuum of available materials in the 1.3 to 1.7 region, materials above
n=1.7 are limited to high Z-number metal oxides such as TiO2, ZrO2, TaO2, amongst others.
Future work might constrain the indices to discreet values of available materials such as
those shown in Table 3-3.
70
The simulations and planar prototype fabrication focused on a low index sphere
embedded in a high index material. There were many reasons to focus on this structure.
Silica (SiO2) spheres are commercially available at radii of every 100 nm. Furthermore, the
Stober process allows spheres to be easily grown or altered with nanometer control. Due
to the difficulty of simultaneously simulating both sphere structures and the ubiquitous
availability of silica spheres, the simulations were constrained early in the project to a low
index (ex: silica) material embedded in a high index material. The low/high (SiO2/TiO2)
simulations made sense when designing a proof of concept planar lens. Unfortunately, this
led to many unforeseen challenges with the back-filling and polishing of the high index
material. Future simulations might also explore inverting the indices of refraction. Earlier
experimental work on hemispherical lenses briefly investigated inverting sphere and matrix
materials. This inverted structure consisted of a high index sphere embedded in a low index
material. The unexplored high/low model offers some advantages for processability of
hemispherical polymer lenses. Polymers are easily processible into hemispheres as
described earlier. Most common polymers have indices of refraction that range from 1.30
to 1.50. In order to ensure fabrication of a lens from homogenous materials, it is
recommended to constrain simulations of the polymer matrix to the common easily
processible region of low index polymers from 1.30 to 1.50.
If the end goal is to create a spectrum splitting tandem solar cell and not just a
spectrum splitting lens, it may be beneficial to approach the simulations from the opposite
direction. This approach would fix the identity and area of solar cells and determine the
minimum scattering ratio necessary to attain an efficiency boost in a tandem architecture.
71
This could be used to identify the most easily polished, host matrix material necessary to
demonstrate a spectrum splitting effect.
72
Conclusions
This work attempted to design and fabricate nanostructured lenses capable of selectively
splitting the solar spectrum. If these goals were realized, they would allow the creation of
side-by-side tandem solar cells which could lower costs or increase efficiencies. Spin
coating and LB trough deposition were used to tune the spacing of sphere layers. It became
apparent that minimizing optical defects is of paramount importance for this application.
Sol-gel and spin-coating methods were presented as cost-effective and scalable techniques.
Unfortunately, these methods introduced too many defects into lenses to be of use for
prototyping lenses for research purposes. They may prove to be the ideal route to
commercialization in the future if these structures are optimized, and the final structural
parameters are more resilient to defects. For the purpose of creating a prototype planar lens,
Table 3-3: Various high refractive index metal oxides
Material Refractive Index
SiO2 1.45
Al2O3 (sapphire) 1.76
HfO2 2.11
TiO2 2.61
Nb2O5 2.34
Ta2O5 2.13
ZrO2 2.15
73
atomic layer deposition (ALD) was used as an adequate method to deposit a host matrix
while minimizing the optical defects that contribute to back scattering loses. Throughout
this work, optical methods have been described for evaluating spectrum splitting lenses.
The final planar prototype lenses exhibit massive back-scattering losses. These losses were
not predicted by theory or simulation. There is also the problem of planarizing these lenses
after the back-fill processes leave rough overlayers. Several of the remaining avenues for
creating structures that might exhibit spectrum splitting phenomena have been outlined.
The most promising direction would be carefully selecting softer host materials, in
consideration of the final finishing polish step.
74
Chapter 4
Synthesis of Ruthenium Polypyridyl Compounds for WS-DSPECs
Abstract
This chapter discusses attempts to synthesize a molecular ruthenium oxygen-
evolving catalyst and a hydroxamate-anchored ruthenium polypyridyl dye. A new class of
molecular ruthenium catalysts have recently been discovered which exhibit
unprecedentedly high turnover frequencies (TOF) that are comparable to the TOFs of
PSII.10,46,65–68 These catalysts are an ideal replacement for the IrOx nanoparticle catalysts
previously used by our group. Past research by our group has shown that electron
scavenging by IrOx nanoparticle catalysts is a significant limitation to the efficiency of
WS-DSPECs.20 In addition to their higher TOF, the molecular catalysts can reduce
scavenging if they are spaced from the electrode surface with a long carbon chain. In
Chatper 2, the limitations of the phosphonate anchoring group for attaching dye molecules
were discussed. The hydroxamate anchoring group has been shown in some limited cases
to be a superior anchoring group with superior injection.19,35,36,69 The second half of this
chapter discusses attempts to incorporate this superior hydroxamate anchoring group into
the synthesis of a Ru(bpy)3 light absorber.
75
Molecular Ruthenium Polypyridyl OER Catalyst
Background
In recent years there have been remarkable reports of photoanodes that produce
high photocurrents for the water-splitting reaction without the intentional attachment of a
catalyst.70,71 The high currents demonstrated without any identifiable catalyst have
reopened many questions about the catalytic processes that were generally accepted as
being true. This has prompted a search for an adventitious catalyst that spontaneously
forms on the electrode surface as a result of trace metal impurities present in the chemicals
used to create the cell. Many groups have attempted to identify the exact nature of the
adventitious catalyst present on the photoanode of WS-DSPECs.71–73 Our groups current
hypothesis is that the adventitious catalyst is probably NiFeO formed from trace nickel
impurities in the buffer and trace iron impurities form the horn sonicator used during the
synthesis of the TiO2 paste.73 Nevertheless, without direct evidence as to the identity of the
adventitious catalyst, any claims about catalytic water-splitting should be carefully
examined. One indirect improvement to this open problem, has been the discovery of a
new class of molecular ruthenium catalysts that exhibit turnover frequencies fast enough
to make the presence of an adventitious catalyst of minor relevance. This new class of
molecular catalysts consist of a ruthenium metal complex with the 2,2′‐bipyridine‐6,6′‐
dicarboxylate (bda) ligand (Figure 4-1). Other groups have also discoveredthat increasing
the rate of hole trapping by the catalyst could indirectly improve the stability of the light
absorber by reducing corrosion or passivation reactions of the vulnerable oxidized dye.74
76
All of these reasons contribute to the attractiveness of including a Ru(bda)L2 catalyst into
WS-DSPECs.
Results
All chemicals and solvents, if not stated, were purchased from Sigma Aldrich and
used without further purification; water used in syntheses and measurements was deionized
by Milli-Q technique. cis-Ru(DMSO)4Cl2 was prepared according to published methods.75
Figure 4-1: A) New class of Ru(bda)L2 catalysts. B) Structure of 2,2′‐bipyridine‐
6,6′‐dicarboxylate (bda) ligand
77
NMR spectra were recorded by Bruker Advance 500 spectrometer and are displayed in
Appendix D.
Synthetic procedures were followed from the literature46,68 to synthesize the silanol
anchored Ru(bda)L2 catalyst (Figure 4-4 and 4-5). The first procedure46 produced low
yields (~10%) of the complex substituted with a single pyridine group. Significant
byproducts were the starting material or the complex substituted with two pyridines. The
second procedure68 refluxes the starting material in stoichiometric methyl pyridine with an
excess of DMSO. This resulted in much higher yields (>60%) of the desired product with
one labile DMSO group that can later be substituted with the anchoring molecule.
During purification of Ru(bda)(Me-py)(DMSO) by column chromatography the
expected brown product decomposed into a blue color. NMR was performed on the crude
brown product and the blue compound and are shown in Figure D-1 and Figure D-2. No
significant peaks were evident to identify the decomposition product. In the absence of new
functional groups on the NMR, it is hypothesized that the ruthenium complex is forming a
dimer with itself, bridged by a water molecule. This hypothesis is supported by two
additional facts. The well-known ruthenium "blue dimer" OER catalyst is structurally
similar and contains a water bridge.76–78 Secondly, the catalytic water-splitting intermediate
of Ru(bda)(Me-py)2 has been isolated and structurally determined with X-ray diffraction
to contain a bridging water molecule.66 Decomposition to the dimer can be avoided by
purifying the compound quickly after the reaction and storing the purified product in a
desiccator under vacuum. The formation of pre-catalyst dimers could be beneficial to the
final goal of creating a water-splitting photoelectrochemical cell. Because the catalytic
mechanism involves two ruthenium centers66, it seems that the pre-catalyst centers need to
78
be anchored to the electrode surface in close proximity to each other. This wouldn't be an
issue at high catalyst loadings, but recent work by Ardo and coworkers21,22 has shown that
dye to catalyst loadings on the order of 100:1 are ideal, to minimize recombination rates.
Therefore, creating pre-catalyst dimers in solution, before anchoring to the surface, could
ensure that the active catalyst is properly paired on the surface. This method of pairing
catalysts before anchoring to the electrode surface could dramatically increase the catalytic
activity of water-splitting photoanodes.
Figure 4-2: Synthetic Scheme for silanol anchor ligand
Figure 4-3: Synthetic Scheme for Ru(bda) catalyst
79
Future Directions
Future work is focused on synthesizing the same catalytic center with a
phosphonate anchor to replace the silanol. The phosphonate anchor can be stored for long
periods of time and does not suffer self-polymerization like the silanol does.46 Multiple
versions of the phosphonate anchor are being synthesized that have varying chain
lengths. Meyer and coworkers have shown that using an anchor with a 13-carbon anchor
chain, spaces the catalytic site from the electrode surface and reduces recombination.79
Future work might also devise a deposition scheme for anchoring a preformed dimer to
the electrode surface. This is one potential way to ensure that catalytic sites are active in
pairs while maintaining a high dye to catalyst loading.
Hydroxamate-Anchored Ruthenium Polypyridyl Light Absorber
Previous Attempts to Synthesize Hydroxamate-Anchored Sensitizers
In Chapter 2, the phosphonate anchor group was presented as a stable anchoring
group, with moderate injection efficiency.4 In dye-sensitized solar cells, carboxylate-
anchored dyes have been shown to inject more efficiently than phosphonates, but the
carboxylate anchor is extremely susceptible to hydrolysis and desorption in the aqueous
environments of WS-DSPECs. Therefore, it is extremely desirable to replace the
80
phosphonate group with an anchor that has equivalent stability and superior injection
properties. Hydroxamate-anchored dyes have been speculated to possess this unique
combination of stability and efficient injection.19,35,36,69,80 They have electron injection
rates comparable to carboxylic-anchored dyes19 and in some cases, a hydroxamate anchor
has been shown to inject even more efficiently than the carboxylate.35 Therefore it is
desirable to find a synthetic pathway that would result in a hydroxamate anchored
ruthenium polypyridyl complex (Figure 4-6).
McNamara and coworkers35,69 successfully synthesized a terpyridine hydroxamate
anchor ligand, but were unable to use the ligand to make a ruthenium polypyridyl
sensitizers.35 When forming the metal complex, the hydroxamate was reduced to an
amide. Terpyridine-based dyes are also much less attractive for incorporation into WS-
DSPECs than bipyridine dyes due to differences in excited state lifetimes.4 Brewster and
coworkers36 also successfully made a series of ruthenium dyes with anchors on the axial
pyridines or phenylimidizoles. However, there were issues with undesirable coordination
of the hydroxamate to ruthenium metal centers.
Figure 4-4: Desired synthesis of hydroxamate anchored Ru(bpy)3 derivative
81
Synthesis of Hydroxamate Anchored Sensitizer
All chemicals and solvents, if not stated, were purchased from Sigma Aldrich and
used without further purification; water used in syntheses and measurements was deionized
by Milli-Q technique. NMR spectra were recorded by Bruker Advance 500 spectrometer
and are displayed in Appendix E.
Nucleophilic substitution of carboxylic acids is commonly achieved by temporarily
converting the carboxylic acid to a more activated anhydride intermediate that is easily
susceptible to nucleophilic attack (Figure 4-7).81–83 Alternatively, the carboxylic acid group
could be activated using a carbonyl diimidizole intermediate, but this route often produces
complications with the imidazole byproduct.84 One route to synthesizing a hydroxamate-
anchored Ru(bpy)3 involves first creating a hydroxamate ligand (Figure 4-8) and then
chelating the bipyridine to the ruthenium center. The advantage of this route is that an entire
library of dyes could be made with different substituents used to tune the absorption
properties.35 The major drawback is the exposed ruthenium metal center can encourage
other side reactions, particularly reduction of the hydroxamic acid group to an amide35 or
parasitic hydroxamate chelation of metal centers.19 For this reason, the experimental
approach was focused on synthesizing the metal complex first and then transforming the
carboxylic acid into a hydroxamic acid (Figure 4-9).
82
Figure 4-5: General scheme for the activation of carboxylic acids for nucleophilic
attack by forming an anhydride intermediate
Figure 4-6: Synthetic scheme for the synthesis of a hydroxamate ligand first, then
the coordination reaction to form a hydroxamate anchored dye.
83
Results
The carboxylic dye was synthesized according to a common literature procedure.85
One difficulty in modifying the existing procedures to synthesize an organometallic
Ru(bpy)3 derivative is finding a compatible solvent. The carboxylic acid bipyridine
((COOH)2-bpy) and Ru(bpy)3 molecules tend to be soluble in very polar solvents such as
Figure 4-7: Scheme for synthesizing a hydroxamate dye from the carboxylic
analogue
84
water or methanol. Like most organic reactions, the literature procedures use relatively
nonpolar solvents such as diethyl ether82 or THF81 for the conversion of the carboxylic acid
to hydroxamic acid. Unfortunately, the full combination of reagents is not soluble in any
of the obvious choices of replacement solvent (THF, MeCN, DCM, DMF). The full set of
reagents is somewhat soluble in methanol so the reactions were carried out in methanol.
The limited solubility of the starting materials may contribute to the lack of success. There
is also some concern that methanol could compete with hydroxyl amine as a nucleophile,
but the hydroxyl amine should be preferred. Neither of these two reasons completely
explains the lack of success for these reactions. 1H NMR was performed on the anhydride
intermediate and final product of the reactions shown in Figure 4-7 and is shown in Figure
E-2 and E-1.
Future Directions: Hydroxamate-Anchored Porphyrin Sensitizers
The work described herein has identified many reasons that a hydroxamate
functional group might not be synthetically compatible with a metal complex. For these
reasons it might be advantageous to attempt the synthesis of a hydroxamate anchored
porphyrin sensitizer. In particular a “free-base” porphyrin (without a metal center) might
avoid many of the complications encountered when attempting a Ru(bpy)3 analogue.
Porphyrins offer several advantages over Ru(bpy)3 derivatives and the synthesis might be
more compatible with installing a hydroxamate anchor. At this time there are no known
literature examples of a hydroxamate-anchored porphyrin.
85
Although some porphyrins have lower hole diffusion constants than RuP,86 there
are other beneficial characteristics that compensate for this drawback. It is well known that
the hole diffusion constant (Dapp) is correlated with sensitizer surface coverage, up to a
certain saturation level.20 Our group has shown that porphyrins are deposited at higher
surface coverages than Ru(bpy)3 sensitizers.86 The higher surface coverage is likely
explained by the spherical nature of the Ru(bpy)3 analogues and the flat, vertical structure
of the porphyrins. Additionally the slower rate of recombination to porphyrin sensitizers
mitigates the low Dapp values.86 In the last year there has been a single report on a hydrazide
anchored porphyrin.87 Although the hydrazide functional group is different from the
hydroxamate functional group, the synthetic steps to create both are very similar, so there
is still hope that a hydroxamate porphyrin can be synthesized in the future.
Conclusions
In the first half of this Chapter, the synthesis of a silanol-anchored Ru(bda)L2 OER catalyst
was accomplished. Future work on replacing the silanol anchor with a more convenient
phosphonate anchor was discussed. In the second half of this chapter, the synthesis of a
hydroxamate-anchored ruthenium polypyridal light absorber was attempted. Multiple
routes were discussed and the most attractive was attempted. Possible explanations for
failure were presented. Lastly, a hydroxamate-anchored porphyrin sensitizer might be an
alternative if a viable synthetic route to a Ru(bpy)3 derivative is not possible.
86
Chapter 5
Conclusions
This thesis has broadly focused on synthesizing molecules and nanostructured materials
with applications for renewable energy. In Chapter 2, a novel oligomeric ruthenium polypyridal
light absorber was synthesized. The structure and photophysical properties were characterized. The
oligomer sensitizer is notable for its increased stability at higher pH and its improved hole diffusion
properties. Both properties have been identified as limitations of the currently used monomer
sensitizer. The hole diffusion is also interesting in that it opens up new pathways to study the hole
diffusion process. In particular, the chain lengths of the linking ligand can be varied to analyze the
effect on the hole diffusion constant. It is likely that the shortest chain length (3 carbon linker) is
not ideal because the intrachain diffusion constant plays a role. Hybrid systems of oligomer and
monomer might provide some additional benefit to mitigate this intrachain gap.
In Chapter 3, a nanostructured lens was fabricated with spheres embedded in a high index
of refraction host matrix. The goal was to create a structure that would selectively scatter short-
wavelength light. This could create new possibilities for side-by-side tandem photovoltaic modules.
The material parameters for the fabricated structures were informed by the simulation of cylindrical
gratings. The bulk of the simulations were focused on low refractive index cylinders embedded in
a high refractive index host matrix. The simulations yielded two significant features. Firstly, a large
difference in refractive index (ΔR ~ 0.8) between the two materials is necessary to selectively
87
scatter light. Secondly, the cylinders should not be in contact with each other. Further simulation
experiments could be done to identify realistic materials for use in this system. In particular, the
reverse system, with high refractive index spheres/cylinders embedded in a low index host matrix
(ex: PMMA) is attractive for the ease of processability.
The experimental structures described and fabricated herein, were composed of SiO2
spheres embedded in a TiO2 host matrix deposited by ALD. SiO2 spheres were chosen for the ready
availability, easy deposition, and refractive index. A Langmuir-Blodgett trough was used to deposit
spheres in a dense, disorganized monolayer. Expensive and time consuming ALD was used to
deposit µm films of dense optically clear TiO2. Early work depositing TiO2 films via inexpensive
sol-gel methods resulted in too many random scattering defects. In the future, specially designed
large-scale chemical vapor deposition (CVD) systems might be the ideal halfway point between
these two methods. The fabricated structures did not display the selective scattering properties
desired. It is hypothesized that cylindrical structures are required to create a spectrum splitting lens.
Future work involving cylindrical structures will have to overcome significant synthetic hurdles of
growing nanostructured cylinders with radii of ~500 nm and lengths in the µm regime. Further
challenges would involve depositing and aligning the cylinders in a periodic manner. In light of
these challenges, the most expedient route to realizing spectrum splitting lenses, may be to explore
further simulations and the reverse index structures.
In recent years, the cost of silicon photovoltaic modules has dropped further than expected
as a result of economies of scale. Studies published by the National Renewable Labs in Colorado
show that the actual cost of the photovoltaic cell is no longer the most expensive real-world cost.
The two largest real-world costs of installing photovoltaics are the human labor and governmental
permits. Although one of the original motivations for this spectrum splitting (cost of silicon PVs)
is no longer an issue, tandem cells remain attractive for many reasons. Because the costs of
installation are dominated by installation and permitting, it is now far more economically attractive
88
to install relatively expensive high-efficiency photovoltaics, such as tandem cells, which more
effectively utilize the available solar spectrum. As a result of the fundamental nature of light and
the Shockley-Queisser limit, a photovoltaic with two complementary absorbers will always harvest
the available light more efficiently. Although dye-sensitized solar cells (DSSCs) have failed to gain
traction as a result of instability issues revolving around their liquid electrolytes, solid state
perovskite solar cells which do not contain a liquid electrolyte are poised to fill this gap. Silicon
photovoltaics routinely have 10-20 year lifetimes and near the end of their lives, experience slow
decay of current efficiencies. Although much work has been done to extend the lifetime of
perovskite cells to many years, when they do reach the end of their lifetime, they dramatically cease
to function. This can be an issue in a tandem cell, where the current flows through both cells. If one
module dies, the entire cell ceases to generate any electricity. As a result, future design of tandem
cells might be engineered so that 5-10 years into their lifecycle a “dead” solid state perovskite cell
might be removed and replaced by a new perovskite cell. The side-by-side architecture of a
spectrum-splitting tandem cell might offer advantages to the replacement of the perovskite cell
component.
In Chapter 4, the synthesis of a molecular OER catalyst and a hydroxamate anchored dye
were attempted. The synthesis of the molecular catalyst was successful, but a phosphonate anchor
was found to be more desirable than the silanol anchor described herein. It is difficult to store and
work with the silanol anchor over long periods of time because the functional group tends to self-
polymerize. Future work is focused on incorporating a longer anchor chain to decrease backwards
recombination to the catalyst. This exploits the distance dependence of electron transfer. An even
better strategy would further distance the catalyst from the surface by attaching the catalyst to the
dye molecule (instead of the electrode surface). This greater degree of control could be achieved at
the cost of more synthetic complexity. Future work might also devise a deposition scheme for
anchoring a preformed dimer to the electrode. This would ensure that catalytic sites are active in
89
pairs at close proximity, even at low catalyst loadings. Whatever the route, it is clear that in the
future the Ru(bda)L2 class of catalysts will be at the forefront of the water-splitting field and
significant effort (synthetic or otherwise) will be exerted to control their location and attachment
in photoelectrochemical cells.
The synthesis of the hydroxamate-anchored dye was not successful. Future efforts might
focus on synthesizing a free-base porphyrin that will not have the synthetic complications
associated with a transition metal center. The significant hurdle to either a ruthenium polypyridal
dye or a porphyrin dye lies in finding a synthetic route to install the delicate hydroxamate group
while synthesizing the dye center and avoiding complicated organometallic redox side reactions.
90
References
(1) Barber, G. D.; Hoertz, P. G.; Lee, S.-H. A.; Abrams, N. M.; Mikulca, J.; Mallouk,
T. E.; Liska, P.; Zakeeruddin, S. M.; Grätzel, M.; Ho-Baillie, A.; et al. Utilization
of Direct and Diffuse Sunlight in a Dye-Sensitized Solar Cell — Silicon
Photovoltaic Hybrid Concentrator System. J. Phys. Chem. Lett. 2011, 2, 581–585.
(2) Fan, L.; Faryad, M.; Barber, G.; Mallouk, T.; Monk, P. B.; Lakhtakia, A.
Optimization of a Spectrum Splitter Using Differential Evolution Algorithm for
Solar Cell Applications. J. Photonics Energy 2015, 5, 1–15.
(3) Lewis, N. S.; Nocera, D. G. Powering the Planet: Chemical Challenges in Solar
Energy Utilization. PNAS 2006, 103, 15729–15735.
(4) Swierk, J. R.; Mallouk, T. E. Design and Development of Photoanodes for Water-
Splitting Dye-Sensitized Photoelectrochemical Cells. Chem. Soc. Rev. 2013, 42,
2357–2387.
(5) Nocera, D. G. The Artificial Leaf. Acc. Chem. Res. 2012, 45, 767–776.
(6) Our Energy Sources: Electricity http://needtoknow.nas.edu/energy/energy-
sources/electricity/ (accessed Jun 20, 2019).
(7) Electricity explained https://www.eia.gov/energyexplained/electricity/electricity-
in-the-us.php (accessed Jun 19, 2019).
(8) Schäfer, A. W.; Evans, A. D.; Reynolds, T. G.; Dray, L. Costs of Mitigating CO2
91
Emissions from Passenger Aircraft. Nat. Clim. Chang. 2015, 6, 412.
(9) Li, Y. C.; Zhou, D.; Yan, Z.; Gonçalves, R. H.; Salvatore, D. A.; Berlinguette, C.
P.; Mallouk, T. E. Electrolysis of CO2 to Syngas in Bipolar Membrane-Based
Electrochemical Cells. ACS Energy Lett. 2016, 1, 1149–1153.
(10) Duan, L.; Bozoglian, F.; Mandal, S.; Stewart, B.; Privalov, T.; Llobet, A.; Sun, L.
A Molecular Ruthenium Catalyst with Water-Oxidation Activity Comparable to
That of Photosystem II. Nat Chem 2012, 4, 418–423.
(11) Dale, M.; Benson, S. M. Energy Balance of the Global Photovoltaic (PV) Industry
- Is the PV Industry a Net Electricity Producer? Environ. Sci. Technol. 2013, 47,
3482–3489.
(12) Holladay, J. D.; Hu, J.; King, D. L.; Wang, Y. An Overview of Hydrogen
Production Technologies. Catal. Today 2009, 139, 244–260.
(13) Surendranath, Y.; Lutterman, D. A.; Liu, Y.; Nocera, D. G. Nucleation, Growth,
and Repair of a Cobalt-Based Oxygen Evolving Catalyst. J. Am. Chem. Soc. 2012,
134, 6326–6336.
(14) Grätzel, M. Photoelectrochemical Cells. Nature 2001, 414, 338–344.
(15) Hanson, K.; Brennaman, M. K.; Luo, H.; Glasson, C. R. K.; Concepcion, J. J.;
Song, W.; Meyer, T. J. Photostability of Phosphonate-Derivatized, RuII
Polypyridyl Complexes on Metal Oxide Surfaces. ACS Appl. Mater. Interfaces
2012, 4, 1462–1469.
(16) Xu, P.; McCool, N. S.; Mallouk, T. E. Water Splitting Dye-Sensitized Solar Cells.
92
Nano Today 2017, 14, 42–58.
(17) Moore, G. F.; Blakemore, J. D.; Milot, R. L.; Hull, J. F.; Song, H.; Cai, L.;
Schmuttenmaer, C. A.; Crabtree, R. H.; Brudvig, G. W. A Visible Light Water-
Splitting Cell with a Photoanode Formed by Codeposition of a High-Potential
Porphyrin and an Iridium Water-Oxidation Catalyst. Energy Environ. Sci. 2011, 4,
2389–2392.
(18) Vagnini, M. T.; Smeigh, A. L.; Blakemore, J. D.; Eaton, S. W.; Schley, N. D.;
D\textquoterightSouza, F.; Crabtree, R. H.; Brudvig, G. W.; Co, D. T.;
Wasielewski, M. R. Ultrafast Photodriven Intramolecular Electron Transfer from
an Iridium-Based Water-Oxidation Catalyst to Perylene Diimide Derivatives.
Proc. Natl. Acad. Sci. 2012, 109, 15651–15656.
(19) Materna, K. L.; Crabtree, R. H.; Brudvig, G. W. Anchoring Groups for
Photocatalytic Water Oxidation on Metal Oxide Surfaces. Chem. Soc. Rev. 2017,
46, 6099–6110.
(20) Swierk, J. R.; McCool, N. S.; Saunders, T. P.; Barber, G. D.; Mallouk, T. E.
Effects of Electron Trapping and Protonation on the Efficiency of Water-Splitting
Dye-Sensitized Solar Cells. J. Am. Chem. Soc. 2014, 136, 10974–10982.
(21) Chen, H.-Y.; Ardo, S. Direct Observation of Sequential Oxidations of a Titania-
Bound Molecular Proxy Catalyst Generated through Illumination of Molecular
Sensitizers. Nat. Chem. 2017, 10, 17–23.
(22) Ardo, S.; Meyer, G. J. Direct Observation of Photodriven Intermolecular Hole
93
Transfer across TiO2 Nanocrystallites: Lateral Self-Exchange Reactions and
Catalyst Oxidation. J. Am. Chem. Soc. 2010, 132, 9283–9285.
(23) Nogueira, A. F.; Longo, C.; De Paoli, M. A. Polymers in Dye Sensitized Solar
Cells: Overview and Perspectives. Coordination Chemistry Reviews. 2004, pp
1455–1468.
(24) Krebs, F. C.; Biancardo, M. Dye Sensitized Photovoltaic Cells: Attaching
Conjugated Polymers to Zwitterionic Ruthenium Dyes. Sol. Energy Mater. Sol.
Cells 2006, 90, 142–165.
(25) Lapides, A. M.; Ashford, D. L.; Hanson, K.; Torelli, D. A.; Templeton, J. L.;
Meyer, T. J. Stabilization of a Ruthenium(II) Polypyridyl Dye on Nanocrystalline
TiO2 by an Electropolymerized Overlayer. J. Am. Chem. Soc. 2013, 135, 15450–
15458.
(26) Fang, Z.; Keinan, S.; Alibabaei, L.; Luo, H.; Ito, A.; Meyer, T. J. Controlled
Electropolymerization of Ruthenium(II) Vinylbipyridyl Complexes in Mesoporous
Nanoparticle Films of TiO2. Angew. Chemie Int. Ed. 2014, 53, 4872–4876.
(27) Dupray, L. M.; Devenney, M.; Striplin, D. R.; Meyer, T. J. An Antenna Polymer
for Visible Energy Transfer. J. Am. Chem. Soc. 1997, 119, 10243–10244.
(28) Moss, J. A.; Yang, J. C.; Stipkala, J. M.; Wen, X.; Bignozzi, C. A.; Meyer, G. J.;
Meyer, T. J. Sensitization and Stabilization of TiO2 Photoanodes with
Electropolymerized Overlayer Films of Ruthenium and Zinc Polypyridyl
Complexes: A Stable Aqueous Photoelectrochemical Cell. Inorg. Chem. 2004, 43,
94
1784–1792.
(29) Wee, K.-R.; Brennaman, M. K.; Alibabaei, L.; Farnum, B. H.; Sherman, B.;
Lapides, A. M.; Meyer, T. J. Stabilization of Ruthenium(II) Polypyridyl
Chromophores on Nanoparticle Metal-Oxide Electrodes in Water by Hydrophobic
PMMA Overlayers. J. Am. Chem. Soc. 2014, 136, 13514–13517.
(30) McCool, N. S.; Swierk, J. R.; Nemes, C. T.; Schmuttenmaer, C. A.; Mallouk, T. E.
Dynamics of Electron Injection in SnO2/TiO2 Core/Shell Electrodes for Water-
Splitting Dye-Sensitized Photoelectrochemical Cells. J. Phys. Chem. Lett. 2016, 7,
2930–2934.
(31) Lee, S.-H. A.; Zhao, Y.; Hernandez-Pagan, E. A.; Blasdel, L.; Youngblood, W. J.;
Mallouk, T. E. Electron Transfer Kinetics in Water Splitting Dye-Sensitized Solar
Cells Based on Core--Shell Oxide Electrodes. Faraday Discuss. 2012, 155, 165–
176.
(32) Sherman, B. D.; Ashford, D. L.; Lapides, A. M.; Sheridan, M. V; Wee, K.-R.;
Meyer, T. J. Light-Driven Water Splitting with a Molecular Electroassembly-
Based Core/Shell Photoanode. J. Phys. Chem. Lett. 2015, 6, 3213–3217.
(33) Xu, P.; Mallouk, T. E. Charge Transfer Dynamics in Aqueous Dye-Sensitized
Photoelectrochemical Cells: Implications for Water Splitting Efficiency. J. Phys.
Chem. C 2019, 123, 299–305.
(34) Koenigsmann, C.; Ripolles, T. S.; Brennan, B. J.; Negre, C. F. A.; Koepf, M.;
Durrell, A. C.; Milot, R. L.; Torre, J. A.; Crabtree, R. H.; Batista, V. S.; et al.
95
Substitution of a Hydroxamic Acid Anchor into the MK-2 Dye for Enhanced
Photovoltaic Performance and Water Stability in a DSSC. Phys. Chem. Chem.
Phys. 2014, 16, 16629–16641.
(35) McNamara, W. R.; Milot, R. L.; Song, H.; Snoeberger III, R. C.; Batista, V. S.;
Schmuttenmaer, C. A.; Brudvig, G. W.; Crabtree, R. H. Water-Stable{,}
Hydroxamate Anchors for Functionalization of TiO2 Surfaces with Ultrafast
Interfacial Electron Transfer. Energy Environ. Sci. 2010, 3, 917–923.
(36) Brewster, T. P.; Konezny, S. J.; Sheehan, S. W.; Martini, L. A.; Schmuttenmaer,
C. A.; Batista, V. S.; Crabtree, R. H. Hydroxamate Anchors for Improved
Photoconversion in Dye-Sensitized Solar Cells. Inorg. Chem. 2013, 52, 6752–
6764.
(37) Gillaizeau-Gauthier, I.; Odobel, F.; Alebbi, M.; Argazzi, R.; Costa, E.; Bignozzi,
C. A.; Qu, P.; Meyer, G. J. Phosphonate-Based Bipyridine Dyes for Stable
Photovoltaic Devices. Inorg. Chem. 2001, 40, 6073–6079.
(38) Evans, I. P.; Spencer, A.; Wilkinson, G. Dichlorotetrakis (Dimethyl Sulphoxide)
Ruthenium (II) and Its Use as a Source Material for Some New Ruthenium (II)
Complexes. J. Chem. Soc., Dalt. Trans. 1973, No. 2, 204–209.
(39) Schmehl, R. H.; Auerbach, R. A.; Wacholtz, W. F.; Elliott, C. M.; Freitag, R. A.;
Merkert, J. W. Formation and Photophysical Properties of Iron-Ruthenium
Tetranuclear Bipyridyl Complexes of the Type {[(Bpy)2Ru(L-L)]3Fe}. Inorg.
Chem. 1986, 25, 2440–2445.
96
(40) Hara, M.; Lean, J. T.; Mallouk, T. E. Photocatalytic Oxidation of Water by Silica-
Supported Tris (4, 4’-Dialkyl-2, 2’-Bipyridyl) Ruthenium Polymeric Sensitizers
and Colloidal Iridium Oxide. Chem. Mater. 2001, 13, 4668–4675.
(41) Xu, P.; Gray, C. L.; Xiao, L.; Mallouk, T. E. Charge Recombination with
Fractional Reaction Orders in Water-Splitting Dye-Sensitized
Photoelectrochemical Cells. J. Am. Chem. Soc. 2018, 140, 11647–11654.
(42) Hanson, K.; Brennaman, M. K.; Ito, A.; Luo, H.; Song, W.; Parker, K. A.; Ghosh,
R.; Norris, M. R.; Glasson, C. R. K.; Concepcion, J. J.; et al. Structure–Property
Relationships in Phosphonate-Derivatized, RuII Polypyridyl Dyes on Metal Oxide
Surfaces in an Aqueous Environment. J. Phys. Chem. C 2012, 116, 14837–14847.
(43) Matsuzaki, H.; Murakami, T. N.; Masaki, N.; Furube, A.; Kimura, M.; Mori, S.
Dye Aggregation Effect on Interfacial Electron-Transfer Dynamics in Zinc
Phthalocyanine-Sensitized Solar Cells. J. Phys. Chem. C 2014, 118, 17205–17212.
(44) Khazraji, A. C.; Hotchandani, S.; Das, S.; Kamat, P. V. Controlling Dye
(Merocyanine-540) Aggregation on Nanostructured TiO2 Films. An Organized
Assembly Approach for Enhancing the Efficiency of Photosensitization. J. Phys.
Chem. B 1999, 103, 4693–4700.
(45) Hyde, J. T.; Hanson, K.; Vannucci, A. K.; Lapides, A. M.; Alibabaei, L.; Norris,
M. R.; Meyer, T. J.; Harrison, D. P. Electrochemical Instability of Phosphonate-
Derivatized, Ruthenium(III) Polypyridyl Complexes on Metal Oxide Surfaces.
ACS Appl. Mater. Interfaces 2015, 7, 9554–9562.
97
(46) Gao, Y.; Ding, X.; Liu, J.; Wang, L.; Lu, Z.; Li, L.; Sun, L. Visible Light Driven
Water Splitting in a Molecular Device with Unprecedentedly High Photocurrent
Density. J. Am. Chem. Soc. 2013, 135, 4219–4222.
(47) McCool, N. S.; Swierk, J. R.; Nemes, C. T.; Saunders, T. P.; Schmuttenmaer, C.
A.; Mallouk, T. E. Proton-Induced Trap States, Injection and Recombination
Dynamics in Water-Splitting Dye-Sensitized Photoelectrochemical Cells. ACS
Appl. Mater. Interfaces 2016, 8, 16727–16735.
(48) Krüger, J.; Plass, R.; Grätzel, M.; Cameron, P. J.; Peter, L. M. Charge Transport
and Back Reaction in Solid-State Dye-Sensitized Solar Cells: A Study Using
Intensity-Modulated Photovoltage and Photocurrent Spectroscopy. J. Phys. Chem.
B 2003, 107, 7536–7539.
(49) Feng, X.; Zhu, K.; Frank, A. J.; Grimes, C. A.; Mallouk, T. E. Rapid Charge
Transport in Dye-Sensitized Solar Cells Made from Vertically Aligned Single-
Crystal Rutile TiO2 Nanowires. Angew. Chemie 2012, 124, 2781–2784.
(50) Knauf, R. R.; Brennaman, M. K.; Alibabaei, L.; Norris, M. R.; Dempsey, J. L.
Revealing the Relationship between Semiconductor Electronic Structure and
Electron Transfer Dynamics at Metal Oxide–Chromophore Interfaces. J. Phys.
Chem. C 2013, 117, 25259–25268.
(51) Sherman, B. D.; Sheridan, M. V; Wee, K.-R.; Marquard, S. L.; Wang, D.;
Alibabaei, L.; Ashford, D. L.; Meyer, T. J. A Dye-Sensitized Photoelectrochemical
Tandem Cell for Light Driven Hydrogen Production from Water. J. Am. Chem.
98
Soc. 2016, 138, 16745–16753.
(52) Yella, A.; Lee, H.-W.; Tsao, H. N.; Yi, C.; Chandiran, A. K.; Nazeeruddin, M. K.;
Diau, E. W.-G.; Yeh, C.-Y.; Zakeeruddin, S. M.; Grätzel, M. Porphyrin-Sensitized
Solar Cells with Cobalt (II/III)--Based Redox Electrolyte Exceed 12 Percent
Efficiency. Science (80-. ). 2011, 334, 629–634.
(53) Chandiran, A. K.; Tetreault, N.; Humphry-Baker, R.; Kessler, F.; Baranoff, E.; Yi,
C.; Nazeeruddin, M. K.; GraÌ<88>tzel, M. Subnanometer Ga2O3 Tunnelling
Layer by Atomic Layer Deposition to Achieve 1.1 V Open-Circuit Potential in
Dye-Sensitized Solar Cells. Nano Lett. 2012, 12, 3941–3947.
(54) Chung, I.; Lee, B.; He, J.; Chang, R. P. H.; Kanatzidis, M. G. All-Solid-State Dye-
Sensitized Solar Cells with High Efficiency. Nature 2012, 485, 486–489.
(55) Liu, M.; Johnston, M. B.; Snaith, H. J. Efficient Planar Heterojunction Perovskite
Solar Cells by Vapour Deposition. Nature 2013, 501, 395–398.
(56) Hardin, B. E.; Snaith, H. J.; McGehee, M. D. The Renaissance of Dye-Sensitized
Solar Cells. Nat. Photonics 2012, 6, 162–169.
(57) Jung, H. S.; Lee, J.-K. Dye Sensitized Solar Cells for Economically Viable
Photovoltaic Systems. J. Phys. Chem. Lett. 2013, 4, 1682–1693.
(58) Mihi, A.; Ocaña, M.; M\’\iguez, H. Oriented Colloidal-Crystal Thin Films by
Spin-Coating Microspheres Dispersed in Volatile Media. Adv. Mater. 2006, 18,
2244–2249.
(59) Terry, K. W.; Tilley, T. D. Trialkoxysiloxy Complexes as Precursors to
99
MO2.Cntdot.4SiO2 (M = Ti, Zr, Hf) Materials. Chem. Mater. 1991, 3, 1001–1003.
(60) Szekeres, M.; Kamalin, O.; Schoonheydt, R. A.; Wostyn, K.; Clays, K.; Persoons,
A.; Dékány, I. Ordering and Optical Properties of Monolayers and Multilayers of
Silica Spheres Deposited by the Langmuir--Blodgett Method. J. Mater. Chem.
2002, 12, 3268–3274.
(61) Lee, S.-H. A.; Abrams, N. M.; Hoertz, P. G.; Barber, G. D.; Halaoui, L. I.;
Mallouk, T. E. Coupling of Titania Inverse Opals to Nanocrystalline Titania
Layers in Dye-Sensitized Solar Cells? J. Phys. Chem. B 2008, 112, 14415–14421.
(62) Malshe, A. P.; Park, B. S.; Brown, W. D.; Naseem, H. A. A Review of Techniques
for Polishing and Planarizing Chemically Vapor-Deposited (CVD) Diamond Films
and Substrates. Diam. Relat. Mater. 1999, 8, 1198–1213.
(63) Nanz, G.; Camilletti, L. E. Modeling of Chemical-Mechanical Polishing: A
Review. IEEE Trans. Semicond. Manuf. 1995, 8, 382–389.
(64) Oganov, A. R.; Lyakhov, A. O. Towards the Theory of Hardness of Materials. J.
Superhard Mater. 2010, 32, 143–147.
(65) Xu, Y.; Åkermark, T.; Gyollai, V.; Zou, D.; Eriksson, L.; Duan, L.; Zhang, R.;
Åkermark, B.; Sun, L. A New Dinuclear Ruthenium Complex as an Efficient
Water Oxidation Catalyst. Inorg. Chem. 2009, 48, 2717–2719.
(66) Duan, L.; Fischer, A.; Xu, Y.; Sun, L. Isolated Seven-Coordinate Ru(IV) Dimer
Complex with [HOHOH]− Bridging Ligand as an Intermediate for Catalytic Water
Oxidation. J. Am. Chem. Soc. 2009, 131, 10397–10399.
100
(67) Sheridan, M. V; Sherman, B. D.; Marquard, S. L.; Fang, Z.; Ashford, D. L.; Wee,
K.-R.; Gold, A. S.; Alibabaei, L.; Rudd, J. A.; Coggins, M. K.; et al. Electron
Transfer Mediator Effects in Water Oxidation Catalysis by Solution and Surface-
Bound Ruthenium Bpy-Dicarboxylate Complexes. J. Phys. Chem. C 2015, 119,
25420–25428.
(68) Li, F.; Fan, K.; Wang, L.; Daniel, Q.; Duan, L.; Sun, L. Immobilizing Ru(Bda)
Catalyst on a Photoanode via Electrochemical Polymerization for Light-Driven
Water Splitting. ACS Catal. 2015, 5, 3786–3790.
(69) McNamara, W. R.; Snoeberger III, R. C.; Li, G.; Richter, C.; Allen, L. J.; Milot, R.
L.; Schmuttenmaer, C. A.; Crabtree, R. H.; Brudvig, G. W.; Batista, V. S.
Hydroxamate Anchors for Water-Stable Attachment to TiO2 Nanoparticles.
Energy Environ. Sci. 2009, 2, 1173–1175.
(70) McCool, N. S.; Swierk, J. R.; Nemes, C. T.; Saunders, T. P.; Schmuttenmaer, C.
A.; Mallouk, T. E. Proton-Induced Trap States, Injection and Recombination
Dynamics in Water-Splitting Dye-Sensitized Photoelectrochemical Cells. ACS
Appl. Mater. Interfaces 0, null.
(71) Roger, I.; Symes, M. D. Efficient Electrocatalytic Water Oxidation at Neutral and
High PH by Adventitious Nickel at Nanomolar Concentrations. J. Am. Chem. Soc.
2015, 137, 13980–13988.
(72) Roger, I.; Shipman, M. A.; Symes, M. D. Earth-Abundant Catalysts for
Electrochemical and Photoelectrochemical Water Splitting. Nat. Rev. Chem. 2017,
101
1, 3.
(73) McCool, N. S. UNDERSTANDING AND CONTROLLING THE KINETICS OF
ELECTRON TRANSFER EVENTS IN THE WATER-SPLITTING DYE-
SENSITIZED PHOTOELECTROCHEMICAL CELL, The Pennsylvania State
University, 2016.
(74) Matheu, R.; Moreno-Hernandez, I. A.; Sala, X.; Gray, H. B.; Brunschwig, B. S.;
Llobet, A.; Lewis, N. S. Photoelectrochemical Behavior of a Molecular Ru-Based
Water-Oxidation Catalyst Bound to TiO2-Protected Si Photoanodes. J. Am. Chem.
Soc. 2017, 139, 11345–11348.
(75) Alessio, E.; Mestroni, G.; Nardin, G.; Attia, W. M.; Calligaris, M.; Sava, G.;
Zorzet, S. Cis- and Trans-Dihalotetrakis(Dimethyl Sulfoxide)Ruthenium(II)
Complexes (RuX2(DMSO)4; X = Cl, Br): Synthesis, Structure, and Antitumor
Activity. Inorg. Chem. 1988, 27, 4099–4106.
(76) Jurss, J. W.; Concepcion, J. C.; Norris, M. R.; Templeton, J. L.; Meyer, T. J.
Surface Catalysis of Water Oxidation by the Blue Ruthenium Dimer. Inorg. Chem.
2010, 49, 3980–3982.
(77) Liu, F.; Concepcion, J. J.; Jurss, J. W.; Cardolaccia, T.; Templeton, J. L.; Meyer,
T. J. Mechanisms of Water Oxidation from the Blue Dimer to Photosystem II.
Inorg. Chem. 2008, 47, 1727–1752.
(78) Concepcion, J. J.; Jurss, J. W.; Templeton, J. L.; Meyer, T. J. Mediator-Assisted
Water Oxidation by the Ruthenium ?Blue Dimer? Cis,Cis-
102
[(Bpy)2(H2O)RuORu(OH2)(Bpy)2]4+. Proc. Natl. Acad. Sci. 2008, 105, 17632–
17635.
(79) Wang, D.; Marquard, S. L.; Troian-Gautier, L.; Sheridan, M. V; Sherman, B. D.;
Wang, Y.; Eberhart, M. S.; Farnum, B. H.; Dares, C. J.; Meyer, T. J. Interfacial
Deposition of Ru(II) Bipyridine-Dicarboxylate Complexes by Ligand Substitution
for Applications in Water Oxidation Catalysis. J. Am. Chem. Soc. 0, null.
(80) Brennan, B. J.; Koenigsmann, C.; Materna, K. L.; Kim, P. M.; Koepf, M.;
Crabtree, R. H.; Schmuttenmaer, C. A.; Brudvig, G. W. Surface-Induced
Deprotection of THP-Protected Hydroxamic Acids on Titanium Dioxide. J. Phys.
Chem. C 2016, 120, 12495–12502.
(81) Ho, C. Y.; Strobel, E.; Ralbovsky, J.; Galemmo, R. A. Improved Solution- and
Solid-Phase Preparation of Hydroxamic Acids from Esters. J. Org. Chem. 2005,
70, 4873–4875.
(82) Reddy, A. S.; Kumar, M. S.; Reddy, G. R. A Convenient Method for the
Preparation of Hydroxamic Acids. Tetrahedron Lett. 2000, 41, 6285–6288.
(83) Rho, H.-S.; Baek, H.-S.; Ahn, S.-M.; Yoo, J.-W.; Kim, D.-H.; Kim, H.-G.
Hydroxamic Acid Derivatives as Anti-Melanogenic Agents: The Importance of a
Basic Skeleton and Hydroxamic Acid Moiety. Bull. Korean Chem. Soc. 2009, 30,
475–478.
(84) Usachova, N.; Leitis, G.; Jirgensons, A.; Kalvinsh, I. Synthesis of Hydroxamic
Acids by Activation of Carboxylic Acids with N,N′-Carbonyldiimidazole:
103
Exploring the Efficiency of the Method. Synth. Commun. 2010, 40, 927–935.
(85) Terpetschnig, E.; Szmacinski, H.; Malak, H.; Lakowicz, J. R. Metal-Ligand
Complexes as a New Class of Long-Lived Fluorophores for Protein
Hydrodynamics. Biophys. J. 1995, 68, 342–350.
(86) Swierk, J. R.; Méndez-Hernández, D. D.; McCool, N. S.; Liddell, P.; Terazono,
Y.; Pahk, I.; Tomlin, J. J.; Oster, N. V; Moore, T. A.; Moore, A. L.; et al. Metal-
Free Organic Sensitizers for Use in Water-Splitting Dye-Sensitized
Photoelectrochemical Cells. Proc. Natl. Acad. Sci. 2015, 112, 1681–1686.
(87) Jia, H.-L.; Peng, Z.-J.; Guan, M.-Y. New Porphyrin Dyes Containing a Hydrazide
Anchor for Dye-Sensitized Solar Cells. New J. Chem. 2018, 42, 13770–13774.
(88) Hanlon, A. D.; Milosavljevic, B. H. Appropriate Excitation Wavelength Removes
Obfuscations from Pyrene Excimer Kinetics and Mechanism Studies. Photochem.
Photobiol. Sci. 2013, 12, 787–797.
(89) Brennaman, M. K.; Patrocinio, A. O. T.; Song, W.; Jurss, J. W.; Concepcion, J. J.;
Hoertz, P. G.; Traub, M. C.; Murakami Iha, N. Y.; Meyer, T. J. Interfacial Electron
Transfer Dynamics Following Laser Flash Photolysis of [Ru(Bpy)2((4,4′-
PO3H2)2bpy)]2+ in TiO2 Nanoparticle Films in Aqueous Environments.
ChemSusChem 2011, 4, 216–227.
104
Appendix A
Intensity Modulated Photovoltage Details
IMVS was conducted using an Autolab potentiostat (PGSTAT128N) in
combination with an Autolab LED Driver. A 470 nm LED light (LDC470, Metrohm),
driven by the LED Driver, was used as the light source. The electrolyte was either 10 mM
acetic acid/sodium acetate solution (pH 4.7) or 10 mM sodium phosphate buffer (pH 5.8,
6.8, or 7.3). All electrolyte solutions were degassed by purging with Ar. IMVS
measurements were carried out at ambient temperature (23−24 °C) in the three-electrode
configuration using Ag/AgCl (3 M NaCl) as the reference electrode and Pt wire as the
counter electrode. All reagents were purchased from Sigma Aldrich and used without
further purification. All potentials reported here are referenced to the reference electrode
unless otherwise noted. In the IMVS experiments, the electrode was held at open-circuit
potential by setting its current zero. The incident light was modulated in a sinusoidal
fashion at a given frequency with a modulation magnitude of 10% of the DC level.
Simultaneously, the electrode potential modulation at the same frequency was analyzed by
the potentiostat. By scanning a range of frequencies, one can generate a Nyquist plot
representing the real and imaginary parts of the frequency responses. Light intensity was
measured by a Si photodiode (Thorlabs, S130C) and ranged from approximately 0.5-7.0
mW/cm2.
105
In a photoelectrode, all the photo-injected electrons must recombine under open-
circuit conditions. With the simplifying assumption of a first-order charge recombination
process and a rate constant of k1, the balance of charge injection and recombination can
be expressed by equation (1):
𝑑𝑛
𝑑𝑡= 𝑞𝐴𝐼 − 𝑞𝑘1𝑛 (1)
where q is the electron charge, A is the fraction of photons that result in electron
injection, I0 is the incident photon flux at steady state, and n is the injected electron
density.
Under constant illumination at an intensity of I0, the electron concentration reaches
steady state and equation (1) can be written as follows:
𝑑𝑛
𝑑𝑡= 0 = 𝑞𝐴𝐼0 − 𝑞𝑘1𝑛0 (2)
Where 𝑛0 is the injected electron density in TiO2 at steady state.
When the light is modulated with a frequency of 𝜔 and an amplitude of M, as represented
in equation (3):
𝐼 = 𝐼0(1 + 𝑀𝑒𝑖𝜔𝑡) (3)
The response of n at this frequency can be expressed as:
𝑛 = 𝑛0(1 + 𝑚𝑒𝑖𝜔𝑡) (4)
Where m is the magnitude of electron density modulation.
Using equations (1)-(4), we can solve analytically for m as follows:
106
𝑚 =𝑀
1+𝑖𝜔
𝑘1
=𝑀𝑘1
2
𝑘12+𝜔2 − 𝑖
𝑀𝑘1𝜔
𝑘12+𝜔2 (5)
This expression for m suggests that when the scanned frequency 𝜔 equals 𝑘1, the
imaginary part of m reaches a maximum. With a small light perturbation, we assume that
the capacitance of the electrode is constant, and thus the modulation of 𝑛 translates to the
frequency response of electrode potential that can be measured.
Figure A-1: IMVS Nyquist plots of the measured sample (black) and the fitted
line using equation (5) for a monomeric sample at a pH of 4.8 and a light intensity of 9.3
mW/cm2. The fitted recombination rate k is also shown.
107
Figure A-2: IMVS Nyquist plots of the measured sample (black) and the fitted
line using equation (5) for a oligomeric samples at various pHs. The fitted recombination
rate k is also shown.
108
Appendix B
Additional Characterization Data for the Oligomeric Dye
MALDI-TOF Mass Spectrometry
Samples were run in DHB (2,5-dihydroxybenzoic acid), dithranol, and trans-2-[3-
(4-tert-butylphenyl)-2-methyl-2-propenylidene]malononitrile (DCTB). The oligomer
suffered the least amount of breakdown in DCTB and the results are shown in the figure
below.
Pore Penetration by the Oligomeric Dye
Samples were soaked in 2 mL solutions of 100 μM of oligomer dye and then the 2
mL of 100 μM monomer dye. Soaking in the monomer solution following the oligomer
adsorption step resulted in an average relative decrease in absorbance of 1.5% across four
Figure B-1: MALDI-TOF Mass spectrum of oligomer
10
81
.61
7
97
1.5
09
22
90
.97
7
14
90
.60
3
18
07
.63
7
16
46
.78
0
11
44
.30
7
13
05
.95
5
18
62
.78
8
15
55
.88
1
23
70
.85
1
22
35
.28
3
25
35
.13
7
30
16
.30
1
35
19
.00
6
27
18
.39
4DCTb1-1-CLG2 0:F2 MS Raw
0.00
0.25
0.50
0.75
1.00
1.25
1.50
4x10
Inte
ns.
[a.u
.]
1000 1500 2000 2500 3000 3500 4000 4500m/z
109
samples. This result demonstrates that after the first oligomer adsorption step there is
little unsensitized area of the electrode available for the smaller monomer to adsorb. In
theory, if the oligomer were not adsorbing into all of the pores, then the second monomer
adsorption step would dramatically increase the absorbance of the electrode. Samples 1
and 2 were made with TiO2 layers from one piece of scotch tape and the samples 3 and 4
were made with 2 layers of scotch tape
Table B-1. Absorbances of samples from sequential oligomer and monomer adsorption
Electrode
Absorbance after
Oligomer Adsorption
(OD)
Electrode
Absorbance after
Oligomer
Adsorption (OD)
Absolute
Difference (OD)
Relative
Difference (%)
Sample 1 0.742 0.776 0.034 4.582
Sample 2 0.712 0.665 -0.047 -6.601
Sample 3 1.416 1.327 -0.089 -6.285
Sample 4 1.332 1.363 0.031 2.327
Average: -1.979
NMR and End Group Calculations
110
Figure B-2. 1H NMR Spectrum. Inset showing comparison of end groups.
111
Figure B-3. 1H NMR spectrum of linker ligand
112
Appendix C
Time-Resolved Emission of Oligomer Electrodes
Motivation
The recombination rate of an electron from TiO2 back to the dye was calculated in
Chapter 1. It is desirable to know the injection rate so that the relationship between these
two rates is known. We attempted to estimate this injection yield by correlating it to the
transient peak during electrolysis. Variation between individual electrodes introduced too
much error into the resulting data. A second method of measuring the injection rate is by
exciting the dye with a laser and monitoring the time resolved emission. Photoanode
samples were prepared as described in Chapter 2. Dye-sensitized electrodes were placed in
phosphate buffer solutions of various pH (4.8-7.3) and the transient emission spectra were
recorded with a photodiode connected to an oscilloscope. Further details of the
instrumental setup can be found in previous publications.88 The injection efficiency
appeared to be correlated to the changes in pH but without a proper control sample it is
difficult to conclusively determine the correlation.
113
ALD ZrO2 Shell Control Samples
An ideal control sample would consist of a metal-oxide with a wide band gap that
the dye cannot inject an electron into. This has been previously done in the literature by
synthesizing ZrO2 nanoparticle that were then suspended in a paste, in the same manner
that we have used for TiO2 herein. We attempted to mimic the effect of a complete ZrO2
electrode by covering the TiO2 electrodes with a thick (~4-6 nm) ALD layer of TaO2. The
TaO2 ALD process was more reliable than the ZrO2 process for ultrathin films. Multiple
ALD runs were performed with parameters given in Table C-1, and a photograph of
resulting electrodes are shown in Figure C-1. Electrodes with a 6 nm shell resulted in
extremely low surface coverage by oligomer and monomer sensitizers. This is likely
explained by the thick (6 nm) shell filling the majority of the porous space. As a result of
the poor surface coverage the time-resolved data from the 6 nm samples is non-sensical.
The samples with a 4 nm shell of TaO2 resulted in reasonable surface coverages of
sensitizer. Despite reasonable coverages the measured emission lifetimes of the control
sample are considerably faster (300 ns) than the expected value for the sensitizer which is
not quenched by injection (RuP solution lifetime = 760 ns).
Precursor Dwell Time (s)
Resulting Shell Thickness (nm)
5 4
12 6
12 6
Table C-1: ALD Deposition Parameters and Resulting Shell Thickness.
114
Results
The transient peaks of the decays were normalized to 1.0. The first 20 ns of the
decays were not included in the fit, to minimize contributions from scattering of the
excitation laser pulse. A Matlab script implementing the lsqcurvefit() function was used to
fit the raw data. The data was fit to the single exponential (eqn. 1) and multi exponential
(eqn. 2). The single exponential did not result in adequate fitting of all of the samples. The
tri-exponential was adequate for fitting all of the samples and the results are shown in
Figure C-1: Photograph of non-injecting TaO2 core-shell control electrodes
115
Figure C-2. It is unclear what the physical significance of the three rate constants that are
extracted from the tri exponential.
[𝐴 ∗] = 𝐴1 𝑒−(𝑡
𝝉𝟏)𝜷
+ 𝐶1 (1)
[𝐴 ∗] = 𝐴1 𝑒−(𝑡
𝝉𝟏) + 𝐴2 𝑒−(
𝑡
𝝉𝟐)+ 𝐴3 𝑒−(
𝑡
𝝉𝟑) + 𝐶1 (2)
Figure C-2: Emission decays fit to a decaying exponential as a function of pH
116
Conclusions
Brennaman and coworkers89 performed a similar time-resolved emission study on RuP at
lower pH and came to the same conclusions. Above pH 3, the time constants cannot be
distinguished as a function of pH due to the resolution of the optical system.
Figure C-3: Extracted excited state lifetimes (A) and average transient peak
values (B) from the single exponential fits of the emission data
117
Appendix D
1H NMR Spectra for the Molecular Ru(bda)L2 Catalyst
Figure D-1: Spectrum for the blue dimer “decomposition” product
118
Figure D-2: Spectrum for the pure asymmetric pre-catalyst Ru(bda)(Me-
pyr)(DMSO)
119
Appendix E
1H NMR Spectra for the Hydroxamate Dye
Figure E-1: Spectrum for the attempted synthesis of a hydroxamate bipyridine
120
Figure E-2: Spectrum for the attempted isolation of the anhydride intermediate
VITA
Christopher L. Gray
Christopher L Gray was born and raised in Peoria, Arizona where he attended
Arizona State University and graduated with a B.S. in Chemistry and a B.S.E. in
Bioengineering in 2012. While there, he did research in Daniel Buttry’s and Devens
Gust’s labs. Chris began working in Thomas Mallouk’s lab in 2013 with a focus on
renewable energy projects. During his time at Penn State, he synthesized and studied a
variety of Ru(bpy)3 compounds and nanomaterials with interesting applications for
renewable energy.