DRAWING INSPIRATION FROM NATURE
FOR SOLAR ENERGY CONVERSION:
MAPPING THE STRUCTURE-FUNCTION RELATIONSHIPS IN
BETALAIN PIGMENTS TO UNDERSTAND NATURAL
LIGHT HARVESTING SYSTEMS
By
NICHOLAS ANDREW TREAT
A dissertation submitted in partial fulfillment of
the requirements for the degree of
DOCTOR OF PHILOSOPHY
WASHINGTON STATE UNIVERSITY
Department of Chemistry
JULY 2016
© Copyright by NICHOLAS ANDREW TREAT, 2016
All Rights Reserved
© Copyright by NICHOLAS ANDREW TREAT, 2016
All Rights Reserved
ii
To the Faculty of Washington State University:
The Members of the Committee appointed to examine the dissertation of NICHOLAS
ANDREW TREAT find it satisfactory and recommend that it be accepted.
_______________________________
Jeanne L. McHale, Ph.D., Chair
_______________________________
John Fellman, Ph.D.
_______________________________
Aurora Clark, Ph.D.
_______________________________
Chulhee Kang, Ph.D.
iii
Acknowledgement
I would like to acknowledge the generous financial support of the Washington State
University College of Arts and Sciences, especially Dean Daryll DeWald whose help was crucial
in securing the funding to complete my degree, Dr. Kirk Peterson for the many times he
facilitated access across the hall, Dr. Fritz Knorr for helping with the “factory”, all the McHale
Group members past and present for their solidarity and Dr. Jeanne McHale whose support
enabled this project to continue despite overwhelming odds. Of course, the most important
support I received was from my friends and family, who have shown me unwavering support
throughout the many years of my Ph.D. career.
iv
Dedication
This dissertation is dedicated to my father, whose support has been unwavering, despite
the odds, over the course of my education and career, and whose guidance has been invaluable
on my journey.
v
DRAWING INSPIRATION FROM NATURE
FOR SOLAR ENERGY CONVERSION:
MAPPING THE STRUCTURE-FUNCTION RELATIONSHIPS IN
BETALAIN PIGMENTS TO UNDERSTAND NATURAL
LIGHT HARVESTING SYSTEMS
Abstract
by Nicholas Andrew Treat, Ph. D.
Washington State University
July 2016
Chair: Jeanne L. McHale
Photophysical properties of betalain pigments from beets (Beta vulgaris L.), amaranth
(Amaranthus cruentus) and pitahaya (Hylocerus polyrhizus) were studied in the context of dye-
sensitized solar cells. Utilizing a nature-inspired approach to solar energy production, natural
pigments were extracted, purified and studied to develop a greater understanding of the light
harvesting systems. The highly efficient light harvesting of the betalains, which is due to their
photoprotective role in the plants, far surpasses the metalorganic dyes commonly used in dye-
sensitized solar cells. Evidence of two-electron oxidation, and ease of preparation and abundant
supply make betalain pigments an ideal candidate for a renewable energy solution.
A process for the extraction and purification of mixtures and pure betalains was
developed based on anion exchange chromatography and high performance liquid
vi
chromatography. The ability to perform experiments on samples with single betalain components
allowed for more precise quantification of the feasibility and performance of betalain-based
devices.
Films sensitized with betanin from non-aqueous solution were shown to have negligible
aggregation of betanin on the TiO2 surface despite the high dye loading. This effect was
observed at all ratios of methanol in water from 4% methanol to neat methanol. No other
changes in the solutions were observed.
Differences in the behavior of two dyes, amaranthin and betanin, were used to develop a
map of the structure-function relationships within betalain pigments. Betanin was shown to
aggregate on TiO2 where amaranthin does not. The aggregate formed by betanin greatly
enhanced the performance and stability of the devices. The fluorescence quantum yields of
betalains were shown to also depend slightly on structure, despite the distance between the
additional sugar or acyl groups and chromophore.
vii
Table of contents Page
Acknowledgement ......................................................................................................................... iii
Dedication ...................................................................................................................................... iv
Abstract ........................................................................................................................................... v
Table of contents ...................................................................................................................... vii
Table of Figures ...................................................................................................................... xii
List of Tables ..................................................................................................................... xvi
1 Introduction ............................................................................................................................... 1
1.1 Research Goals ...................................................................................................... 4
1.2 Specific Aims ........................................................................................................ 4
1.2.1 Purify samples of betalain pigments for spectroscopic studies ....................... 4
1.2.2 Investigate dye absorption spectral broadening on TiO2 and aggregation ...... 4
1.2.3 Quantify the performance of aggregated compared to monomeric
samples ........................................................................................................... 5
1.2.4 Compare the behavior of amaranthin to that of betanin .................................. 5
1.3 Background ........................................................................................................... 5
1.3.1 DSSCs ............................................................................................................. 5
1.3.2 Principles ......................................................................................................... 5
1.3.3 Characterization .............................................................................................. 8
1.4 Betalains .............................................................................................................. 11
1.4.1 Properties and Sources .................................................................................. 11
1.4.2 Stability: ........................................................................................................ 14
1.4.3 Photophysical Properties: .............................................................................. 16
viii
1.4.4 Comparisons with Anthocyanins .................................................................. 16
1.5 References ........................................................................................................... 19
2 Preparation of Betalain Pigments............................................................................................ 23
2.1 Introduction: Sources and Previous Techniques ................................................. 23
2.2 Preparatory Techniques ....................................................................................... 25
2.2.1 Extraction ...................................................................................................... 25
2.2.2 HPLC ............................................................................................................. 25
2.2.3 MPLC ............................................................................................................ 29
2.2.4 Anion-Exchange Column Chromatography .................................................. 30
2.3 Conclusions ......................................................................................................... 32
2.4 References ........................................................................................................... 35
3 Templated Assembly of Betanin Chromophore on TiO2: Aggregation-Enhanced Light
Harvesting and Efficient Electron Injection in a Natural Dye-Sensitized Solar Cell ............. 38
3.1 Introduction ......................................................................................................... 38
3.2 Experimental Methods ........................................................................................ 41
3.2.1 Purification .................................................................................................... 41
3.2.2 HPLC ............................................................................................................. 42
3.2.3 Film preparation and sensitization (TiO2 and ZrO2) ..................................... 42
3.2.4 Absorption and fluorescence spectra............................................................. 44
3.2.5 IPCE and APCE ............................................................................................ 44
3.2.6 Solar Cell Construction. ................................................................................ 45
3.2.7 Numerical methods. ...................................................................................... 45
ix
3.3 Results and Discussion ........................................................................................ 46
3.3.1 Absorption Spectra and Numerical Modeling of Betanin on TiO2 and
ZrO2.............................................................................................................. 46
3.3.2 Quantum Efficiency of Photon-to-Current Conversion ................................ 50
3.3.3 Fluorescence of Betanin on TiO2 and ZrO2 .................................................. 53
3.3.4 Power Conversion Efficiency of Dimer-based Solar Cells ........................... 54
3.3.5 Theory of Transition Dipole-Dipole Coupling in Betanin Dimers ............... 56
3.4 Conclusions ......................................................................................................... 59
3.5 Supporting Information ....................................................................................... 61
3.6 References ........................................................................................................... 66
4 Tuning the Aggregation of Betanin on TiO2 Using Non-Aqueous Solvents and
Phosphate Additives................................................................................................................ 72
4.1 Introduction ......................................................................................................... 72
4.2 Experimental Methods ........................................................................................ 75
4.2.1 Film Preparation ............................................................................................ 75
4.2.2 Film Sensitization .......................................................................................... 75
4.2.3 Absorbance Spectroscopy ............................................................................. 75
4.2.4 Solar Cell Measurements .............................................................................. 75
4.3 Results and Discussion ........................................................................................ 76
4.3.1 Phosphate ...................................................................................................... 76
4.3.2 Methanol........................................................................................................ 80
4.3.3 Performance of Cells ..................................................................................... 84
4.4 Conclusions ......................................................................................................... 86
x
4.5 References ........................................................................................................... 87
5 An Amaranthin Based Solar Cell: Comparison of Efficiency and Properties to
Betanin Based Dye Sensitized Solar Cells.............................................................................. 88
5.1 Introduction ......................................................................................................... 88
5.2 Experimental Methods ........................................................................................ 90
5.2.1 Preparation of Amaranthin Samples ............................................................. 90
5.2.2 Film Preparation ............................................................................................ 91
5.2.3 Sensitization of TiO2 films with Amaranthin ................................................ 91
5.2.4 Absorbance Spectroscopy ............................................................................. 91
5.2.5 Fluorescence Quantum Yield Measurements ................................................ 91
5.2.6 Preparation of Colloidal Silver Nanoparticles .............................................. 92
5.2.7 Surface Enhanced Raman Scattering ............................................................ 92
5.3 Results and Discussion ........................................................................................ 92
5.3.1 Comparison of Aggregation Behavior Based on Absorption Spectra........... 92
5.3.2 Solvents and pH Differences ......................................................................... 94
5.3.3 Fluorescence .................................................................................................. 96
5.3.4 Raman............................................................................................................ 97
5.3.5 Device Performance ...................................................................................... 98
5.4 Conclusions ....................................................................................................... 101
5.5 References ......................................................................................................... 102
6 Summary and Conclusion ..................................................................................................... 103
6.1 Future Work ...................................................................................................... 104
xi
6.2 Final Remarks ................................................................................................... 105
7 Appendix A: MATlab Code.................................................................................................. 106
7.1 BuildRC ............................................................................................................. 106
7.2 LeeSeung ........................................................................................................... 106
7.3 Lee ..................................................................................................................... 111
7.4 Dcalc.................................................................................................................. 111
8 Appendix B: HPLC Methods ................................................................................................ 112
xii
Table of Figures Page
Figure 1.1: Structure of N3, Ru(4,4’-dicarboxylicacid-2,2’-bipyridine)2(NCS)2(left), and
its tert-butyl ammonium (TBA) salt analogue N719 (right).5................................. 3
Figure 1.2: Cartoon schematic of a dye sensitized solar cell.5 reprinted with permission
from Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-
Sensitized Solar Cells. Chem. Rev. 2010, 6595–6663.
DOI: 10.1021/cr900356p ........................................................................................ 6
Figure 1.3: Energy level schematic for a DSSC in operation showing the various
reactions (arrows) that occur and the associated rate constants. ............................. 8
Figure 1.4: Reaction pathways for betacyanin (BcN) and betaxanthin (BxN) formation
from betalamic acid (BtA). ................................................................................... 11
Figure 1.5: Betacyanin structures from the natural sources studied. ............................................ 13
Figure 1.6: Structures of betaxanthins that are relevant to the sources of betalains in this
research. ................................................................................................................ 14
Figure 1.7: Two possible structures proposed for neobetanin: traditional structure shown
in most literature (left XIVa) and structure suggested recently by
Wybraniec et al. (right XIVb). .............................................................................. 15
Figure 1.8: The basic structure of anthocyanins is based on anthocyanidin. In
anthocyanidin all of the R groups are hydrogens.................................................. 17
Figure 1.9: Structures of common copigments found throughout the betalain and
anthocyanin species. The binding site to the base dye structure is denoted
by “R”. .................................................................................................................. 18
Figure 1.10: Aggregates of anthocyanins consisting of units of succinylcyanin and
malonylflavone shown schematically and in wire-frame form (B).
Reprinted with permission from Ellestad, G. A. Structure and Chiroptical
Properties of Supramolecular Flower Pigments. Chirality, 2006, 18, 134-
144. DOI: 10.1002/chir.20228 .............................................................................. 18
Figure 2.1: Preparatory HPLC of amaranth extract with amaranthin (1) and isoamaranthin
(2) the predominant pigments, but small amount of decarboxylated
betacyanins (3 &4) being present. ........................................................................ 29
Figure 2.2: Schematic of the preparatory process developed in this research. ............................. 31
xiii
Figure 2.3: HPLC chromatogram of purified amaranth (a) and that of filtered amaranth
juice (b) monitored at 254 nm. .............................................................................. 32
Figure 2.4: HPLC chromatogram of purified samples of amaranth (black), pitahaya (red),
beets (green) and a combination of all the sources. The primary
betacyanins are marked as amaranthin (A), betanin (B), phylocactin (C)
and hylocerinin (D). .............................................................................................. 33
Figure 3.1: Structure of betanin (left) and 3D representation of optimized geometry
(right) obtained using DFT as described in Ref. 42. ............................................. 39
Figure 3.2: Absorption spectra of Bt on A) ZrO2 and B) TiO2 as a function of
sensitization time in a saturated betanin solution at pH 2. A 900s film on
TiO2 was prepared, but was too dark to determine the absorbance. ..................... 48
Figure 3.3: Results of NMF analysis of Bt/TiO2 at A) low and B) high dye loading. The
films in A) and B) were sensitized for 300s and overnight (21 hours) and
contain 25% and 65% aggregate, respectively. The components are dimer
(green), monomer (blue), TiO2 extinction (cyan), experimental absorbance
(black), and total calculated absorbance (red dashed). ......................................... 49
Figure 3.4: A) IPCE and B) APCE of Bt-based DSSC consisting of 25% aggregate (red),
and 65% aggregate (black). ................................................................................... 52
Figure 3.5: Fluorescence spectra, excited at 514.5 nm and corrected for absorptance at
514.5 nm, of Bt on A) TiO2 and B) ZrO2 as a function of sensitization
time. ...................................................................................................................... 52
Figure 3.6: Photocurrent versus photovoltage for several Bt-based dye-sensitized solar
cells. ...................................................................................................................... 54
Figure 3.7: Scanning electron micrographs of TiO2 films showing 3.5µm thickness (left),
7µm thickness (middle) and the nanoparticles (right). ......................................... 61
Figure 3.8: UV-Vis absorption spectra of the sensitized TiO2 film (black) and the
sensitizing solution (red) showing the dramatic blue-shift from 537nm to
482nm. .................................................................................................................. 61
Figure 3.9: (A) HPLC chromatogram, monitored at 525nm, of the extracted pigments of
films sensitized for 10min (red) and 16hours (green) shown with the
sensitizing solution (black) for comparison. (B) Absorption spectra of film
extract (black) and sensitizing solution (red). ....................................................... 62
Figure 3.10: Calibration curve of betanin vs. calculated concentration. ...................................... 62
xiv
Figure 3.11: UV-Vis absorption spectra of films sensitized without phosphate (A, Black)
and with 1 mM phosphate (B, red). ...................................................................... 63
Figure 3.12: Unit components from NMF analysis showing the peaks of the aggregate
(green), monomer (blue), TiO2, and the sum of all components (red).................. 63
Figure 3.13: NMF component analysis showing three subsets of the samples used for:
fluorescence measurements (A), Beer’s Law analysis of solutions (B), and
DSSC characterization (C). ................................................................................... 64
Figure 3.14: Linear combination of the aggregate and monomer unit components from
the NMF modeling showcasing the representative peak locations for
aggregation amounts in 10% intervals. ................................................................. 65
Figure 3.15: Characteristic current-voltage plots for films with no pretreatment (black),
pretreatment with 0.5M hydrochloric acid in ethanol (red), and purified
water (blue). .......................................................................................................... 65
Figure 4.1: The molecular structure of betanin showing the atomic labeling of the various
functional groups. ................................................................................................. 73
Figure 4.2: Speculative structure of betanin dimer on TiO2 (101) surface. Reprinted with
permission from Treat, N. A.; Knorr, F. J.; McHale, J. L., Templated
Assembly of Betanin Chromophore on TiO2: Aggregation-Enhanced
Light-Harvesting and Efficient Electron Injection in a Natural Dye-
Sensitized Solar Cell. J. Phys. Chem. C, 2016, 120 (17), 9122-9131.
Copyright 2016 American Chemical Society. ...................................................... 74
Figure 4.3: Absorbance spectra of betanin films sensitized overnight in with various
amounts of phosphate, 1.0 mM (black), 0.1 mM (blue), 10 mM (red) and
0.0 mM (pink), and 1.0 mM betanin all exhibit blue-shifted absorbance
maxima around 485 nm. ....................................................................................... 77
Figure 4.4: Absorption spectra of films sensitized with 5 mM betanin (black and blue)
and 1 mM betanin (red and magenta) in a buffer of 1 mM phosphate. ................ 78
Figure 4.5: Sequential absorption spectra of betanin films sensitized in 10 mM (a), 1 mM
(b), 0.1 mM (c) and 0.0 mM (d) phosphate taken at 1 minute intervals for
up to 20 minutes, except (a) which were taken at 10 minute intervals for
90 minutes. ............................................................................................................ 79
Figure 4.6: Absorbance spectra of TiO2 films sensitized with 1 mM betanin in 1 mM
sodium phosphate at pH 3 taken sequentially at regular intervals. ....................... 80
xv
Figure 4.7: Absorbance spectra of betanin films sensitized for 5 minutes (black) and 10
minutes (red) from methanol (left) and water (right) with otherwise
identical conditions. .............................................................................................. 81
Figure 4.8: Absorption spectra of films sensitized with betanin solutions containing 4%
(black), 52% (red) and 96% MeOH (green) for 10 minutes and 4% MeOH
for 30 min (blue). .................................................................................................. 82
Figure 4.9: Absorbance spectra of TiO2 films sensitized with betanin in 100% methanol
sequentially from 1 to 40 minutes, with spectra taken at regular intervals. .......... 83
Figure 4.10: IPCE spectra of betanin solar cells produced with various amounts of
phosphate in the sensitizing solutions. .................................................................. 84
Figure 4.11: J-V curves of solar cells produced from betanin in methanol (black) and
water (red). ............................................................................................................ 85
Figure 5.1: Molecular structures for betanin (left), amaranthin (right) and their aglycone
betanindin (center). ............................................................................................... 89
Figure 5.2: Absorbance spectra of films sensitized with amaranthin from aqueous
solutions of varying pH (a), and from methanol (MeOH) and acetonitrile
(MeCN) (b). .......................................................................................................... 94
Figure 5.3: Absorbance spectra of solutions of amaranthin in methanol and aqueous
solvents show no significant difference, though amaranthin in acetonitrile
(purple) shows a broadening and blue-shift. ......................................................... 95
Figure 5.4: Emission spectra of amaranthin (black) and betanin (red) in aqueous solution
with excitation at 514.5 nm normalized to show that the shapes are
identical. ................................................................................................................ 96
Figure 5.5: Surface enhanced resonance Raman spectra of betanin and amaranthin show
that the chromophore vibrational modes are nearly identical excited at 458
nm. ........................................................................................................................ 97
Figure 5.6: Replicate IPCE spectra of amaranthin based solar cells sensitized from MeOH
(a) and MeCN (b) solutions. ................................................................................. 99
Figure 5.7: A comparison of the J-V curves of amaranthin from aqueous (a) and non-
aqueous solution (b). ........................................................................................... 100
xvi
List of Tables Page
Table 3.1: Power conversion efficiency, open circuit voltage Voc, short-circuit current Jsc,
wavelength of maximum absorption for sensitized film λmax, film
thickness, and percent aggregation for each of the solar cells in Figure 3.6. ....... 56
Table 5.1: Energy conversion efficiencies of amaranthin based solar cells sensitized from
acetonitrile (left), methanol (center) and water (right). The efficiency of
cell 1 in methanol was excluded for the calculation of the average. .................. 100
Table 8.1: Binary gradient developed specifically for analysis of amaranth samples. ............... 112
Table 8.2: Binary gradient developed for analysis of samples from beets. ................................ 112
Table 8.3: Binary gradient developed for preparatory methods of all betalain sources. ............ 112
1
1 Introduction
Over the past few decades, it has become apparent that our over utilization of fossil fuels
for energy production is having a deleterious effect on the global ecosystem. In addition to their
effect on the environment, research indicates that the global stores of fossil fuels are being
depleted. However, the demand for energy is increasing as we depend increasingly on electronics
in our daily lives. These factors combined have led to an emergent global energy crisis.
Therefore, an effective and ecologically conscientious method of energy production is a rapidly
expanding area of interest in chemical and materials research. Of the several ecologically
friendly energy production methods, solar power is perhaps the most promising. Solar radiation
is a dependable source of energy that nature has adapted to harness. Perhaps we can take a lesson
from nature in our pursuit of reliable, cost effective, and ecologically friendly energy.1,2
Plants have adapted to harness solar energy through photosynthesis involving various
photo-reactive chemical species and complex molecular scaffolds. An understanding of the
principle of light harvesting chemicals and structures in these highly adapted organisms could
lay the foundation for an innovative approach to the energy crisis. Of primary interest is the
underlying principle of the structure-function relationship in these unique systems.1,2
Betalains, a family of photo-protective molecules, are found exclusively in plants of the
order Caryophyllales and a few species of fungi. In recent years, research has shown that
betalains offer a unique potential in the field of dye sensitized solar cells (DSSCs). By utilizing
the photochemical oxidation of these pigments, DSSCs have been constructed to perform at an
efficiency of up to 3%.3 While the ultimate implementation of DSSCs would require higher
efficiencies, the rough and non-optimal engineering of these particular cells indicates a very
promising class of solar energy device. In a comparison study, N3 dye was tested, without
2
significant optimization of the cell, and produced a maximum efficiency of 2.6%.4 This
similarity exemplifies the potential of betalain dyes as sensitizers. In other labs, with
optimization, N3 solar cells have performed with efficiencies as high as 9%.5 Additionally, these
small organic molecules offer several specific advantages over traditional ruthenium based
DSSCs. First, these dyes are environmentally friendly and readily abundant. Since they are
produced naturally in plants, they are easily obtained without difficult and costly synthetic
pathways and do not require the use of ruthenium, which is of limited supply. Secondly, the
organic molecular nature of these pigments indicates that possibility of a two electron oxidation
mechanism.6,7
Multiple electron oxidation mechanisms would allow for an increase in the
thermodynamic efficiency determined by the Schockley-Queisser limit.4,8
Third, the wavelength-
dependent incident-photon-to-electron conversion efficiency (IPCE) of DSSCs based on betanin,
from red beet root (Beta vulgaris L.), has been found to approach 100%.9 These measurements
also show good response at red wavelengths which fits well with the solar spectrum. Finally, due
to the allowed nature of their absorption transition, compared to the marginally allowed d-d
transition of metal-based dyes, betalains allow for DSSC construction with thinner
semiconductor films. Thinner films would allow electrons to more effectively be transferred into
the circuit by decreasing their diffusion length in titanium dioxide.5
3
Unfortunately, betalains suffer from several drawbacks that need to be overcome before
effective implementation of betalain-based DSSCs is feasible. The primary hurdle in betalain
research is stability.10-14
In our lab, one successful solar cell system was studied for a period of a
few months and was functional during daylight hours for the majority of the time. However, the
current produced by the arrangement decreased by approximately 30% over the first few days. It
appears that irreversible photo-degradation occurred and hindered device performance. While
betalain stability is a current topic of research in the food sciences, a systematic approach to
understanding the structural aspects of betalain stability has yet to be performed with regards to
the interfacial chemistry occurring at solution-titanium dioxide surfaces. Furthermore, a study
delineating the relative stability and photophysical properties of different betalain dyes could
improve our understanding of the differing aspects of structure and function of photo-protective
systems in nature.15
Figure 1.1: Structure of N3, Ru(4,4’-dicarboxylicacid-2,2’-bipyridine)2(NCS)2(left), and its tert-butyl ammonium (TBA) salt analogue N719 (right).5
4
1.1 Research Goals
Overall the main objective of this research is to develop an understanding of structure-
function relationships in betalain pigments and investigate how these relationships can be
harnessed in solar energy conversion technology. By utilizing a spectroscopically focused
research plan, these relationships will be probed in four specific ways.
1.2 Specific Aims
1.2.1 Purify samples of betalain pigments for spectroscopic studies
Previous work on betalain-based DSSCs was significantly hampered by the lack of
experimental control.9 The biggest variable in any natural product based research is purity.
Without a pure sample, a thorough understanding of experimental outcomes cannot be
established. Therefore, this research developed a method to produce purified dye samples which
can then be used to carefully study the behavior of betalains in DSSCs.
1.2.2 Investigate dye absorption spectral broadening on TiO2 and aggregation
Spectral characterization of betanin adsorbed on TiO2 shows a significant broadening of
the absorption spectrum in both the blue and red regions with the most prominent being in the
blue. The overall change produces a spectrum with an absorption maximum near 480 nm, nearly
50 nm blue of the solution phase maximum. Whether this phenomenon is the product of
degradation, multiple species adsorbing or aggregation of the betanin is of great interest due to
the near 100% IPCE achieved at this blue shifted wavelength.9 Based on some preliminary work,
it is believed that this blue-shift is associated with chromophore aggregation on the TiO2, but
other research shows that degradation products would likely produce similar spectra.6 Using
5
purified dye, spectroscopy and HPLC we aimed to determine the cause of the spectral
broadening.
1.2.3 Quantify the performance of aggregated compared to monomeric samples
The broadened spectra seem to produce more efficient solar cells, and quantification of
this enhancement could lead to inspired design of future DSSCs. Therefore, additives commonly
used to inhibit aggregation were used to control the spectral broadening and develop a model for
the mechanism of aggregation.16
1.2.4 Compare the behavior of amaranthin to that of betanin
The small differences in structure between amaranthin and betanin provide a good
starting point to begin mapping structure-function relationships in betanin. Prior to this research,
no studies had been done utilizing amaranthin in a solar cell. We wished to determine what effect
the steric bulk of an additional sugar group has on the behavior of betalains in DSSCs.17
1.3 Background
1.3.1 DSSCs
Dye sensitized solar cells (DSSCs) are a form of photoelectrochemical cell that harnesses
the sun’s energy to produce electricity. Unlike traditional photovoltaic cells, DSSCs utilize a dye
sensitizer to perform the duty of absorbing the light. Photosynthesis is also a
photoelectrochemical process, though much more intricate, and can serve as inspiration in DSSC
design. Photosynthesis, however, converts the solar energy into chemical energy, whereas
DSSCs convert the solar energy directly into electricity.5,18
1.3.2 Principles
6
The essential components of a DSSC are a semiconductor film, dye sensitizer, platinum
or carbon catalyst, and two electrical contacts. Figure 1.2 shows the components of a typical
DSSC. Titanium dioxide is used as the semiconductor and platinum as the cathode catalyst,
though carbon may also be used. These materials are bonded to conductive glass to form the
foundation of the solar cell. Several different electrolytes are used, but the most common to date
is the iodide-triiodide redox couple dissolved in an organic solution.5
The operation of a DSSC can be broken down into three essential steps: light absorption,
dye oxidation/electron injection and dye regeneration. The light absorption step is pretty
straightforward: a photon is absorbed by the sensitizer which causes the molecule to enter into an
electronically excited state. Once in this state, the dye can either inject the excited electron into
the semiconductor, step 2, or it can relax back to the ground state. Reverse electron injection into
the ground state can also occur, which is called recombination. Recombination results in dye
molecules that have returned to their ground state without the electron completing its journey
through the circuit. Electrons that have been injected into the semiconductor, and are not
Figure 1.2: Cartoon schematic of a dye sensitized solar cell.5 reprinted with permission from Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-Sensitized Solar Cells. Chem. Rev. 2010, 6595–6663. DOI: 10.1021/cr900356p
7
recombined with the dye, are free to move, and are now able to flow through the electrical
circuit. After electron injection the dye molecule is returned to its ground state but in an oxidized
form. The oxidized dye is then reduced by the iodide-triiodide couple in the third and final step.
The platinum catalyst serves the function of regenerating the electrolyte at the cathode and
completing the circuit. Figure 1.3 shows an energy level diagram for the cell and each step. Each
process occurs at a given rate, which is determined by the nature of the dye, semiconductor and
the electrolyte. The maximum voltage for the device is determined by the conduction band and
electrolyte energies and not by the sensitizer. The rate at which the dye undergoes electron
injection is determined by the difference in energy, on an electrochemical scale, between the
conduction band of the semiconductor and the excited state of the dye. A large difference,
therefore higher the driving force, results in a faster rate of injection.5,18
Figure 1.3: Energy level schematic for a DSSC in operation showing the various reactions (arrows) that occur and the associated rate constants.
8
The most unique component of the solar cell is the dye that is used to sensitize the
titanium dioxide. Many different types of molecules have been used a sensitizers and can range
from organometallics to simple organic dyes. The most efficient dyes that have been researched
recently are ruthenium based organometallic compounds. These metal based dyes have shown up
to 11% total energy conversion efficiency in a laboratory setting.5 Natural pigments, on the other
hand, have only shown efficiencies up to 1% with the exception of betanin which has achieved
3% in our lab. While this efficiency may not directly compare with those shown by other
research labs with organic and metalorganic dyes, it compares well with the efficiency that our
lab has achieved with N3 of 4-5%. In light of this comparison, the natural sensitizers are quite
promising.5,9,17
1.3.3 Characterization
The efficiency of a solar cell is measured in two ways each giving crucial information on
the performance of the device. Most important to this research is the external quantum
efficiency, alternatively known as the incident photon to current conversion efficiency (IPCE).
Most simply put this quantity measures the ratio of the of electrons injected to photonsincident
on the sample:5
IPCE =𝑁𝑒
𝑁𝑝 Equation 1.1
Here Ne and Np are the number of electrons injected into the circuit and the number of photons
incident on the surface respectively. Experimentally, the device is illuminated with
monochromatic light of a known intensity, and the resulting current density is measured. By
scanning over the range of absorption of the device, the wavelength specific conversion
efficiency can be determined by:5
9
𝐼𝑃𝐶𝐸 =𝐽𝑠𝑐(𝜆)
𝑒Φ(λ)= 1240
𝐽𝑠𝑐(𝜆)[𝐴 𝑐𝑚−2]
λ(nm)𝑃𝑖𝑛(λ)[𝑊𝑐𝑚−2] Equation 1.2
JSC(λ) is the short circuit current when illuminated at wavelength λ, e is the charge of an electron,
Φ(λ) is the incident photon flux. The denominator can be separated into the wavelength λ and
incident power Pin with a constant conversion factor of 1240 with units of V*nm. Essentially the
IPCE can be considered analogous to the spectral response function of the device. Knowing how
a device performs at varying wavelengths can inform the researcher of conditions and variables
that are most favorable to device optimization. 5
The other primary method of device characterization is the total energy conversion
efficiency. The total conversion efficiency can be measured under simulated solar conditions
(AM1.5) and the resulting efficiency is calculated with:5
𝜂 =𝐽𝑠𝑐𝑉𝑜𝑐𝐹𝐹
𝑃𝑖𝑛=
𝑃𝑚𝑎𝑥
𝑃𝑖𝑛 Equation 1.3
Once again JSC is the short circuit current, VOC is the open circuit voltage, FF is the fill factor, Pin
is the incident light power and Pmax is the maximum power of the device. This method of
characterization is essential for characterizing device performance issues that are correlated to
construction and engineering as well as those caused by the specific sensitizer used.5
Much like the ideal Carnot Cycle governs a heat engine; solar energy conversion is
limited in the maximum conversion efficiency it can achieve by the Schockley-Queisser Limit.
Based upon the physical laws that govern light and its interaction with matter, the maximum
efficiency of solar energy conversion is given by a detailed balance treatment of the conditions
of the device. Many different forms of this calculation are given in the literature; each with
specific modifications for the researchers’ approach to photovoltaic construction. The generally
10
accepted fundamental limit for the maximum energy conversion efficiency for a single junction
semiconductor photovoltaic device is 31%.8 This principle is based on several assumptions:
The device absorbs all photons which have energy greater than the band gap of the
semiconductor;
One electron is generated for each photon that is absorbed;
The spectral response of the device is constant over all absorbed photon energies; and
The excess energy of the photons above the band gap is lost as heat.
For the most part, these assumptions are appropriate and provide an accurate estimation
of the maximum achievable efficiency, however, for betalains, the proposed oxidation reaction is
a two-electron process.6,7
By doubling the photon to electron ratio, the fundamental maximum
efficiency increases to approximately 41.9%.4
1.4 Betalains
1.4.1 Properties and Sources
Betalains consist of two subclasses of pigments: yellow betaxanthins and magenta
betacyanins, which are derived from the parent molecule betalamic acid. Both classes are forms
of condensation products of amino compounds with betalamic acid. Betacyanins are specifically
the condensation products involving cyclo-dopa derivatives, see Figure 1.3. The purple
betacyanins show the greatest overlap with the solar spectrum compared to the betaxanthins.
Thus, the betacyanins show the most promise in DSSCs. Furthermore; the betacyanins show a
wide range of acylated derivatives, which have varying stability, redox chemistry, and
photophysical properties.19-39
11
Betalains have been studied extensively in the food sciences for the purposes of food
coloring. Many sources of betalains have been identified. The primary sources include: red beets
(Beta vulgaris L.), cactus pears (Optuntia sp.) purple pitahaya (Hylocereus sp.), amaranth
(Amaranthus sp.), and bougainvillea (Bougainvillea sp.).26
Each source has a unique palette of
betalains with varying ratios of betaxanthins to betacyanins. Of particular interest to this research
are the betalains from red beets, amaranth, cactus pears and pitahaya. Figure 1.5 shows the
betacyanin pigments from these sources which are derivatives of betanin (I) and amaranthin
(IV). All of these Betacyanins show different stability characteristics and could be used in lieu of
betanin in DSSCs. Hylocerenin (III), for example has shown increased thermal stability when
the carboxylic acid moiety at the C2 is removed.15
Of the Betaxanthins present in these sources
only indicaxanthin (VIII), portulacaxanthin II and III (IX and X respectively), and vulgaxanthin
I (XI) are in sufficient quantity to be extracted efficiently, see Figure 1.6.26,30,31,33,35,37
Figure 1.4: Reaction pathways for betacyanin (BcN) and betaxanthin (BxN) formation from betalamic acid (BtA).
12
Figure 1.5: Betacyanin structures from the natural sources studied.
13
1.4.2 Stability:
Many factors have been identified to affect betalain stability in solution, and degradation
pathways have been proposed. However, observation of these degradation pathways has not been
thoroughly studied and the influence of substitutions on degradation is still not understood.
Certain aspects of betacyanin structural degradation are ambiguous in the literature. For example,
the proposed structure of neobetanin varies within the literature between a zwitterionic structure
with a positive charge on the nitrogen (XIVa), and a structure with the double bond shifted to
form a pyridine ring at the base of the molecule (XIVb), see Figure 1.7.40
The structure of this
degradation product is of interest to our lab because the neobetanin structure XIVb may have
Figure 1.6: Structures of betaxanthins that are relevant to the sources of betalains in this research.
14
more favorable properties for electron injection into TiO2. Further, it is interesting to note that
none of the literature has been able arrive at a consensus on the absorption maximum of
neobetanin. Some sources cite 450-460 nm and other 470-490. The solution conditions of these
measurements could cause the neobetanin to preferentially exist in one form or the other.
Another interesting aspect of neobetanin behavior is that the addition of a double bond, which
would normally cause the absorption maximum to shift to the red, causes the absorption to shift
significantly blue. We believe that this blue shift is based on a change to the charge transfer
characteristics of the ground to excited state transition and may provide insight into the
underlying photophysics of betalains.
Figure 1.7: Two possible structures proposed for neobetanin: traditional structure shown in most literature (left XIVa) and structure suggested recently by Wybraniec et al. (right XIVb).
15
1.4.3 Photophysical Properties:
Recently, a study by Wybraniec et al showed interesting photophysical properties of
indicaxanthin. 41
This study indicated that the fluorescence quantum yield depends on solvent
viscosity. They showed that as the viscosity of the solution increased, so did the quantum yield.
It was proposed that this increase is due to restriction of the internal rotation of the molecule.
This fluorescence increase is consistent with preliminary work performed in our lab on
amaranthin. In combination, these results indicate the potential for tunable photophysical
properties in betalains.42
Restriction of internal rotation can also be achieved by aggregation, and
comparing the fluorescence properties of aggregated dye and monomeric dye should yield
similar results, however, this trend was not observed upon dye aggregation as will be discussed
in Chapter 3.
1.4.4 Comparisons with Anthocyanins
Anthocyanins are the familiar plant pigments that are found in all plants not in the order
Caryophyllales. These pigments are found in everything from berries to roses and their light
absorption spans almost the entire visible spectrum. While anthocyanins are a different class of
pigments, many of their photophysical properties are determined by their copigments, or acyl
groups, which are the same groups found in betalains. Figure 1.8 shows the basic unit of all
anthocyanins, anthocyaninidin, and through functionalization at any of the R positions, unique
pigments are formed. A significant body of research exists on the nature of these pigments, and
parallels can be drawn between the two dye classes since they share a common function and a set
of common building blocks.
16
What is perhaps most interesting to this study is the evidence for highly ordered
aggregates of anthocyanins in nature. Many plants utilize the interactions of covalently bound
acyl and sugar groups to generate unique aggregates of pigment that can drastically alter the
photophysical properties. In fact, it is only through aggregation that plants can produce blue
colors. 43
Many anthocyanins favor intermolecular interactions so strongly that they can only be
observed in dimeric states, even in solution, unless conditions are specifically constructed to de-
aggregate the dyes. As mentioned above these copigments are the same for betalains and
anthocyanins. A few examples of the acyl groups which are found throughout both classes of
natural pigments are shown in Figure 1.9. Figure 1.10 shows one particularly complex aggregate
formation of anthocyanins that even exhibits chiral character. 44,45
Figure 1.8: The basic structure of anthocyanins is based on anthocyanidin. In anthocyanidin all of the R groups are hydrogens.
17
Figure 1.9: Structures of common copigments found throughout the betalain and anthocyanin species. The binding site to the base dye structure is denoted by “R”.
Figure 1.10: Aggregates of anthocyanins consisting of units of succinylcyanin and malonylflavone shown schematically and in wire-frame form (B). Reprinted with permission from Ellestad, G. A. Structure and Chiroptical Properties of Supramolecular Flower Pigments. Chirality, 2006, 18, 134-144. DOI: 10.1002/chir.20228
18
1.5 References
1 Croce, R.; van Amerongen, H.; Natural Strategies for Photosynthetic Light Harvesting. Nature
Chem. Bio. 2014, 10, 492-501.
2 Fassioli, F.; Dinshaw, R.; Arpin, P. C.; Scholes, G. D. Photosynthetic Light-Harvesting:
Excitons and Coherence. J. Royal Soc. Interface 2014, 11, 20130901/1-22.
3 Treat, N. A.; Knorr, F. J.; McHale, J. L., Templated Assembly of Betanin Chromophore on
TiO2: Aggregation-Enhanced Light-Harvesting and Efficient Electron Injection in a Natural
Dye-Sensitized Solar Cell. J. Phys. Chem. C 2016, 120 (17), 9122-9131.
4 Hanna, M. C.; Nozik, A. J. Solar Conversion Efficiency of Photovoltaic and Photoelectrolysis
Cells with Carrier Multiplication Absorbers. J. Appl. Phys. 2006, 100, 074510–074510.
5 Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-Sensitized Solar
Cells. Chem. Rev. 2010, 110, 6595–6663.
6 Knorr, F. J.; McHale, J. L.; Clark, A. E.; Marchioro, A.; Moser, J.-E. Dynamics of Interfacial
Electron Transfer from Betanin to Nanocrystalline TiO2: The Pursuit of Two-Electron
Injection. J. Phys. Chem. C 2015, 119, 19030-19041.
7 Knorr, F. J.; Malamen, D. J.;McHale, J. L.; Marchioro, A.; Moser, J.-E. Two-Electron Photo-
Oxidation of Betanin on Titanium Dioxide and Potential for Improved Dye-Sensitized Solar
Energy Conversion. Proc. SPIE 9165, Physical Chemistry of Interfaces and Nanomaterials
XIII 2104, 9165-0N.
8 Shockley, W.; Queisser, H. J. Detailed Balance Limit of Efficiency of p-n Junction Solar
Cells. J. Appl. Phys. 1961, 32, 510–510.
9 Sandquist, C.; Mchale, J. L. Improved Efficiency of Betanin-Based Dye-Sensitized Solar
Cells. Journal of Photochemistry and Photobiology A: Chemistry 2011, 221, 90–97.
10 Cai, Y.; Sun, M.; Corke, H. Colorant Properties and Stability of Amaranthus Betacyanin
Pigments. J. Agric. Food Chem. 1998, 46, 4491–4495.
11 Elbe, J. V.; Attoe, E. Oxygen Involvement in Betanine Degradation—Measurement of
Active Oxygen Species and Oxidation Reduction Potentials. Food Chemistry 1984, 16, 49–
67.
12 Reynoso, R.; Garcia, F. A.; Morales, D.; Mejia, E. G. D. Stability of Betalain Pigments from
a Cactacea Fruit. J. Agric. Food Chem. 1997, 121, 2884–2889.
19
13 Schliemann, W.; Strack, D. Intramolecular Stabilization of Acylated Betacyanins.
Phytochemistry 1998, 49, 585–588.
14 Herbach, K. M.; Stintzing, F. C.; Carle, R. Betalain Stability and Degradation: Structural and
Chromatic Aspects. J Food Science 2006, 71, R41-R50.
15 Herbach, K. M.; Stintzing, F. C.; Carle, R. Identification of Heat-Induced Degradation
Products from Purified Betanin, Phyllocactin and Hylocerenin by High-Performance Liquid
Chromatography/Electrospray Ionization Mass Spectrometry. Rapid Commun. Mass
Spectrom. 2006, 19, 2603-2616.
16 Cai, M.; Pan, X.; Lui, W.; Sheng, J.; Fang, X.; Zhang, C.; Huo, Z.; Tian, H.; Xiao, S.; Dai, S.
Multiple Absorption of Tributyl Phosphate Molecule at the Dyed-TiO2/Electrolyte Interface
to Suppress the Charge Recombination in Dye-Sensitized Solar Cell. J. Mater. Chem. A
2013, 1, 4885-4892.
17 Narayan, M. R. Review: Dye Sensitized Solar Cells Based on Natural
Photosensitizers. Renewable and Sustainable Energy Reviews 2012, 16, 208-215.
18 O’Regan, B.; Grätzel, M. A Low-Cost, High-Efficiency Solar Cell Based on Dye-Sensitized
Colloidal Titanium Dioxide Films. Nature 1991, 353, 737-40.
19 Bartoloni, F. H.; Gonçalves, L. C. P.; Rodrigues, A. C. B.; Dörr, F. A.; Pinto, E.; Bastos, E.
L. Photophysics and Hydrolytic Stability of Betalains in Aqueous Trifluoroethanol. Monatsh
Chem 2013, 144, 567–571.
20 Cai, Y.; Sun, M.; Corke, H. Colorant Properties and Stability of Amaranthus Betacyanin
Pigments. J. Agric. Food Chem. 1998, 46, 4491–4495.
21 Elbe, J. V.; Attoe, E. Oxygen Involvement in Betanine Degradation—Measurement of
Active Oxygen Species and Oxidation Reduction Potentials. Food Chemistry 1984, 49–67
22 Reynoso, R.; Garcia, F. A.; Morales, D.; Mejia, E. G. D. Stability of Betalain Pigments from
a Cactacea Fruit. J. Agric. Food Chem. 1997, 45, 2884–2889.
23 Wybraniec, S.; Starzak, K.; Skopińska, A.; Szaleniec, M.; Słupski, J.; Mitka, K.; Kowalski,
P.; Michałowski, T. Effects of Metal Cations on Betanin Stability in Aqueous-Organic
Solutions. Food Sci Biotechnol 2013, 353–363
24 Moßhammer, M. R.; Rohe, M.; Stintzing, F. C.; Carle, R. Stability of Yellow-Orange Cactus
Pear (Opuntia ficus-indica [L.] Mill. cv. ‘Gialla’) Betalains as Affected by the Juice Matrix
and Selected Food Additives. Eur. Food Res. Technol. 2006, 225, 21–32.
20
25 Herbach, K. M.; Maier, C.; Stintzing, F. C.; Carle, R. Effects of Processing and Storage on
Juice Colour and Betacyanin Stability of Purple Pitaya (Hylocereus polyrhizus) Juice. Eur
Food Res Technol 2006, 224, 649–658.
26 Mabry, T.; Taylor, A.; Turner, B. The Betacyanins and Their Distribution.
Phytochemistry 1963, 2, 61–64.
27 Kanner, J.; Harel, S.; Granit, R. Betalains: A New Class of Dietary Cationized
Antioxidants. J. Agric. Food Chem. 2001, 49, 5178–5185.
28 Kugler, F.; Stintzing, F. C.; Carle, R. Identification of Betalains from Petioles of Differently
Colored Swiss Chard (Beta vulgaris L. ssp. cicla [L.] Alef. Cv. Bright Lights) by High-
Performance Liquid Chromatography−Electrospray Ionization Mass Spectrometry J. Agric.
Food Chem. 2004, 52, 2975–2981.
29 Wybraniec, S.; Mizrahi, Y. Generation of Decarboxylated and Dehydrogenated Betacyanins
in Thermally Treated Purified Fruit Extract from Purple Pitaya (Hylocereus polyrhizus)
Monitored by LC-MS/MS. J. Agric. Food Chem. 2005, 53, 6704–6712.
30 Stintzing, F. C.; Schieber, A.; Carle, R. Identification of Betalains from Yellow Beet (Beta
vulgaris L.) and Cactus Pear [ Opuntia ficus-indica (L.) Mill.] by High-Performance Liquid
Chromatography−Electrospray Ionization Mass Spectrometry. J. Agric. Food Chem. 2002,
50, 2302–2307.
31 Stintzing, F. C.; Schieber, A.; Carle, R. Betacyanins in Fruits from Red-Purple Pitaya,
Hylocereus polyrhizus (Weber) Britton & Rose. Food Chemistry 2002, 77, 101–106.
32 Wybraniec, S.; Mizrahi, Y. Fruit Flesh Betacyanin Pigments in Hylocereus Cacti. J. Agric.
Food Chem. 2002, 50, 6086–6089.
33 Cai, Y.; Sun, M.; Corke, H. HPLC Characterization of Betalains from Plants in the
Amaranthaceae. Journal of Chromatographic Science 2005, 43, 454–460.
34 Cai, Y.-Z.; Sun, M.; Corke, H. Characterization and Application of Betalain Pigments from
Plants of the Amaranthaceae. Trends in Food Science & Technology 2005, 16, 370–376.
35 Sun, M.; Corke, H. Identification and Distribution of Simple and Acylated Betacyanins in the
Amaranthaceae. J. Agric. Food Chem. 2001, 49, 1971–1978.
36 Heuer, S.; Wray, V.; Metzger, J. W.; Strack, D. Betacyanins from flowers of Gomphrena
globosa. Phytochemistry 1992, 31, 1801–1807.
21
37 Kugler, F.; Stintzing, F. C.; Carle, R. Characterisation of Betalain Patterns of Differently
Coloured Inflorescences from Gomphrena Globosa L. and Bougainvillea sp. by HPLC–
DAD–ESI–MS. Anal Bioanal Chem 2006, 387, 637–648.
38 Stintzing, F. C.; Carle, R. Functional Properties of Anthocyanins and Betalains in Plants,
Food, and in Human Nutrition. Trends in Food Science & Technology 2004, 15, 19–38.
39 Gandía-Herrero, F.; Escribano, J.; García-Carmona, F. Structural Implications on Color,
Fluorescence, and Antiradical Activity in Betalains. Planta 2010, 232, 449–460.
40 Wybraniec, S.; Starzak, K.; Skopińska, A.; Nemzer, B.; Pietrzkowski, Z.; Michałowski, T.
Studies on Nonenzymatic Oxidation Mechanisms in Neobetanin, Betanin, and
Decarboxylated Betanins. J. Agric. Food Chem. 2013, 61, 6465–6476.
41 Wendel, M.; Szot, D.; Starzak, K.; Tuwalska, D.; Prukala, D.; Pedzinski, T.; Sikorski, M.;
Wybraniec, S.; Burdzinski, G. Photophysical Properties of Indicaxanthin in Aqueous and
Alcoholic Solutions. Dyes and Pigments 2015, 113, 634–639.
42 Lakowicz, J. R. Principles of Fluorescence Spectroscopy, 3rd
ed.; Plenum Press: New York
2006.
43 Socaciu, C.; Food Colorants: Chemical and Functional Properties; CRC Press LLC Taylor
and Francis 2007.
44 Ellestad, G. A. Structure and Chiroptical Properties of Supramolecular Flower Pigments.
Chirality 2006, 18, 134-144.
45 Fernandes, A.; Bras, N. F.; Mateus, N.; de Freitas, V. A Study of Anthocyanin Self-
Association by NMR Spectra. New J. Chem. 2015, 39, 2602-2611.
22
2 Preparation of Betalain Pigments
2.1 Introduction: Sources and Previous Techniques
Betalains are natural photoprotective pigments found in plants in the order
Caryophyllales. The pigments are all based on Schiff’s base condensations with betalamic acid
and can be separated into two subclasses: betacyanins and betaxanthins as shown in Figure 1.4.
As discussed previously, there are many different structures of betalains based on the amine
source for the condensation, but the primary distinction is that betacyanins are condensation
products of betalamic acid and a cyclo-dopa derivative. All betacyanins are derivatives of
betanidin, which is referred to as the aglycone. Other betacyanins have glucose or other sugar
moieties at either the C5 or C6 position as well as an array of additional acyl groups attached to
these. Figure 1.5 shows a few the betacyanin pigments. Betaxanthins are a much more diverse set
of pigments as they are any non-cyclo-dopa amine derived condensation product with betalamic
acid. The range of betaxanthins found in natural sources comprises the entire spectrum of amino
acids and similar structures. Only a few betaxanthins are expressed in significant quantities in the
sources used in this research and their structures are shown in Figure 1.6. Despite their
occurrence in nature, betaxanthins have shown significant instability and have proven difficult to
isolate. Therefore, betacyanins will be the primary focus of this research.1-23
The stability of the betacyanins is dependent on many factors and is not very well
understood. While food science research has identified factors that influence the stability of these
pigments, their primary focus was on matrix elements.14
Some of this research, however,
23
indicates that the degree and nature of acylation can affect stability.8,13
A study on the stability of
amaranthin derivatives in acidic solution showed that certain acyl groups significantly increased
pigment retention over several hours.13
Another study showed that decarboxylation of
hylocerenin, a betacyanin from pitahaya, increased the thermal stability of the compound to
temperatures approaching 100˚C.15
For practical reasons, the stability of extracted pigments is of
great interest for DSSC applications, and so methods for isolation of these compounds were
developed in order to elucidate these structure-function properties.
A significant amount of research has been done in the field of betacyanin extraction and
purification, and many different methods have been developed over the 40+ years of inquiry.15,21
The methods used in this research were developed off of an assortment of recent publications by
Carle et al. (references 15-18) and Wybraniec et al (reference 26). Since betacyanins are quite
hydrophilic and have multiple charge centers (three carboxylic acid groups, two nitrogen
heteroatoms and plenty of hydroxyl groups), development of a separation technique was a
difficult proposal. These two research groups have developed techniques which combine high
performance liquid chromatography and solid phase extraction to produce ultra-pure pigment
samples as well as a more bulk material process involving ion-exchange chromatography to
produce mixtures of betacyanins removed from other matrix components.22
This research uses
these previously developed methods as a starting point for our purification process.
24
2.2 Preparatory Techniques
2.2.1 Extraction
Raw materials were purchased (beets and pitahaya) from local supermarkets, or harvested
from our personal gardens (amaranth) for use in this study. All of the raw materials were
processed in a similar manner and will be discussed simultaneously with exceptions noted. Raw
materials were juiced in a macerating juicer to produce a colored liquid. The pulp by-product
was re-processed by soaking in 60% ethanolic solution then passed through the juicer again. Raw
juice was centrifuged for 20 minutes at 10,000 rpm then filtered through three consecutive filters,
Whatman #4, #1, #42 respectively. The filtered juice was then diluted with ethanol to a total
concentration of 60% ethanol and allowed to sit at -20˚C overnight to denature proteins. The
supernatant was decanted and filtered through a series of glass microfiber filters at 2.7µm
(GF/D), 1.6µm (GF/A), 0.4µm and 0.2µm. Once filtered the juice was adjusted to pH 3.5 for
further purification. For samples intended for HPLC or MPLC purification, ethanol was removed
via rotary evaporation at 30˚C and reduced pressure.
2.2.2 HPLC
Reversed phase high performance liquid chromatography (RP-HPLC) is the standard
separatory method to begin with for method development. In RP-HPLC the stationary phase is
hydrophobic and a moderate degree of hydrophobicity is required for analyte retention. The
analytes are then eluted from the column by using a mixture of aqueous and non-aqueous
solvents as the mobile phase.24
For the purposes of this discussion the mobile phase will be
limited to two components labeled A for aqueous and B for organic. A mixture of the two
25
components is used to tune the retention of the analytes to the desired level. The simplest form of
mixture is called an isocratic elution which maintains a constant ratio of A and B for the duration
of the experiment. Each analyte is eluted in order of increasing hydrophobicity based on its
partitioning between mobile and stationary phase. Isocratic experiments can be useful if the
analytes have similar hydrophobicity, but are not very effective at separating components over a
wide range of character. In this instance a binary gradient is used, whereby the amount of B is
increased over time to cause the more highly retained analytes to elute sooner. By using a binary
gradient samples with components comprising a wide range of hydrophobic character can be
efficiently separated in a short period of time. Solvent selection also plays a crucial role in tuning
the retention characteristics of the analytes. The aqueous phase typically consists of a buffer to
maintain a pH that produces the most ideal (hydrophobic) conditions of the analyte. For example,
analytes that are acidic in nature will be separated using a low-pH buffer whereas those that are
basic will use a high-pH buffer. The organic phase is typically chosen based on the protonation
consideration of the analyte. There are two primary solvents used in most RP-HPLC methods:
methanol (MeOH) and acetonitrile (MeCN). The main distinction between these two solvents is
the presence of a hydrogen atom on MeOH that can be involved in chemistry with the analytes.
If pH is a significant factor in the efficiency of separation, methanol can change the behavior of
organic acids and should be avoided. In this scenario MeCN is ideal because it will have little
effect on the proton activity in the solvent, and minimize detrimental effects due to pH changes.
It is worth noting, however, that cost and scaling-up can promote use of MeOH over MeCN in
certain circumstances, but additional controls to the pH may be necessary. Other organic solvents
26
can be used as long as they are miscible with water; however, MeOH and MeCN are the main
choices for starting method development.
Additional measures may be necessary to generate the proper hydrophobic environment
for analyte retention when the species have hard to neutralize charge centers. In these cases, an
ion-pairing agent can be used. Typically, a perfluorinated organic acid is used as an ion pairing
agent for positively charged species and a tertiary amine for negatively charged species. These
ion-pairing agents are added to the mobile phase and associate strongly with the analytes to
produce a greater hydrophobic character, and increase retention on the stationary phase. In some
cases the addition of ion-pairing agents can significantly enhance the separation of similar
species. In addition to ion-pairing agents, chiral-pairing agents can be used to enhance
separations of chiral molecules.
For our research with beets, amaranth and pitahaya, RP-HPLC methods were developed
on a C18 stationary phase using a binary gradient with an anionic ion-pairing agent and MeCN.
The first hurdle was to achieve hydrophobicity in the betacyanins. The carboxylic acid groups
were protonated by using an acidic buffer with a pH from 2-3 based on acetic or formic acid. It
was found that formic acid performed ideally by producing a stable buffer at pH 2 whereas the
acetic acid produced a buffer at approximately pH 3. While the initial studies used acetic acid,
for the later studies, I settled on formic acid because of the lower pH and lower boiling point.
Despite the higher purity product that HPLC garners, this method required less technical
development, especially with the large body of literature that has contributed to the methodology
of HPLC to betalain isolation. Some minor adjustments to the mobile phase and gradient were
necessary for use with the specific column we obtained from Phenomenex, which consisted of
27
superficially porous particles compared to the fully porous particles used in most previous
betalain research. The primary components of the mobile phase were 0.1% (V/V) trifluoroacetic
acid (TFA), as the ion-pairing agent, and 1% (V/V) formic acid (FA) in the aqueous phase and
pure acetonitrile (MeCN) as the organic phase. These components were chosen based on the
evidence that perfluorinated acids significantly enhanced the retention characteristics of betalains
in C18 RP-HPLC as discussed previously.
Originally, a simple binary gradient was developed using 10% MeCN and 90% aqueous
(5% acetic acid), which produced good results for betanin from beets, but could not resolve
amaranthin. The amaranthin was not retained at all. For betanin, the retention was sufficient to
isolate the primary pigment, though analysis at 254 nm showed significant residual impurities.
Thus, a better gradient or mobile phase was required. Then I tried a mobile phase consisting of
0.25% TFA as the aqueous component. The switch to TFA did not in itself alleviate the issue,
and the B% had to be reduced to 7%. The high concentration of TFA was a concern given its
high acidity, and so a combined aqueous phase of FA and TFA was developed. The weaker
acidity of the formic acid creates a buffer to keep the mobile phase at pH 2 while the TFA
generates the added retention characteristics discussed above. Being an organic acid, FA should
not produce significant salty deposits on/in the instrument and its high volatility decreases the
amount of residual acid in the final product. Despite the high volatility of TFA, I suspect that the
majority exists as the salt in the eluent and remains in the product despite low pressure rotary
evaporation.
28
2.2.3 MPLC
Previous work in the McHale lab had focused on medium pressure liquid
chromatography (MPLC) as the purification process. The previous researchers used an isocratic
flow scheme with 80% MeOH and 20% purified water on a reversed phase C18 column. This
method provided minimal purification, and needed to be performed multiple times to produce
semi-pure dye. In retrospect, it is clear that the mobile phase was too polar and the betacyanin
was eluted with the solvent front while the less hydrophilic betaxanthins were partially retained.
While the resulting pigment could be purified to be almost betaxanthin free, it did not allow for
separation from sugars and salts that would also elute with the solvent front.
Figure 2.1: Preparatory HPLC of amaranth extract with amaranthin (1) and isoamaranthin (2) the predominant pigments, but small amount of decarboxylated betacyanins (3 &4) being present.
29
With the help of the HPLC methods, an appropriate MPLC method could be developed
for the amaranth samples. Similar to the HPLC method, an acidic buffer with TFA was used, and
the organic phase was chosen to be methanol for environmental and cost concerns. It was found
that an isocratic mixture of 7% MeOH and 93% aqueous buffer provided sufficient retention of
the amaranthin to be isolated. The main advantage of the MPLC technique over that of HPLC is
that it allowed for significant amounts of secondary pigments to be purified. In particular, one
secondary betacyanin from amaranthin was isolated using this procedure. While MPLC allowed
for a significant scale increase over HPLC, it was ultimately determined that the most cost
effective mass production scheme was ion-exchange chromatography as discussed below. It is
worth noting, however, that MPLC provides isolated betalain compounds whereas the ion-
exchange method provides mixtures of betacyanins and could be used in future research to
evaluate secondary betacyanin pigments.
2.2.4 Anion-Exchange Column Chromatography
For large scale production, the HPLC system in our lab is insufficient given the small
injection volume and long run time. Therefore, a bulk processing procedure was developed based
on an anion exchange protocol, shown schematically in Figure 2.2. By using Sephadex DEAE-
A25 the betacyanins can be separated from the betaxanthins and proteins. The raw solution is
adjusted to pH 3.5 with 100 mM sodium phosphate buffer. 10-15 g of Sephadex DEAE-A25 is
prepared by forming a slurry in 500 mL of pH 3.5 buffer and allowing to swell for a minimum of
5 hours. After equilibrating with three 500 mL aliquots of pH 3.5 buffer, the raw beet/amaranth
juice is added to the column until the stationary phase has been saturated with pigment (a small
amount of red pigment begins eluting). Following loading, the column is rinsed with a minimum
30
of 5 volumes of pH 3.5 buffer, or until the eluent runs clear with no residual yellow color. To
desorb the pigment, the column is washed with pH 7.5 100 mM sodium phosphate buffer.
Following exchange chromatography, the product is desalted with solid phase extraction on a
C18 cartridge. The cartridge is prepped with five volumes of methanol, and then equilibrated
with 3 volumes of TFA/FA buffer. The pigment solution is brought to a pH of 2 before being
applied to the cartridge. Once the cartridge is loaded it is rinsed with 10 volumes of TFA/FA
buffer or until eluent runs clear. Betanin/amaranthin is eluted with pure methanol. The resultant
solution is dried under reduced pressure at 30˚C.
As validation of this method to produce purified mixtures of betacyanins, HPLC analysis
was conducted on the purified powder and compared with the micro filtered juice. By monitoring
the chromatogram at 254 nm we were able to see all components that would be contaminates of
Figure 2.2: Schematic of the preparatory process developed in this research.
Juice beets
Denature proteins
Microfilter
Acidify: pH 3.5
Anion Exchange: DEAE-A25
pH 3.5 PO4
Load Elute pH 7.5
<pH 2
SPE
MeOH 0.1% TFA/1%FA
Load Elute MeOH
Dry @ 30˚C
Filter
#4 #1 #42/5
GF/D GF/A 0.44µm 0.22µm
31
our sample. Figure 2.3 shows two chromatograms that evidence the efficacy of our anion
exchange protocol. The purified amaranth consists of only one species and no residual absorbing
contaminate. The filtered juice, on the other hand, has so many components that it is difficult to
tell which peak corresponds to the amaranthin. The retention times on these two chromatograms
do not exactly match because they were run with different protocols as discussed above. Similar
quality analysis procedures were run on the samples from beets and pitahaya and the purified
sample chromatograms are show in Figure 2.4 below.
2.3 Conclusions
The preparation of isolated betacyanin compounds has led to significant increases in the
control of experimental variables in exploring the photophysical properties of betalains. The
three primary preparatory techniques discussed here allow for gram-scale quantities of betanin,
amaranthin and other betacyanins to be produced from their raw material sources. In their
Figure 2.3: HPLC chromatogram of purified amaranth (a) and that of filtered amaranth juice (b) monitored at 254 nm.
32
powdered form, these pigments are stable indefinitely, and can be used in many different studies.
It is likely that the impure nature of the pigments used in previous research has led to
irreproducibility in the data. Careful control of sample purity is essential for laying the ground
work for mapping the structure-function relationships in betalain pigments and constitutes a great
achievement toward understanding their fundamental properties.
Although we were unable to use mass spectrometry to positively identify the individual
betacyanins present in our sources, we can compare the relative retention rates of the betacyanins
using HPLC. Figure 2.4 shows HPLC chromatograms of the betacyanin mixtures obtained from
the anion exchange preparation of amaranth, beets and pitahaya. By comparing with literature,
the primary components could be identified as amaranthin, betanin, phylocactin and hylocerinin.
Figure 2.4: HPLC chromatogram of purified samples of amaranth (black), pitahaya (red), beets (green) and a combination of all the sources (blue). The primary betacyanins are marked as amaranthin (A), betanin (B), phylocactin (C) and hylocerinin (D).
33
The betacyanins from amaranth and beets were used in further studies as discussed below.
Unfortunately, time constraints did not allow for further study of the pigments from pitahaya.
34
2.4 References
1 Wybraniec, S.; Mizrahi, Y. Fruit Flesh Betacyanin Pigments in Hylocereus Cacti. J. Agric.
Food Chem. 2002, 50, 6086–6089.
2 Cai, Y.; Sun, M.; Corke, H. HPLC Characterization of Betalains from Plants in the
Amaranthaceae. Journal of Chromatographic Science 2005, 43, 454–460.
3 Cai, Y.-Z.; Sun, M.; Corke, H. Characterization and Application of Betalain Pigments from
Plants of the Amaranthaceae. Trends in Food Science & Technology 2005, 16, 370–376.
4 Sun, M.; Corke, H. Identification and Distribution of Simple and Acylated Betacyanins in the
Amaranthaceae. J. Agric. Food Chem. 2001, 49, 1971–1978.
5 Heuer, S.; Wray, V.; Metzger, J. W.; Strack, D. Betacyanins from Flowers of Gomphrena
globosa. Phytochemistry 1992, 31, 1801–1807.
6 Kugler, F.; Stintzing, F. C.; Carle, R. Characterisation of Betalain Patterns of Differently
Coloured Inflorescences from Gomphrena globosa L. and Bougainvillea sp. by HPLC–
DAD–ESI–MS n. Anal Bioanal Chem 2006, 387, 637–648.
7 Stintzing, F. C.; Carle, R. Functional Properties of Anthocyanins and Betalains in Plants,
Food, and in Human Nutrition. Trends in Food Science & Technology 2004, 15, 19–38.
8 Gandía-Herrero, F.; Escribano, J.; García-Carmona, F. Structural Implications on Color,
Fluorescence, and Antiradical Activity in Betalains. Planta 2010, 232, 449–460.
9 Bartoloni, F. H.; Gonçalves, L. C. P.; Rodrigues, A. C. B.; Dörr, F. A.; Pinto, E.; Bastos, E. L.
Photophysics and Hydrolytic Stability of Betalains in Aqueous Trifluoroethanol. Monatsh
Chem 2013, 113, 567–571
10 Cai, Y.; Sun, M.; Corke, H. Colorant Properties and Stability of Amaranthus Betacyanin
Pigments. J. Agric. Food Chem. 1998, 46, 4491–4495.
11 Elbe, J. V.; Attoe, E. Oxygen Involvement in Betanine Degradation—Measurement of
Active Oxygen Species and Oxidation Reduction Potentials. Food Chemistry, 1984, 16, 49–
67.
12 Reynoso, R.; Garcia, F. A.; Morales, D.; Mejia, E. G. D. Stability of Betalain Pigments from
a Cactacea Fruit. J. Agric. Food Chem. 1997, 121, 2884–2889.
35
13 Schliemann, W.; Strack, D. Intramolecular Stabilization of Acylated
Betacyanins. Phytochemistry 1998, 49, 585–588.
14 Wybraniec, S.; Starzak, K.; Skopińska, A.; Szaleniec, M.; Słupski, J.; Mitka, K.; Kowalski,
P.; Michałowski, T. Effects of Metal Cations on Betanin Stability in Aqueous-Organic
Solutions. Food Sci Biotechnol 2013, 22, 353–363.
15 Herbach, K. M.; Stintzing, F. C.; Carle, R. Identification of Heat-Induced Degradation
Products from Purified Betanin, Phyllocactin and Hylocerenin by High-Performance Liquid
Chromatography/Electrospray Ionization Mass Spectrometry. Rapid Commun. Mass
Spectrom. 2006, 19, 1822–1822.
16 Herbach, K. M.; Stintzing, F. C.; Carle, R. Betalain Stability and Degradation˗˗Structural and
Chromatic Aspects. J Food Science, 2006, 71, R41-R50.
17 Moßhammer, M. R.; Rohe, M.; Stintzing, F. C.; Carle, R. Stability of Yellow-Orange Cactus
Pear (Opuntia ficus-indica [L.] Mill. cv. ‘Gialla’) Betalains as Affected by the Juice Matrix
and Selected Food Additives. Eur. Food Res. Technol. 2006, 225, 21–32.
18 Herbach, K. M.; Maier, C.; Stintzing, F. C.; Carle, R. Effects of Processing and Storage on
Juice Colour and Betacyanin Stability of Purple Pitaya (Hylocereus polyrhizus) Juice. Eur.
Food Res. Technol. 2006, 224, 649–658.
19 Mabry, T.; Taylor, A.; Turner, B. The Betacyanins and Their Distribution.
Phytochemistry 1963, 2, 61–64.
20 Kanner, J.; Harel, S.; Granit, R. Betalains: A New Class of Dietary Cationized
Antioxidants. J. Agric. Food Chem. 2001, 49, 5178–5185.
21 Kugler, F.; Stintzing, F. C.; Carle, R. Identification of Betalains from Petioles of Differently
Colored Swiss Chard (Beta vulgaris L. ssp. cicla [L.] Alef. Cv. Bright Lights) by High-
Performance Liquid Chromatography−Electrospray Ionization Mass Spectrometry J. Agric.
Food Chem. 2004, 52, 2975–2981.
22 Wybraniec, S.; Mizrahi, Y. Generation of Decarboxylated and Dehydrogenated Betacyanins
in Thermally Treated Purified Fruit Extract from Purple Pitaya (Hylocereus polyrhizus)
Monitored by LC-MS/MS. J. Agric. Food Chem. 2005, 53, 6704–6712.
23 Stintzing, F. C.; Schieber, A.; Carle, R. Identification of Betalains from Yellow Beet (Beta
vulgaris L.) and Cactus Pear [Opuntia ficus-indica (L.) Mill.] by High-Performance Liquid
36
Chromatography−Electrospray Ionization Mass Spectrometry. J. Agric. Food Chem. 2002,
50, 2302–2307.
24 Wybraniec, S.; Mizrahi, Y. Influence of Perfluorinated Carboxylic Acids on Ion-Pair
Reversed-Phase High-Performance Liquid Chromatographic Separation of Betacyanins and
17-Decarboxy-Betacyanins. Journal of Chromatography A 2004, 1029, 97–101.
37
3 Templated Assembly of Betanin Chromophore on TiO2: Aggregation-
Enhanced Light Harvesting and Efficient Electron Injection in a
Natural Dye-Sensitized Solar Cell
Reprinted and adapted with permission from Treat, N. A.; Knorr, F. J.; McHale, J. L.,
Templated Assembly of Betanin Chromophore on TiO2: Aggregation-Enhanced Light-
Harvesting and Efficient Electron Injection in a Natural Dye-Sensitized Solar Cell. J. Phys.
Chem. C, 2016, 120 (17), 9122-9131. DOI:10.1021/acs.jpcc.6b02532
3.1 Introduction
Ongoing efforts to optimize solar energy conversion are frequently inspired by Nature,
which uses finely tuned assemblies of chromophores to harness sunlight.1,2
In photosynthetic
organisms, light-harvesting complexes of chlorophyll and bacteriochlorophyll derivatives are
optimized to collect and funnel solar energy to reaction centers where charge separation takes
place. TiO2-based dye-sensitized solar cells (DSSCs), on the other hand, depend on adsorbed
dyes which perform both the light-harvesting and interfacial charge-separation steps.3,4,5
While
self-assembly of dyes on the metal oxide surface is prevalent owing to their high surface
density,6,7
aggregation of dyes on TiO2 is frequently reported to lower the rate and yield of
electron injection, resulting in lower photocurrents.8,9,10,11,12
Consequently, high-performing
DSSCs often employ spacer molecules,13,14,15,16
competitively binding molecules such as
organophosphates,17
functionalized dyes with steric hindrance,18,19,20
or surface treatments21
to
prevent dye aggregation. The reduced photocurrents that result from sensitizer aggregation are
38
variously attributed to decreased excited state lifetime, self-quenching, attenuation of light by a
thicker dye layer, and weak electronic coupling of dyes to TiO2 as a result of greater distance
from the surface.
On the other hand, dye aggregation can be beneficial to solar energy conversion in that it
is often accompanied by spectral broadening which can improve the overlap of the device optical
absorption with the solar spectrum. Structured biomimetic chromophore assemblies are being
pursued with the goal of exploiting energy transfer in solar energy conversion,22,23,24,25,26
and
there is additional incentive to replace costly synthetic sensitizers with natural pigments.27,28,29
Unfortunately, none of these biomimetic approaches have resulted in energy conversion
efficiencies that rival those of DSSCs using monomeric organic or metallorganic sensitizers.
Figure 3.1: Structure of betanin (left) and 3D representation of optimized geometry (right) obtained using DFT as described in Ref. 42.
39
Better understanding of the influence of aggregation on TiO2-based solar energy conversion
should permit the enhanced light harvesting of dye assemblies to be exploited without
diminished yield of photoelectrical conversion.
Recently, we have used betanin (Bt) extracted from beet root as a sensitizer and achieved
power conversion efficiencies up to 2.7% and high incident photon-to-current conversion
efficiency (IPCE).30
Bt (Figure 3.1) is a light-harvesting plant pigment belonging to the betalain
family, the pigments of which serve photoprotective and anti-oxidant roles in plants of the order
Caryophyllales.31,32
Betalain pigments can be separated into two subclasses of molecules based
on Schiff’s base condensation with betalamic acid and a nitrogen hetero atom. The simpler
subclass of betalains is the betaxanthins which are condensation products with various amino
acids and amines. Betacyanins, including betanin, are more specifically the condensation
products of cyclo-dopa derivatives with betalamic acid. While betalamic acid and the
betaxanthins exhibit a strong absorption in the blue, with molar absorptivity ε = 45,000 M-1
cm-1
at 470-500 nm, betacyanins absorb in the green, ε = 65,000 M-1
cm-1
at 525-535 nm.33,34
Like
their cousins the anthocyanin plant pigments, betalains have been explored as natural sensitizers
in DSSCs by our group30,35,36
and others.37,38,39
Though anthocyanins are known to self-assemble
in solution and in vivo,40, 41
we are not aware of any studies pointing to aggregation of betanin or
other betalains in vivo or in vitro. However, we have noted previously that the absorption
spectrum of Bt on TiO2 is considerably broader than that of aqueous Bt, and have speculated that
dye aggregation may be responsible for enhanced light-harvesting.36
Though previously we have
considered that preferential absorption of betaxanthins on TiO2 might have contributed to the
blue shift, in the present work HPLC analysis confirms the absence of betaxanthins or other
40
yellow pigments. As shown in Figure 3.1, the optimized ground state geometry of Bt as obtained
by a DFT calculation is decidedly nonplanar, which would inhibit intermolecular interactions via
π-stacking. However, the presence of three carboxylic acid functions and two nitrogen
heteroatoms on Bt provides ample opportunity for intermolecular interactions though hydrogen-
bond formation, perhaps assisted by templated adsorption on the TiO2 surface.
We have reported femtosecond and nanosecond transient absorption spectroscopy of Bt-
sensitized TiO2 along with spectroelectrochemical measurements and DFT calculations,
revealing that excited state Bt undergoes a two-electron, one-proton oxidation.42,43
In the course
of investigating the electrochemical oxidation of Bt on TiO2 as a function of dye-loading, we
observed spectral changes suggestive of aggregation. In this work, we use internal and external
quantum efficiencies for electron collection, fluorescence measurements, and numerical
modeling to understand the evolution of quantum efficiency and light-harvesting as a function of
dye-loading. We report here a record high power conversion efficiency, 3%, for a natural-dye
based DSSC, sensitized with aggregated Bt. Evidence from numerical modeling and theory will
be shown to lead to the conclusion that an excitonically coupled dimer of Bt forms on the surface
of TiO2, leading to enhanced light harvesting, electron injection, and energy conversion
compared to analogous DSSCs containing monomeric Bt.
3.2 Experimental Methods
3.2.1 Purification
Betanin was extracted from red beet (Beta vulgaris L.) via a modified procedure from
Ref. 44. Red beet root was crushed and juiced by a macerating juicer. The juice was mixed 1:1
41
with ethanol, and filtered to remove proteins. This mixture was adjusted to pH 4 with HCl.
Betanin was isolated, on a Sephadex DEAE A25 column at pH 4, and washed with a 0.05 M pH
4 acetate buffer. Betanin was desorbed from the column with a 0.05 M pH 7 phosphate buffer.
All excess salts in the eluent were removed and the product further purified by solid phase
extraction on a Phenomenex Strata C18 SPE column and washed with five volumes of water
acidified with HCl to a pH of 2. The betanin was desorbed with methanol acidified with a small
amount of HCl and dried under reduced pressure at 30 ºC. Purity was confirmed by analytical
HPLC.44
3.2.2 HPLC
Analytical HPLC was performed via a modified method using a binary gradient with
solution A (0.25% aqueous trifluoroacetic acid) and solution B (acetonitrile). The gradient is as
follows: 0-1min 7.2% B, then ramped to 10% B at 5 min, then to 15% B at 7 min, followed by a
rinse of 100% B at 8 min for 1 min, then equilibrated for 2 min at 7.2% B.45,46,47
3.2.3 Film preparation and sensitization (TiO2 and ZrO2)
Films of TiO2 were prepared on a conductive substrate, fluoride doped tin oxide (FTO),
for DSSC construction, while films of TiO2 and ZrO2 for absorption and emission spectroscopy
were prepared on glass microscope slides. For DSSC construction, a blocking layer of titanium
dioxide was prepared by heating FTO coated slides (Tech15, Hartford Glass) in a solution of
dilute 0.1 M TiCl4 to 60°C for 30min. The coated films were dried in an oven at 100°C for
30min before being heated in a muffle furnace to 500°C for 30 min. Films of titanium dioxide
were prepared by the doctor blade method with Dyesol 18NR-T paste on FTO glass slides. The
film consists of 20 nm nanoparticles of TiO2 in the anatase phase. Typical SEM images of the
42
resulting films are shown in Figure 3.7, along with side views which show the film thicknesses.
3.5 μm thick films were prepared with a 50% dilution of the TiO2 paste with ethanol, 7μm films
were prepared by using undiluted paste, 14 μm films were prepared by 2 successive coatings of
paste. The films were placed in a drying oven at 100°C for 1hour to smooth then heated in a
muffle furnace to 500°C for 30 min. Some of the films were selected for an additional treatment
with 0.1 M TiCl4 after cooling, as in Ref 30. Prior to sensitization films were equilibrated either
in purified water or in 0.5 M hydrochloric acid in ethanol.
For absorption and fluorescence measurements, films of TiO2 and ZrO2 were prepared by
doctor-blading directly onto glass slides. TiO2 films for spectroscopy were 3.5 μm, leading to
sufficient transmission for accurate absorption measurement. Owing to lower tendency of Bt to
adsorb on ZrO2 compared to TiO2, 7 μm thick ZrO2 films were used in order to obtain sufficient
dye loading for spectroscopy measurements. Zirconium dioxide films were prepared from Zr-
Nanoxide Z/SP paste purchased from Solaronix. Prior to sensitization, films for fluorescence
measurements were equilibrated in a solution of 0.25% aqueous trifluoroacetic acid. Since the
fluorescence measurements (described below) were done using front face emission and corrected
for absorptance, the film thickness is not critical to the comparison of fluorescence of Bt on TiO2
and ZrO2. Bt-sensitized ZrO2 films were used only for absorption and emission, not for
fabrication of dye-sensitized solar cells.
Films were sensitized in solutions with varying amounts of betanin powder dissolved in
25 mL of purified water and the pH was adjusted by dropwise addition of aqueous 1 M
hydrochloric acid. Films were sensitized for varying times to produce a range of aggregation. To
produce films with very low amounts of aggregation, 1 mM sodium phosphate was used in place
43
of purified water to prepare the betanin sensitizing solution. Phosphate ions are known to bind
strongly to TiO2 in contact with aqueous solutions,48
thus we expected competitive binding
would limit self-assembly in the presence of sodium phosphate.
3.2.4 Absorption and fluorescence spectra
Fluorescence spectra of Bt-sensitized TiO2 and ZrO2 films were excited using a Lexel
95L-UV laser operating at 514.5 nm, using a 45° front facing geometry. The incident laser power
was attenuated to 20 ± 2μW using a neutral density filter. Elastically scattered laser radiation was
rejected with a holographic notch filter. Emission from 400-900 nm was detected with a
thermoelectrically cooled CCD camera. All spectra were taken within 10 min of removal of films
from the sensitizing solution. Films were otherwise kept in the dark to minimize photo-
degradation of betanin.
Absorption spectra were taken on a Shimadzu UV-2550 spectrometer before and after
fluorescence measurements to check for dye degradation. Fluorescence spectra were scaled by
the absorptance (1−10−A
, where A is the absorbance of the film at the excitation wavelength) to
correct for the variations in the percent of light absorbed for different samples.
3.2.5 IPCE and APCE
Incident photon-to-current conversion efficiency was recorded as a function of
wavelength, IPCE(λ), as described in a previous publication.30
Adsorbed photon-to-current
conversion efficiency APCE(λ) was obtained from IPCE(λ) by dividing by the absorptance.
IPCE(λ)=Jsc(𝜆)
eΦ(𝜆) Equation 3.1
APCE(λ)=IPCE(λ)
1−10−A(λ) Equation 3.2
44
η=JscVocFF
Pin=Pmax
Pin Equation 3.3
In the above, Jsc is the short-circuit current, Φ(λ) is the incident photon flux at wavelength λ, A(λ)
is the absorbance at wavelength λ, η is the total energy conversion efficiency, Voc is the open
circuit voltage, FF is the fill factor, and Pin is the incident radiation power.
3.2.6 Solar Cell Construction.
Platinum counter electrodes were prepared by heating an FTO glass slide to 40˚C and
adding 6 drops of platinum hexachloride dissolved in isopropyl alcohol. Once dry, the films were
sintered at 450˚C for 30 min. Solar cells were constructed by sandwiching the sensitized TiO2
film and the platinum electrode, and then sealed on 3 sides with hot-melt glue. On the fourth side
a drop of 0.50 M lithium iodide and 0.05 M iodine dissolved in acetonitrile was applied and
allowed to fill the gap between electrodes. Hot melt glue was then applied to the remaining side.
The films were tested via the method published in Ref. 30.
3.2.7 Numerical methods.
Non-negative matrix factorization (NMF) was used to determine the number of
components and their spectra for Bt-sensitized films as a function of dye-loading. NMF is a self-
modeling data interpretation method that resolves a data set into its individual components. The
structure of the data must be a linear combination of the components, which in the present case
are spectra of distinct forms of Bt on the TiO2 surface and the background spectrum of TiO2.
The component spectra and their weights are required to have positive values. In this case, the
absorbance of the betanin sensitized TiO2 films is found to be the linear combination of three
components as discussed below. To apply this method, we construct an 801 x 92 data matrix, D,
45
whose 92 columns are the measured absorbance spectra of the samples and whose 801 rows are
the wavelengths (measured every 0.5 nm). Using the algorithm of Lee and Seung49
, D is
factored into smaller all-positive matrices: R, an 801 x 3 matrix which contains the normalized
spectra of the three individual components in its columns and C, a 3 x 92 matrix which contains
the concentration and magnitude information; i.e., D = RC. The spectra in the R matrix are
normalized as part of the algorithm, so it is not possible to separate the magnitude of the molar
absorptivity from the concentration in the C matrix. The number of components was determined
to be three, based on an examination of the residuals of the model of the data and the
reproducibility of the components. The residuals are the difference between the measured data
and the modeled D matrix, equal to RC calculated using the factored R and C matrices. If not
enough components are used, then the data cannot be accurately reproduced, if too many
components are used, then experimental noise is mistaken for a component.
3.3 Results and Discussion
3.3.1 Absorption Spectra and Numerical Modeling of Betanin on TiO2 and ZrO2
Figure 3.2A and B show the absorption spectrum of Bt adsorbed on films of
nanocrystalline ZrO2 and TiO2, respectively, as a function of sensitization time. The absorption
spectrum of the sensitizing solution is shown in Figure 3.8, compared to that of a sensitized TiO2
film at high dye-loading. For low dye-loading, the absorption spectrum of Bt on TiO2 resembles
that of the aqueous solution with an absorption maximum near 530 nm. Films exposed to the
sensitizing solution for longer times are seen to exhibit blue-shifted absorption maxima and
spectral broadening. Previous evidence showed that the absorption spectrum of Bt on ZrO2 did
46
not exhibit as significant a blue-shift and was therefore used to model a film with higher
monomer character. Another factor that we considered was that the blue-shift resulted from a
degradation of the betanin chromophore to a yellow oxidation product.50
This
degradation/oxidation product was observed during time-resolved absorption experiments. We
have previously shown that Bt/ZrO2 films are stable to photooxidation owing to the more
negative conduction band potential of ZrO2 compared to TiO2.42,43
Though Bt appears to absorb
less strongly on ZrO2 than on TiO2, it is clear that in both cases longer sensitization times lead to
a blue shift that accompanies the increased absorbance of the film. Thus the spectral changes
observed at high dye-loading are not the result of photodegradation of the dye. To confirm that
degradation was not the source of the blue-shift, Bt was desorbed from the TiO2 films and the
extract submitted to HPLC analysis. As shown in Figure 3.9, HPLC analysis of solutions of
material desorbed from the films confirms that betanin is the only component giving rise to any
visible light absorption. Similarly, the absorption spectrum of the extract following dye
desorption strongly resembles that of the original sensitizing solution. These results strongly
indicate that betanin undergoes reversible templated aggregation on the surface of TiO2 and ZrO2
films. It should also be noted that repeated efforts to observe Bt aggregation in aqueous solution
were not successful: no deviations from Beer’s law were observed in the absorption spectrum at
the highest achievable Bt concentrations, Figure 3.10. We conclude that intermolecular
interactions of closely packed molecules affect the absorption properties of the dye on the
surface of nanocrystalline TiO2 and ZrO2 at high dye loadings. For similar staining times Bt dye
loading is higher and aggregation is more readily observed on TiO2 than on ZrO2. Further
47
confirmation of dye aggregation was obtained by using phosphate ions as co-absorbents.48
As
shown in Figure 3.8, these inhibit the spectral changes otherwise observed at high dye loadings.
We considered the possibility that spectral shifts and inhomogeneous broadening might
result from a distribution of binding sites which perturb the transition frequency, versus the
formation of discrete aggregates with shifted spectra. We therefore used nonnegative matrix
factorization (NMF) to discern the number of components and their resolved spectra which
contribute to the total absorption spectrum.51
Ninety-two UV-visible absorption spectra of
betanin on TiO2 and ZrO2 films and in solution were compiled and analyzed by NMF. Initial
attempts to analyze the data of sensitized films considered a variable number of components and
revealed that three components were sufficient to account for the total spectrum. Note that the
NMF analysis included a series of spectra: 37 sensitized TiO2 and 19 sensitized ZrO2 films, plus
Figure 3.2: Absorption spectra of Bt on A) ZrO2 and B) TiO2 as a function of sensitization time in a saturated betanin solution at pH 2. A 900s film on TiO2 was prepared, but was too dark to determine the absorbance.
48
20 bare films and 16 solution phase spectra. The spectral decomposition and residuals for the
entire series is shown in Supporting Information, Figure 3.12-Figure 3.14. The NMF method
cannot separate each component spectrum into the amount and relative intensity of each
contribution. Rather, it provides a basis set of three spectra for which each experimental
spectrum can be expressed as a linear combination thereof.
Figure 3.3 displays the results of NMF analysis for a film containing a relatively low
(Figure 3.3A) and high (Figure 3.3B) contribution of aggregate to the total absorption of Bt on
TiO2. The three individual components are accounted for as the absorption/scattering of TiO2,
and the absorption of monomer and aggregated forms of Bt. The data in Figure 3.3A are for a
sample for which NMF reveals 75% monomer and 25% aggregate contributions, while Figure
3.3B corresponds to 35% monomer and 65% aggregate. As seen below, the aggregate spectrum
Figure 3.3: Results of NMF analysis of Bt/TiO2 at A) low and B) high dye loading. The films in A) and B) were sensitized for 300s and overnight (21 hours) and contain 25% and 65% aggregate, respectively. The components are dimer (green), monomer (blue), TiO2 extinction (cyan), experimental absorbance (black), and total calculated absorbance (red dashed).
49
is accounted for well if it is attributed to a dimer of Bt. Since spectra were obtained in
transmission, the TiO2 contribution consists of the band edge absorption (at ~390 nm) as well as
scattering at visible wavelengths. NMF analysis of solution phase spectra produced only minor
amounts of dimer (Figure 3.13B), consistent with adherence to Beer’s law. We believe the small
apparent dimer contribution in the resolved solution phase spectra is a consequence of the
inability of the model to separate the relatively similar spectra of the monomer and dimer. (See
Figure 3.8.) Blank films produced only the TiO2 (or ZrO2) component spectrum (Figure 3.13C).
The consistency with which the NMF method produced component spectra similar to those of
the bare films and the monomer in solution lends confidence that the third component is an
aggregated form of Bt. Based on the NMF, the absorption spectrum of aggregated betanin
consists of two bands which are respectively blue- and red-shifted relative to the monomer
spectrum, at about 470 and 607 nm respectively. This is in complete accord with excitonic
splitting of a dimer (vide infra), in which transition dipole moment coupling leads to two new
absorption bands. Dimer aggregates are quite common in the plant pigments of the anthocyanin
family, and it is reasonable to speculate that betalains would exhibit similar behavior.40,41
At long
staining-times the amount of aggregate and monomer on TiO2 is approximately equal.
3.3.2 Quantum Efficiency of Photon-to-Current Conversion
Figure 3.4A and B show the wavelength-dependent incident photon-to-current conversion
efficiency (IPCE) and absorbed photon-to-current conversion efficiency (APCE) for Bt/TiO2
based solar cells containing mostly (75%, red curve) monomer and mostly (65%, black curve)
dimer. IPCE is the ratio of the number of electrons collected in the external circuit to the number
of incident photons and is the product of three efficiencies: IPCE = ηLHηinjηcoll, where ηLH, ηinj,
50
and ηcoll are the efficiencies of light-harvesting, electron injection, and electron collection,
respectively. The light harvesting efficiency, also called the absorptance, is a function of
absorbance A: ηLH = 1 – 10−A(λ)
. APCE is obtained from IPCE by dividing by the absorptance at
each wavelength and is thus the product of the efficiencies of electron injection and collection.
The wavelength-dependent ηLH is of course larger for the film containing more aggregates,
owing to higher dye loading. In addition, IPCE(λ) of the aggregate-rich DSSC is broader and
attains a higher maximum value than that of the monomer-rich DSSC. Further, IPCE(λ) of the
monomer-rich DSSC follows the absorbance much more closely than that of the aggregate-rich
DSSC. Comparison of APCE(λ) for the two solar cells, Figure 3.4B, is more revealing. While
APCE(λ) for the monomer-rich solar cell is similar to the corresponding absorption spectrum,
that of the aggregate-rich solar cell has pronounced maxima at 450 and 675 nm. While these are
close to the maxima in the two bands of the aggregate component absorption spectrum, the red
peak in APCE(λ) is almost as intense as the blue peak, whereas in the aggregate absorption
spectrum, the blue band is considerably more intense than the red. This correlates to fast
relaxation from the higher excited state to the lowest-lying excited state according to Kasha’s
rule. The overall higher values of APCE for the aggregate-rich DSSC indicate that improvements
to photoconversion efficiency are not merely the result of better light harvesting. Rather, the
aggregate exhibits higher efficiency for electron injection and collection than the monomer.
Unfortunately, quantification of this phenomena by NMF analysis is not possible since the
IPCE(λ) is not a linear combination of the IPCE(λ) of the individual components. However, the
increased IPCE correlates very well with the amount of aggregation determined by NMF
51
analysis on the absorbance. Since efforts to develop sensitizers with enhanced red absorption are
sometimes thwarted by lower driving force for electron injection,5 it is particularly significant
that the lower energy excited state of the dimer has sufficient reducing power for efficient
electron injection.
Figure 3.4: A) IPCE and B) APCE of Bt-based DSSC consisting of 25% aggregate (red), and 65% aggregate (black).
Figure 3.5: Fluorescence spectra, excited at 514.5 nm and corrected for absorptance at 514.5 nm, of Bt on A) TiO2 and B) ZrO2 as a function of sensitization time.
52
3.3.3 Fluorescence of Betanin on TiO2 and ZrO2
Ordinarily, a rough estimate of the injection efficiency of photoexcited dye can be
determined from the quenching of its fluorescence on TiO2 compared to on ZrO2 where no
injection takes place, owing to the higher absolute energy of the ZrO2 conduction band.52
However, there are complications to this approach for the present system. In aqueous solution,
Bt displays a weak fluorescence spectrum with a maximum at about 640 nm.30,43,53
Wendel et al.
recently reported that the fluorescence quantum yield φfl of Bt in aqueous solution, about 0.0007
in water, increases with solvent viscosity, while the Stokes shift decreases.53
This points to an
excited state conformational change as a significant nonradiative relaxation pathway responsible
for the weak fluorescence. Clearly, electron injection from excited state Bt to TiO2 quenches the
fluorescence, such that the emission of Bt is quite weak on TiO2. Nevertheless, the fluorescence
spectra of sensitized films are readily determined if stray light is carefully rejected. In addition,
independent of fluorescence quenching that results from electron injection to TiO2, there are
potentially changes to the fluorescence spectrum and quantum yield on aggregation. To
investigate these, the fluorescence spectrum of Bt on both TiO2 and ZrO2 was determined as a
function of dye loading as shown in Figure 3.5. As seen there, in both cases higher dye loadings
lead to reduced intensity and increasing red shift in the fluorescence spectrum. Considering the
front-face emission geometry used here and the large Stokes shift (about 3,000 cm-1
in solution),
the inner filter effect is expected to be negligible. Thus the results indicate an increasing Stokes
shift for the aggregate compared to the monomer. This is expected based on Kasha’s rule that
emission takes place from the lowest energy excited state, i.e., the red band of the dimer. In
addition, the decrease in intensity that accompanies increased dye-loading on ZrO2 reveals that
53
the dimer has a lower fluorescence quantum yield than the monomer, since fluorescence
quenching by electron injection is precluded on ZrO2. Aggregation tends to enhance nonradiative
pathways leading to lower fluorescence yields compared to the monomer. Despite this
complication, it is clear from the scale in Figure 3.5A and B that the fluorescence is lower for Bt
on TiO2 compared to ZrO2 for similar sensitization times, despite higher dye-loadings on TiO2.
Though we can’t make a quantitative estimate of the injection efficiency from this data, the
reduced fluorescence on TiO2 compared ZrO2 is evidence of efficient electron injection by both
the monomer and the aggregate.
3.3.4 Power Conversion Efficiency of Dimer-based Solar Cells
Several films were prepared at conditions which produced varying amounts of
aggregation in the IPCE study, but were too thick to allow direct absorbance spectra to be taken,
for efficiency testing with a solar simulator. These films produced efficiencies ranging from 0.5-
Figure 3.6: Photocurrent versus photovoltage for several Bt-based dye-sensitized solar cells.
54
3.0% power conversion efficiency. Shown in Figure 3.6 are the results for several DSSCs
prepared using films in a range of thicknesses from 3.5 to 14 μm. Table 3.1 displays further
information on the percent aggregation, and maximum photocurrent and photovoltage for each
solar cell. By using linear combinations of the monomer and aggregate components, an
approximate aggregate content could be estimated for films that were too dark to effectively
analyze with the NMF technique. The darkest films often showed a distortion in the spectrum
due to detector saturation, and the median point of the saturated peak was taken as the
approximate maximum. All DSSCs were constructed using TiO2 films that had been pre-treated
with acidic ethanol. The highest efficiencies were produced by films prepared under conditions
that produced larger amounts of aggregation as judged by the blue shift in the absorption
maximum. The lowest efficiency, 1.3%, was obtained for the DSSC having an absorption
maximum (536 nm) consistent with the prevalence of monomers. Unfortunately, despite high
current densities, Bt-based DSSCs are limited by low Voc values of about 300-375 mV, which is
due to recombination effects. The low voltage does not appear to be a result of acid pretreatment
of the films as similar voltages were obtained with films that were not treated with acid prior to
sensitization as shown in supporting information Figure 3.15. Despite the low power conversion
efficiency, it is quite remarkable that a DSSC created from a naturally occurring pigment can be
used to moderate effect in solar energy applications. In our lab DSSCs constructed similarly to
the ones tested in this study but using N3 as the sensitizer produced a maximum conversion
efficiency of ~5%. With better understanding of the role of aggregation in enhancing the
efficiency of Bt-based DSSC, we are now able to pursue optimization of efficiency through
improvements to electrolyte composition and surface treatment. Compared to the typical
55
maximum efficiency obtained with N3 of 11%, it is reasonable to postulate that betanin based
solar cells can be significantly improved by engineering modifications to the solar cell assembly.
The total conversion efficiency of 3.0% overtops the maximum obtained for a DSSC based on a
natural dye of 2.7% obtained by our lab in 2011.30
3.3.5 Theory of Transition Dipole-Dipole Coupling in Betanin Dimers
Analysis of the spectra of Bt on TiO2 using NMF reveals a component with two
absorption bands at about 470 and 607 nm (21,300 and 16,500 cm-1
, respectively), with an
intensity ratio of approximately 4:1. The more intense band is shifted 2400 cm-1
to the blue of
the monomer absorption band at 530 nm (18,900 cm-1)
, while the less intense component is red-
shifted from the monomer by 2400 cm-1
. The NMF analysis reveals that these blue- and red-
shifted absorption bands belong to a single species, and given the remarkably similar magnitudes
of the shifts relative to the monomer, it is reasonable to propose that this species is a dimer that
assembles on the TiO2 surface. The coupling of the transition dipole moments μge of two
Table 3.1: Power conversion efficiency, open circuit voltage Voc, short-circuit current Jsc, wavelength of maximum absorption for sensitized film λmax, film thickness, and percent aggregation for each of the solar cells in Figure 3.6.
DSSC Efficiency, % Voc, mV
Jsc, mA/cm2
λmax, nm
Film Thickness, μm
% Aggregation
black 2.5 375 14 501 3.5 50
red 2.4 350 15 523 7 25
green 2.6 350 15 --- 14 75
blue 3.0 375 14 486 7 80
cyan 1.3 350 7 536 3.5 0
56
molecules leads to two new excited electronic states which are shifted from the monomer by an
amount dependent on intermolecular distance, magnitude of the transition dipole and relative
orientations μge on the two molecules. The relative intensities of the blue- and red-shifted dimer
bands depend on the angle between the transition dipoles. As shown next, a reasonable model for
this dimer is one in which the transition dipole moments of the two molecules make an angle of
about 50º.
Excitonic coupling of the excited states of the dimer leads to wavefunctions which are in-
phase and out-of-phase combinations of the basis states )2()1( eg and )1()2( eg , which represent
excitation of either molecule 1 or 2 of the dimer with the other molecule remaining in the ground
state. These two new dimer wavefunctions are:
φ±=
1
√2[φg(1)φ
e(2) ± φ
e(1)φ
g(2)] Equation 3.4
Excitonic coupling of the basis states leads to two new states split by 2V12, where V12 is the
matrix element of the coupling perturbation with respect to the two basis states:
)1()2(ˆ)2()1( egeg H . Using the crude assumption that the coupling results from the
interaction of point dipoles, the coupling strength V12 is expressed in wavenumber units as:
𝑉12 =𝜇𝑔𝑒
2
ℎ𝑐𝑟3[��1 ∙ ��2 − 3(��1 ∙ ��)(��2 ∙ ��)] Equation 3.5
Here, r is the intermolecular distance, r is a unit vector in the direction of r
, μge is the magnitude
of the monomer transition moment, and 1u and 2u are unit vectors in the direction of the
transition dipole moments of molecule 1 and 2 respectively. Since the blue-shifted dimer state
57
corresponds to the more intense band, V12 is positive and the dimer is more H-like than J-like.
The transition dipole moment connecting the ground state )2()1( gg to the two excited states is
𝜇± =1
√2(��𝑔𝑒(1) ± ��𝑔𝑒(2)) Equation 3.6
Here, ge
is the transition dipole moment of the monomer. The energies of the two dimer states
are:
𝐸± = 𝐸𝑚𝑜𝑛 ± 𝑉12 Equation 3.7
where Emon/hc = 18,900 cm-1
is the energy of the monomer excited state. The intensity of each
band of the dimer is proportional to the square of μ+ or μ−, such that the ratio of the band
intensities for the blue- and red-shifted components is
𝐼+
𝐼−=
𝜇+2
𝜇−2 =
1+cos 𝛼
1−cos 𝛼 Equation 3.8
where α is the angle between the transition dipoles on molecules 1 and 2. The results of the NMF
analysis result in a blue band for the dimer which is about four times as intense as the red band,
based on peak height. This leads to an estimated angle α of approximately 53º.
There is no evidence that dimerization takes place in aqueous solution, thus we
hypothesize that the TiO2 surface acts as a template for the assembly. The distance between
bridging oxygens on the ideal (101) surface of anatase TiO2 is on the order of 5 Å, as is the
distance between five-fold coordinated Ti. Therefore, we chose r = 5 Å to estimate the coupling
strength V12. Using the integrated absorption of the solution phase spectrum /)( d to
estimate the transition dipole results in μge = 8.7 Debye. (This estimate is based on a maximum
molar absorptivity εmax of 65,000 L/mol cm.54
) This gives
58
𝜇𝑔𝑒2
ℎ𝑐𝑟3 ≈ 3050 𝑐𝑚−1 Equation 3.9
Thus an angular factor )ˆˆ)(ˆˆ(3ˆˆ2121 ruruuu on the order of 0.8 is consistent with the intensities
and frequencies of the two dimer absorption bands. Further efforts to estimate the orientation of
the dipole moments relative to the intermolecular vector r
are not warranted in this crude model;
however, it is clear that transition dipole coupling provides a reasonable explanation of the
observed spectrum of the dimer.
3.4 Conclusions
This research represents the first evidence of an aggregated sensitizer improving
performance in a DSSC. Based on APCE(λ) data, the aggregate outperformed the monomer by
broadening the absorption spectrum and increasing electron injection and collection over a wider
range of wavelengths than the monomer. In addition to this unprecedented result, we have
achieved a record efficiency for a DSSC based on a natural dye. As reported previously, betanin-
based DSSCs produce large current densities (~15 mA/cm2). Unfortunately, the Voc is limited by
recombination. The large current densities and the broad absorption reinforces the potential of
betanin based DSSCs. Since we have shown that the electron injection of Bt to TiO2 is a two-
electron, one-photon process,42,43
we are presently pursuing alternative proton-coupled redox
mediators, unlike the iodide/triiodide redox couple used here, which may be more efficient at dye
regeneration, inhibiting the recombination that appears to limit Voc. In addition, as shown in Ref.
30, there are efficiency losses over the course of several hours for Bt-based DSSCs, and more
work is needed to improve the stability of the dye.
59
It is interesting to consider the mechanism for improved electron injection and collection
in DSSCs with more aggregated as compared to monomeric Bt. One factor is the strong splitting
that results from coupling of the transition dipole moments, which allows light harvesting over a
broader range of wavelengths. The visible absorption band of Bt in solution has an oscillator
strength near unity, and the fairly large transition dipole moment facilitates excitonic coupling
that broadens the absorption band. It is also reasonable to speculate that, with three carboxylate
functions and two nitrogen heteroatoms, there is ample opportunity for hydrogen bonding to play
a role in the assembly, in addition to templating on the TiO2 surface. It may be significant that
assembly apparently stops at the dimer stage. Larger aggregates present drawbacks mentioned in
the introduction, such as reduced electronic coupling to TiO2 and formation of insulating layers
of dye. Smaller aggregates would not exhibit these detrimental characteristics. Importantly, we
find that it is not mere improvement in light harvesting that makes for better performance of
dimer-based DSSCs. However, we are not able to separate the roles of improved electron
injection and collection as of yet. We have shown that the photo-oxidation of Bt on TiO2 is a
two-electron, one-proton process.42,43
It is conceivable that dimerization enables proton transfer,
enhancing electron injection. Alternatively, there may be more spatial separation of the HOMO
and LUMO in the dimer than in the monomer, inhibiting electron recombination and enhancing
collection.
60
3.5 Supporting Information
Figure 3.7: Scanning electron micrographs of TiO2 films showing 3.5µm thickness (left), 7µm thickness (middle) and the nanoparticles (right).
Figure 3.8: UV-Vis absorption spectra of the sensitized TiO2 film (black) and the sensitizing solution (red) showing the dramatic blue-shift from 537nm to 482nm.
61
Figure 3.9: (A) HPLC chromatogram, monitored at 525nm, of the extracted pigments of films sensitized for 10min (red) and 16hours (green) shown with the sensitizing solution (black) for comparison. (B) Absorption spectra of film extract (black) and sensitizing solution (red).
Figure 3.10: Calibration curve of betanin vs. calculated concentration.
62
Figure 3.11: UV-Vis absorption spectra of films sensitized without phosphate (A, Black) and with 1 mM phosphate (B, red).
Figure 3.12: Unit components from NMF analysis showing the peaks of the aggregate (green), monomer (blue), TiO2, and the sum of all components (red).
63
Figure 3.13: NMF component analysis showing three subsets of the samples used for: fluorescence measurements (A), Beer’s Law analysis of solutions (B), and DSSC characterization (C).
64
Figure 3.14: Linear combination of the aggregate and monomer unit components from the NMF modeling showcasing the representative peak locations for aggregation amounts in 10% intervals.
Figure 3.15: Characteristic current-voltage plots for films with no pretreatment (black), pretreatment with 0.5M hydrochloric acid in ethanol (red), and purified water (blue).
65
3.6 References
1 Croce, R.; van Amerongen, H.; Natural Strategies for Photosynthetic Light Harvesting. Nature
Chem. Bio. 2014, 10, 492-501.
2 Fassioli, F.; Dinshaw, R.; Arpin, P. C.; Scholes, G. D. Photosynthetic Light-Harvesting:
Excitons and Coherence. J. Royal Soc. Interface 2014, 11, 20130901/1-22.
3 O’Regan, B.; Grätzel, M. A low-Cost, High-Efficiency Solar Cell Based on Dye-Sensitized
Colloidal Titanium Dioxide Films. Nature 1991, 353, 737-40.
4 Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-Sensitized Solar Cells.
Chem. Rev. 2010, 110, 6595-6663.
5 Hamann, T.; Jensen, R. A.; Martinson, A. B. F.; Ryswyk, H. V.; Hupp, J. T. Advancing
Beyond Current Generation Dye-Sensitized Solar Cells. Energy & Environ. Science 2008, 1,
66-78.
6 Kamat, P. V. Photochemistry on Nonreactive and Reactive (Semiconductor) Surfaces. Chem.
Rev. 1993, 93, 267-300.
7 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.
8 Wenger, B.; Grӓtzel, M.; Moser, J.-E. Rational for Kinetic Heterogeneity of Ultrafast Light-
Induced Electron Transfer from Ru(II) Complex Sensitizers to Nanocrystalline TiO2. J. Am.
Chem. Soc. 2005, 127, 12150-12151.
9 Planells, M.; Forneli, A.; Martinez-Ferrero, E.; Sanchez-Diaz, A.; Sarmentero, M. A.;
Ballester, P.; Palomeres, E.; O’Regan, B. C. The Effect of Molecular Aggregates over the
Interfacial Charge Transfer Process on Dye Sensitized Solar Cells. Appl. Phys. Lett. 2008,
92, 13506/1-3.
10 Tatay, S.; Haque, S. A.; O’Regan, B.; Durrant, J. R.; Verhees, W.J. H.; Kroon, J. M.; Vidal-
Ferran, A.; Gavino, P.; Palomeres, E. J. Kinetic Competition in Liquid Electrolyte and Solid
State Dye-Sensitized Solar Cells. Mater. Chem. 2007, 17, 3037-3044.
11 Teuscher, J.; Decoppet, J.-D.; Punzi, A.; Zakeeruddin, S. M.; Moser, J.-E; Grӓtzel, M.
Photoinduced Interfacial Electron Injection in Dye-Sensitized Solar Cells under Photovoltaic
Operating Conditions. J. Phys. Chem. Lett. 2012, 3, 3786-3790.
66
12 Koops, S. E.; Barnes, P. R. F.; O’Regan, B. C.; Durrant, J. R. Kinetic Competition in a
Coumarin Dye-Sensitized Solar Cell: Injection and Recombination Limitations upon Device
Performance. J. Phys. Chem. C 2010, 114, 8054-8061.
13 Salvatori, D.; Marotta, G.; Cinti, A.; Mosconi, E.; De Angelis, F. Supramolecular
Interactions of Chenodeoxycholic Acid Increase the Efficiency of Dye-Sensitized Solar
Cells. J. Phys. Chem. C 2013, 117, 3874-3887.
14 Hu, K.; Robson, K. C. D.; Berlinguette, C. P. Meyer, G. J. Donor-π-Acceptor Organic
Hybrid TiO2 Interactions for Solar Energy Conversion. Thin Solid Films 2014, 560, 49-54.
15 Burke, A.; Schmidt-Mende, L.; Ito, S.; Grӓtzel, M. A Novel Blue Dye for Near-IR ‘Dye-
Sensitized’ Solar Applications. Chem. Comm. 2007 234-236.
16 Wang, Z. S.; Cui, Y. Dan-oh, Y.; Kasada, C.; Shinpo, A.; Hara, K. Thiophene-Functionalized
Coumarin Dye for Efficient Dye-Sensitized Solar Cells. Electron Lifetime Improved by Co-
Adsorption of Deoxycholic Acid. J. Phys. Chem. C 2007, 111, 7224-7230.
17 Cai, M.; Pan, X.; Lui, W.; Sheng, J.; Fang, X.; Zhang, C.; Huo, Z.; Tian, H.; Xiao, S.; Dai, S.
Multiple Absorption of Tributyl Phosphate Molecule at the Dyed-TiO2/Electrolyte Interface
to Suppress the Charge Recombination in Dye-Sensitized Solar Cell. J. Mater. Chem. A,
2013, 1, 4885-4892.
18 He, J.; Benkö, G.; Korodi, F.; Polívka, T.; Lomoth, R.; Ǻkermark, B.; Sun, L.; Hagfeldt, A.;
Sundström, V. Modified Phthalocyanines for Efficient Near-IR Sensitization of
Nanostructured TiO2 Electrodes. J. Amer. Chem. Soc. 2002, 124, 4922-4932.
19 Yang, L.-N.; Li, S.-C.; Li, Z.-S.; Li, Q.-S. Molecular Engineering of Quinoxaline Dyes
Toward More Efficient Sensitizers for Dye-Sensitized Solar Cells. RSC Advances 2015, 5,
25079-25088.
20 Hua, Y.; Lin Lee, L. T.; Zhao, J.; Chen, T.; Wong, W.-Y.; Wong, W.-K.; Zhu, X. Co-
sensitization of 3D Bulky Phenothiazine-Cored Photosensitizers with Planar Squaraine Dyes
for Efficient Solar Cells. J. Mater. Chem. A 2015, 3, 13848-13855.
21 Son, H.-J.; Kim. C. H.; Kim, D. W.; Jeong, N. C.; Prasittichai, C.; Luo, L.; Wu, J. T.; Farha,
O. K.; Wasielewski, M. R.; Hupp, J. T. Post-assembly Atomic Layer Deposition of Ultrathin
Metal-Oxide Coatings Enhances the Performance of an Organized Dye-Sensitized Solar Cell
by Suppressing Aggregation. ACS Appl. Mater. Interfaces 2015, 1, 5150-5159.
67
22 Chappaz-Gillot, C.; Marek, P. L.; Blaive, B. J.; Canard, G.; Bürke, J.; Garab, G.; Hahn, H.;
Jávorfi, T.; Kelemen, L.; Krupke, R.; et al. Anisotropic Organization and Microscopic
Manipulation of Self-Assembling Synthetic Porphyrin Microrods that Mimic Chlorosomal
Bacterial Light-Harvesting Systems. J. Amer. Chem. Soc. 2012, 134, 944-954.
23 Huijser, A.; Marek, P. L.; Savenije, T. J.; Siebbeles, L. D. A.; Scherer, T.; Hauschild, R.;
Szmytkowski, J.; Kalt, H.; Hahn, H.; Balaban, T. S. Photosensitization of TiO2 and SnO2 by
Artificial Self-Assembling Mimics of the Natural Chlorosomal Bacteriochlorophylls. J. Phys.
Chem. C 2007, 111, 11726-11733.
24 Yang, Y.; Jankowiak, R.; Lin, C.; Pawlak, K.; Reus, M.; Holzwarth, A. R.; Li, J. Effect of
the LHCII Pigment-Protien Complex Aggregation on Photovoltaic Properties of Sensitized
TiO2 Solar Cells, Phys. Chem. Chem. Phys. 2014, 16, 20856-20865.
25 Wang, X.-F.; Kitao, O.; Zhou, H.; Tamiaki, H.; Sasaki, S.-I. Efficient Dye-Sensitized Solar
Cells Based on Oxy-Bacteriochlorin Sensitizer with Broadband Absorption. J. Phys. Chem.
C 2009, 113, 7954-7961.
26 Marek, P. L. Hahn, H.; Balaban, T. S. On the Way to Biomimetic Dye Aggregate Solar
Cells. Energy Environ. Science 2011, 4, 2366-2378.
27 Tsui, L.-K.; Huang, J.; Sabat, M.; Zangari, G. Visible Light Sensitization of TiO2 Nanotubes
by Bacteriochlorophyll C Dyes for Photoelectrochemical Solar Cells. ACS Sustainable Chem.
Eng. 2014, 2, 2097-2101.
28 Su, Y.-H.; Teoh, L. G.; Lee, J.-H.; Tu, S.-L.; Han, M. H. Photoelectric Characteristics of
Natural Pigments Self-Assembly Fabricated on TiO2/FTO Substrate. Nanoscience and
Nanotechnology 2009, 9, 960-964.
29 Calogero, G., di Marco, G., Caramori, S., Cazzanti, S., Argazzi, R. and Bignozzi, C. A.
Natural Dye-Sensitizers for Photoelectrochemical Cell. Energy Environ. Science 2009, 2,
1162-1172.
30 Sandquist, C.; McHale, J. L. Improved Efficiency of Betanin-Based Dye-Sensitized Solar
Cells,” J. Photochem. Photobiol. A 2011, 221, 90-97.
31 Gandía-Herrero, F.; Escribano, J.; García-Carmona, F. Structural Implication on Color,
Fluorescence, and Antiradical Scavenging Activity in Betalains, Planta 2010, 232, 449-460.
32 Khan, M. I.; Giridhar, P. Plant Betalains: Chemistry and Biochemistry. Phytochemistry 2015,
117, 267-295.
68
33 Mabry, T.; Taylor, A.; Turner, B. The Betacyanins and their Distribution.
Phytochemistry 1963, 2, 61–64.
34 Stintzing, F. C.; Schieber, A.; Carle, R. Identification of Betalains from Yellow Beet (Beta
vulgaris L.) and Cactus Pear [ Opuntia ficus-indica (L.) Mill.] by High-Performance Liquid
Chromatography−Electrospray Ionization Mass Spectrometry. J. Agric. Food Chem. 2002,
2302–2307
35 Zhang, D.; Lanier, S.; Downing, J. A.; Avent, J.; Lum, J.; McHale, J. L. Betalain Pigments
for Dye-Sensitized Solar Cells. J. Photochem. Photobio. A 2008, 195, 72-80.
36 McHale, J. L. Light-Harvesting Chromophore Aggregates and Their Potential for Solar
Energy Conversion. J. Phys. Chem. Lett. 2012, 3, 587-597.
37 Calogero, G.; Yum, J.-H.; Sinopoli, A.; Di Marco, G.; Grätzel, M.; Nazeeruddin, M. K.
Anthocyanins and Betalains as Light-Harvesting Pigments for Dye-Sensitized Solar Cells,”
Solar Energy 2012, 86, 1562-1575.
38 Calogero, G., di Marco, G., Caramori, S., Cazzanti, S., Argazzi, R. and Bignozzi, C. A.
Natural Dye-Sensitizers for Photoelectrochemical Cells,” Energy Environ. Science 2009, 2,
1162-1172.
39 Hernandez-Martinez, A. R.; Estevez, M.; Vargas, S.; Rodriguez, R. Stabilized Conversion
Efficiency and Dye-Sensitized Solar Cells from Beta vulgaris Pigment. Int. J. Mol. Sci. 2013,
14, 4081-4093.
40 Ellestad, G. A. Structure and Chiroptical Properties of Supramolecular Flower Pigments.
Chirality 2006, 18, 134-144.
41 Fernandes, A.; Bras, N. F.; Mateus, N.; de Freitas, V. A Study of Anthocyanin Self-
Association by NMR Spectra. New J. Chem. 2015, 39, 2602-2611.
42 Knorr, F. J.; McHale, J. L.; Clark, A. E.; Marchioro, A.; Moser, J.-E. Dynamics of Interfacial
Electron Transfer from Betanin to Nanocrystalline TiO2: The Pursuit of Two-Electron
Injection. J. Phys. Chem. C 2015, 119, 19030-19041.
43 Knorr, F. J.; Malamen, D. J.;McHale, J. L.; Marchioro, A.; Moser, J.-E. Two-Electron Photo-
Oxidation of Betanin on Titanium Dioxide and Potential for Improved Dye-Sensitized Solar
Energy Conversion. Proc. SPIE 9165, Physical Chemistry of Interfaces and Nanomaterials
XIII, 2104, 9165-0N.
69
44 Nemzer, B.; Pietrzkowski, Z.; Spórna, A.; Stalica, P.; Thresher, W.; Michałowski, T.;
Wybraniec, S. Betalainic and Nutritional Profiles of Pigment-Enriched Red Beet Root (Beta
vulgaris L.) Dried Extracts. Food Chemistry 2011, 42–53.
45 Herbach, K. M.; Stintzing, F. C.; Carle, R. Identification of Heat-Induced Degradation
Products from Purified Betanin, Phyllocactin and Hylocerenin by High-Performance Liquid
Chromatography/Electrospray Ionization Mass Spectrometry. Rapid Commun. Mass
Spectrom. 2006, 1822–1822.
46 Cai, Y.; Sun, M.; Corke, H. HPLC Characterization of Betalains from Plants in the
Amaranthaceae. Journal of Chromatographic Science 2005, 454–460
47 Wybraniec, S.; Mizrahi, Y. Influence of Perfluorinated Carboxylic Acids on Ion-Pair
Reversed-Phase High-Performance Liquid Chromatographic Separation of Betacyanins and
17-Decarboxy-betacyanins. Journal of Chromatography A 2004, 97–101.
48 Moss, R.; Pérez-Roa, R. E.; Anderson, M. A. Electrochemical Response of Titania, Zirconia,
and Alumina Electrodes to Phosphate Adsorption. Electrochimica Acta, 2013, 104, 314-321.
49 Lee, D. D.; Seung, H. S. Learning the Parts of Objects by Non-Negative Matrix
Factorization. Nature 1999, 401, 788-791.
50 Wybraniec, S.; Starzak, K.; Skopińska, A.; Nemzer, B.; Pietrzkowski, Z.; Michałowski, T.
Studies on Nonenzymatic Oxidation Mechanisms in Neobetanin, Betanin, and
Decarboxylated betanins. J. Agric. Food Chem. 2013, 61, 6465-6476.
51 Rex, R. E.; Knorr, F. J.; McHale, J. L. Surface Trap States of TiO2 Nanosheets and
Nanoparticles as Illuminated by Spectroelectrochemical Photoluminescence. J. Phys. Chem.
C 2014, 118, 16831-16841.
52 Tachibana, Y.; Moser, J. E; Grätzel, M. Klug, D. R.; Durrant, J. R. Subpicosecond Interfacial
Charge Separation in Dye-Sensitized Nanocrystalline Titanium Dioxide Films. J. Phys.
Chem. 1996, 100, 20056-20062.
53 Wendel, M.; Nizinski, S.; Tuwalska, D.; Starzak, K.; Szot, D.; Prukula D.; Marek, S.;
Wybraniec, S.; Burdzinski, G. Time-Resolved Spectroscopy of the Singlet Excited State of
Betanin in Aqueous and Alcoholic Solutions. Phys. Chem. Chem. Phys. 2015, 17, 18152-
18158.
70
54 Schwartz, S. J.; Von Elbe, J. H. Quantitative Determination of Individual Betacyanin
Pigments by High-Performance Liquid Chromatography, J. Agric. Food Chem. 1980, 28,
540-543.
71
4 Tuning the Aggregation of Betanin on TiO2 Using Non-Aqueous
Solvents and Phosphate Additives
4.1 Introduction
In previous research we studied the aggregation of betanin on TiO2 and a speculative
structure of that aggregate is shown in Figure 4.2. Higher amounts of aggregation of betanin on
TiO2 have been shown to increase the efficiency of DSSCs, but making a control film with no
aggregation has proven difficult using our standard sensitizing solutions. Phosphates have
commonly been used in other DSSC systems to inhibit aggregation and are a good starting point
to create films of pure monomer character. 1
Of other interest is sensitization from non-aqueous
solvents as some research in our lab has shown that the VOC may be lowered by water as a result
of enhanced recombination.2
Our initial hypothesis for aggregation is simply that betanin occupies adjacent binding
sites on the TiO2 which are spaced approximately 5 Å apart. As discussed in chapter 3, this
approximation is theoretically consistent with the splitting observed in the absorption band of
betanin aggregates. Based on this theory, we approximate an angle between the transition dipole
moments of 50 degrees. Since the binding orientation of betanin on TiO2 is still highly
speculative, we approximated a structure of the betanin occupying two adjacent binding sites,
and found that a possible orientation included one betanin binding at the carboxylic acid group at
C2 and C17 sites while the other was bound through those at C17 and C15. While this
orientation is highly speculative, it does take into consideration the significant steric bulk of the
72
glucose group and still satisfies the requirements of the theory. Aside from removal of the C2
acid group, the only test to this hypothesis is to control the protonation state of the acid.1-4
Based on our observed optimal sensitization at pH of 2.0, and the pKa of the three
carboxylic acid groups, 3.5 for C15 and C17 and <2 (~1.5) for C2, it is reasonable to assume that
the C15 and C17 acids are mostly protonated while the C2 acid is mostly unprotonated. As
discussed earlier, these conditions provide the highest efficiency of betanin adsorption and
aggregation, and appear to be crucial to proper sensitization. Full protonation of the C2 acid is
not feasible under operational conditions since the betanin degrades at pH values below 2.0.
However, research has shown that methanol can increase the degree of protonation of organic
acids by changing the proton affinity of the solution. For example, one study reported a common
Figure 4.1: The molecular structure of betanin showing the atomic labeling of the various functional groups.
73
carboxylic acid group with a pKa of 2.0 would have an effective pKa of 5.0 in methanolic
solution.3,4
While non-aqueous solvents may have additional effects on the binding of betanin to
TiO2, the impact of the protonation state of C2 carboxylic acid can be probed by using a panel of
organic solvents. Ethanol has been shown to have the opposite effect, though to a lesser
magnitude, than methanol on the acidity of carboxylic acids in solution, and MeCN should
provide a useful control as it should have no effect on the available protons in solution. If
variations in the aggregation behavior are observed over the range of different solvents, the C2
carboxylic acid may be implicated in the formation of aggregates, and further (more
complicated) studies can be done to elucidate its influence.5,6
Figure 4.2: Speculative structure of betanin dimer on TiO2 (101) surface. Reprinted with permission from Treat, N. A.; Knorr, F. J.; McHale, J. L., Templated Assembly of Betanin Chromophore on TiO2: Aggregation-Enhanced Light-Harvesting and Efficient Electron Injection in a Natural Dye-Sensitized Solar Cell. J. Phys. Chem. C, 2016, 120 (17), 9122-9131. Copyright 2016 American Chemical Society.
74
4.2 Experimental Methods
4.2.1 Film Preparation
Films of TiO2 were prepared via the method described previously with the addition of a
1:3 dilution of the paste purchased from Solaronix. While these films were not characterized by
SEM, as in the past work, this dilution should produce films approximately 2-2.5 µm in
thickness.1
4.2.2 Film Sensitization
The TiO2 films were pretreated with 0.5 M HCl in ethanol, as described previously, prior
to sensitization. Films were immersed in solutions containing 0.2 mM HCl in the respective
solvent or buffer with either saturated betanin, 5 mM or 1 mM betanin. Phosphate concentrations
were carefully controlled by adding small amounts of a 100 mM sodium phosphate buffer
titrated to a starting pH of 2.0. Solutions for studying various methanol concentrations were
prepared by adding 1 mL of 10 mM aqueous betanin stock and the prescribed amount of
methanol and then the solutions were diluted to a final volume of 10.0 mL.
4.2.3 Absorbance Spectroscopy
Films were removed from the sensitizing solutions and immediately rinsed with 100%
ethanol to remove physisorbed betanin and dried with nitrogen. The films were kept in the dark
and immediately transferred to the UV-Vis spectrometer. For sequential sensitization
experiments, the films were returned to the sensitizing solutions immediately after the spectra
were taken.1
4.2.4 Solar Cell Measurements
75
Solar cell measurements and IPCE were taken as discussed previously with the
modification that the stepping voltage was reduced to 10 mV over the range from 0-500 mV.
Total energy conversion efficiency was calculated via the reported method based on 13 mW
illumination intensity.1
4.3 Results and Discussion
4.3.1 Phosphate
In Chapter 2 we reference the use of phosphate to produce films of high monomeric
character and discuss the many reasons why phosphates are an ideal candidate for inhibiting
aggregation of DSSC sensitizers. Phosphate binds very strongly with TiO2 and would compete
with betanin for binding sites on TiO2 therefore reducing the likelihood of adjacent sites being
occupied by dye. The unfortunate side effect of using phosphate to reduce aggregation is that it
also reduces total dye loading.
By adding various amounts of phosphates to the sensitizing solution, betanin aggregation
could be slowed, but could not be completely inhibited. When sensitized overnight (~21 hours)
films produced the same degree of blue-shift and only the dye loading was different. For high
phosphate concentrations, the dye was very limited in adsorption, but at 0.1 mM phosphate, TiO2
films exhibited almost identical dye loading with the film produced in the absence of phosphate
for the same sensitization time as shown in Figure 4.3. Further investigation showed that the
degree of aggregation is only limited by phosphate when it is in comparable concentrations with
that of betanin. A study in which two solutions were prepared of high (5 mM, Film5) and low (1
mM, Film6) betanin concentration both with 1 mM phosphate showed that the 5 mM betanin
76
solution aggregated within 5 minutes of staining time. The 1 mM betanin solution, however,
remained monomeric at 5 minutes based on an absorption maximum near 530 nm, but had
completely aggregated after 60 minutes as shown in Figure 4.4. This dependence on betanin
concentration was further confirmed by two additional films sensitized for 5 minutes in the same
solutions. For comparable concentrations, phosphate (inorganic) delays aggregation of betanin in
aqueous solution, but does not inhibit aggregation completely. For solutions with higher betanin
than phosphate concentration, very minimal effects on aggregation were observed and at higher
phosphate concentrations, betanin binding was greatly reduced.
Figure 4.3: Absorbance spectra of betanin films sensitized overnight in with various amounts of phosphate, 1.0 mM (black), 0.1 mM (blue), 10 mM (red) and 0.0 mM (pink), and 1.0 mM betanin all exhibit blue-shifted absorbance maxima around 485 nm.
77
Based on these observations, phosphate is not an ideal choice for developing a model of
the monomeric behavior of betanin in DSSCs, because high dye loading cannot be achieved
without aggregation. A few very interesting conclusions can be drawn from these data though.
First, the binding affinity of betanin to TiO2 is similar but slightly higher than that of inorganic
phosphate. Since initially the phosphate appears to effectively eliminate dye aggregation the two
must be energetically similar, but the tendency towards aggregation at longer times indicates that
the phosphate’s binding energy is insufficient to outcompete betanin. Secondly, it appears that
betanin aggregation occurs as a two-step process by first depositing a monolayer of dye and then
the extra dye aggregates with the dye that is initially attached to the surface. This two-step
behavior is seen in the time-lapse absorbance spectra taken of the TiO2 films sensitized from
solutions with various amounts of phosphate shown in Figure 4.5. Figure 4.6 shows the step-wise
Figure 4.4: Absorption spectra of films sensitized with 5 mM betanin (black and blue) and 1 mM betanin (red and magenta) in a buffer of 1 mM phosphate.
78
aggregation of betanin in more detail. Monomer character dominates these absorbance spectra
until about 5 minutes of sensitization time when it begins to shift toward aggregate. These
spectra show the step-wise behavior that was observed with the films shown in Figure 4.4. This
behavior was also observed, but to a lesser extent, in the non-aqueous solutions discussed below.
Figure 4.5: Sequential absorption spectra of betanin films sensitized in 10 mM (a), 1 mM (b), 0.1 mM (c) and 0.0 mM (d) phosphate taken at 1 minute intervals for up to 20 minutes, except (a) which were taken at 10 minute intervals for 90 minutes.
79
4.3.2 Methanol
Films sensitized with betanin in pure methanolic solution produced monomeric behavior,
even at long staining times as shown in Figure 4.7. While it is unclear whether the inhibition of
aggregation from methanolic solution is a result of the changing acidity of the sample, it is clear
that the methanol does not inhibit binding in general as can be seen by comparing the magnitudes
of the absorbances of films sensitized from methanol and water.
Figure 4.6: Absorbance spectra of TiO2 films sensitized with 1 mM betanin in 1 mM sodium phosphate at pH 3 taken sequentially at regular intervals.
80
Interestingly, this inhibition of aggregation does not require the solution to be pure
methanol, as little as 4% methanol is sufficient to reduce aggregation significantly as can be seen
in Figure 4.8. Increasing the concentration of methanol in the solution similarly inhibits
aggregation with 96% MeOH producing films with an absorption maximum of 527 nm, and
therefore aggregation is inhibited at all MeOH concentrations. Interestingly, for the same
sensitization time, it was observed that increasing the concentration of methanol increased the
dye-loading. It is unclear what caused this increase in dye loading, as shown in Figure 4.8, but it
is reasonable to assume that betanin is less favorably solvated in the high methanol
concentrations, and therefore partitions onto the surface of the TiO2. This effect could, however,
be dependent on the acidity of the solutions, because compared to aqueous sensitization at
optimal pH, there appears to be little difference in dye loading between the two solvents.
Figure 4.7: Absorbance spectra of betanin films sensitized for 5 minutes (black) and 10 minutes (red) from methanol (left) and water (right) with otherwise identical conditions.
81
Over longer staining times, up to 40 minutes, the slight shift of the absorbance maximum
from 530 nm to 520 nm can be observed to occur in a delayed fashion as was observed with
phosphate additives. Figure 4.9 shows that for the first ten minutes, the absorbance of betanin on
TiO2 from MeOH solutions remains in the 530-540 nm range. After 10 minutes, though, the
maximum shifts very slowly to about 525 nm. While this minor blue-shift is not as large as that
of the aggregates observed from aqueous solutions which absorb at 480 nm, it might indicate a
small amount of aggregate is forming. It is more likely, however, that the sequential staining of
the film results in small amounts of degradation upon exposure to air. The films themselves were
kept in the dark when removed from solution, but the presence of a slight shoulder in the 450 to
475 nm range is likely this degradation product. We reported previously that betanin degrades in
Figure 4.8: Absorption spectra of films sensitized with betanin solutions containing 4% (black), 52% (red) and 96% MeOH (green) for 10 minutes and 4% MeOH for 30 min (blue).
82
the presence of light and oxygen to form a product that absorbs in this range and this is the most
plausible explanation.
Results that are consistent with those of the methanolic solutions are produced when
ethanol or acetonitrile is used. MeCN does seem to promote degradation of the betanin more so
than ethanol or methanol, but still does not promote aggregation of the dye molecules. It would
appear that the initial assumption of the acidity modulation in organic solutions does not
adequately explain the lack of aggregation in non-aqueous solvents. Therefore, it can be
postulated that water plays a crucial role in the structure of the aggregates on TiO2 and while the
studies did not provide the intended insight into the involvement of C2 carboxylic acid, they
enabled monomeric films of betanin to be produced in the absence of other inhibiting factors
such as phosphate.
Figure 4.9: Absorbance spectra of TiO2 films sensitized with betanin in 100% methanol sequentially from 1 to 40 minutes, with spectra taken at regular intervals.
83
4.3.3 Performance of Cells
Devices produced from betanin under the conditions with methanol and phosphate
described above were tested and unfortunately showed reduced performance when compared to
those sensitized under normal aqueous conditions as discussed in Chapter 3. As shown in Figure
4.10, the IPCE of betanin solar cells decreased with the amount of phosphate present in the
solution. The broadened and blue-shifted IPCEs correlate very nicely with the absorbance
spectra of the films which are shown in Figure 4.10 and also resemble the aggregate films shown
previously. The reduced IPCE is easily explained by the lowering of dye loading in the presence
of phosphate.
Figure 4.10: IPCE spectra of betanin solar cells produced with various amounts of phosphate in the sensitizing solutions.
84
For the MeOH based films, the difference in device performance is most notable in the J-
V curve. The JSC is reduced to about half of that of films sensitized from aqueous solution with
high aggregate concentrations. Despite the belief that sensitization from water leaves residual
water on the TiO2 surface and is the source of the low VOC, sensitizing from MeOH did not
provide any improvement. Therefore, it is most likely the recombination, caused by the nature of
the dye and device engineering, which limits the VOC and MeOH does not inhibit this effect.
Figure 4.11: J-V curves of solar cells produced from betanin in methanol (black) and water (red).
85
4.4 Conclusions
The mechanism of inhibiting dye aggregation in MeOH is not clear, however, research
has shown that methanol binds very strongly with the surface of TiO2, and therefore could
compete for binding sites as was expected with the phosphate additives.7 Similarly, the binding
affinities of ethanol and MeCN must be higher than that of betanin to produce the same
monomeric behavior. It is interesting to note that organic solvents have higher binding affinities
with TiO2 than phosphate, and provide more efficient inhibition of betanin aggregation. In other
DSSC applications, organophosphates are used to this effect, and perhaps would provide a better
method for inhibiting aggregation in aqueous solution.
Despite our high hopes of increasing the VOC by sensitizing from non-aqueous solution in
order to reduce water-mediated recombination, the recombination effect on VOC appears to be
independent of sensitization solvent. This recombination is likely due to the properties of the dye
rather than the sensitization conditions as it was observed in all the solvents tested. While a VOC
of 375 mV produced a cell with a conversion efficiency of 3.0% a much higher voltage would be
required to achieve feasible device implementation. The current reduction from the methanol
sensitized cells is consistent with the observation of monomeric films producing lower currents
as shown with phosphate, in Chapter 3, and amaranthin to be shown in Chapter 5.
86
4.5 References
1 Treat, N. A.; Knorr, F. J.; McHale, J. L., Templated Assembly of Betanin Chromophore on
TiO2: Aggregation-Enhanced Light-Harvesting and Efficient Electron Injection in a Natural
Dye-Sensitized Solar Cell. J. Phys. Chem. C 2016, 120 (17), 9122-9131.
2 Rex, R. E.; Knorr, F. J.; McHale, J. L. Surface Trap States of TiO2 Nanosheets and
Nanoparticles as Illuminated by Spectroelectrochemical Photoluminescence. J. Phys. Chem.
C 2014, 118, 16831-16841.
3 Mabry, T.; Taylor, A.; Turner, B. The Betacyanins and their Distribution.
Phytochemistry 1963, 2, 61–64.
4 Khan, M. I.; Giridhar, P. Plant Betalains: Chemistry and Biochemistry. Phytochemistry 2015,
117, 267-295.
5 Rived, F.; Canals, I.; Bosch, E.; Rosé, M. Acidity in Methanol-Water. Analytical Chimica
Acta 2001, 439, 315-333.
6 Rived, F.; Rosé, M.; Bosch, E. Dissociation Constants of Neutral and Charged Acids in
Methyl Alcohol. The Acid Strength Resolution. Analytical Chimica Acta, 1998, 374, 309-
324.
7 Cai, M.; Pan, X.; Lui, W.; Sheng, J.; Fang, X.; Zhang, C.; Huo, Z.; Tian, H.; Xiao, S.; Dai, S.
Multiple Absorption of Tributyl Phosphate Molecule at the Dyed-TiO2/Electrolyte Interface
to Suppress the Charge Recombination in Dye-Sensitized Solar Cell. J. Mater. Chem. A
2013, 1, 4885-4892.
87
5 An Amaranthin Based Solar Cell: Comparison of Efficiency and
Properties to Betanin Based Dye Sensitized Solar Cells
5.1 Introduction
Photosynthetic organisms can serve as an inspiring role-model when developing solar
energy conversion technology. Solar energy conversion has the potential to supplant a large
portion of the world’s reliance on fossil fuels for energy production, but has yet to harness this
capability. By studying natural light harvesting systems, scientists can unlock the remaining
potential of solar energy. Ongoing research has taken this nature-inspired approach in the form
of dye-sensitized solar cells (DSSCs). While the electrochemical principles remain similar to
those of photosynthesis, DSSCs rely on dyes adsorbed to a semiconductor surface and the dye
performs the light-harvesting and charge separation processes simultaneously. In contrast, in
photosynthesis, different molecules perform the light harvesting and charge separation steps.
While DSSC technology has been around for quite some time (1991), it has yet to harness the
potential of natural dyes with a large degree of success. However, our group recently showed
that use of betanin, extracted from red beets (Beta vulgaris L.) resulted in a solar cell with an
efficiency of 3%, rivaling the efficiency of N3 (5%, unpublished) obtained in our lab. Further,
this work also showed that the natural pigment underwent templated aggregation on the TiO2
surface, which resulted in an increase in efficiency. In DSSCs, an increase in efficiency upon dye
88
aggregation was entirely unprecedented, but gives crucial insight into the mechanisms of light
harvesting in the natural sources.12
Other natural pigments used in DSSCs have failed to produce high conversion
efficiencies, and it is of great interest to understand what separates betanin from these other
pigments. The anthocyanins, a different form of plant pigment found in all other plants other
than those of the order Caryophyllales, have been studied extensively elsewhere. Other betalain
pigments, however, have not been studied in DSSC applications. Amaranthin, extracted from
Amaranthus cruentus, is an ideal candidate for comparison studies with betanin. Amaranthin is
readily available in the inflorescences, leaves, and stems of A. cruentus up to 200mg/100g fresh
weight. The differences in structure between betanin and amaranthin are very simple and do not
involve the base aglycone structure, which likely means little effect to the chromophore.
Amaranthin possesses an additional sugar group which increases the steric bulk of the molecule
and the hydrophilicity to a small degree.3,4
Figure 5.1: Molecular structures for betanin (left), amaranthin (right) and their aglycone betanindin (center).
89
The main interest in studying the differences in DSSCs with betanin vs. with amaranthin
is to probe the nature of the dye interaction with the TiO2 surface. In this work we use absorption
spectroscopy, IPCE, Raman scattering, and fluorescence to understand how the structure of
betalains impacts the performance of DSSCs. We report a decrease in dye-loading efficiency,
aggregation and device performance in amaranthin based DSSCs compared to betanin cells.
Since betanin aggregates so readily on TiO2, production of monomeric films has required the use
of competitive binders, such as phosphate, which could alter the performance of the device.
However, the steric bulk of the amaranthin molecule helps to inhibit aggregation, without the
addition of phosphate, and therefore provides a good model for the performance of monomeric
betanin in DSSCs.1
5.2 Experimental Methods
5.2.1 Preparation of Amaranthin Samples
Amaranth plant matter was homogenized in a blender with 80% aqueous methanol and
filtered through a stainless steel mesh. The remaining plant matter was soaked for 30 minutes in
a small amount of 80% methanol then passed through a macerating juicer to remove all
remaining pigment. Several passes were made until the remaining pulp contained no more red
pigment. The resulting solution was centrifuged at 10,000 rpm for 20 min. Clarified juice was
then filtered through a series of filters with decreasing porosity: 27 µm (Whatman #4), 11 µm
(Whatman #1), 2.5 µm (Whatman #42). Filtered juice was stored in a freezer at -20˚C overnight
to allow protein precipitation then filtered through glass microfiber filters of 2.7 µm (Whatman
GF/D) and 1.6 µm (Whatman GF/A) pore size then through micro filters of 0.4 µm and then 0.2
90
µm pore size. Amaranthin was then purified from the micro filtered juice via the methods
mentioned previously in Chapter 3.
5.2.2 Film Preparation
TiO2 films were prepared via the method discussed in Chapter 3.
5.2.3 Sensitization of TiO2 films with Amaranthin
Films of TiO2 were sensitized with 1 mM amaranthin in 0.2 mM HCl in the respective
solvents. For studies with MeCN, the solutions were slightly less concentrated due to saturation
of the amaranthin in solution and the presence of small amounts of residual solid.
5.2.4 Absorbance Spectroscopy
Absorption spectra of amaranthin films were obtained via transmission spectroscopy on a
UV-Vis spectrometer. For sequential sensitization, the films were returned to the solution
immediately after spectra were taken. Films were kept in the dark until the respective
measurements were taken.
5.2.5 Fluorescence Quantum Yield Measurements
Solution phase fluorescence quantum yield measurements were taken with a 45˚ back
scattering arrangement with 0.1 mm path length demountable quartz cuvettes and 514.5 nm laser
excitation modulated to 20 µW with a neutral density filter. Solutions of betanin and amaranthin
were prepared in 100 mM pH 3.5 sodium phosphate buffers and diluted to an absorbance of less
than 0.1. Two independent fluorescence standards were used, DCM and LDS 751 in MeOH, with
standard fluorescence quantum yields of 0.453 and 0.01 respectively.5,67
The quantum yield
values obtained using LDS 751 as a reference were determined to be more accurate based on the
91
ability to open the entrance slit wider, 5 µm, than the 1 µm required for the highly fluorescent
DCM.
5.2.6 Preparation of Colloidal Silver Nanoparticles
Silver nanoparticles were prepared via the method described in reference 8 and were used
without further purification.
5.2.7 Surface Enhanced Raman Scattering
Samples of betanin and amaranthin were added to a solution of colloidal silver
nanoparticles and titrated to a pH of 1.9 using concentrated HCl. The particles and dye were
allowed to settle, then were analyzed using laser excitation at 454 nm in a confocal back
scattering arrangement. Absolute wavenumber assignments were made by comparing to the
known vibrational frequencies of cyclohexane.
5.3 Results and Discussion
5.3.1 Comparison of Aggregation Behavior Based on Absorption Spectra
Compared to the betanin films, amaranthin films do not exhibit a blue-shift in absorbance
on adsorption on TiO2 and the films retain the absorption maximum at about 530 nm. The
absence of a blue-shift strongly indicates that amaranthin does not form aggregates on the TiO2
surface. Differences are also observed in the kinetics of dye-loading. While specific experiments
were not performed to elucidate the kinetic constants associated with dye loading, the basic
observation that films sensitized for similar time under similar conditions produced slightly
lighter films with amaranthin vs. betanin. These differences in behavior indicate that the steric
bulk of amaranthin may be a factor in reducing the dye-dye and dye-surface interactions. The
92
additional sugar group provides slightly more hydrophilicity and so the dye itself may have a
lower tendency to partition onto the surface. Based on HPLC analysis, however, the
hydrophilicity difference is not consistent with the magnitude of the observed changes. Further,
the hydroxyl groups of the sugar could provide additional binding sites for the dye on TiO2,
however, that does not seem to be observed. Thus, we can speculate that the primary binding
sites are the carboxylic acid groups, which is consistent with the current theory of dye binding in
DSSCs. Other than binding differences, amaranthin does not exhibit significant spectral changes
from those of betanin, and since the chromophore is fairly removed from the sugar groups,
observation of behavior of amaranthin can give a rough approximation of betanin monomeric
behavior. It is worth mentioning, though, that the redox properties of amaranthin are slightly
different than those of betanin, but only to the extent that one would assume a minor decrease in
DSSC performance.9 The oxidation potential of amaranthin is slightly less negative than that of
betanin, which would increase the rate of recombination thereby lowering the VOC.
Figure 5.2: Absorbance spectra of films sensitized with amaranthin from aqueous solutions of varying pH (a), and from methanol (MeOH) and acetonitrile (MeCN) (b).
93
5.3.2 Solvents and pH Differences
Sensitization of amaranthin on TiO2 films does not appear to depend greatly on whether
the solutions are aqueous or non-aqueous, though pH plays a significant role in binding just as is
observed with betanin. Solutions of amaranthin in all solvents show an absorption maximum
around 545 nm, which is slightly red-shifted from that of betanin. On films, there is a slight blue-
shift of the absorbance to about 535 nm. This blue-shift is not associated with a broadening of
the spectrum as it is with betanin, and so aggregation is not considered as the source of the
spectral change.
Figure 5.3: Absorbance spectra of solutions of amaranthin in methanol and aqueous solvents show no significant difference, though amaranthin in acetonitrile (purple) shows a broadening and blue-shift.
94
5.3.3 Fluorescence
In solution, the fluorescence of amaranthin is slightly less than that of betanin, which
stands to reason given the additional hydroxyl groups and the assumed greater flexibility of the
molecule. The measured fluorescence quantum yields for the dyes are at 0.68% and 0.88% for
amaranthin and betanin rescpectively. The reduced fluorescence implies that amaranthin may not
perform as well in a DSSC device due to the increased non-radiative decay, but the ease of
preparation of the pigment compared to betanin serves as a significant impetus to explore its
functionality. On the surface of TiO2, the fluorescence from films doesn’t show a significant
deviation from those with betanin, and so the changes may not affect the performance of DSSCs
because both dyes have high injection rate into TiO2. The spectral shape of the fluorescence is
the same for both betanin and amaranthin indicating identical chromophores.
Figure 5.4: Emission spectra of amaranthin (black) and betanin (red) in aqueous solution with excitation at 514.5 nm normalized to show that the shapes are identical.
95
5.3.4 Raman
Using surface enhanced resonance Raman scattering, solutions of both betanin and
amaranthin were observed to have very similar vibrational frequencies meaning their
chromophores are identical. Of particular note is the vibrational frequency at approximately 1600
cm -1
which corresponds to the C=N stretch according to reference 10. The intense enhancement
of this mode points to the excitation frequency being resonant with an excited electronic state
which is displaced along the normal coordinate for this vibration. In fact, all of the enhanced
modes are of nearly identical frequency in amaranthin and betanin which further emphasizes that
the chromophores of betanin and amaranthin behave in the same manner.
Figure 5.5: Surface enhanced resonance Raman spectra of betanin and amaranthin show that the chromophore vibrational modes are nearly identical excited at 458 nm.
96
5.3.5 Device Performance
Solar cells produced from amaranthin pigment did not perform quite as efficiently as
those from betanin. The IPCE, as shown in Figure 5.6, show a maximum of about 45% compared
to the 50% observed for betanin. The maximum IPCE, however, lies in the red range around 550
nm to 600 nm and correlates more closely with a monomeric betanin film. When this aspect is
taken into consideration, the IPCE of amaranthin is slightly higher than that produced by our
betanin monomeric film discussed in chapter 3. An increase in IPCE for the amaranthin cells is
likely correlated with the lack of aggregation at higher dye loading and so better light harvesting
is achieved by the monomer. Interestingly, when sensitized from non-aqueous solvents,
amaranthin shows some anomalous behavior as is shown in Figure 5.6. When sensitized from
methanol, amaranthin starts to behave similarly to that of betanin aggregates with an increase in
light harvesting in the blue region from 450 nm to 475 nm. However, from acetonitrile, there is
no evidence of increased light harvesting in the blue and the IPCE closely resembles the
absorptance of amaranthin solution.
97
Figure 5.7 shows the J-V curves for amaranthin based cells from both aqueous and non-
aqueous sensitizing solutions. These cells produce high JSC much like the betanin cells, but have
an overall lower average VOC. The reduced VOC is likely due to larger amounts of recombination
due to energetic differences in the amaranthin oxidation potentials. The efficiencies of the cells
are shown in Table 5.1. The maximum conversion efficiency obtained for a single amaranthin
cell was roughly 1.5%. This is half of the maximum efficiency obtained for a betanin based cell,
but compares well with the monomeric betanin cell mentioned in Chapter 3.
Figure 5.6: Replicate IPCE spectra of amaranthin based solar cells sensitized from MeOH (a) and MeCN (b) solutions.
98
Table 5.1: Energy conversion efficiencies of amaranthin based solar cells sensitized from acetonitrile (left), methanol (center) and water (right). The efficiency of cell 1 in methanol was excluded for the calculation of the average.
Amaranthin in MeCN Amaranthin in MeOH Amaranthin in Water
Cell Efficiency
(%) Cell
Efficiency (%)
Cell Efficiency
(%)
1 1.33 1 0.47 1 0.60
2 1.25 2 1.03 2 0.61
3 1.23 3 0.91 3 0.99
4 0.94 4 1.49 4 1.17
5 1.16 5 1.38 5 1.12
AVG 1.18 AVG 1.20 AVG 0.90
Figure 5.7: A comparison of the J-V curves of amaranthin from aqueous (a) and non-aqueous solution (b).
99
5.4 Conclusions
Despite the lower overall efficiency of DSSCs produced with amaranthin, there are
several factors that show this dye might still have promise as a sensitizer. Mainly, the JSC values
obtained for cells of comparable structure to those of betanin studied in chapter 3 are very
similar. The high currents show that amaranthin can easily inject electrons into the TiO2, but
suffers from some additional recombination losses. We have not attempted any engineering
measures to reduce recombination. There are many different methods that have been
implemented in other DSSC research such as adding a tunneling layer or changing redox
mediator. Furthermore, the pigment source of amaranthin is not a food source and can even be
harvested simultaneously with the food source from this crop. Betanin production from beets is
exclusive of their use in food products. While amaranth is not a widely grown and used grain
crop in the United States, it is a highly nutritious and easily cultivated crop that provides benefits
over rice and wheat.
The differences observed in the performance of DSCCs between amaranthin and betanin
are not related to fundamental differences in the photophysics as is shown by the fluorescence
and Raman data. Since these data indicate that the chromophores of the two molecules are nearly
identical, amaranthin is a good companion molecule to study with betanin. Since the sugar
groups effectively inhibit aggregation, amaranthin can be substituted for betanin in order to gain
insight into the nature of the aggregate structure and its effects on DSSC performance.
100
5.5 References
1 Treat, N. A.; Knorr, F. J.; McHale, J. L., Templated Assembly of Betanin Chromophore on
TiO2: Aggregation-Enhanced Light-Harvesting and Efficient Electron Injection in a Natural
Dye-Sensitized Solar Cell. J. Phys. Chem. C. 2016, 120 (17), 9122-9131.
2 O’Regan, B.; Grätzel, M. A low-Cost, High-Efficiency Solar Cell Based on Dye-Sensitized
Colloidal Titanium Dioxide Films., Nature 1991, 353, 737-40.
3 Cai, Y.; Sun, M.; Corke, H. Colorant Properties and Stability of Amaranthus Betacyanin
Pigments. J. Agric. Food Chem. 1998, 46, 4491–4495.
4 Mabry, T.; Taylor, A.; Turner, B. The Betacyanins and their Distribution.
Phytochemistry 1963, 2, 61–64.
5 Crosby, G. A.; Demas, J. N. Measurement of Photoluminescence Quantum Yields. Review. J.
Phys. Chem. 1971, 75, 991–1024.
6 Rurack, K. Fluorescence Quantum Yields: Methods of Determination and Standards. Springer
Series on Fluorescence Standardization and Quality Assurance in Fluorescence
Measurements I 2008, 5, 101–145.
7 Melhuish, W. H. Measurement of Quantum Efficiencies of Flourescence and Phosphorescence
and Some Suggested Luminescence Standards. J. Opt. Soc. Am. 1954, 54, 183–183
8 Lee, P. C.; Meisel, D. Adsorption and Surface-Enhanced Raman of Dyes on Silver and Gold
Sols. J. Phys. Chem. 1952, 88, 3391-3395.
9 Gandía-Herrero, F.; Escribano, J.; García-Carmona, F. Structural Implication on Color,
Fluorescence, and Antiradical Scavenging Activity in Betalains, Planta 2010, 232, 449-460.
10 Cai, Y.; Sun, M.; Wu, H.; Huang, R.; Corke, H. Characterization and Quantification of
Betacyanin Pigments from Diverse Amranthus Species. J. Agric. Food Chem. 1998, 46,
2063-2070.
101
6 Summary and Conclusion
Betalain pigments were studied for use in dye-sensitized solar cell applications. By
developing a bulk production method for pure samples of betanin and amaranthin, the
photphysical properties of these pigments could be studied without the potential interference of
matrix components and other secondary pigment species. Utilizing absorbance and fluorescence
spectroscopy, Raman scattering, incident photon to electron conversion efficiency and total
energy conversion efficiency measurements, two primary betalain pigments were thoroughly
analyzed.
By using purified betanin, our lab was able to show that the dye molecule undergoes
templated assembly into dimers on the surface of TiO2 which causes significant spectral changes.
The discovery of this dimer is the first reported case of an aggregated dye outperforming the
monomer in a DSSC and produced a record efficiency of 3%, compared to the 1.7% obtained in
other labs, for a natural dye based solar cell. Increases in light harvesting ability and stability
were observed in aggregated samples and this fact may provide some insight into the
mechanisms that are used in nature to produce high-efficiency light harvesting for
photoprotective purposes.
Amaranthin, grown and extracted by our group, provides a nice step for building a model
of the structure-function relationships in betalains. With minor differences in fluorescence and
hydrophilicity, we do not have to make any serious assumptions when comparing the behavior of
the two dyes. One extra sugar group added onto the betanin molecule does however impact the
102
ability of amaranthin to form aggregates. Therefore, amaranthin is used to model what we might
expect from a monomeric betanin DSSC. Furthermore, the practical implications of amaranthin
provide good reasoning for investigating the pigment.
Some basic understanding the aggregation of the chromophore was gained by using the
additives phosphate and methanol to the sensitizing solutions. These additives allowed us to
determine the relative binding affinity of betanin to the surface of TiO2, discover that
aggregation is a two-step process and eliminate variables in the formation of aggregates.
6.1 Future Work
Betalains are a diverse class of molecules and the next step in mapping the structure-
function relationships is to systematically study the different structures. Of particular interest to
future studies are the malonyl and glutaryl acylated betanin structures phylocactin and
hylocerenin respectively. These molecules have been shown by food scientists to have more
favorable stability than betanin and do not have significantly increased steric bulk such as
amaranthin. Some amaranthin derivatives are of interest as well since the addition of acyl groups
has shown significant improvement in pigment retention in unfavorably acidic conditions. Other
studies could focus on the betaxanthin class of pigments, which absorb in the blue region.
Ultimately, a combination of both betaxanins and betacyanin may prove to be the most effective
in DSSC applications since these molecules are suspected to undergo effective energy transfer in
the form of FRET from yellow to red. This energy transfer advantage could broaden the effective
range of the solar cells, especially if aggregation between the two pigments were achieved.
103
6.2 Final Remarks
It is unfortunate that time constraints can play such a major role in the amount of work
that can be accomplished. For example, we set out on this project intending to analyze the
betacyanins phylocactin and hylocerinin as well as the degradation product neobetanin. The
analysis of these structures would provide an excellent foundation for the mapping of structure-
function relationships in betalain pigments. Some of our precious time was spent on trying to
effectively extract betaxanthins from yellow beets, an endeavor that proved to be much more
complicated and cumbersome than we had anticipated. Ultimately, our intention was to generate
an understanding of these dye structures to enable a nature-inspired design of future sensitizers.
While an overall conversion efficiency of 3% will not be a large-scale production system in the
near future, the understanding of natural light harvesting systems can provide inspiration in our
search for more renewable energy sources. While our research may be driven by a particular goal
or application, it is often the things we pick up on the side that change the game the most.
104
7 Appendix A: MATlab Code
7.1 BuildRC
function [ Rn,Cn ] = BuildRC( n,m,j )
Rn=rand(m,n);
Cn=rand(j,n);
End
7.2 LeeSeung
function [ R2,R3,R4,R5,R6,C2,C3,C4,C5,C6 ] = LeeSong( D,X,N )
[m,j]=size(D);
[R2,C2]=BuildRC(2,m,j);
[R3,C3]=BuildRC(3,m,j);
[R4,C4]=BuildRC(4,m,j);
[R5,C5]=BuildRC(5,m,j);
[R6,C6]=BuildRC(6,m,j);
for i=1:100
figure(N)
for i=1:100
[R2,C2] = Lee(R2,C2,D);
end
subplot(5,2,1)
plot(X,R2)
axis tight
subplot(5,2,2)
plot(C2)
axis tight
105
for i=1:100
[R3,C3] = Lee(R3,C3,D);
end
subplot(5,2,3)
plot(X,R3)
axis tight
subplot(5,2,4)
plot(C3)
axis tight
for i=1:100
[R4,C4] = Lee(R4,C4,D);
end
subplot(5,2,5)
plot(X,R4)
axis tight
subplot(5,2,6)
plot(C4)
axis tight
for i=1:100
[R5,C5] = Lee(R5,C5,D);
end
subplot(5,2,7)
plot(X,R5)
axis tight
subplot(5,2,8)
plot(C5)
axis tight
for i=1:100
[R6,C6] = Lee(R6,C6,D);
end
subplot(5,2,9)
plot(X,R6)
axis tight
106
subplot(5,2,10)
plot(C6)
axis tight
end
figure(N+1)
[ D21 ] = Dcalc( R2,C2,1 );
[ D22 ] = Dcalc( R2,C2,2 );
subplot(2,1,1)
plot(X,D21)
axis tight
subplot(2,1,2)
plot(X,D22)
axis tight
figure(N+2)
[ D31 ] = Dcalc( R3,C3,1 );
[ D32 ] = Dcalc( R3,C3,2 );
[ D33 ] = Dcalc( R3,C3,3 );
subplot(3,1,1)
plot(X,D31)
axis tight
subplot(3,1,2)
plot(X,D32)
axis tight
subplot(3,1,3)
plot(X,D33)
axis tight
figure(N+3)
[ D41 ] = Dcalc( R4,C4,1 );
[ D42 ] = Dcalc( R4,C4,2 );
107
[ D43 ] = Dcalc( R4,C4,3 );
[ D44 ] = Dcalc( R4,C4,4 );
subplot(2,2,1)
plot(X,D41)
axis tight
subplot(2,2,2)
plot(X,D42)
axis tight
subplot(2,2,3)
plot(X,D43)
axis tight
subplot(2,2,4)
plot(X,D44)
axis tight
figure(N+4)
[ D51 ] = Dcalc( R5,C5,1 );
[ D52 ] = Dcalc( R5,C5,2 );
[ D53 ] = Dcalc( R5,C5,3 );
[ D54 ] = Dcalc( R5,C5,4 );
[ D55 ] = Dcalc( R5,C5,5 );
subplot(2,3,1)
plot(X,D51)
axis tight
subplot(2,3,2)
plot(X,D52)
axis tight
subplot(2,3,3)
plot(X,D53)
axis tight
subplot(2,3,4)
plot(X,D54)
axis tight
108
subplot(2,3,5)
plot(X,D55)
axis tight
figure(N+5)
[ D61 ] = Dcalc( R6,C6,1 );
[ D62 ] = Dcalc( R6,C6,2 );
[ D63 ] = Dcalc( R6,C6,3 );
[ D64 ] = Dcalc( R6,C6,4 );
[ D65 ] = Dcalc( R6,C6,5 );
[ D66 ] = Dcalc( R6,C6,6 );
subplot(2,3,1)
plot(X,D61)
axis tight
subplot(2,3,2)
plot(X,D62)
axis tight
subplot(2,3,3)
plot(X,D63)
axis tight
subplot(2,3,4)
plot(X,D64)
axis tight
subplot(2,3,5)
plot(X,D65)
axis tight
subplot(2,3,6)
plot(X,D66)
axis tight
% Automatically generates a Lee and Seung NMF analysis of your data in
% 2-6 components. Including showing the calculated totals which can be
% directly compared to the input spectra.
end
109
7.3 Lee
function [R,C] = Lee(R,C,D)
C=C';
[r,c]=size(D);
[n,x]=size(C);
DD = R*C;
Err = D./DD;
ER = Err*C';
R = R.*ER;
sr = sum(R);
for i = 1:n
R(:,i)=R(:,i)/sr(i);
end
EC=R'*Err;
C = C.*EC;
C = C';
% Lee and Seung Algorithm for non-negative matrix factorization
end
7.4 Dcalc
function [ D ] = Dcalc( R,C,n )
C=C';
D=R(:,n)*C(n,:);
%Generates the calculated spectra from components
end
110
8 Appendix B: HPLC Methods
Table 8.1: Binary gradient developed specifically for analysis of amaranth samples.
Amaranth Analytical Run Time (min) % B
0 7.2
1 8
3.5 10
5 12.5
7 15
9 100
11 100
12 7.2
Table 8.2: Binary gradient developed for analysis of samples from beets.
Beets Analytical Run Time (min) % B
0 10
1 12
3.5 13
5 15
7 100
10 100
11 10
Table 8.3: Binary gradient developed for preparatory methods of all betalain sources.
Preparatory
Run Time (min) % B 0 7.2
10 7.2
12 9
15 9.5
17 10
20 12
22 15
25 100
30 100
31 7.2