Pyridinium p-DSSC dyes: An old acceptor learns new tricks

Article abstract of DOI:10.1016/j.dyepig.2019.02.044

A family of six (five new) thiophenyl bridged triarylamine-donor based dyes with pyridine anchoring groups have been synthesized and studied as sensitizers for the p-type dye-sensitized solar cell (p-DSSC). They comprise bis-dicyano acceptor systems with

Full text of DOI:10.1016/j.dyepig.2019.02.044

Accepted Manuscript
Pyridinium p-DSSC dyes: An old acceptor learns new tricks
Anil Reddy Marri, Fiona A. Black, John Mallows, Elizabeth A. Gibson, John Fielden
PII:
S0143-7208(18)32390-8
https://doi.org/10.1016/j.dyepig.2019.02.044
DYPI 7378
DOI:
Reference:
To appear in: Dyes and Pigments
Received Date: 29 October 2018
Revised Date: 22 February 2019
Accepted Date: 24 February 2019
Please cite this article as: Marri AR, Black FA, Mallows J, Gibson EA, Fielden J, Pyridinium p-DSSC
dyes: An old acceptor learns new tricks, Dyes and Pigments (2019), doi: https://doi.org/10.1016/
j.dyepig.2019.02.044.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Pyridinium p-DSSC Dyes: An Old Acceptor Learns New Tricks
Anil Reddy Marri,a Fiona A. Black,b John Mallows,b Elizabeth A. Gibsonb and John
Fieldena*
a
School of Chemistry, University of East Anglia, Norwich Research Park, Norwich, NR4
7TJ, United Kingdom.
b School of Chemistry, Newcastle University, Newcastle-upon-Tyne, NE1 7RU, UK.
Abstract
A family of six (five new) thiophenyl bridged triarylamine-donor based dyes with pyridine
anchoring groups have been synthesized and studied as sensitizers for the p-type dye-
sensitized solar cell (p-DSSC). They comprise bis-dicyano acceptor systems with a single
pyridyl binder incorporated directly into the triarylamine (1), or separated by a phenyl group
(2); a mono-dicyano with two phenyl pyridine binders (3); and respective homologues 4 to 6
with pyridinium acceptors. In all cases, compared to their dicyano counterparts, the
pyridinium based dyes have higher extinction coefficients and smaller HOMO-LUMO gaps
that give broader spectrum absorption. Thus, despite lower dye uptake, devices based on
pyridiniums 4 and 6 have identical power conversion efficiencies (η) to the equivalent
dicyano systems 1 and 3. However, the best performing device (η = 0.06%) is based on the
known bis-acceptor dicyano system 2, as the large size and double positive charge of 5 leads
to a substantial disadvantage in loading on NiO. Absorbed-photon-to-current efficiencies for
5 are competitive with or higher than those of 2, implying a better per dye performance
consistent with the absorption profile, and DFT calculations suggesting better charge
separation. Thus, pyridiniums may provide a new, and easily accessible high performance
acceptor for p-DSSC dyes, but are likely better paired with anionic binding groups such as
carboxylates.

ACCEPTED MANUSCRIPT
Introduction
Dye Sensitized solar cells have attracted great attention since the first report of Grätzel and
co-workers in 1991,[1] as low cost alternatives to conventional silicon solar cells.[2,3] To
date, the best efficiency obtained is around 13%, reported by Grätzel et. al, based on
panchromatic dyes with cobalt electrolytes in n-type DSSCs.[4] In the n-DSSC architecture,
the ability of dyes to usefully harvest low energy photons is limited by the need to combine a
HOMO low enough in energy to be regenerated by the redox mediator, and a LUMO high
enough to efficiently inject electrons into the conduction band.[5] A potential solution is to
build tandem DSSCs, where a photoactive cathode would replace the platinised counter
electrode of the Grätzel cell and harvest lower energy photons, increasing both spectral
coverage and photovoltage (VOC) .[5-8]
Over the past decade, extensive research has been conducted into the p-type dye sensitized
solar cells (p-DSSCs) needed to pair with n-DSSCs in tandem devices.[5-15] In a p-type
DSSC, the operating mechanism is the reverse of that of the n-type DSSC (i.e, Grätzel solar
cells): instead of injecting electrons into an n-type semiconductor upon photoexcitation, the
sensitizer dye injects holes to the valence band (VB) of a p-type semiconductor (typically
NiO). Then the redox electrolyte oxidizes the reduced dye molecule and transfers electrons to
the counter electrode (or n-type DSSC photoanode). However, the power conversion
efficiencies (PCEs) of p-DSSCs have remained very low, preventing construction of useful
tandem devices. This is due to a number of factors, including the small energy difference
between the VB of NiO and the redox mediator, rapid back transfer of photogenerated holes
from NiO to the electrolyte, and slow hole transfer through NiO.[5-8] Changing redox
-
electrolyte from I-/I3 to Co2+/3+ [9,10] and latterly Fe2+/3+ [11] based systems has led to greatly

ACCEPTED MANUSCRIPT
increased VOC (up to 709 mV) and the highest p-DSSC power conversion efficiencies (PCE,
η) to date (up to 2.5%), but the negative redox potentials associated with these mediators
would severely reduce the photovoltage produced by the TiO2 side of any tandem cell. With
-
the more TiO2 compatible I-/I3 , the p-DSSC record VOC remains 350 mV,[12] and record η
0.61%.[13], and there are only a few reports of pn-tandem systems whose overall
performance exceeds that of the n-type component alone.[14]
Dye-sensitized solar cells are complex devices whose overall performance can be
unpredictable, so development relies on a combination of design, and exploration of new
materials. To date, nearly all sensitizers for p-DSSCs have used the carboxylic acid anchoring
groups employed in n-DSSCs, but one report has suggested pyridyl binders could be
competitive, and give high VOC.[15] Pyridyl groups may also be more suitable for binding
alternative p-SCs. Similarly, while dicyano acceptor groups have featured in many p-DSSC
dyes, cationic acceptor groups – such as the indoliums that have achieved record JSC[16] –
have been little explored. One large, and easily accessible family of cationic organic acceptor
groups are the pyridiniums, whose absorption properties can easily be tailored by adjusting
the quaternising group – for example by replacing alkyl groups with phenyl, or
dinitrophenyl.[17]
Herein, we investigate a family of six pyridyl functionalized triaryl amine donor-acceptor
dyes with both dicyano and methyl pyridinium acceptor groups, with either two anchors and
one acceptor or one anchor and two acceptors, and pyridine groups either directly forming
part of the triaryl amine or appended to one of the phenyl groups (Fig. 1). While the
pyridiniums give higher extinction coefficients and absorb across a broader spectrum of light,
and DFT calculations suggest better spatial separation of HOMO and LUMO (and thus
charge) the best device performance is obtained from the bis-dicyano, mono-pyridyl
architecture. This appears to be a consequence of poor uptake of the larger, and cationic

ACCEPTED MANUSCRIPT
pyridinium dyes at the positively charged NiO surface, thus suggesting that charge
compensation (for example from carboxylate binders) could yield a new class of easily
accessible and high performance p-DSSC dyes.
Fig. 1. Chemical structures of the dyes 1 to 6 used in this study. Compound 2 was previously
published.14
Experimental
Materials and Procedures. The starting materials 4-bromotriphenylamine, pyridine-4-
boronic
acid
dihydrate,
tetrakis(triphenylphosphine)
palladium(0),
bis(diphenylphosphinoferrocene) palladium(II) chloride, formylthiophene boronic acid, N-
bromosuccinimide, 4-methyl pyridine, iodomethane and malononitrile were purchased from
Sigma–Aldrich and Alfa Aesar. Dry dimethylformamide (DMF) for electrochemical
experiments was purchased from Sigma-Aldrich in SureSeal® bottles. Solvents were purified
by standard procedures and purged with argon or nitrogen before use. All other chemicals
used in this work were of analytical grade and were used without further purification until or
unless stated, and all reactions were performed under an inert atmosphere of argon or
nitrogen. N-methyl-4-picolinium iodide,[18] 5-[4-(diphenylamino)phenyl]thiophene-2-

ACCEPTED MANUSCRIPT
carbaldehyde (P7)[19], 5-[4-bis-(4-bromophenyl)aminophenyl]thiophene-2-carbaldehyde
(P8)[20] and the electrolyte N(C4H9-n)4]BF4 [21] were synthesized according to literature
procedures. Chromatographic separations were carried out on silica gel (60–120 mesh).
1
13
General Physical Measurements. H NMR and C NMR spectra were recorded a Bruker
Ascend 500 MHz spectrometer. Elemental analyses and mass spectrometry were outsourced
to London Metropolitan University, and the UK National Mass Spectrometry Service at
Swansea University respectively. Electronic absorption spectra were recorded on an Agilent
Cary60 UV-Vis spectrophotometer.. Electrochemical data were obtained using Autolab
PGSTAT30 and PGSTAT302 potentiostat/galvanostats. A single-compartment cell was used
with an Ag wire pseudo reference electrode, platinum working electrode and Pt wire auxiliary
electrode, the supporting electrolyte was [N(C4H9-n)4]BF4 in dry DMF. Solutions containing
ca. 10–3 M analyte (0.1 M electrolyte) were degassed by purging with nitrogen. All E1/2
values were calculated from (Epa + Epc)/2 at a scan rate of 100 mV s–1 and referenced to
Fc/Fc+.
Solar Cell Assembly and Measurements. A NiO precursor solution was prepared by
dissolving anhydrous NiCl2 (1 g) and the tri-block co-polymer F108 (poly(ethylene glycol)-
block-poly (propylene glycol)-block-poly(ethylene glycol)) (1 g) in ethanol (6 g) and distilled
water (5 g). The solution was allowed to rest for 10 days and then centrifuged to separate out
any remaining solids.
For the photocathodes, conducting glass substrates (Pilkington TEC15, sheet resistance 15
Ω/square) were cut to size (approx. 2 × 2 cm) and then cleaned by sonic bath for 15 minutes,
first in soapy water, second in 0.1 M HCl, and finally in ethanol. The NiO electrodes were
made by applying the precursor solution onto the conducting glass substrates by doctor blade
using Scotch Magic tape as a spacer (0.2 cm2 active area), followed by sintering in a

ACCEPTED MANUSCRIPT
Nabertherm B150 oven at 450 °C for 30 min. Additional layers of precursor solution were
applied and sintered to increase the film thickness. The NiO electrodes were soaked in a
dichloromethane (or acetonitrile) solution of the dye (0.3 mM) for 16 h at room temperature.
For the counter electrodes, the conducting glass substrates (Pilkington TEC8, sheet resistance
8 Ω/square) were cut and cleaned as described above. A small (ca. 1 × 1 mm) hole was
drilled into a corner for the electrolyte solution to be introduced at the assembly stage.
Platinized counter electrodes were prepared by applying chloroplatinic acid solution (4.4
mM, 10 µL cm-2) in ethanol onto the conducting glass substrate. The electrodes were then
heated in a Nabertherm B150 oven at 400 °C for 15 minutes and then allowed to cool. The
dyed electrodes were removed from the dye bath and washed with acetonitrile. Cells were
assembled face-to-face with the platinized counter electrode using a 30 µm thick
thermoplastic frame (Surlyn 1702). The electrolyte contained LiI (1.0 M) and I2 (0.1 M) in
dry acetonitrile, and was introduced through the predrilled hole in the counter electrode by
vacuum back filling. The cells were then sealed using a 30 µm thick thermoplastic square
(Surlyn 1702) and a glass cover slip.
Current density vs. voltage measurements were measured using an Ivium CompactStat
potentiostat either in the dark or under AM 1.5 simulated sunlight from an Oriel 150 W solar
simulator, calibrated using a Si diode to give light with an intensity of 100 mW cm−2. DSCs
were held in place using a custom made cell holder with a circular aperture with a diameter 1
mm wider than the working area of the cell. An applied voltage was swept from slightly
above the given open-circuit voltage, at a scan rate of 3 mV s-1. Incident photon-to- current
conversion efficiencies (IPCEs) were recorded by passing the light from the solar simulator
through an Oriel Cornerstone 130 1/8 m monochromator over the range of 400-700 nm at 10
nm increments and recording the current from the solar cell with an Ivium CompactStat
potentiostat which was calibrated against a Si photodiode.

ACCEPTED MANUSCRIPT
Dye loading measurements were performed by soaking four NiO films for each dye in a ca.
0.3 mM solutions of the dye in acetonitrile. UV-vis absorption spectra were taken of the films
before and after soaking. To minimise the effect of any variable scattering, both bare and
dyed NiO films were normalised to 0 abs at 800 nm, and the spectrum of the bare film
subtracted from the dyed film. The average loading was then obtained using the dye
extinction coefficient.
DFT calculations
DFT calculations were carried out using the ADF suite of programs.[22] Both geometry
optimsation and property calculations were carried out using the B3LYP functional (20%
Hartree-Fock exchange)[23] and the ADF double-ζ, single polarisation function DZP basis
set. This is comparable to the commonly used 6-31g* basis set, and initial optimization of the
calculations on compounds 1 and 4 indicated that the larger triple-ζ and augmented basis sets
(TZP, AUG-DZP) increased computational expense, without signficantly changing results. A
‘small’ frozen core was employed for the geometry optimizations. Calculations of the
electronic spectra were carried out by TD-DFT using the EXCITATION modules
implemented in the ADF program [24] and were based on the optimized geometries, with no
frozen core. The EXCITEDGO module was used to calculate geometries of the first excited
states, and ground-to-excited state dipole moment changes (∆µ12) calculated by subtracting
the ground state dipole moment from the excited state dipole moment. Solvent (acetonitrile)
was introduced using COSMO with Klamt atomic Radii. The subkey NOCSMRSP was
applied to prevent induced electronic charges (which respond on a much slower timescale
than electronic excitations) from influencing COSMO surface charges, together with the
ALLPOINTS key.

ACCEPTED MANUSCRIPT
Syntheses
N,N-diphenylpyridin-4-amine (P1)
Diphenylamine (0.500 g, 2.954 mmol), 4-bromopyridine hydrochloride (0.690 g, 3.545
mmol), and Pd(OAc)2 (0.045 g, 0.12 mmol), (t-Bu)3P (0.04 g, 0.12 mmol), and t-BuONa
(0.852 g, 8.86 mmol) were dissolved in toluene (40 mL) and stirred for 10 h at 100 oC. After
concentrating under reduced pressure, the resulting residue was dissolved in dichloromethane
and washed with water. The residue was chromatographed on silica gel (hexane/ethylacetate
1
= 3:7 as eluent) to give P1 as colourless solid with 34 % yield. H NMR (CDCl3, 500 MHz,
δ): 8.23 (d, J = 6.6 Hz, 2H), 7.35 (t, J = 7.8 Hz, 4H), 7.17-7.20 (m, 6H), 6.73 (d, J = 6.6 Hz,
2H). 13C NMR (CDCl3, 125 MHz, δ): 153.70, 150.18, 145.25, 129.79, 126.69, 125.55,
112.78. ESI-MS: m/z: 247.12. Elemental analysis for C17H14N2, Anal. (Calcd) %: C 82.78
(82.90); H 5.60 (5.73); N 11.25 (11.37).
N,N-bis(4-bromophenyl)pyridin-4-amine (P2)
Compound P1 (0.300 g, 1.218 mmol) and N-bromosuccinimide (0.477 g, 2.68 mmol) were
added to chloroform (30 mL) and several drops of acetic acid were then added. The mixture
was stirred at room temperature for an hour and refluxed overnight, before cooling to room
temperature and addition of water (50 mL). The mixture was extracted with CH2Cl2 (3 × 50
mL) and the combined organic layers dried over MgSO4. The solvent was removed under
reduced pressure and the residue was purified by column chromatography with
hexane/ethylacetate (3:7) as the eluent. The product P2 was obtained as a colourless solid
with 75% yield. 1H NMR (CDCl3, 500 MHz, δ): 8.27 (d, J = 6.8 Hz, 2H), 7.51 (d, J =8.8 Hz,
13
4H), 7.05 (d, J = 8.8 Hz, 4H), 6.77 (d, J = 6.8 Hz, 2H). C NMR (CDCl3, 125 MHz, δ):
153.01, 150.54, 144.08, 133.05, 127.88, 118.75, 113.53. ESI-MS (m/z): 404.94. Elemental

ACCEPTED MANUSCRIPT
analysis for C17H12N2Br2•0.1C6H14, Anal. (Calcd) %: C, 51.10 (51.22); H 3.31 (3.27); N 6.61
(6.78).
N,N-bis(5-formylthiophenyl)pyridin-4-amine (P3)
A 50 mL Schlenk tube was charged with 5-formylthiophene boronic acid (0.085 g, 0.544
mmol), Pd(PPh3)4 (0.006 g, 0.005 mmol), dimethoxyethane (6 mL) and 2 M aqueous sodium
carbonate (1.5 mL), before purging with argon gas for 5 evacuate/refill cycles. Then
compound P2 (0.100 g, 0.247 mmol) was added. The tube was sealed and heated at 90 °C
with very vigorous stirring for 16 h. Upon cooling to ambient temperature, the organics were
removed by extracting three times with DCM. The combined organic fractions were washed
with brine and dried over MgSO4. The solvent was removed under reduced pressure and the
residue was purified by silica gel column chromatography using hexane/ethylacetate (3/7,
v/v) as eluent to give P3 as a yellow powder in 52% yield. 1H NMR (300 MHz, CDCl3, δ):
9.89 (s, 2H), 8.36 (d, J = 7.0 Hz, 2H), 7.75 (d, J = 4.0 Hz, 2H), 7.67 (d, J = 9.0 Hz, 4H), 7.39
13
(d, J = 4.0 Hz, 2H), 7.24 (d, J = 9.0 Hz, 4H), 6.91 (d, J = 7.0 Hz, 2H). C NMR (75 MHz,
CDCl3, δ): 182.71, 153.09, 152.90, 150.60, 146.03, 142.53, 137.47, 130.13, 127.92, 126.43,
124.78, 124.08, 114.78. ESI-MS (m/z): 467.0872. Elemental analysis for C27H18N2O2S2,
Anal. (Calcd) %: C (69.62) 69.51; H 3.95 (3.89); N 5.91 (6.00).
N,N-diphenyl-4-(pyridin-4-yl)aniline (P4)
A 50 mL Schlenk tube was charged with 4-pyridyl boronic acid (0.568 g, 4.626 mmol),
Pd(PPh3)4 (0.178 mg, 0.154 mmol), dimethoxyethane (10 mL) and 2 M aqueous sodium
carbonate (2.5 mL), and was purged with argon gas for 5 evacuate/refill cycles. Then 4-
bromotriphenylamine (0.500 g, 1.542 mmol) was added. The tube was sealed and heated at
90 °C with very vigorous stirring for 16 h. Upon cooling to ambient temperature, the mixture

ACCEPTED MANUSCRIPT
was extracted three times with DCM. The combined organic fractions were washed with
brine and dried over MgSO4. The solvent was removed under reduced pressure and the
residue was purified by silica gel column chromatography using hexane/ethylacetate (7/3,
1
v/v) as eluent to give P4 (77%) as a pale yellow powder. H NMR (500 MHz, CDCl3, δ):
8.61 (d, J = 6.2 Hz, 2H), 7.52 (d, J = 8.8 Hz, 2H), 7.47 (d, J = 6.2 Hz, 2H), 7.29 (m, 4H),
7.16-7.12 (m, 6H), 7.08 (tt, J = 7.4 Hz, 2H). 13C NMR (125 MHz, CDCl3, δ): 150.14, 148.99,
147.71, 147.26, 130.92, 129.45, 127.66, 125.01, 123.63, 122.91, 120.89. ESI-MS (m/z):
323.1542. Elemental analysis for C23H18N2, Anal. (Calcd) %: C 85.61 (85.68); H 5.58 (5.63);
N 8.78 (8.69).
4-iodo-N-(4-iodophenyl)-N-(4-(pyridin-4-yl)phenyl)aniline (P5)
A 1 M solution of iodine monochloride in DCM (6.20 mL, 6.20 mmol) was added to 5 mL of
dioxane. Zinc acetate was added to the above solution, which was stirred for 15 minutes at
room temperature. Compound P4 (0.500 g, 1.551 mmol) was dissolved in 5 mL of dioxane
and added slowly, and stirring was continued at room temperature overnight. After the
reaction completed (monitored by TLC), the reaction mixture was poured into sodium
thiosulfate solution (1 M, 30 mL) and extracted twice with dichloromethane. The combined
organic fractions were dried over MgSO4, then filtered and the volume reduced in vacuo. The
crude product was further purified by silica gel column chromatography using
hexane/ethylacetate (7/3, v/v) as eluent to give compound P5 (91%) as a pale yellow powder.
1H NMR (500 MHz, CDCl3, δ): 8.63 (d, J = 6.2 Hz, 2H), 7.57 (d, J = 8.8 Hz, 4H), 7.53 (d, J
= 8.8 Hz, 2H), 7.47 (d, J = 6.2 Hz, 2H), 7.13 (d, J = 8.7 Hz, 2H), 6.87 (d, J = 8.8 Hz, 4H). 13C
NMR (125 MHz, CDCl3, δ): 150.06, 147.81, 147.57, 146.62, 138.54, 132.41, 128.02, 126.44,
123.95, 121.04, 86.88. ESI-MS (m/z): 574.95. Elemental analysis for C23H16N2I2, Anal.
(Calcd) %: C 48.19 (48.11); H 2.81 (2.81); N 4.94 (4.88).

ACCEPTED MANUSCRIPT
4-{Bis-[4-(5-formyl-thiophen-2-yl)-phenyl]-amino}-N,N-diphenyl-4-(pyridin-4-yl)aniline
(P6)
A 50 mL Schlenk tube was charged with 5-formylthiophene boronic acid (0.060 g, 0.383
mmol), Pd(PPh3)4 (0.040 mg, 0.035 mmol), dimethoxyethane (8 mL) and 2 M aqueous
sodium carbonate (2.0 mL), and purged with argon gas for 5 evacuate/refill cycles. Then
compound P5 (0.100 g, 0.174 mmol) was added. The tube was sealed and heated at 90 °C
with very vigorous stirring for 16 h. Upon cooling to ambient temperature, the mixture was
extracted three times with DCM. The combined organic fractions were washed with brine and
dried over MgSO4. The solvent was removed under reduced pressure and the residue was
purified by silica gel column chromatography using hexane/ethylacetate (3/7, v/v) as eluent
1
to give compound P6 (56%) as a yellow powder. H NMR (500 MHz, CDCl3, δ): 9.88 (s,
2H), 8.61-8.70 (br s, 2H), 7.73 (d, J = 3.9 Hz, 2H), 7.61 (m, 6H), 7.52 (d, J = 6.2 Hz, 2H),
13
7.35 (d, J = 3.9 Hz, 2H), 7.26 (d, J = 8.7 Hz, 2H), 7.19 (d, J = 8.7 Hz, 4H). C NMR (125
MHz, CDCl3, δ): 182.68, 153.73, 149.91, 147.76, 147.68, 147.55, 142.00, 137.62, 133.17,
128.28, 128.23, 127.65, 125.00, 124.53, 123.53, 121.19. ESI-MS (m/z): 543.12. Elemental
analysis for C33H22N2O2S2•1.2H2O, Anal. (Calcd) %: C 70.18 (70.24); H 4.16 (4.36); N 5.03
(4.96).
5-{4-[bis(4-pyridin-4-yl)phenyl]aminophenyl}-thiophene-2-carbaldehyde (P9)
A 50 mL of Schlenk tube was charged with 4-pyridyl boronic acid (0.096 g, 0.78 mmol),
Pd(PPh3)4 (0.045 g, 0.039 mmol), dimethoxyethane (8 mL) and 2 M aqueous sodium
carbonate (2.0 mL), then the tube was purged with argon gas for 5 evacuate/refill cycles.
Compound P8 (0.100 g, 0.195 mmol) was and the tube was sealed and heated at 90 °C with
very vigorous stirring for 16 h. Upon cooling to ambient temperature, the mixture was

ACCEPTED MANUSCRIPT
extracted three times with DCM. The combined organic fractions were washed with brine and
dried over MgSO4. The solvent was removed under reduced pressure and the residue was
purified by silica gel column chromatography using hexane/ethylacetate (2/8, v/v) as eluent
1
to give compound P9 (65%) as a yellow powder. H NMR (500 MHz, CDCl3, δ): 9.88 (s,
1H), 8.66 (br s, 4H), 7.73 (d, J = 4.1 Hz, 1H), 7.59-7.62 (m, 6H), 7.50 (d, J = 5.2 Hz, 4H),
13
7.35 (d, J = 4.0 Hz, 1H), 7.26 (d, J = 8.7 Hz, 4H), 7.20 (d, J = 8.8 Hz, 2H). C NMR (125
MHz, CDCl3, δ): 182.66, 153.84, 150.33, 148.04, 147.64, 147.32, 141.93, 137.62, 133.19,
128.14, 127.98, 127.59, 124.90, 124.28, 123.44, 121.12. ESI-MS (m/z): 510.16. Elemental
analysis for C33H23N3OS, Anal. (Calcd) %: C 77.65 (77.78); H 4.61 (4.55); N 8.13 (8.25).
4-(Bis-{4-[5-(2,2-dicyano-vinyl)-thiophene-2-yl]-phenyl}-amino)-pyridine (1)
A mixture of compound P3 (0.050 g, 0.107 mmol), malononitrile (0.142 g, 2.14 mmol) and
pyridine (0.2 mL) in anhydrous toluene (6 mL) was deoxygenated five times by placing
under vacuum and backfilling with argon, and then heated at 70 oC for 16 h. After cooling to
room temperature, the solvent was removed and the residue was purified by column
chromatography over silica with a dichloromethane/methanol mixture (9:1) as the eluent to
1
give 1 as a black solid (0.035 g, 60%). H NMR (500 MHz, CDCl3, δ) 8.40 (d, J = 6.3 Hz,
2H), 7.80 (s, 2H), 7.7 (d, J = 4.2 Hz, 2H), 7.7 (d, J = 8.7 Hz, 4H), 7.42 (d, J = 4.1 Hz, 2H),
13
7.24 (d, J = 8.7 Hz, 4H), 6.93 (d, J = 6.3 Hz, 2H). C NMR (125 MHz, CDCl3, δ): 155.15,
152.62, 150.87, 150.44, 146.68, 140.09, 134.31, 129.15, 128.19, 126.26, 124.54, 115.46,
114.11, 113.33. ESI-MS (m/z): 563.11. Elemental analysis for C33H18N6S2, Anal. (Calcd) %:
C 70.29 (70.44); H 3.31 (3.22); N 15.09 (14.94).
4-(Bis-{4-[5-(2,2-dicyano-vinyl)-thiophene-2-yl]-phenyl}-amino)-4-phenylpyridine (2)

ACCEPTED MANUSCRIPT
A mixture of compound P6 (0.060 g, 0.110 mmol), malononitrile (0.146 g, 2.21 mmol) and
pyridine (0.2 mL) in anhydrous toluene (6 mL) was deoxygenated five times by placing
under vacuum and backfilling with argon, and then heated at 70 oC for 16 h. After cooling to
room temperature, the solvent was removed and the residue was purified by column
chromatography over silica gel by using dichloromethane/methanol mixture (9:1) as the
eluent to give 2 (75%) as a red solid. 1H NMR (500 MHz, (CD3)2CO, δ): 8.63 (d, J = 6.2 Hz,
2H), 8.41 (s, 2H), 7.97 (d, J = 4.2 Hz, 2H), 7.88 (m, 6H), 7.72 (d, J = 4.2 Hz, 2H), 7.68 (d, J
13
= 6.2 Hz, 2H), 7.35 (d, J = 8.7 Hz, 2H), 7.28 (d, J = 8.8 Hz, 4H). C NMR (125 MHz,
(CD3)2CO, δ: 155.22, 151.65, 150.43, 148.40, 147.21, 146.63, 141.82, 134.05, 133.79,
128.29, 127.94, 127.43, 125.68, 124.59, 124.42, 120.80, 114.38, 113.74, 75.25. ESI-MS
(m/z): 639.14. Elemental analysis for C39H22N6S2•0.6H2O, Anal. (Calcd): C 72.07 (72.11); H
3.74 (3.60); N 12.54 (12.94).
5-{4-[bis(4-pyridin-4-yl)phenyl]aminophenyl}-thiophene-2-vinyldinitrile (3)
A mixture of compound P9 (0.100 g, 0.196 mmol), malononitrile (0.130 g, 1.962 mmol) and
pyridine (0.2 mL) in anhydrous toluene (6 mL) was deoxygenated five times by placing
under vacuum and backfilling with argon, and then heated at 70 oC for 16 h. After cooling to
room temperature, the solvent was removed and the residue was purified by column
chromatography over silica with a dichloromethane/methanol mixture (9:1) as the eluent to
give compound 3 (75%) a brown solid. 1H NMR (500 MHz, CDCl3, δ): 8.66 (d, J = 5.8 Hz,
4H), 7.77 (s, 1H), 7.70 (d, J = 4.2 Hz, 1H), 7.61-7.63 (m, 6H), 7.50 (d, J = 6.2 Hz, 4H), 7.38
(d, J = 4.2 Hz, 1H), 7.27 (d, J = 8.7 Hz, 4H), 7.19 (d, J = 8.8 Hz, 2H). 13C NMR (125 MHz,
CDCl3, δ): 156.23, 150.35, 148.82, 147.37, 147.29, 140.29, 133.66, 133.61, 128.23, 127.86,
126.71, 125.23, 123.82, 123.72, 121.12, 114.35, 113.54. ESI-MS (m/z): 558.1735. Elemental

ACCEPTED MANUSCRIPT
analysis for C36H23N5S•H2O, Anal. (Calcd) %: C 75.16 (75.10); H 4.35 (4.38); N 12.18
(12.17).
4-(Bis-{4-[5-(2-[N-methyl-4-pyridiniumyl-vinyl)-thiophene-2-yl]-phenyl}-amino)-
pyridine dihexafluorophosphate (4)
A 50 mL round bottom flask was charged with compound P3 (0.050 g, 0.107 mmol) and N-
methylpicolinium iodide (0.063 g, 0.268 mmol) and a 1:1 mixture of chloroform and ethanol
was added (10 mL). A catalytic amount of piperidine was added and the mixture was refluxed
o
at 80 C for 4 hrs, before cooling to room temperature and addition of an excess of solid
NH4PF6. The mixture was stirred until the product precipitated. Compound 4 (0.025 g, 25%,
pale brown) was collected by vacuum filtration and washed several times with water and then
diethyl ether. 1H NMR (500 MHz, CD3CN, δ): 8.41 (d, J = 7.0 Hz, 4H), 8.30 (dd, J = 4.8 Hz,
2H), 7.96 (d, J = 16.1 Hz, 1H), 7.93 (d, J = 7.1 Hz, 2H), 7.74 (d, J = 8.6 Hz, 4H), 7.47 (m,
2H), 7.27 (d, J = 8.7 Hz, 4H), 7.07 (d, J = 16.0 Hz, 2H), 6.89 (dd, J = 4.7 Hz, 2H), 4.17 (s,
3H). 13C NMR (125 MHz, (CD3CN, δ): 153.82, 149.80, 148.68, 148.59, 148.40, 147.86,
145.05 139.83, 134.70, 134.46, 132.76, 129.19, 128.88, 127.69, 125.39, 125.18, 125.09,
123.84, 121.69, 121.53, 47.67. ESI-MS (m/z): 323.11 [M - 2PF6]2+. Elemental analysis for
C41H34F12N4P2S2, Anal. (Calcd) %: C 52.57 (52.49); H 3.66 (4.49); N 5.98 (5.37).
4-(Bis-{4-[5-(2-[N-methyl-4-pyridiniumyl-vinyl)-thiophene-2-yl]-phenyl}-amino)-4-
phenylpyridine dihexafluorophosphate (5)
A 50 mL round bottom flask was charged with compound P6 (0.060 mg, 0.110 mmol) and N-
methyl picolinium iodide (0.057 g, 0.1841 mmol), which were then dissolved in 50:50
chloroform:ethanol (10 mL). A catalytic amount of piperidine was added and the mixture was
refluxed at 80 oC for 4 h. After cooling to room temperature, excess solid NH4PF6 was added

ACCEPTED MANUSCRIPT
and the mixture was stirred until precipitation occurred. The precipitate was collected by
vacuum filtration and washed several times with water and diethyl ether to yield 5 (45%) as a
dark red solid. 1H NMR (500 MHz, CD3CN, δ): 8.61 (d, J = 6.2 Hz, 2H), 8.39 (d, J = 6.9 Hz,
4H), 7.96 (d, J = 16.1 Hz, 2H), 7.92 (d, J = 6.9 Hz, 4H), 7.74 (d, J = 8.8 Hz, 2H), 7.69 (d, J =
8.8 Hz, 4H), 7.61 (d, J = 6.2 Hz, 2H), 7.44 (m, 4H), 7.25 (d, J = 8.9 Hz, 2H), 7.18 (d, J = 8.7
13
Hz, 4H), 7.05 (d, J = 16.0 Hz, 2H), 4.16 (s, 6H). C NMR (125 MHz, CD3CN, δ): 150.45,
148.43, 145.05, 134.71, 134.47, 128.77, 127.67, 125.30, 125.20, 125.05, 123.84, 121.53,
47.67. ESI-MS(m/z): 361.13 [M - 2PF6]2+. Elemental analysis for C47H38N5F12N4P2S2, Anal.
(Calcd) %: C 55.73 (55.80); H 3.78 (3.92); N 5.53 (5.45).
5-{4-[bis(4-pyridin-4-yl)phenyl]aminophenyl}-thiophene-2-N-methylpyridinium
hexafluorophosphate (6)
A 50 mL round bottom flask was charged with compound P9 (0.050 g, 0.098 mmol) and N-
methyl picolinium iodide (0.057 g, 0.245 mmol) and 1:1 chloroform:ethanol (10 mL) was
added. A catalytic amount of piperidine was added to the above mixture, which was refluxed
o
at 80 C for 4 h. The mixture was cooled to room temperature and excess solidNH4PF6 was
added, before stirring until precipitation occurred. The precipitate was collected by vacuum
1
filtration and washed several times with water and ether to yield 6 as a red solid (52%). H
NMR (500 MHz, CD3CN, δ): 8.60 (d, J = 6.1 Hz, 4H), 8.37 (d, J = 6.8 Hz, 2H), 7.93 (d, J =
16.1 Hz, 1H), 7.91 (d, J = 7.0 Hz, 2H), 7.75 (d, J = 8.7 Hz, 4H), 7.69-7.66 (m, 6H), 7.44 (m,
2H), 7.24 (d, J = 8.7 Hz, 4H), 7.19 (d, J = 8.7 Hz, 2H), 7.04 (d, J = 15.9 Hz, 2H), 4.16 (s,
3H). 13C NMR (125 MHz, (CD3CN, δ): 153.82, 149.80, 148.68, 148.59, 148.40, 147.86,
145.05 139.83, 134.70, 134.46, 132.76, 129.19, 128.88, 127.69, 125.39, 125.18, 125.09,
123.84, 121.69, 121.53, 47.67. ESI-MS (m/z): 599.23 [M - PF6]+. Elemental analysis for
C40H31F6N4PS•H2O, Anal. (Calcd) %: C 62.89 (62.99); H 4.32 (4.36); N 7.41 (7.34).

ACCEPTED MANUSCRIPT
Results and discussion
Synthesis
Scheme 1 Synthetic route to short pyridyl anchored dyes 1 and 4: (a) Pd(OAc)2, (t-Bu)3P, t-BuONa,
toluene, 100 oC, 10h, 34%; (b) NBS, CHCl3, reflux, overnight, 75%; (c) 5-formyl-2-thiophene boronic
acid, Pd(PPh3)4, DME, 2M Na2CO3, reflux, 16h, 53%; (d) malononitrile, pyridine, toluene, 70 oC, 16h,
60%; (e) N-methyl picolinium iodide, piperidine, CH3Cl/EtOH, 80 oC, 4 h, then NH4PF6, 25%.
Scheme 2 Synthetic route to extended pyridyl anchored, bis acceptor dyes 2 and 5: (a) 4-
pyridinylboronic acid, Pd(PPh3)4, DME, 2M Na2CO3, reflux, 16h, 77%; (b) ICl, Zn(OAc)2, 1,4-
dioxane, RT, 5h, 91%; (c) 5-formyl-2-thiophene boronic acid, Pd(dppf)Cl2, toluene/methanol, K2CO3,

ACCEPTED MANUSCRIPT
o
reflux, 16h, 56%; (d) malononitrile, pyridine, toluene, 70 C, 16h, 75%; (e) 4-methyl picolinium
iodide, CHCl3/EtOH, 80 oC, 4h, 45%.
Scheme 3 Synthetic route to extended bis-pyridyl anchored, mono-acceptor dyes 3 and 6: (a) 5-
formyl-2-thiophene boronic acid, Pd(dppf)Cl2, toluene/methanol, K2CO3, reflux, 16 h, 89%; (b) NBS,
rt, 2 h, 95%; (c) 4-pyridinylboronic acid, Pd(PPh3)4, DME, 2M Na2CO3, reflux, 16 h, 65%; (d)
o
malononitrile, pyridine, toluene, 70 C, 16h, 75%; (e) 4-methyl picolinium iodide, CHCl3/ethanol, 80
oC, 4 h, 52%.
The synthetic routes to 1 to 6 are shown in Schemes 1 to 3. Of the six dyes, five are new
while 2 has been published before.[15] Our approach to 2 is different to the literature,
however. Rather than making 4-diphenylamino-phenylboronic acid, we chose to shorten the
synthesis by coupling commercially available 4-bromotriphenylamine and 4-pyridyl boronic
acid via a standard Suzuki approach. The resulting pyridine derivative, P4, was then
iodinated rather than brominated before reaction with formylthiophene boronic acid. From
the resulting aldehyde, both dicyano (2) and pyridinium (5) acceptor dyes could be obtained
by straightforward Knoevenagel condensations. Access to the other four dyes is also based on
the same reaction of a formyl thiophene derivative (P3 or P9). In the case of the “short” dyes
1 and 4 (where the pyridine anchor is directly part of the triarylamine donor group) the

ACCEPTED MANUSCRIPT
starting point was N-aryl coupling of diphenylamine with 4-bromopyridine hydrochloride to
obtain P1. To access the bis-anchor, mono acceptor systems 3 and 6 the formylthiophene unit
was introduced first, followed by the pyridines, rather than the other way round. This enabled
the routes to 2, 3, 5, and 6 all to start from the same precursor, 4-bromotriphenylamine.
Photophysical and Electrochemical Properties
UV-vis absorption and emission: Absorption spectra of 1 to 6 in acetonitrile are shown in Fig.
2 (top). All six show two major absorption bands in the visible and near-UV. The stronger,
lower energy band between 400 and 600 nm can be assigned to intramolecular charge transfer
(ICT) from the donor to the acceptor, while the higher energy band that has a peak at ca. 350
nm can be assigned to π-π* electronic transitions. The λmax of the ICT band shifts very
significantly (30 – 46 nm) to the red upon replacing the pyridyl binding group of 1 and 4 with
the extended phenyl pyridyl unit, likely because it reduces the electron withdrawing influence
of the pyridyl ring on the donor group. Generally, the dicyano acceptor gives rise to a more
red-shifted λmax than the pyridinium acceptor (with the exception of the pyridyl systems 1 and
4), and this difference is significant in 2 vs 5, but the band has a broader shape in the
pyridiniums and is also more intense (ε of up to 62300 M-1 cm-1, 12 to 43% higher than for
the comparable dicyano). Consequently, the pyridiniums absorb more strongly at long
wavelengths, their absorption edge is at lower energy and lower E0-0 values are found for
these systems. Compared to other dyes with the indolium cationic acceptor,[16] however,
λmax for 5 is over 100 nm lower and the ε lower by around 50%, illustrating the effect of
extended conjugation onto the larger indolium acceptor. The higher energy π-π* transition
varies significantly in strength relative to the ICT band. In the pyridyl (rather than phenyl
pyridyl) binder based systems 1 and 4, it is very weak, appearing only as a small bump, while

ACCEPTED MANUSCRIPT
in 3 and 6 (two phenyl pyridyl groups) its peak ε exceeds that of the ICT band; the situation is
intermediate (relative to ICT) for 2 and 5 which each have one phenyl pyridyl binding group.
Thus, it appears this peak is associated with the pyridyl or phenyl pyridyl units. Emission
spectra (Table 1, Figure S2) show red-shifts in λem from 1 to 2 and then 3 consistent with the
shifts in the absorption maxima, the pyridinium dyes however are non-emissive.
Table 1. Optical and electrochemical parameters of 1 to 6.
a
a
b
Dye λmax
/ nm M-1 cm-1
ɛ / 103
λem
/
D0/D+, Epa or
1/2 vs NHE /
D0/D-, Epc or
1/2 vs NHE /
E0-0
Computed
HOMO-LUMOc /
E
nm
E
/ V
V (∆E, mV)
1.42
V (∆E, mV)
-0.68
eV
445
491
475
449
478
471
41.7
46.8
23.6
45.3
62.3
33.8
626
673
642
2.27
2.07
2.21
2.19
1.99
2.10
2.30
2.12
2.23
2.11
1.90
2.04
1
2
3
4
5
6
1.32
-0.72
-0.71
1.34(100)
d
e
-
-
-0.63
d
-
1.26
-0.52 (120)f
-0.69
d
e
-
-
(a) Absorption and emission spectra recorded in CH3CN solution. (b) The HOMO-LUMO gap E0-0 estimated
from the absorption edge. (c) Lowest energy TD-DFT calculated UV-vis transition (HOMO-LUMO), see
experimental for details. (d) All three pyridinium dyes are non-emissive. (e) Dye oxidation peak cannot be
clearly distinguished from solvent oxidation. (f) Apparently quasi-reversible process overlapped with
irreversible peak at -0.68 V. Electrochemical data were obtained using 0.1 M NBu4BF4 in DMF as electrolyte,
with an analyte concentration of ca. 1 mM. Potentials referenced to internal Fc/Fc+ and converted to NHE using
Fc/Fc+ = 0.69 V vs NHE.

ACCEPTED MANUSCRIPT
Fig. 2. Absorption spectra of 1 to 6. (a) In acetonitrile solutions at 298 K. (b) Absorbed onto the NiO
films used for solar cell studies.
Fig. 2 (bottom) shows the absorption spectra of the dyes on NiO, by obtained by soaking NiO
films in the dye solutions for 20 hours. The spectra are broader than in solution, and only 2
and 5 have clearly defined absorbance peaks on NiO, at 487 and 463 nm respectively. These
blue shifts could indicate electronic communication between the dyes and the NiO surface,
although contribution from the NiO background is also possible. The absorbances obtained in
the films are generally rather low and follow a different pattern to the solution extinction

ACCEPTED MANUSCRIPT
coefficients. With the short, pyridyl anchored 1 and 4, stronger light absorption is seen for
dicyano-acceptor 1, despite a lower ε. Similarly, but more pronounced, the peak absorbance
for dicyano 2 on NiO is over 50% greater than for pyridinium 5, even though its ε is 25%
lower. For the bis-anchor dyes 3 and 6, there is little difference in light absorption. This
implies that in every case significantly less of the cationic dye is adsorbed on the NiO
surface, and the trend seems to be more dramatic for the dicationic species. This suggests it is
likely a consequence of repulsion of the cationic dyes by the positive surface charge
associated with NiO, as well as the larger size of the pyridinium architectures.
Cyclic Voltammetry: Electrochemical data are also summarised in Table 1 and were obtained
in DMF, rather than MeCN, as the compounds were not sufficiently soluble in MeCN at the
concentrations required for electrochemistry. In most cases, oxidation of the donor and
reduction of the acceptor appear irreversible on the timescale of these measurements,
although quasi-reversiblity is seen in the oxidation of 3, and reduction of 5 (Fig. S1, SI).
Moreover, for two of the three pyridinium compounds, the oxidation of the donor cannot be
clearly observed above solvent background. This means that only a limited interpretation can
be made of the data, but some trends are apparent. Firstly, in all cases the reduction potential
of the pyridinium dye is less negative than its nitrile counterpart, implying that it is a stronger
acceptor and electron transfer to it is easier. This is consistent with the smaller HOMO-
LUMO (E00) gaps estimated for the pyridiniums from the absorption edges of 4 to 6, though
not the blue shifted λmax values. Interestingly, though, for the only pyridinium with a clearly
observed oxidation potential (5), the oxidation potential is less positive than for any of the
nitrile systems. This is consistent with the low E00 observed for this compound. The nature of
the binder also seems to influence the potential of the HOMO (donor), with the highest
oxidation potential being seen for 1, where the electron deficient ring is directly connected to

ACCEPTED MANUSCRIPT
the aryl amine N. In all cases, the dye oxidation potentials are comfortably more positive than
the top of the NiO VB (0.54 V vs NHE), indicating that hole transfer to NiO is energetically
favourable – similarly, the reduction potentials of the dye acceptors are comfortably more
-
negative than that of the I-/I3 redox electrolyte.
DFT calculations: DFT and TD-DFT calculations were performed on 1 to 6 using the
Amsterdam Density Functional (ADF) software package, with the B3LYP functional and
DZP basis set.[22–24]. Acetonitrile solvent was added using COSMO. The trend in the
lowest energy TD-DFT calculated transitions (all HOMO‰LUMO) perfectly matches that of
the experimental E00 values within the individual dicyano (1 to 3) and pyridinium (4 to 6)
series (Table 1), and also mirrors experimental findings by showing that the pyridinium
HOMO‰LUMO transitions are in all cases lower energy than those of their dicyano
analogues. However, the computed lowering in transition energy for pyridiniums (0.19 to
0.22 eV) is greater than that estimated from the UV-vis and electrochemical measurements
(0.08 to 0.11 eV) – this is reflected in the in the position of the computed lowest energy
transitions which are always a few nanometres to the blue of the measured dye absorption
edge for the dicyano dyes (Fig. S3, SI). The computed spectra also echo experiment by
consistently finding higher oscillator strengths (fos) for the pyridinium dyes, which are shown
to have higher extinction coefficients.
Analysis of the orbital-to-orbital transitions (Tables S1 to S6, Figures S4 to S9, SI) indicates
that the two families of dyes have fundamentally similar electronic structures. In all cases, the
lowest energy (and usually strongest) transition involves charge transfer (CT) from a HOMO
based on the triarylamine unit, to a LUMO largely based on the thiophene bridge and dicyano
or pyridinium acceptor. The highest energy transitions are largely π‰π* in character.

ACCEPTED MANUSCRIPT
However, the structural variations lead to differences in the spatial distribution and energies
of the computed orbitals which are consistent with the observed spectroscopic properties
(vide supra) and solar cell performance (vide infra) of the dyes.
Firstly, in both the dicyano and pyridinium series, clear trends can be seen on moving from
the mono-pyridine (1, 4), to mono-phenylpyridine (2, 5) and then bis-phenylpyridine binding
units. In 1 and 4 (Tables S1, S4; Fig.s S4, S7), the HOMO levels extend onto the electron
deficient pyridine ring, lowering their energy and resulting in larger HOMO-LUMO gaps (by
at least 0.07 eV, Table 1) than the other dyes in each series. In 2 and 5 (Tables S2, S5; Fig.s
S5, S8), however, the phenyl spacer reduces the influence of the pyridine group on the
triarylamine donor and participation of the pyridine ring is absent not only from the HOMO,
but also the HOMO-1 and HOMO-2. Adding a second phenylpyridine binding unit can be
expected to have a detrimental effect on charge separation performance (as found
experimentally) because in both 3 and 6 (Tables S3, S6; Fig.s S6, S9) their presence
introduces new acceptor orbitals at accessible energies (LUMO+1/+2 for 3, LUMO+2 for 6)
that are expected to interact strongly with the NiO surface. The excited state HOMOs (Fig.s
S6, S9) are also located on the phenylpyridine binding groups. This will encourage
recombination, moreover in both dyes there are calculated CT transitions in the near UV from
the HOMO, to these orbitals –sending electrons in an undesired direction. This is consistent
with the experimentally observed, comparatively stronger absorption between 300 and 400
nm for 3 and 6.

ACCEPTED MANUSCRIPT
Fig. 4. Selected DFT calculated frontier orbitals of 2 and 5

ACCEPTED MANUSCRIPT
Secondly, analysis of the orbitals involved in the computed UV-vis transitions reveals likely
advantages for the pyridinium acceptor. For both acceptor systems, the LUMOs generally
involve a combination of the acceptor, and thiophene bridge (Fig. 4). For the pyridiniums,
this LUMO is lower energy (leading to transitions covering more of the visible spectrum),
and larger due to the increased size of the pyridinium, vs dicyano group. Thus, the pyridinium
is able to delocalise charge further away from the triarylamine donor and NiO surface. The
pyridiniums also show less participation of the acceptor in the HOMO and other donor
orbitals (HOMO-1/-2), thus, the pyridinium systems are likely to separate charge more
effectively and direct it away from the surface of the p-type substrate. Participation of the
relatively electron deficient pyridine/phenylpyridine units in the HOMO orbitals is also
suppressed in the pyridiniums, raising the HOMO energy and helping absorb more visible
light. Lastly, in 5 and 6 (vs 2 and 3), there is less tendency to transfer charge to the
phenylpyridines – in compound 5 the LUMO/+1/+2 are all based on the pyridinium and only
the LUMO+3 involves the phenylpyridine (Fig. S8) whereas in 2 the LUMO+2 is
phenylpyridine based (Fig. S5), and the transition to it is about three times stronger than the
equivalent in 2. A similar situation pertains for 6 vs 3 (See SI, Tables S3 and S6, Fig.s S6 and
S9).

ACCEPTED MANUSCRIPT
Fig. 5. DFT calculated HOMOs of the excited states of 2 and 5. It can be seen that in 5, less electron
density remains on the phenylpyridine unit, and more on the bridge and the acceptor, providing more
effective charge separation.
Calculation of the structures of the first excited states reflects these findings, through excited
state HOMOs for the pyridiniums that tend to distribute less electron density onto the
pyridine/phenylpyridine binding groups, and more onto the bridge and acceptor (Fig.s S4 to
S9). This is seen most clearly for the excited states of 5 vs 2 (Fig. 5). Computed ground-to-
excited state dipole moment changes (Table S7) shows that in every pyridinium/dicyano pair,
charge is moved further away from the donor (and thus the NiO surface) by the pyridinium.
Thus, electronically the pyridinium group appears to have the advantage over the dicyano for
light absorption and charge separation.
Photovoltaic performance
Table 2. Photovoltaic parameters of NiO p-DSSCs based on 1 to 6.
Dye
1
2
3
4
Loading / 10-10 mola
Jsc / mA cm-2
0.83
Voc / mV
50
FF (%)
43
η (%)
6.38
11.8
5.14
3.30
6.23
4.96
0.018
0.060
0.014
0.018
0.036
0.014
1.6
0.87
0.83
1.11
103
49
66
86
70
36
32
33
37
5
6
0.84
23
Electrolyte: 0.1 M I2, 1 M LiI in MeCN. Counter electrode: platinized FTO glass. a Loading per 0.25 cm2, 1 µM
thick film, average of four films.
-
Photovoltaic parameters of devices based on the six sensitizers on NiO (with I-/I3 ), under
AM 1.5 solar simulated light, are shown in Table 2 (for J-V curves see Fig. S10, SI) and
IPCE spectra are displayed in Fig. 6. In our work, compound 2 shows a slightly weaker
performance (η = 0.06%) than previously published in similar conditions (η = 0.093%),[15]

ACCEPTED MANUSCRIPT
due to a lower Jsc, but is still the best performing dye in this series. The next best performance
(η = 0.036%) comes from the other phenyl pyridyl, bis-acceptor structure 5, which has both
lower Jsc and lower Voc than 2, and weaker IPCE across the spectrum from 400 to 700 nm.
The obvious explanation is a lower loading of 5, indicated by the weaker absorption of the
NiO films prepared with this compound (Fig. 2) and confirmed in a loading study (Table 2,
Fig. S11). Otherwise, the higher extinction coefficient of 5 coupled with its similar redox
properties (and the better electronic structure for charge separation suggested by DFT, vide
supra) might be expected to yield a higher power conversion efficiency. Indeed, absorbed
photon to current efficiencies (APCEs, Fig. S12, SI) shows a very similar performance for 5
vs 2 across the visible range. This suggests that the performance of 5 in p-DSSCs is held back
by the ability to load dicationic dyes onto the positively charged NiO surface, rather than the
fundamental electronics and light absorption properties of the dye.
Fig. 6. IPCE spectra of p-DSSCs assembled with 1 to 6
The other dyes (1, 3, 4 and 6) produce overall efficiencies at best 50% of that of 5. This is
likely because addition of a second anchoring group (as in 3 and 6) in place of an acceptor

ACCEPTED MANUSCRIPT
sacrifices light absorption, as seen in the UV-vis spectra in solution and on the NiO surface
and reflected in the IPCE spectra. In addition, the DFT calculations suggest that the
additional phenylpyridine leads to these anchoring groups acting as acceptors, close to and
interacting strongly with the NiO surface. The same absorption profile argument, and
weakened donor capability in the triphenylamine, can be used to explain the weak
performance of 1 and 4 – note that 4 also has the lowest loading of any of the dyes. In
addition, these systems all show lower VOC than seen for 2 or 5, which could imply that these
dyes, that are either shorter or have fewer/less optimally positioned electron acceptor groups,
do less to mitigate recombination than 2 or 5 – again consistent with the DFT calculated
electronic structures. It is also consistent with the prior observation that extension of 2 with a
further phenyl spacer increases VOC by ca. 20 mV.[15] Lastly the pyridinium systems 4 and
6, despite the clearly lower dye loading of 4, yield higher VOCs (and very similar JSCs) to their
dicyano counterparts 1 and 3, although η values are identical due to lower fill factors. IPCEs
for 4 are slightly higher than for 1, and for 6 and 3 are very similar. Conversion to APCE
(Fig. S12) shows a clear advantage for 4 over 1 across the visible range, and for 6 over 3 at
longer wavelengths. Thus, the overall picture from the solar cell measurements suggests that
the simple pyridinium acceptor performs just as well, or better than the widely used dicyano
group in p-DSSCs, but that in the specific example here, inadequate dye loading (around 50%
lower for bis-acceptor, dicationic systems vs dicyano analogues) holds back performance. As
the difference in loading is much smaller for the mono-acceptor, mono-cationic 6 compared
to its dicyano analogue, loadings would likely improve in systems with carboxylate binders
where deprotonation can mitigate repulsion between positively charged dyes, and the
positively charged NiO surface.
Conclusion