The role of para-alkyl substituents on meso-phenyl porphyrin sensitised TiO2 solar cells: Control of the eTiO2/electrolyte + recombination reaction

Article abstract of DOI:10.1039/b717081e

We aim to investigate the effect of adding hydrophobic alkyl chains substituents to unsymmetrical free base tetra-phenyl porphyrins used for the preparation of dye sensitised solar cells (DSSC). We have used two different unsymmetrical meso-tetraphenyl substituted free base porphyrins attending to two objectives: (1) to observe how the substitution of three para positions of the meso-phenyl groups with hydrophobic alkyl chains influences the formation of molecular aggregates onto the semiconductor nanoparticles and (b) to deduce the influence that the substitution exerts over the e_TiO2/ electrolyte⁺ recombination reaction in operating devices. To achieve these goals we have focussed on the study of the electron transfer processes that take place at the different interfaces of the photovoltaic device using electrochemistry, steady-state and time resolved spectroscopic techniques. The Royal Society of Chemistry 2008.

Full text of DOI:10.1039/b717081e

www.rsc.org/materials
Volumeꢀ18ꢀꢀ|ꢀꢀNumberꢀ14ꢀꢀ|ꢀꢀ14ꢀAprilꢀ2008ꢀꢀ|ꢀꢀPagesꢀ1605–1712
ISSNꢀ0959-9428
PAPER
EmilioꢀPalomaresꢀet al.
Theꢀroleꢀofꢀpara-alkylꢀsubstituentsꢀonꢀ
meso-phenylꢀporphyrinꢀsensitisedꢀ
HIGHLIGHT
TiO ꢀsolarꢀcells:ꢀcontrolꢀofꢀtheꢀeTiO2/
HeꢀTianꢀandꢀYanliꢀFeng
2
+
electrolyte ꢀrecombinationꢀreaction
Nextꢀstepꢀofꢀphotochromicꢀswitches?

PAPER
www.rsc.org/materials | Journal of Materials Chemistry
The role of para-alkyl substituents on meso-phenyl porphyrin sensitised TiO2
solar cells: control of the eTiO2/electrolyte recombination reaction
+
a
a
a
Amparo Forneli, Miquel Planells, Maria Angeles Sarmentero, Eugenia Martinez-Ferrero,
a
b
ac
ac
Brian C. O’Regan, Pablo Ballester* and Emilio Palomares*
Received 5th November 2007, Accepted 29th January 2008
First published as an Advance Article on the web 19th February 2008
DOI: 10.1039/b717081e
We aim to investigate the effect of adding hydrophobic alkyl chains substituents to unsymmetrical free
base tetra-phenyl porphyrins used for the preparation of dye sensitised solar cells (DSSC). We have
used two different unsymmetrical meso-tetraphenyl substituted free base porphyrins attending to two
objectives: (1) to observe how the substitution of three para positions of the meso-phenyl groups with
hydrophobic alkyl chains influences the formation of molecular aggregates onto the semiconductor
+
nanoparticles and (b) to deduce the influence that the substitution exerts over the eTiO2/electrolyte
recombination reaction in operating devices. To achieve these goals we have focussed on the study of
the electron transfer processes that take place at the different interfaces of the photovoltaic device using
electrochemistry, steady-state and time resolved spectroscopic techniques.
Introduction
the use of alkyl chain substituents in the porphyrin structure.
Other authors, using ruthenium bis-thiocyanate polypyridyl
dyes, have already used such a strategy in order to increase not
only the device efficiency but also the stability of the photovol-
1
Since the seminal paper of O’Regan and Gr a¨ tzel, about the use
of mesoporous semiconductor metal oxides sensitised with
a monolayer of photo-active molecules to convert solar irradia-
tion into electrical power, there has been an intensive effort
towards the synthesis of efficient dyes that could push the effi-
ciency above 10%. In fact, this has already been achieved by
8
taic cells under illumination. However, there is still some contro-
versy about the effect of the hydrophobic alkyl chains on the
control of the electron recombination dynamics between the
photo-injected electrons at the semiconductor metal oxide and
2
several groups using ruthenium dyes, where the optical and
9
the electrolyte. In principle, porphyrins, due to their tendency
electrochemical properties have been designed to achieve optimal
electron injection into the semiconductor, slow back-electron
transfer kinetics and efficient regeneration of the light-induced
oxidized molecule by the solution electrolyte.
to form aggregates, are ideal candidates to study whether the
presence of hydrophobic alkyl chain substituents in their mole-
cular structure not only inhibits or reduces the formation of
molecular aggregates, but also creates a compact hydrophobic
layer at the nanoparticle surface that may increase the distance
The electron transfer (ET) reactions taking place at the
different interfaces of the dye sensitized solar cell device have
an extraordinary resemblance to the ET reactions that occur at
the photosynthetic reaction centre (PSC) of several bacteria
ꢀ
ꢀ
with the I /I redox electrolyte and, therefore, slow the back
3
electron transfer reaction. Hence, we decided to use such
‘‘biological like’’ dyes for our study.
3
and plants. In fact, an extensive number of scientists have
As molecular sensitizers we synthesised the unsymmetrical
porphyrins illustrated in Fig. 1.
synthesized dyads and triads based on porphyrins, or even
more complex structures such as dendrimers, to mimic, in
4
solution, the ET processes at the PSC. Only recently, the use
Experimental
of such biological dyes, i.e. Zn-porphyrins, in molecular photo-
voltaic devices have given impressive light-to-energy efficiencies
5
when irradiated at 1 sun (100 mW cm ) although free-base
Materials and reagents
ꢀ2
All reagents were purchased from Aldrich and the solvents from
SDS. All were used as received without further purification,
except pyrrole which was distilled under vacuum prior to use.
porphyrins have only reached efficiencies higher than 5% but
ꢀ
2 6
at much lower light intensities (1.5 mW cm ). Several authors
have assigned the low performance of porphyrin based devices
to the dye desorption process and to the formation of molecular
6a,7
Synthesis of porphyrins
aggregates onto the surface of the nanoparticles. To avoid or
limit the aggregation inconvenience we have decided to explore
The methodology for the synthesis of unsymmetrical free base
10
porphyrins has been described by Lindsey et al. Scheme 1 illus-
trates the synthetic steps for the preparation of FbP1 and FbP2.
a
Institute of Chemical Research of Catalonia (ICIQ), Avda. Pa ı¨s os
Catalans, 16 Tarragona, Spain. E-mail: epalomares@iciq.es; pballester@
iciq.es; Fax: +34 977 920 224; Tel: +34 977 920 241
b
Synthesis of 4-pentylbenzaldehyde
Centre for Electronic Materials and Devices, Dept. Chemistry. Imperial
College of London, London, UK, SW7 2AZ
4-Pentylbencene (25 mL; 0.145 mol) and 20.4 g of hexamethy-
lenetetramine (0.145 mol) were added in a round bottom flask
c
Catalan Institution for Research and Advanced Studies (ICREA), Spain
1
652 | J. Mater. Chem., 2008, 18, 1652–1658
This journal is ª The Royal Society of Chemistry 2008

under vacuum. Finally, the product was purified by distillation
ꢁ
of the residue under diminished pressure (130 C, 1 mbar). The
compound was isolated as a colourless liquid in 39% yield.
1
3 H
H-NMR (400 MHz, CDCl ) d : 9.96 (1 H, s); 7.79 (2 H, d,
J ¼ 8.0 Hz); 7.32 (2 H, d, J ¼ 8.0 Hz); 2.67 (2 H, t, J ¼ 8.0
Hz); 1.64 (2 H, m); 1.33 (2 H, m); 1.32 (2 H, m); 0.89 (3 H, t,
J ¼ 6.8 Hz).
13
3 C
C-NMR (400 MHz, CDCl ) d : 191.93; 150.45; 134.43;
1
29.86; 129.07; 36.17; 31.42; 30.75; 22.48; 13.97.
Synthesis of 5-(4-carboxyphenyl)-10,15,20-trisphenylporphyrin
FbP1) and 5-(4-carboxyphenyl)-10,15,20-tris(4-
(
pentylphenyl)porphyrin (FbP2)
Methyl 5-formylbenzoate (1 g; 6.1 mmol), the appropriate
aldehyde (either 0.95 mL of benzaldehyde (9.1 mmol) or 1.45 g
of 4-pentylbenzaldehyde (9.1 mmol)) and 0.95 mL of freshly
distilled pyrrole (15.2 mmol) were added to a solution consisting
of 530 mL of dichloromethane and 4 mL of ethanol. The result-
ing solution was stirred under argon for 10 min. The reaction
mixture was protected from light and 0.69 mL of boron trifluor-
ide dietherate (6.1 mmol) was added. The solution was stirred for
Fig. 1 Molecular structures of FbP1 and FbP2.
6
0 min. Subsequently, 3.11 g of 2,3-dicyano-5,6-dichloropara-
benzoquinone (15.2 mmol) were added and the reaction was
stirred for another 90 min. After, the reaction was quenched
by the addition of 3 mL triethylamine. The solvent was removed
under vacuum and the crude was purified by column chromato-
graphy on silica gel using a 1 : 1 hexane–dichloromethane solvent
mixture. The mono-carboxymethylporphyrins were isolated as
purple solids in 10% (FbP1-COOMe) and 7.5% (FbP2-COOMe)
yields. Finally, the ester group was hydrolyzed to the carboxylic
acid anchoring group. In brief, 90 mg of mono-carboxymethyl-
porphyrin FbP1-COOMe (0.14 mmol) or FbP2-COOMe
(
0.1 mmol) were added to a mixture of 8 mL of THF (tetrahydro-
furan) and 2 mL of 2 M aqueous KOH. The reaction was stirred
ꢁ
for 14 h at 90 C. After, the THF was removed under vacuum
and 80 mL of water were added to the resulting concentrate.
The pH of the solution was adjusted to pH 4.0–5.0 using
a solution of HCl (2 M), followed by the addition of 100 mL
of chloroform. The organic layer was separated, washed with
water, dried over sodium sulfate and filtered. The organic solvent
was removed under vacuum, giving the pure free base monocar-
boxylic acid porphyrins as purple solids with 95% yield in both
cases. The porphyrins were dried overnight before being used
in the construction of the devices.
1
FbP1: H-NMR (400 MHz; CHCl
) d
3 H
: 8.77 (8 H, m); 8.37
(
2 H, d, J ¼ 7.9 Hz); 8.23 (2 H, d, J ¼ 7.9 Hz); 8.14 (6 H, m);
7
.70 (9 H, m); ꢀ2.89 (2 H, br s).
+
ESI/MS (m/z): [M + H] ¼ 659.2442 (calc. M for C45
30 4 2
H N O
¼
658.2369).
Scheme 1 Synthesis route to FbP1 and FbP2.
ꢀ1
FT-IR (cm ): 2952, 2922(w), 2852(w), 1684, 1606, 1556, 1422,
1
288, 797, 722, 700.
1
to 250 mL of TFA (trifluoroacetic acid). The mixture was heated
ꢁ
FbP2: H-NMR (400 MHz, CDCl ) d : 8.92 (2 H, d, J ¼ 4.8
3
H
to 100 C for 12 h. Then, TFA was removed under vacuum and
Hz); 8.89 (4 H, s); 8.80 (2 H, d, J ¼ 4.8 Hz); 8.51 (2 H, d,
J ¼ 8.0 Hz); 8.36 (2 H, d, J ¼ 8.0 Hz); 8.13 (6 H, d, J ¼ 8.0 Hz);
7.59 (6 H, d, J ¼ 8.0 Hz); 2.98 (6 H, t, J ¼ 8.0 Hz); 1.95 (6 H,
m); 1.56 (12 H, m); 1.08 (9 H, t, J ¼ 6.8 Hz); ꢀ2.73 (2 H, br s).
2
50 mL of ice-water were added to the concentrate. Under
stirring, sodium carbonate was added to the solution until
a pH value ꢂ8 was achieved. Diethyl ether was added and the
organic layer was separated and washed with water. The organic
layer was dried over sodium sulfate, filtered and concentrated
ꢀ
MALDI/MS (m/z): [M ꢀ H] ¼ 867.4623 (calc. M for
C
60
H
60
4
N O
2
¼ 868.4716).
This journal is ª The Royal Society of Chemistry 2008
J. Mater. Chem., 2008, 18, 1652–1658 | 1653

ꢀ
1
FT-IR (cm ): 2953, 2922(s), 2852(s), 1688, 1606, 1420, 1278,
178(s), 965(s), 797, 729.
Scanning electron microscopy (SEM) images were obtained
using a JEOL JSM 6400 microscope. Before the measurements,
the samples were covered with a thin layer of gold to increase
the conductivity for superior image quality. Molecular Imaging
Pico SPMII apparatus was used to obtain atomic force micros-
copy (AFM) images. Transmission electron microscopy (TEM)
measurements were performed with a JEOL 1011 microscope.
The surface analysis was carried out with an Autosorb 1-MP
Quantachrome apparatus and the X-ray power diffraction
measurements were acquired with a Bruker Siemens Smart
CCD diffractometer.
1
Titanium nanoparticles synthesis
TiO anatase nanoparticles were prepared as previously reported
2
11
by Palomares et al. Titanium isopropoxide (40 mL, 0.13 mol)
was added to glacial acetic acid (9.12 g) under stirring. The resul-
ting solutionwas cooled in anice-bath and240 mLof 0.1 MHNO
3
were added under vigorous stirring. The mixture was heated at 80
ꢁ
C for 8 h. Then it was allowed to cool to room temperature and
filtered through a 0.45 mm filter. The filtrate was diluted to 5%
weight of TiO by the addition of water. Then solution was placed
2
ꢁ
Electrochemical and device characterization
into a high-pressure autoclave at 220 C for 12 h. After centrifuga-
tion, the aqueous phase was decanted and the solid residue was
rinsed with ethanol twice. The aggregates were broken up using
an ultrasonic bath and the solvent was removed. The nanopar-
ticles were dissolved in a solution of terpineol containing 5% of
ethylcellulose and the paste was homogenised by ball milling.
The photoelectrochemical measurement were carried out with
a ORIEL 150 W xenon light source equipped with the correct
set of filters to achieve the solar spectrum AM 1.5 G. The light
ꢀ2
intensity was adjusted to 100 mW cm , the equivalent of 1
sun, using a calibrated Si photodiode. The applied potential
and cell current were measured with a Keithley model 2600
digital source meter. The current to voltage (I–V curve) was
measured automatically with home-built Labview software.
The charge extraction and photovoltage decay were carried out
The TiO
powder analysis confirming the presence of 95% of anatase and
% of brookite TiO phase.
2
nanoparticle crystallinity was characterized by X-ray
5
2
12
Alumina nanoparticles synthesis
with a system set-up as previously reported by O’Regan et al.
using a CHI Instruments model 600B potentiostat-galvanostat
which was used for the cyclic voltammetry experiments too.
Al
2 3
O was prepared using a commercial water based solution of
Al O nanoparticles with uniformdiameter of20nm (Alfa-Aesar).
2
3
Device preparation
Spectroscopic characterization techniques
Nanocrystalline TiO
2
(19 nm particle size) was deposited onto
The UV-visible spectra were recorded using a Shimadzu
UV-1700 spectrophotometer. When the absorption spectrum
was recorded for dye sensitised films we used the same film
without dye to measure the instrument blank and ensure that
the light scattering effects had been minimised. For the steady-
state fluorescence emission spectra we used an Aminco-Bowman
Series 2 fluorimeter with adequate support for solid samples.
Time correlated single photon counting (TCSPC) experiments
were carried out with a Lifespec-red picosecond fluorescence
lifetime spectrophotometer from Edinburgh Instruments equip-
ped with lasers as excitation sources. The instrument response
was always shorter than 300 ps measured at full width half
maximum (FWHM).
a FTO conducting glass substrate (Hafordt glass TEC 15 U
ꢀ2
cm resistance) by the well known doctor blade technique.
The resulting thickness was 4 mm and the device active area
2
was 1 cm . The resulting electrodes were gradually heated under
ꢁ
ꢁ
ꢁ
airflow at 325 C for 5 min, 375 C for 5 min, 450 C for 15 min
ꢁ
and 500 C for 15 min. The heated electrodes were soaked into
ꢁ
TiCl
4
aqueous solution 0.04 M at 70 C for 15 min and then
ꢁ
washed with ethanol. Electrodes were heated again at 500 C
for 30 min and cooled before dye adsorption. The platinised
ꢀ3
counter electrode was made applying a drop of 5 ꢃ 10 M of
PtCl in 2-propanol dry solution and spreading onto the
H
2
6
conducting glass substrate (FTO). The coated glass was heated
ꢁ
under airflow at 390 C for 15 min.
The transient absorbance experiments were recorded using
A solution (5 ꢃ 10ꢀ M) of FbP1 or FbP2 in THF was
prepared and the film was immersed in the solution at room
temperature until desired absorbance values were reached. The
sensitised electrodes were assembled by sandwiching the working
and the counter electrode using a thin thermoplastic (Surlyn)
4
1
1
a home-built system as reported before.
1
13
H- and C-NMR spectra were recorded with a Bruker
Avance 400 Ultrashield NMR spectrometer, equipped with
a BBI (broadband inverse) probe for proton spectra and
a BBO (broadband obverse) probe for carbon spectra.
ꢁ
frame that melts at 100 C.
FT-IR spectra were recorded using a FT-IR ThermoNicolet
The liquid electrolyte was elaborated using 0.6 M of
5
700 spectrometer with solid-state samples.
1
-propyl-2,3-dimethylimidazolium iodide (DMPII), 0.025 M
MS spectra were recorded with a Waters LCT Premier, which
of lithium iodide (LiI), 0.04 M of iodine (I ) in a mixture of
2
operates with a Bruker Autoflex instrument. The ionization tech-
nique was matrix assisted laser desorption/ionization (MALDI)
and the mass was measured by the time of flight (TOF) method.
acetonitrile and valeronitrile (85 : 15). This electrolyte is denoted
AF4.
The cells were filled with the electrolyte through a hole
previously made in the back of a platinised counter electrode.
Then, the hole was sealed with a thermoplastic polymer and
a cover slide glass. Finally, a drop of silver conductive paint
was spread at the electrode contacts to increase the conductivity.
Materials characterization
The mesoporous nanocrystalline titanium dioxide films were
characterised by scanning probe microscopy techniques.
1
654 | J. Mater. Chem., 2008, 18, 1652–1658
This journal is ª The Royal Society of Chemistry 2008

ꢀ4
for the films after immersion on a 5 ꢃ 10 M solution of each
Results and discussion
dye.
UV-visible measurements
As can be seen in Fig. 3, both films reached different maximum
absorbance at 520 nm under the same conditions. In the case of
The visible spectra of FbP2 in solution and on a TiO film are
2
FbP1 (e520 nm ¼ 17250 Mꢀ cm ) the film reached a maximum
1
ꢀ1
illustrated in Fig. 2. For the FbP2 dye, we found that after 2
min of sensitisation, a 4 mm thick mesoporous sensitised film
shows an ‘‘optimised’’ (less aggregated) visible spectrum with
a sharp Soret band (lSORET ¼ 415 nm). The broadening
observed between the solid state and solution measurements,
as reported before, is mainly due to the adsorption of the
molecules onto the mesoporous film and not to the formation
absorbance of 0.8 units while for FbP2 (with the same e520 nm
)
the maximum absorbance was 0.6 units. This can be understood
6a
in terms of porphyrin spatial distribution if we take into
account that the presence of alkyl chains in FbP2 implies that
the dye occupies a higher surface are at the nanoparticle and
also avoids the adsorption of other molecules nearby, preventing
the formation of aggregates. Moreover, we also found that the
presence of the alkyl chains does not have a large influence
over the dye adsorption kinetics. The rate constant for adsorp-
tion was studied fitting the experimental adsorption data
13
of typical aggregates.
In order to compare the adsorption kinetics of both porphy-
rins, we monitored the change in absorbance at l ¼ 520 nm of
two identical 4 mm thick TiO films until the absorbance reached
2
14
following the pseudo-first order Lagergren’s equation (eqn (1))
a plateau. Fig. 3 shows the adsorption kinetics comparison
A ¼ Amax(1ꢀexp(ꢀKadst)) (1)
where Amax represents the absorbance maxima and Kads is the
rate constant of dye adsorption. For both FbP1 and FbP2 Kads
ꢀ1
was found to be 0.65 min . Taking into account these data
ꢀ1
2
and the calculated surface area of the TiO
2
films, 99.4 m g ,
we have also estimated the porphyrin density in the films: 2.1
2
FbP1 nm molecule and 2.8 FbP2 nm molecule .
ꢀ1
2
ꢀ1
Hence, we can assume that the presence of alkyl groups on the
molecule affects the formation of porphyrin aggregates under
controlled dye adsorption conditions but does not have
a significant influence over the dye adsorption kinetics.
Dye solar cell characterization
We focused our attention on the device performance of FbP2
using for comparison purposes FbP1. We have already demon-
strated that the presence of pentyl groups on the porphyrin
ring does not really have an influence over the dye adsorption
kinetics. However, as can be seen in Fig. 4 the cell characteristics
are different.
Fig. 2 UV-Vis spectra of FbP1 (black) and FbP2 (grey) on a transparent
4
2
mm thick mesoporous TiO film and in THF solution (inset).
In both devices the dye adsorption was controlled to achieve
the same film absorbance (0.6 a.u. at l ¼ 520 nm), which explains
the similar photocurrent in both devices despite the presence of
alkyl groups in FbP2, but more interesting is the fact that the
voltage increased too. In spite of this result we decided to study
further the different charge transfer processes taking place at the
device which have been previously identified in the literature as:
(
i) electron injection from the dye excited state into the semicon-
ductor conduction band, (ii) electron recombination between the
+
photo-injected electrons and the oxidised dye (eTiO2/dye ), (iii)
+
dye regeneration by the redox active electrolyte (dye /electrolyte)
and (iv) electron recombination between the photo-injected
+
electrons and the electrolyte (eTiO2/electrolyte ).
Time correlated single photon counting measurements. The
measurement of the excited-state emission lifetime decay of the
sensitised films is a straightforward measurement to estimate
the electron injection yield of a dye adsorbed onto a semicon-
15
ductor. As a control sample, a mesoporous wide band-gap metal
oxide such as Al is used to prevent electron injection from the
Fig. 3 Absorbance kinetics for FbP1 (dots) and FbP2 (squares) on
a transparent 4 mm thick mesoporous TiO film at room temperature.
2 3
O
2
excited state since the conduction band of the metal oxide is above
This journal is ª The Royal Society of Chemistry 2008
J. Mater. Chem., 2008, 18, 1652–1658 | 1655

Kohlrausch–Williams–Watts (KWW) function (eqn (2)). As an
example FbP1 sensitised films can be fitted with a monoexponen-
tial time of t ¼ 2.95 ns for Al O and a t ¼ 0.67 ns for TiO while
2
3
2
FbP2 results in t ¼ 3.12 ns for Al
2
O
3
and t ¼ 1.10 ns for TiO
2
a
decay f exp[ꢀ(t/t) ]
(2)
16
Recent work of Koops and Durrant, using DSSC sensitised
with ruthenium polypyridyl complexes, proposes that the fitting
to the KWW function is consistent with the presence of inhomo-
geneous broadening of the energetics of electron injection.
Moreover, in the case of DSSC sensitised with either FbP1 or
FbP2 the initial decay amplitude is considerably higher when
2
compared to the TiO sensitised films which is indicative that,
in complete devices with redox active electrolyte, the electron
injection is much slower in good agreement with the case of
ruthenium bis-thiocyanate polypyridyl complexes studied
17
elsewhere. Following the methodology described Koops and
Durrant, we estimated that the injection half-time for FbP1
and FbP2 occurs in less than 100 ps.
Fig. 4 Current vs. voltage characteristics for 4 mm transparent thick
2
DSSC devices. The cell active area was 1 cm and the electrolyte was AF4.
Laser transient absorption spectroscopy (L-TAS). We utilized
L-TAS as a technique that allows us to carry out the measure-
the LUMO of the molecule. As an example, Fig. 5 shows the
ments of the electron recombination dynamics for 4 mm TiO
2
emission lifetime measurements for FbP1 adsorbed onto nano-
films sensitized with either FbP1 or FbP2 in the presence or
ꢀ
crystalline mesoporous films of Al
2
O
3
and TiO
2
. Both films
absence of redox active I
3
/I electrolyte. The signal decay after
were sensitized to achieve identical optical absorption (0.6 a.u.
at l ¼ 520 nm). Under the same measurement conditions (the
acquisition time was 900 s) FbP1 and FbP2 show almost compa-
rable kinetics. In fact, we would like to note that both on Al O
the laser pulse is assigned to charge recombination of the dye
cation with the electrons in the trap/conduction band states of
18
the TiO semiconductor as reported before (Fig. 6).
2
2
3
As expected, all recombination decay dynamics can be fitted to
eqn (2) as a consequence of their dependence on the electron
transparent mesoporous films or in tetrahydrofuran solvent the
kinetics were fitted to identical monoexponential decay.
trapping/detrapping process at the nanocrystalline TiO particles
2
Further analysis of the emission dynamics reveals that the
decays can be also fitted using an extension of the
with a ¼ 0.42 for both FbP1 and FbP2. The estimated recombi-
nation half time for both porphyrins was in the range of 100–150
ms. From these data it seems difficult to assign the slight
differences in electron recombination dynamics between the
Fig. 5 Time correlated single photon counting measurements for FbP1
Fig. 6 Electron recombination dynamics for a 4 mm thick dye sensitised
on Al
6). The lex was 635 nm and the lem was 720 nm. Acquisition time
was 900 s. The curve fits were obtained using eqn (2).
2
O
3
(C), TiO
2
(,) and a 4 mm thick DSSC with AF4 electrolyte
2
mesoporous TiO film. The sensitized film optical absorbance at the laser
(
excitation wavelength (lex ¼ 525 nm) was 0.3 a.u. for each film. The
probe wavelength for the measurement was fixed at lprobe ¼ 715 nm.
1
656 | J. Mater. Chem., 2008, 18, 1652–1658
This journal is ª The Royal Society of Chemistry 2008

Fig. 7 The electron recombination dynamics for two 4 mm thick DSSCs.
The sensitized film optical absorbance at the laser excitation wavelength
Fig. 8 Charge density as a function of Voc for 4 mm DSSCs sensitised
with FbP1 and FbP2. The electrolyte was AF4. Measurements performed
at 1 sun.
(
lex ¼ 525 nm) was 0.3 a.u. The probe wavelength for the measurements
was lprobe ¼ 715 nm.
dyes and the TiO
2
to the presence of the alkyl chains. In spite of
dynamics between different devices the charge density on both
cells must be equal. As illustrated in Fig. 8, in our case, this condi-
tion is achieved when the cell voltage is 415 mV.
this, we decided to study the electron recombination dynamics in
the presence of the redox active electrolyte.
As illustrated in Fig. 7, in the presence of a redox active
electrolyte the electron recombination kinetics become biphasic.
As can be seen in Fig. 8 for DSSCs sensitised with both FbP1
and FbP2 the experimental points emerge along the same curve.
Note, as indicated on the figure, that the cells have different Voc
when illuminated at 1 sun as expected from the results illustrated
before (Fig. 4). The differences in voltage may be due to: (a) a shift
19
Previous work demonstrated that the second phase observed in
the presence of electrolyte arises from slower formation of
by-products as a result of the regeneration of the dye ground
ꢀ
state by subsequent electron transfer from the I /I electrolyte
3
2
on the TiO conduction band with respect to the electrolyte poten-
+
after the electron injection from the dye into the semiconductor
conduction band. The fastest component of the optical transient
is assigned to the dye cation, which is rapidly regenerated by the
electrolyte, and therefore, its lifetime becomes shorter as can be
confirmed when compared with the kinetics in Fig. 6, where the
tial or (b) differences in the eTiO2/electrolyte recombination reac-
tion. The former hypothesis can be discarded because a change in
the TiO conduction band position also implies a shift in the cell
2
19
charge density as demonstrated previously by other authors.
In our case, the hydrophobic pentyl group on FbP2 does not shift
ꢀ4
signal reaches a plateau at DO.D. of 2 ꢃ 10 units while in Fig. 7
the decay has almost disappeared at times shorter than 1 ms.
Attending to the results detailed above, we can conclude that
the curve, thus, the differences in voltage must be due to the
+
increase in the eTiO2/electrolyte recombination reaction in FbP1.
Fig. 9 shows the Voc decay for both devices measured at the
same charge density (Voc ¼ 415 mV). It is noticeable that the
presence of the pentyl groups on the free-base tetra-phenyl
porphyrin (FbP2) results in slower recombination dynamics
in both devices the electron regeneration dynamics compete
+
efficiently with the TiO /dye electron recombination process
2
under the cell operating conditions and the presence of hydro-
phobic pentyl groups on the porphyrin ring does not affect
+
2
between the photo-injected electrons at the TiO and the AF4
significantly the kinetic competition between the TiO
ꢀ
2
/dye
electrolyte. We would like to mention that although the measure-
ments were performed on sensitised films with the same
absorbance (0.6 a.u. at l ¼ 520 nm) and, thus, we could expect
that the lack of complete coverage would also play an important
role in the accelerated recombination dynamics, experiments
performed using complete full-monolayer coverage, according
to Fig. 3, showed identical performance.
+
recombination dynamics and the dye /I regeneration reaction.
Transient photovoltage measurements. Recently, several
authors have employed transient photovoltage measurements,
also known as Voc decays, to evaluate the electron recombination
2
dynamics between the photo-injected electrons at the TiO and the
20
oxidised electrolyte. For DSSC employing liquid redox electro-
lytes, such dynamics generally occur on the millisecond time scale
and, therefore, no longer compete with the regeneration reaction
studied above. On the other hand, it is worth noting that Voc
decays are measurements strongly dependent on the accumulated
charge at the semiconductor (charge density), and hence to obtain
Conclusions
It can be concluded from our studies that the presence of
hydrophobic alkyl chains on the molecular structures of porphy-
rins decreases the recombination between the photo-injected
+
a fair comparison of the eTiO2/electrolyte recombination
electrons at the TiO
2
nanoparticles and the electrolyte. However,
This journal is ª The Royal Society of Chemistry 2008
J. Mater. Chem., 2008, 18, 1652–1658 | 1657

Tecnolog ´ı a (CTQ2005-08989C01-C02/BQU and CSD2006-
003), Generalitat de Catalunya DURSI (2005SGR00108) and
0
ICIQ Foundation for financial support.
References
1
2
B. O’Regan and M. Gr a¨ tzel, Nature, 1998, 353, 737.
(a) M. Gr a¨ tzel, Inorg. Chem., 2005, 44, 6841; (b) L. Han, N. Koide,
Y. Chib, A. Islam, R. Komiya, N. Fuke, A. Fukui and
R. Yamakada, Appl. Phys. Lett., 2005, 86, 213501;
(
c) N. Robertson, Angew. Chem., Int. Ed., 2006, 45, 2338.
(a) J. R. Durrant, S. A. Haque and E. Palomares, Chem. Commun.,
006, 3279; (b) D. Gust, T. A. Moore and A. L. Moore, Acc.
3
4
5
2
Chem. Res., 2004, 34, 40; (c) H. Imahori, Y. Mori and Y. Matano,
J. Photochem. Photobiol., C: Photochem. Rev., 2003, 4, 51.
(a) M. S. Choi, T. Yamazaki, I. Yamazaki and T. Aida, Angew.
Chem., Int. Ed., 2004, 43, 150; (b) S. J. Lee and J. T. Hupp, Coord.
Chem. Rev., 2006, 250, 1710; (c) F. R. Kelley, R. H. Goldsmith and
M. R. Wasielewski, J. Am. Chem. Soc., 2007, 129, 6384.
(a) W. M. Campbell, K. W. Jolley, P. Wagner, K. Wagner,
P.
J.
Walsh,
Ke.
C.
Gordon,
L.
Schmidt-Mende,
Md. K. Nazeeruddin, Q. Wang, M. Gr a¨ tzel and D. L. Officer,
J. Phys. Chem. C, 2007, 111, 11760; (b) S. Eu, S. Hayashi,
T. Umeyama, A. Oguro, M. Kawasaki, N. Kadota, Y. Matano and
H. Imahori, J. Phys. Chem. C, 2007, 111, 3528; (c) J. Rochford,
D. Chu, A. Hagfeldt and E. Galoppini, J. Am. Chem. Soc., 2007,
129, 4655; (d) J. R. Stromberg, A. Marton, H. L. Kee, C. Kirmaier,
J. R. Diers, C. Muthiah, M. Taniguchi, J. S. Lindsey, D. F. Bocian,
G. J. Meyer and D. Holten, J. Phys. Chem. C, 2007, 111, 15464;
Fig. 9 Photovoltage transients for two 4 mm DSSCs sensitised with
FbP1 and FbP2. The electrolyte was AF4.
their presence does not have a severe influence on the adsorption
kinetics, and it suggests that they have little influence over the
electron recombination between the photo-injected electrons at
(
e) M. Tanaka, S. Hayashi, S. Eu, T. Umeyama, Y. Matano and
H. Imahori, Chem. Commun., 2007, 2069.
6 (a) S. Cherian and C. C. Wamser, J. Phys. Chem. B, 2000, 104, 3624;
b) F. Odobel, E. Blart, M. Lagree, M. Villieras, H. Boujtita, N. El
Murr, S. Camaroni and C. A. Bignozzi, J. Mater. Chem., 2003, 13,
02; (c) J. Jasieniak, M. Johnston and E. R. Waclawik, J. Phys.
(
the TiO
2
nanoparticles and the oxidized porphyrin. This is in
good agreement with the ‘‘distance-dependence’’ electron recom-
5
21
bination theory, if we consider that the highest occupied molec-
ular orbital (HOMO) in the porphyrins is delocalized on the
aromatic ring, which implies that the distance between the
Chem. B, 2004, 108, 12962.
7
8
A. Kay and M. Gr a¨ tzel, J. Phys. Chem., 1993, 97, 6272.
(a) J. E. Kroeze, N. Hirata, S. Koops, Md. K. Nazeeruddin,
L. Schmidt-Mende, M. Gr a¨ tzel and J. R. Durrant, J. Am. Chem.
Soc., 2006, 128, 16376; (b) L. Schmidt-Mende, J. E. Kroeze,
J. R. Durrant, Md. K. Nazeeruddin and M. Gr a¨ tzel, Nano Lett.,
HOMO of the oxidised porphyrin and the surface of the TiO
2
22
nanoparticles must be identical. Moreover, both porphyrin
dyes showed almost identical decays for electron recombination
and electron regeneration dynamics in the presence of an active
redox couple. This implies that the presence of the alkyl chain
does not affect the regeneration reaction, which is key for the
efficient performance of the devices. The yields of regeneration
products are similar with and without the presence of hydro-
2
S. N. Mori, W. Kubo, T. Kanzaki, N. Masaki, Y. Wada and
005, 5, 1315.
9
S. Yanagida, J. Phys. Chem. C, 2007, 111, 3522.
10 J. S. Lindsey, H. C. Hsu, I. C. Shreiman, P. C. Kearney and
A. M. Marguerettaz, J. Org. Chem., 1987, 52, 827.
1 E. Palomares, J. N. Clifford, S. A. Haque, T. Lutz and J. R. Durrant,
J. Am. Chem. Soc., 2003, 125, 475.
2 B. C. O’Regan, S. Scully, A. C. Meyer, E. Palomares and
J. R. Durrant, J. Phys. Chem. B, 2005, 109, 4616.
13 (a) T. Ma, K. Inoue, H. Noma, K. Yao and E. Abe, J. Photochem.
Photobiol., A: Chem., 2002, 152, 207; (b) M. Cervaldo, F. Fungo,
E. N. Durantini, J. J. Silber, L. Sereno and L. Otero, J. Phys.
Chem. B, 2005, 109, 20953.
14 G. Annadurai, M. Chellapandian and M. R. V. Krishnan, Environ.
Monit. Assess., 1999, 59, 111–119.
5 Y. Tachibana, I. V. Rubtsov, I. Montanari, K. Yoshihara,
D. R. Klug and J. R. Durrant, J. Photochem. Photobiol., A: Chem.,
2001, 142, 215.
1
1
phobic alkyl chains, and the regeneration occurs before the
+
TiO2/dye recombination takes place. However, the analysis of
e
the photovoltage transients revealed that, in fact, the effect of
the hydrophobic alkyl chains on the free base porphyrin is to
slow the electron recombination dynamics between the photo-
injected electrons at the semiconductor and the oxidised electro-
lyte. The control of this wasteful reaction is key for improving
the performance of the DSSC and requires careful attention.
In summary, we have demonstrated using free base porphyrins
1
1
6 S. Koops and J. R. Durrant, Inorg. Chim. Acta, 2008, 361, 663.
17 S. A. Haque, E. Palomares, M. B. Cho, A. N. M. Green, N. Hirata,
+
that the eTiO2/electrolyte recombination can be tuned by mole-
D. R. Klug and J. R. Durrant, J. Am. Chem. Soc, 2005, 127, 3456.
cular engineering of the dyes and its consequences at the mole-
cular level have important implications for the device efficiency.
18 Y. Tachibana, S. A. Haque, I. P. Mercer, J. R. Durrant and
D. R. Klug, J. Phys. Chem. B, 2000, 104, 1198.
19 I. Montanari, J. Nelson and J. R. Durrant, J. Phys. Chem. B, 2002,
1
06, 12203.
20 (a) B. C. O’Regan and J. R. Durrant, J. Phys. Chem. B, 2006, 110,
544; (b) A. Zaban, M. Greenshtein and J. Bisquert,
Acknowledgements
8
EMF and MP thanks the Spanish MEC for a Juan de la Cierva
Fellowship and a Ph.D. grant respectively. EP and AF also
thank the MEC for the CONSOLIDER-HOPE 0007–2007
project and OrgaPVNet FP6 EU project. PB and MAS thank
Direcci o´ n General de Investigaci o´ n, Ministerio de Ciencia y
ChemPhysChem, 2003, 4, 859.
21 J. N. Cliford, E. Palomares, Md. K. Nazeeruddin, M. Gr a¨ tzel,
J. Nelson, X. Li, N. J. Long and J. R. Durrant, J. Am. Chem. Soc.,
2
004, 126, 5225.
2
2 O. Cramariuc, T. I. Hukka and T. T. Rantala, Chem. Phys., 2004,
305, 13.
1
658 | J. Mater. Chem., 2008, 18, 1652–1658
This journal is ª The Royal Society of Chemistry 2008