Ultra-fast co-sensitization and tri-sensitization of dye-sensitized solar cells with N719, SQ1 and triarylamine dyes

Article abstract of DOI:10.1039/c2jm31314f

This paper describes the synthesis of a new, yellow triphenylamine dye, 4-[2-(4-diphenylaminophenyl)vinyl]benzoic acid] (6), with a sorption maximum at 380 nm in solution for which EQE data show shifts to 420 nm on sorption to TiO₂. The performance of this dye has been measured in dye-sensitized solar cell (DSC) devices, showing η = 2.6% for 1 cm² devices. Light soaking of (6) shows excellent long-term stability with <10% variation in device performance over 1800 h. Full characterization data are reported for (6) and the intermediates used in its synthesis including single-crystal X-ray structural analysis of all compounds. The paper also describes the ultra-fast dye sensitization and co-sensitization of TiO₂ photo-electrodes in 5 minutes using one or two dyes and the first example of ultra-fast tri-sensitization. The dyes tested include the ruthenium dye N719, the squaraine dye SQ1, the red triphenylamine dye 2-cyano-3-{4-[2-(4-diphenylaminophenyl) vinyl]phenyl}acrylic acid (5) and (6). DSC efficiencies of 7.5% have been achieved for 1 cm² devices co-sensitized using (6) and N719. These efficiencies exceed those recorded for single dye devices and EQE measurements confirm efficient photon capture from two or more dyes in a single photo-electrode. Photo-acoustic calorimetry (PAC) has also been used to measure the energy of the charge separation states formed for (6) and N719, showing a larger value (1.47 eV) for (6) compared to N719 (1.08 eV), whilst a TiO ₂ film co-sensitized with both (6) and N719 gave an intermediate value (1.28 eV). These data have been used to calculate dye HOMO, LUMO and λ_max levels for (6) and N719 leading to important insights for future successful co-sensitization.

Full text of DOI:10.1039/c2jm31314f

View Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links <
C
Journal of
Materials Chemistry
Cite this: J. Mater. Chem., 2012, 22, 13318
www.rsc.org/materials
PAPER
Ultra-fast co-sensitization and tri-sensitization of dye-sensitized solar cells
with N719, SQ1 and triarylamine dyes†
a
a
a
a
a
Peter J. Holliman,* Moneer Mohsen, Arthur Connell, Matthew L. Davies, Kareem Al-Salihi,
b
Mateusz B. Pitak, Graham J. Tizzard, Simon J. Coles, Ross W. Harrington, William Clegg, Carlos Serpa,
b
b
c
c
d
d
e
e
Oct ꢀa vio H. Fontes, Cecile Charbonneau and Matthew J. Carnie
Received 2nd March 2012, Accepted 2nd May 2012
DOI: 10.1039/c2jm31314f
This paper describes the synthesis of a new, yellow triphenylamine dye, 4-[2-(4-diphenylaminophenyl)
vinyl]benzoic acid] (6), with a sorption maximum at 380 nm in solution for which EQE data show shifts
2
to 420 nm on sorption to TiO . The performance of this dye has been measured in dye-sensitized solar
2
cell (DSC) devices, showing h ¼ 2.6% for 1 cm devices. Light soaking of (6) shows excellent long-term
stability with <10% variation in device performance over 1800 h. Full characterization data are
reported for (6) and the intermediates used in its synthesis including single-crystal X-ray structural
analysis of all compounds. The paper also describes the ultra-fast dye sensitization and co-sensitization
of TiO photo-electrodes in 5 minutes using one or two dyes and the first example of ultra-fast
2
tri-sensitization. The dyes tested include the ruthenium dye N719, the squaraine dye SQ1, the red
triphenylamine dye 2-cyano-3-{4-[2-(4-diphenylaminophenyl)vinyl]phenyl}acrylic acid (5) and (6).
2
DSC efficiencies of 7.5% have been achieved for 1 cm devices co-sensitized using (6) and N719. These
efficiencies exceed those recorded for single dye devices and EQE measurements confirm efficient
photon capture from two or more dyes in a single photo-electrode. Photo-acoustic calorimetry (PAC)
has also been used to measure the energy of the charge separation states formed for (6) and N719,
showing a larger value (1.47 eV) for (6) compared to N719 (1.08 eV), whilst a TiO film co-sensitized
2
with both (6) and N719 gave an intermediate value (1.28 eV). These data have been used to calculate
dye HOMO, LUMO and lmax levels for (6) and N719 leading to important insights for future successful
co-sensitization.
2
high-efficiency DSC devices with examples including N3, N719
Introduction
4
ref. 3) and C101. It is well known that Ru-bipy dyes such as
(
Dye-sensitized solar cells (DSCs) are a promising form of low-
1
cost photovoltaic technology. O’Regan and Gr €a tzel first
these are highly efficient at ca. 550 nm but that their response
falls away dramatically between 650 and 700 nm. Attempts have
been made to improve the spectral response of individual dyes by
2
reported using nanoparticulate TiO sensitized with Ru-bipyr-
2
idyl dye to drastically increase DSC device efficiency. Since this
report in 1991, thousands of dyes have been synthesized for
DSCs with the aim of improving device efficiency. For many
years, the ruthenium-bipyridyl family of dyes have dominated
5
synthesizing panchromatic dyes such as ‘‘black dye’’ but, to
date, increasing Jsc has been achieved at the expense of Voc,
limiting the benefit of this approach. In addition, over the last
decade, significant progress has been made in the synthesis of
6
7
more effective organic dyes such as coumarins, indolines and
8
triarylamines, although the predominant spectral response of
a
School of Chemistry, Bangor University, Gwynedd, LL57 2UW, UK.
E-mail: p.j.holliman@bangor.ac.uk; Fax: +44 (0)1248 370528; Tel: +44
these dyes has also tended to be centred around 550 nm where
AM1.5 solar intensity is greatest. In more recent times, the search
for effective near-infrared (NIR) DSC dyes, which are capable of
absorbing photons above ca. 650 nm where most Ru-bipy and
organic dyes are less efficient, has led to several reports of
(
0)1248 382375
b
EPSRC National Crystallography Service, School of Chemistry,
University of Southampton, Southampton SO17 1BJ, UK
c
School of Chemistry, Newcastle University, Newcastle upon Tyne, NE1
7
RU, UK
9
10
11
d
phthalocyanine, squaraine and cyanine dyes. Recently, Ko
Chemistry Department, University of Coimbra, P3004-535 Coimbra,
Portugal
et al. have reported the first example of a panchromatic squar-
12
e
SPECIFIC, College of Engineering Swansea University, Baglan Bay
Innovation and Knowledge Centre, Port Talbot, SA12 7AZ, UK
aine dye. However, the performance of this dye (h ¼ 6.3%) still
does not match that of Ru-bipy dyes.
†
8
Electronic supplementary information (ESI) available. CCDC
38485–838488. For ESI and crystallographic data in CIF or other
Another approach to extending DSC spectral response is
co-sensitization where two or more dyes are used to absorb light
electronic format see DOI: 10.1039/c2jm31314f
1
3318 | J. Mater. Chem., 2012, 22, 13318–13327
This journal is ª The Royal Society of Chemistry 2012

View Online
from different parts of the solar spectrum. Durrant et al.
reported a step-wise approach to co-sensitization by first sorbing
a Ru-bipyridyl dye onto TiO , then treating with aluminium
2
isopropoxide to deposit an Al
2
O
3
layer before sorbing a
Ru-phthalocyanine dye. This approach was developed further
by Choi et al. using JK2, then Al
13
2
O
3
and then SQ1 to give h ¼
1
4
8
.65%. Co-sensitization has been reported using a number of
15
other dyeing techniques including the use of supercritical fluids,
16,17
the use of specialist photo-electrodes,
and slow dyeing
1
8,19
procedures.
Recent reports have included co-sensitizing
20
chlorophylls to give h ¼ 5.4% or co-sensitizing macrocyclic and
organic dyes such as the Zn-phthalocyanine TT1 with organic
dye D2 giving h ¼ 4.1% and a V value in between those of the
oc
21
individual dyes. Step-wise dyeing of the zinc porphyrin LD12
2
2
for 3 h and then the organic dye CD5 for 2 h has given h ¼ 9.0%
whilst h > 12% has been achieved by overnight sensitization with
a Zn-porphyrin and a triphenylamine dye using a Co complex as
23
the redox couple. These reports show some of the potential of
co-sensitization to increase device efficiency, although the dyeing
timescales are too long for DSC devices to be manufactured at
the scale necessary for them to become a truly low-cost photo-
voltaic technology. In this context, recently we have reported
ultra-fast (5 min) sensitization and co-sensitization of N719 and
24
SQ1 using simple procedures which should be scalable. In this
paper, we also report the wider applicability of this rapid co-
sensitization procedure for other dye combinations. As such, this
paper describes the synthesis of a new yellow triarylamine DSC
dye (6). Ultra-fast sensitization and co-sensitization of TiO
2
photo-electrodes have successfully been applied to this new dye
on its own and in combination with N719, SQ1 and the litera-
25
ture red triarylamine dye (5). However, successful co-sensiti-
zation requires more than spectral matching; interference
between dye HOMO and LUMO levels must also be avoided to
optimize device performance. In this context, time-resolved
photoacoustic calorimetry (PAC) has previously been shown to
be a powerful tool for measuring the energy of the charge-
Scheme 1 Synthetic routes to (2), (3), (4), (5) and (6).
ꢀ1
1681 n/cm for the acid. Single-crystal X-ray structural analyses
(Fig. 1, Tables 1 and 2) verified the expected structural
arrangements of (2), (4) and (6). Importantly this confirms the
desired donor–spacer–linker arrangement for (6) with the
strongly donating triarylamine group separated from the sorbing
carboxylate group by the vinyl spacer. In addition, these data
also show that these three groups are approximately collinear
and the conjugated nature of the bonding suggests that the
structure should be relatively rigid. This type of atomic
arrangement should be preferable to optimize efficient electron
26
separated state (Ecs) of dyes bound to TiO
Here, we have used PAC to measure Ecs for (6), N719 and
a co-sensitized mixture of (6) combined with N719 on TiO films
2
for use in DSCs.
2
and used these data to calculate dye HOMO, LUMO and lmax
energy levels.
Results and discussion
A new triarylamine dye (6) has been synthesized (Scheme 1) with
injection into the TiO . The high EQE measured for (6) (Fig. 3) is
2
a
donor–spacer–acceptor (triarylamine–vinyl–benzyl linker)
in line with this.
27
arrangement effective in many DSC dyes. The new dye (6) has
Compound (5) was synthesized by coupling (2) with 4-bromo-
1
benzaldehyde to give (3) which was confirmed by the H NMR
25
reduced conjugation compared to the literature red DSC dye
(
5) with a carboxylate linker replacing cyanoacetate but
aldehyde singlet resonance at 9.97 ppm as described by Kim
25
et al. Finally, a Knoevenagel condensation of (3) with cyano-
28
increased conjugation compared to the literature yellow dye
with a benzene ring in place of a cyclohexadiene closest to the
linker. Compound (6) was synthesized by coupling ethyl-4-bro-
mobenzoate to diphenyl(4-vinyl)amine (2) to produce the
acetic acid–base produced (5). This was confirmed by mass
1
spectrometry and also the loss of the H NMR aldehyde signal,
13
the shift of the C NMR aldehyde carbonyl signal at 192.25 pm
1
benzoic acid ethyl ester (4), then de-esterifying with HCl. H
for (3) to 165.00 ppm for the carbonyl of the acid, and the new
13
NMR data for (4) showed triphenylamine and benzoate (d 6.9–
nitrile signals in the C NMR and FTIR data at 118.00 ppm and
ꢀ1
7
.2 ppm), vinyl (d ca. 4.3 ppm) and alkyl (d ca. 1.3 ppm) signals.
2251 n/cm , respectively. Single-crystal X-ray structural
analysis of (3) confirmed the expected structural arrangement
(Fig. 1) but we have not yet been able to grow a suitable single
crystal of (5).
De-esterification of (4) to produce (6) was confirmed by mass
1
spectrometry, the absence of H NMR alkyl ester signals and
ꢀ1
a shift in the FTIR carbonyl signal from 1709 n/cm to
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 13318–13327 | 13319

View Online
and from the EQE are coincident. However, efficiency
improvements are seen due to (5) and (6) having significantly
higher molar extinction coefficients than N719. For example,
ꢀ1
ꢀ1
taking (6) and N719 as examples, (6) has 3 of 32 349 L cm mol
ꢀ1
ꢀ1
compared to ca. 13 000 L cm mol for N719. However, it must
be remembered that dyes are adsorbed onto TiO particles within
2
a photo-electrode film such that dye loadings are limited by the
number of available adsorption sites. Thus, if the molar extinc-
tion coefficients for (6) and N719 are calculated based on mass,
ꢀ1
ꢀ1
the respective values are 82.6 and 11.2 cm
g
L meaning that,
for the same loading, (6) absorbs over 7 times as many photons at
its lmax than N719. Alternatively, within the limited space of the
photo-electrode, a relatively lower dye loading of (6) can still
absorb more light at 420 nm than N719 resulting in improved
EQE in this part of the spectrum. A similar approach has just
been reported by Yella et al. who have co-sensitized with 2 dyes
23
within a similar spectral region to achieve h > 12%.
The optical properties of (5) and (6) may explain the differ-
ences in performance for DSC devices prepared using electrodes
dyed passively (overnight) with these dyes. Thus Table 3 shows
ꢀ2
photocurrents of 8.43 versus 2.64 mA cm and overall efficien-
cies of 3.4 versus 1.2% for Devices A and B, prepared using (5)
and (6) respectively. Devices D and E show ultra-fast dyeing of
2
4
(
5) and (6) in 5 min as described previously, which interestingly
yields improved efficiencies (h ¼ 2.6% and 4.4%) partly due to
ꢀ2
increased Jsc values (6.2 and 8.8 mA cm ) but also improved Voc
versus passive dyed devices.
Importantly, dye loading data measured by desorption show
that ultra-fast dyeing of (6) results in less dye uptake when
compared to passively dyed devices (Fig. 5). It has often been
assumed that increasing dye loading is the key to increasing Jsc
,
whereas these data clearly show that the issue is more complex
than this. One issue could be that fast dyeing has been shown to
prevent photo-bleaching of air-sensitive dyes such as SQ1
24
because dyeing takes place within pre-sealed devices. However,
our experimental observations suggest that (6) is remarkably air-
stable compared to SQ1, so the improved efficiency for fast-dyed
(
6) does not seem to be due to this.
Another issue is that fast dyeing works by pumping dye
2
solution through the mesoporous TiO film to overcome the
24
mass transfer issues of passive diffusion. This means that
sensitization is mainly controlled by the rate of adsorption with
the whole process being completed within a matter of minutes.
This, and the lower dye loading, imply a more random and
spaced arrangement of dye molecules on the surface compared to
passive dyeing which aims to produce a complete monolayer of
Fig. 1 Molecular structures of (2), (3), (4) and (6). Displacement ellip-
soids are drawn at the 50% probability level. For (2), (3) and (4), only one
of two independent molecules is shown.
UV-visible spectroscopy (Fig. 2) shows that (5) and (6) have
absorption maxima at 370 and 390 nm, respectively, but that the
red dye (5) has an additional shoulder compared to (6) which
appears at ca. 440 nm and which absorbs up to ca. 490 nm.
Whilst the UV-visible bands for (5) and (6) at 400 nm appear at
similar wavelengths to the secondary band of N719, the EQE
data (Fig. 3 and ESI†) show that N719 harvests photons less
efficiently than (5) or (6) in this region of the spectrum. Inter-
estingly, a comparison of the EQE profiles of (5) and (6) shows
significantly greater differences than the UV data, with (5) har-
vesting photons across a much wider range (350–560 nm) than
29
dye on the TiO surface. It is of course possible to fast dye for
2
much shorter periods of time (1–3 min), which results in signifi-
cantly lower dye coverage; these devices do show lower effi-
ciencies which suggests that a threshold dye coverage is
30
required. This threshold will vary for different dyes depending
on their molar extinction coefficient and efficiency of electron
injection. Indeed it is well known that similar Jsc can be recorded
for highly absorbing dyes in thin photo-electrode layers
31
(2–3 mm) where dye loadings must be considerably lower than
for 10–15 mm films.
(
6) which absorbs in the range 350–500 nm. At first sight, using
Fig. 3 confirms the beneficial effects of ultra-fast co-sensiti-
zation in terms of the EQE for selected dyes. For instance, the
yellow dye (6) exhibits high EQE between ca. 380 and 500 nm
either (5) or (6) in combination with N719 does not seem
a promising approach because the absorption bands in solution
1
3320 | J. Mater. Chem., 2012, 22, 13318–13327
This journal is ª The Royal Society of Chemistry 2012

View Online
Table 1 Crystal data and structural refinement for compounds (2), (3), (4) and (6)
2) (3)
(
(4)
(6)
Formula
Formula weight
Temperature (K)
C
271.35
120
Synchrotron, 0.6889
Colourless plate
0.05 ꢁ 0.04 ꢁ 0.01
Monoclinic
20
H
17
N
C
375.45
120
0.71073
27
H
21NO
C
29
H
25NO
2
27 2
C H21NO
419.50
120
0.71073
391.45
120
0.71073
Light yellow block
0.20 ꢁ 0.14 ꢁ 0.06
Monoclinic
ꢁ
Wavelength (A)
Crystal colour and habit
Crystal size (mm)
Crystal system
Yellow needle
0.41 ꢁ 0.07 ꢁ 0.05
Triclinic
Green prism
0.20 ꢁ 0.05 ꢁ 0.04
Monoclinic
/c
ꢂ
P1
Space group
ꢁ
P2
1
/n
P2
1
1
P2 /c
a (A)
ꢁ
10.000(2)
18.679(4)
15.789(4)
10.1895(10)
11.4764(16)
17.763(3)
102.487(5)
96.931(8)
101.360(8)
1959.1(4)
4
11.2472(4)
13.4462(4)
30.0363(11)
17.1248(5)
9.6815(2)
12.9988(3)
b (A)
ꢁ
c (A)
ꢂ
a ( )
ꢂ
b ( )
ꢂ
91.675(2)
90.567(1)
110.727(2)
g ( )
ꢁ
3
V (A )
2948.0(11)
8
2
1.6–25.5
25 445
5998
4542.2(3)
8
2
3.0–25.0
32 084
7961
2015.63(9)
4
1
3.1–27.5
28 268
4613
Z
0
Z
q range ( )
2
ꢂ
3.0–25.0
30 858
6734
Reflections collected
Independent reflections
R
int
0.0530
4684
99%
380
1.094
0.0584
0.1707
0.26 and ꢀ0.24
0.1570
3732
98%
523
1.074
0.1076
0.2430
0.26 and ꢀ0.28
0.0685
5368
99%
579
1.167
0.0843
0.1605
0.26 and ꢀ0.26
0.0564
3640
99%
272
1.052
0.0482
0.1141
0.21 and ꢀ0.25
2
Reflns with F > 2s
Data completeness
Refined parameters
Goodness of fit (F )
2
R (F, F > 2s)
2
2
(F , all data)
R
Max diff peak and hole (e A
w
ꢁ
ꢀ3
)
Table 2 Selected bond lengths and angles for compounds (2), (3), (4)
and (6)
(
2)
(3)
(4)
(6)
ꢁ
Bonds [A]
N1–C1
N1–C7
1.428(3)
1.424(3)
1.411(3)
1.470(3)
1.329(3)
1.413(6)
1.425(6)
1.412(6)
1.472(6)
1.331(6)
1.457(6)
1.469(7)
1.413(4)
1.443(4)
1.417(4)
1.464(5)
1.344(5)
1.458(5)
1.485(5)
1.4351(18)
1.4153(19)
1.4156(18)
1.470(2)
1.338(2)
1.466(2)
1.477(2)
N1–C13
C16–C19
C19–C20
C20–C21
C24–C27
Angles [ ]
C1–N1–C7
C1–N1–C13
C7–N1–C13
C16–C19–C20
C19–C20–C21
—
—
ꢂ
117.55(17)
119.73(17)
122.67(18)
127.4(2)
—
120.7(4)
122.0(4)
116.9(4)
126.0(5)
128.2(5)
120.2(3)
121.6(3)
116.1(3)
127.3(3)
125.7(3)
119.35(12)
117.78(12)
122.80(12)
122.73(14)
128.66(14)
Fig. 2 UV-visible spectroscopy of N719 (solid line), (5) (dashed line),
SQ1 (dotted line) and (6) (dash dot line) in ethanol.
whilst N719 is most effective between ca. 350 and 650 nm. The
benefit of co-sensitizing (6) with N719 occurs because the EQE of
sensitization with N719 (lmax ¼ 550 nm) and SQ1 (lmax
¼
(
6) is most effective at ca. 410 nm where the EQE of N719 is
640 nm) showed enhanced absorption of NIR photons from
24
relatively poor, improving the overall EQE profile. Hence, the
co-sensitized device harvests more photons than either single dye
on its own, although the effect is not entirely additive because of
the overlapping EQE profiles of the dyes individually.
SQ1 but here we report greater absorption of blue photons
from either (5) or (6). Device J shows that mixing (5) and (6) does
not significantly increase h; this is expected as both dyes absorb
at similar wavelengths and have similar 3. Mixing yellow (6) and
blue SQ1 (Device K) gives a green device and a small increase in
Ultra-fast co-sensitization applied to combinations of (5), (6)
and N719 (Table 3) shows increased Jsc (from 12.4 to ca.
ꢀ
2
Jsc (from 8.16 to 9.23 mA cm ) and a similar V for both
oc
ꢀ2
1
5.0 mA cm ) and h from 6.0% to 7.2% (Devices H, I and J)
devices.
which is in line with the EQE data showing increased photon
conversion across a wider range of the spectrum. Successful co-
sensitization is defined here as increased h relative to the most
efficient single dye (N719 in this case). Our previous ultra-fast
However, a slightly poorer FF results in similar h to SQ1.
Finally, Devices L and M show what we believe are the first
reported examples of successful ultra-fast tri-sensitization
with enhanced Jsc and h arising from 3 dyes within the
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 13318–13327 | 13321

View Online
sensitized devices, the order of dyeing is also important. Taking
(
(
6) and N719 as an example, dyeing with (6) and then N719
2
Device H1) gives loadings of 57 and 217 mg dye per cm ,
respectively, and h ¼ 7.5% whilst the opposite order (Device H2)
ꢀ2
ꢀ2
gives a similar loading of N719 (206 mg cm ) but only 28 mg cm
of (6) which may explain the slightly lower efficiency of 6.7%.
Fig. 4 shows photographs of DSC devices (2 ꢁ 0.5 cm)
sensitized with different dye combinations showing that (5) and
(
(
6) result in red and yellow devices, respectively as expected
Fig. 4a and b). Combinations of dyes result in mixed colours
again as expected. For instance, co-sensitizing with (6), (5) and
N719 produces a darker coloured film than any of the dyes
individually (Fig. 4c) and a similar effect is observed when co-
sensitizing with (6) and N719 (Ftig. 4d) with the yellow (6) and
blue SQ1 producing a green colour again as expected (Fig. 4e).
Colour photographs of all devices including images of sequen-
tially dyed devices are held in the ESI†. Fig. 4f shows a scanning
electron micrograph of a cross-section of a typical photo-elec-
Fig. 3 Percentage external quantum efficiency (EQE) versus wavelength.
The dashed line is for (6), the dotted line is for N719 and the solid line is
for co-sensitized (6) + N719.
trode film showing a layer of smaller TiO
thick) with a layer of larger, light scattering TiO
ca. 5 mm thick).
The changes in dye loading as a function of sensitization time
2
particles (ca. 16–17 mm
2
particles on top
Table 3 I–V testing data for selected DSC devices
(
ꢀ
2
Device
h/%
Fill factor
J
sc/mA cm
Voc/V
have been measured for 3 systems; Fig. 5 and Table 4 show data
for the fast and passive dyeing of (6) and Table 5 shows data for
the fast dyeing of a mixture of (6) and N719. The data show that,
Passive dyed
A
B
C
(6)
(5)
N719
1.2
3.4
5.6
0.65
0.63
0.56
2.64
8.43
13.10
0.70
0.65
0.77
when passively dyeing (6), saturation of the TiO
2
film occurs
after ca. 6 h after which the dye loading remains reasonably
Fast dyed
ꢀ2
D
E
(6)
(5)
2.6
4.4
3.5
6.0
7.5
6.7
7.2
3.2
3.7
6.5
4.4
6.5
0.66
0.72
0.67
0.63
0.65
0.63
0.63
0.57
0.63
0.55
0.51
0.68
6.19
8.80
8.16
12.42
15.45
14.08
14.92
8.89
0.64
0.70
0.64
0.77
0.74
0.76
0.77
0.63
0.64
0.73
0.64
0.73
constant at 75–80 mg cm . However, the data for films fast dyed
with (6) show that a similar dye loading can be achieved in 7 min.
Interestingly, the corresponding device performance data (Table
4) show that h ¼ 2.0% is reached after 4 min of fast dyeing
F
SQ1
N719
(6) + N719
N719 + (6)
(5) + N719
(6) + (5)
G
H1
H2
I
2
corresponding to ca. 56 mg (6) per cm and whilst dye loading
J
K
L
M
N
(6) + SQ1
(6) + N719 + SQ1
(6) + (5) + SQ1
(6) + (5) + N719
9.23
16.05
13.56
13.04
photo-electrode. Indeed Device L shows the highest Jsc of any of
ꢀ2
the devices reported here (16.05 mA cm ). However, h is slightly
lower due to a decrease in Voc and FF. The latter may reflect
some slight increase in recombination between these three dyes
when present in a single photo-electrode. By comparison, Device
M, which is a mixture of (5), (6) and SQ1, shows significantly
ꢀ2
higher Jsc (13.56 mA cm ) than either combinations of (5) and
6) in Device J, a combination of (6) and SQ1 in Device K, or any
of the (5), (6) or SQ1 dyes individually, Devices D–F. In addition,
(
32
although tri-sensitization has been reported before, we believe
this is the first report of ultra-fast tri-sensitization.
Dye loadings have been measured for the fast dyed devices
showing relatively high loadings for individual dyes (e.g. Device
D ¼ 171, Device E ¼ 137, Device F ¼ 206 and Device G ¼ 284 mg
2
dye per cm for (6), (5), SQ1 and N719, respectively) which is
expected because all sorption sites are available in a singly dyed
photo-electrode. Again as expected, co-sensitized devices show
lower dye loadings of each component dye. For example, co-
sensitizing with (6), SQ1 and N719 (Device L) gives dye loadings
Fig. 4 Photographs of DSC devices dyed (a) with (5), (b) with (6), (c)
with (5), (6) and N719, (d) with (6) and N719, (e) with (6) and SQ1 and (f)
2
of 74, 16 and 142 mg dye per cm , respectively. For sequentially
SEM of a typical TiO
2
photo-electrode film.
1
3322 | J. Mater. Chem., 2012, 22, 13318–13327
This journal is ª The Royal Society of Chemistry 2012

View Online
interesting trends which reflect the fact the dye sorption is taking
place within a competitive environment with molecules of both
N719 and (6) capable of sorbing at the same time. As expected
the uptake of both dyes is fast within the first minute as the
maximum number of dye sorption sites are available on the TiO
2
surface (uptake is 36 mg N719 per cm and 25 mg (6) per cm ).
2
2
However, the uptake of (6) slows down after this and only
2
a further ca. 20 mg (6) per cm absorbs over the next 11 min. By
comparison, N719 uptake continues such that, after 12 min,
2
230 mg N719 per cm has sorbed. At the same time, the device
>
efficiency climbs steadily to 7.1% after 12 min dyeing time with
the FF remaining fairly constant during this time whilst the Jsc
ꢀ2
increases steadily >15 mA cm after 7 min. The Voc fluctuates in
the first 7 min of dyeing which presumably reflects the changing
surface arrangement of dyes before reaching a maximum of 0.71
V after 12 min of dyeing. This may suggest a more stable surface
arrangement in line with a higher surface coverage of dye.
Fig. 6 shows light soaking data for (6) and N719 over 1800 h.
Both dyes show a slight increase in efficiency from 0 to 200 h
(
which is ascribed to improved electrolyte pore filling) after
which the efficiency remains stable up to 1800 h. These slight
changes are reflected in the short-circuit current whilst Voc and
fill factor remain fairly constant over the time period measured.
This shows that the new dye (6) exhibits excellent long-term
stability which compares favourably with N719.
Fig. 5 Dye uptake profiles of (6) using (a) passive and (b) fast dyeing.
Table 4 I–V and dye loading data for devices fast dyed with 1.2 mM
The use of time-resolved photo-acoustic calorimetry (PAC) to
measure the energy of the charge-separated state (Ecs) of dyes
ꢀ
1
solution of (6) with 10 mM at a rate of 100 mL min
2
6
bound to TiO
2
for use in DSC has recently been reported. Here,
films sensitized
PAC has been applied to glass-supported TiO
2
Dyeing
time/min
Desorbed
(6)/mg cm
ꢀ
2
ꢀ2
h/%
V
oc/V
Jsc/mA cm
FF
with either N719 or (6) to measure (Ecs), that is the energy
difference between the charge separated state with the electron at
1
2
3
4
5
7
1.2
1.5
1.6
2.0
2.1
2.2
0.62
0.62
0.63
0.65
0.65
0.66
2.81
3.48
3.82
4.56
4.88
5.18
0.66
0.67
0.67
0.66
0.66
0.65
30.6
38.2
50.7
55.7
64.7
73.2
the conduction band edge of TiO
2
and ground state of the elec-
tron in its HOMO (Fig. 7). This ignores small changes in entropy
26
and volume that tend to compensate each other.
Table 6 shows the data for these two dyes with a larger Ecs
value for (6) (1.47 eV) compared to N719 (1.08 eV). Using a value
3
3
of ꢀ4.4 eV vs. vacuum for the conduction band of TiO
2
,
this
gives a value of ꢀ5.87 eV for the HOMO of (6) and ꢀ5.48 eV for
Table 5 I–V and dye loading data for devices fast dyed with a solution
containing 0.75 mM (6), 0.67 mM N719 and 10 mM CDCA
Desorbed
dye/mg cm
ꢀ
2
Dyeing
time/min
ꢀ
2
h/%
V
oc/V
Isc/mA cm
FF
N719
(6)
1
2
3
4
5
7
3.2
4.2
4.7
4.9
5.3
5.9
7.1
0.67
0.60
0.64
0.60
0.63
0.62
0.71
7.37
0.66
0.64
0.63
0.62
0.64
0.62
0.63
36.3
96.3
24.6
26.4
33.3
36.0
37.6
42.8
44.3
11.02
11.55
12.99
13.18
15.32
15.76
117.5
122.5
162.5
165.0
236.3
1
2
2
increases to ca. 73 mg (6) per cm after 7 min the efficiency only
increases to 2.2%. This suggests that there are dye sorption sites
on the TiO surface which might generate more photo-current if
2
they were loaded with a different dye to (6); in essence this is the
logic behind co-sensitization.
Leading from this, Table 5 shows devices which have been
co-sensitized with a mixture of (6) and N719. The data show
Fig. 6 Light soaking of DSC devices for up to 1800 h at 1 Sun for N719
(open circles) and (6) (black squares) showing h, Voc, Jsc and fill factor.
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 13318–13327 | 13323

View Online
shows the HOMO and LUMO of (6) and N719 calculated from
these PAC data.
PAC measurements have also been carried out on TiO films
2
co-sensitized with (6) and N719. We believe this is the first
example of measurements of this type with two dyes in a single
2
TiO film. Interestingly, the data for both (6) and N719 give an
intermediate value for Ecs (1.28 eV) between the extremes of the
single dyes on their own. Unfortunately, this suggests that, at this
time, it may not be possible to delineate the two PAC signals
from the separate dyes when there is an intimate mixture of both
dyes within the same TiO film and both have ultra-fast injection
2
rates. The distinction between the dyes in the mixture would
probably require PAC with picosecond resolution.
Electron impedance spectroscopy has been used to measure
electron lifetimes for devices containing (5), (6), and N719 and
also for devices co-sensitized with a combination of these dyes
Fig. 7 Charge separation energies (Ecs), HOMO, LUMO and lmax
energy levels for (6) and N719.
(
(
Fig. 8). The data show that devices dyed with the organic dyes
5) and (6) either on their own with a combination of (5) and (6)
exhibit shorter electron lifetimes across a range of voltages than
for devices dyed with N719 alone. For co-sensitized devices
involving N719, the device dyed with a combination of (5) and
N719 shows electron lifetime values which are intermediate
between (5) and N719. However, devices which have been dyed
with a combination of (6) and N719 exhibit electron lifetime data
which is comparable to N719 on its own. These data suggest that
having a combination of an organic dye in the same photo-
electrode as N719 does not necessarily have to have a negative
impact on electron lifetimes but rather that this effect is specific
to particular dye combinations. Interestingly, there is some
evidence to suggest that the electron lifetimes are slightly longer
when (6) is dyed first followed by N719 (labelled (6)–N719)
compared to the opposite (labelled N719–(6)). This is in line with
the IV data which show slightly higher device performance when
(6) is dyed before N719 (Device H1, h ¼ 7.5%) rather than the
other way round (Device H2, h ¼ 6.7%). As discussed earlier this
is mainly due to a higher dye loading of (6) for Device H1 giving
slightly higher Jsc. In the context of electron lifetimes, this
suggests that a higher loading of the organic dye (6) in combi-
nation with N719 is not detrimental. The data from the order of
dyeing may also suggest that the organisation of the dyes on the
surface is important for multiply sensitized devices and that some
Table 6 Energy level data for selected dyes derived from PAC
measurements (error estimated as ꢃ0.1 eV)
Energy (eV)
(6)
N719
(6) + N719
E
cs
1.47
ꢀ5.87
1.04
ꢀ5.48
1.28
—
—
HOMO
LUMO
ꢀ3.62
ꢀ3.83
l
max
ꢀ2.92 (420 nm)
ꢀ2.60 (430 nm)
ꢀ3.18 (540 nm)
N719. This value for N719 is in good agreement with the liter-
34
ature value of ꢀ5.45 eV, obtained from cyclic voltammetry of
the isolated dye in acetonitrile solution, whereas the present
value was directly calculated from a PAC experiment where the
dye is adsorbed onto TiO films immersed in acetonitrile. These
2
data also confirm that the HOMO energies of both dyes lie below
ꢀ
ꢀ
35
that of the I –I
3
redox couple (ꢀ4.85 eV), facilitating efficient
reduction of the excited-state dyes during device operation.
The LUMO energies for each dye have been calculated using
the absorption onsets for each dye to calculate the minimum
excitation energy between HOMO and LUMO. These are 750
nm for N719 and 550 nm for (6) corresponding to 1.65 eV and
2
.25 eV, respectively. These data give the LUMO at ꢀ3.83 eV for
N719 which again is in good agreement with the literature value
34
of ꢀ3.85 eV for N719 and at ꢀ3.62 eV for (6). The energies of
the maximum photon harvesting have also been calculated from
the IPCE data as ꢀ2.92 eV corresponding to 420 nm for (6)
and ꢀ2.60 and ꢀ3.18 eV corresponding to 430 and 540 nm,
respectively for N719. Hence the LUMO for both dyes is above
2
the conduction band for TiO , facilitating efficient electron
injection which is reflected in the high EQE values for each dye.
Interestingly, the LUMO of (6) is an off-set of ca. 0.2 eV from
N719. We believe this is why the Voc does not drop for devices co-
sensitized with N719 and (6). This is important because Voc is
often reported to decrease for co-sensitized devices which may be
due to a lowering of the TiO Fermi level if lower dye energy
2
levels are introduced. The energy levels associated with
maximum photon harvesting are also off-set between (6) and
N719 which potentially may be helpful in reducing inter-dye
recombination processes, also preventing any Voc losses. Fig. 6
Fig. 8 Electron lifetimes (s) vs. voltage for (5), (6) and N719 and selected
co-sensitized combinations of these dyes.
1
3324 | J. Mater. Chem., 2012, 22, 13318–13327
This journal is ª The Royal Society of Chemistry 2012

View Online
ꢂ
surface dye arrangements may result in improved electron
injection and/or reduced recombination processes.
(0.699 g, 0.83 mmol) was added. The mixture was stirred at 45 C
for 18 h. HCl(aq) (5%, 1.61 M) was then added to adjust the pH to
between 5 and 6. Ethyl acetate was added and the organic layer
was separated using a separating funnel, dried over anhydrous
magnesium sulfate and the solvent removed in vacuo to give
Experimental
ꢂ
Instrumentation and chemicals
a yellow solid. Yield 0.30 g, 93%, mp 214–216 C. Suitable
crystals for X-ray analysis were prepared by dissolving (6) in
All chemicals were purchased from Aldrich and used as supplied
unless otherwise stated. Anhydrous solvents were used as
purchased except for THF which was dried with Na wire. NMR
spectra were recorded on a Bruker AC500 instrument operating at
CH
2
Cl
2
and layering with diethyl ether (1 : 1, v/v).
H NMR (500 MHz, DMSO-d
) d 6.95 (d, J ¼ 8.5 Hz, 2H,
Ph–N–Ph ), 7.08 (t, J ¼ 7.25
), 7.05 (d, J ¼ 7.9 Hz, 4H, –N–Ph
1
6
–
2
2
Hz, 2H, –N–Ph ), 7.17 (d, J ¼ 16.4 Hz, 1H, –CH]CH–Ph–),
1
13
2
5
00MHzfor H and 125MHzfor C. Chemical shifts(d)are given
7
.33 (m, 5H), 7.54 (d, J ¼ 8.85 Hz, 2H, –Ph–N–Ph ), 7.66 (d, J ¼
1
13
2
in ppm relative to tetramethylsilane (for H and C). J values are
given in Hz and refer to JH,H unless otherwise indicated. Mass
spectra were recorded using electron impact, chemical ionisation
8
.5 Hz, 2H, CO–Ph–CH]), 7.91 (d, J ¼ 8.5 Hz, 2H, CH]Ph–
13
CO). C NMR (500 MHz, DMSO-d ) d 122.60, 123.44, 124.33,
6
1
26.09, 127.95, 129.59, 129.70, 146.84, 183.46. UV-vis (EtOH):
ꢀ1
(
3
NH ), fast atom bombardment, electrospray, matrix-assisted
1
ꢀ1
l
max. 296 nm (3 ¼ 21 100 Mꢀ cm ) and 376 nm (3 ¼ 22 600 M
laser desorption ionisation and laser desorption ionisation at the
EPSRC National Mass Spectrometry Service at the University of
Swansea. Infrared spectra were recorded on a Perkin-Elmer 1600
series FTIR spectrometer. Elemental analysis was performed on
a Carlo Erba EA1108 elemental analyser.
ꢀ1
3
ꢀ1
ꢀ1
cm ), 3 values are precise to ꢃ1 ꢁ 10 M cm . FT-IR (KBr):
ꢀ1
n/cm
FTMS -MS: m/z (accurate mass), reference compound:
NH OAc, calcd for C27 392.1645, found 392.1641
M + H] . Anal. calcd for C27 : C, 82.84; H, 5.41; N, 3.58;
3421, 3029, 2251, 1681, 1589, 1493, 1283, 1179.
+
4
H
21NO
2
+
[
H21NO
2
O, 8.17. Found: C, 81.89; H, 5.60; N, 3.41; O, 8.99%.
Preparation of 4-[2-(4-diphenylamino-phenyl)vinyl]benzoic acid
ethyl ester (4)
X-ray crystallography
Compound (2) (1 g, 3.7 mmol) was dissolved in anhydrous N,N-
dimethylacetamide (DMAC) (20 ml) with trans-di(m-acetato)bis
Single-crystal analysis of (2) was performed with the use of
synchrotron X-rays at Diamond Light Source UK, beamline I19
ꢁ
[
O-(di-o-tolylphosphoino)benzyl]dipalladium(II) (0.015 g, 0.05
(
l ¼ 0.6889 A) on a Crystal Logic diffractometer and Rigaku
mmol), ethyl-4-bromobenzoate (0.38 g, 1.66 mmol), Na CO
2
3
Saturn 724+ detector equipped with an Oxford Cryosystems
(
0.44 g, 4.1 mmol) and 2,6 di-tert-butylcresol (0.073 g,
ꢂ
Cryostream device. Data were processed using the Rigaku
0
.33 mmol) and the solution was stirred at 130 C for 24 h. After
O–CH Cl
20 ml, 1 : 1 by vol) and the mixture separated using a separating
36
CrystalClear package and Bruker APEX2 software.
cooling, the solution was poured into distilled H
2
2
2
Single-crystal X-ray diffraction analyses of (3), (4) and (6) were
performed using a Bruker APEXII CCD diffractometer moun-
ted at the window of a Bruker FR591 rotating anode (Mo Ka,
(
funnel. The organic layer was dried over anhydrous magnesium
sulfate and the solvent removed in vacuo before purification by
column chromatography on silica gel with petroleum spirit/
diethyl ether (95 : 5, v/v). The product was identified by TLC and
the solvent removed in vacuo followed by precipitation in MeOH
ꢁ
l ¼ 0.71073 A) and equipped with an Oxford Cryosystems
Cryostream device. Data were processed using the Collect
37
package and unit cell parameters were refined against all data.
ꢂ
An empirical absorption correction was carried out using
to give a yellow solid (4). Yield 0.4 g, 26%, mp 120–122 C.
38
SADABS.
Suitable crystals for X-ray analysis were prepared by dissolving
2
All structures were solved by direct methods and refined on Fo
(
4) in CH
1
2
Cl
H NMR (500 MHz, DMSO-d
COO–CH –CH
), 4.31 (dd, J ¼ 14.15, 6.9 Hz, 2H, Ph–COO–CH
CH ), 7.05 (d, J¼7.55 Hz, 4H,
), 6.95 (d, J¼8.5 Hz, 2H, –Ph–N–Ph
N–Ph ), 7.18 (d, J ¼ 16.4 Hz,
), 7.08 (t, J ¼ 7.25 Hz, 2H, –N–Ph
H, –CH]CH–Ph–), 7.33 (m, 5H), 7.54 (d, J¼ 8.5 Hz, 2H, –Ph–N–
), 7.69 (d, J ¼ 8.55 Hz, –CO–Ph–CH]), 7.93 (d, J ¼ 8.5 Hz, 2H,
CH–Ph–CO–). C NMR (500 MHz, DMSO-d ) d 122.36,
2
and layering with diethyl ether (1 : 1, v/v).
by full-matrix least-squares refinements using programs of the
6
) d 1.32 (d, J ¼ 15 Hz, 3H, Ph–
39
SHELX family. All non-hydrogen atoms were refined with
anisotropic displacement parameters. All hydrogen atoms were
added at calculated positions and refined using a riding model
with isotropic displacement parameters based on the equivalent
isotropic displacement parameter (Ueq) of the parent atom.
2
3
2
–
3
2
–
1
2
2
Ph
2
40
13
Figures were produced using ORTEP3 for Windows. The CIFs
for the crystal structures of (2), (3), (4) and (6) have been
deposited with the CCDC and have been given the deposition
numbers CCDC 838485–838488, respectively.
]
6
1
1
2
1
23.54, 124.44, 125.40, 126.66, 128.13, 129.62, 130.00, 130.40,
31,51, 134.69, 143.47, 146.75, 147.44, 192.25. UV-vis (EtOH): lmax
ꢀ1
96 nm, 388 nm. FT-IR (KBr): n/cm 3023, 2251, 1709, 1589, 1493,
+
278, 1178, 1108. FTMS -MS: m/z (accurate mass), reference
compound: NH OAc, calcd for C29
4
H
25NO 420.1958 found
2
25NO : C, 83.03; H, 6.01;
2
Device manufacture and testing
+
20.1956 [M + H] . Anal. calcd for C29
4
H
ꢀ
2
Photo-electrodes (0.5 ꢁ 2 cm ¼ 1.0 cm ) were prepared by
N, 3.34; O, 7.63. Found: C, 82.99; H, 6.08; N, 3.36; O, 7.60%.
doctor blading two layers of TiO colloid (DSL18NRT, Dyesol)
2
followed by one layer of scattering TiO2 colloid (WER4-O,
Preparation of 4-[2-(4-diphenylamino-phenyl)vinyl]benzoicacid(6)
Dyesol) onto fluorine tin oxide-coated glass (TEC15, 15 U per
layer was calcined at 723 K for 30 min, then
: THF in deionised water (50 mM) at 343 K
Compound (4) (0.35 g, 0.83 mmol) was dissolved in
,). Each TiO
2
THF–MeOH–H
2
O (12 ml, 10 : 1 : 1, v/v/v) and LiOH 20 eq.
immersed in TiCl
4
2
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 13318–13327 | 13325

View Online
for 30 min, rinsed with de-ionised water, air-dried for 10 min and
re-sintered at 723 K for 30 min.
achieved by the output of an EKSPLA PG122/SH OPO pumped
by an EKPLA NL301G Nd:YAG laser third harmonic at
a frequency of 10 Hz. A small fraction of the laser beam was
reflected to a photodiode and used to trigger the transient
Two dyeing methods were used. For passive dyeing, sintered
photo-electrodes were cooled to ca. 343 K and submersed in dye
solution for 18 h. A solution of N719 (Dyesol, 0.5 mM) was
prepared in acetonitrile–tert-butyl alcohol (1 : 1) whilst SQ1
ꢀ1
recorder (Tektronix DSA 601, 1 Gs s ). The photoacoustic
waves, detected with a 2.25 MHz Panametrics transducer (model
5676) and captured by the transient recorder, were transferred to
a PC for data analysis. Typically, 200 PAC waves from the
4
1
(
0.05 M) was prepared as described in the literature and dis-
solved in ethanol containing chenodeoxycholic acid (5 mM).
After dyeing, a Surlynꢀ (DuPont) gasket was placed around the
photo-electrode and a Pt-coated TEC 8 (8 U per ,) counter
electrode which had been treated with Pt paste (PT1, Dyesol) and
heated to 673 K was placed on top and the two electrodes were
2
sample (TiO with (6), N719 or (6) + N719), reference
(TiO with 5,10,15,20-tetrakis(4-sulfonylphenyl)porphyrinate
2
manganese(III) acetate or MnTPPS) or background (TiO only)
2
were recorded and averaged under the same conditions. Four sets
of averaged sample and background waves were used for data
analysis at each laser intensity and four laser intensities were used
in each experiment by interposing neutral density filters with %T
between 25% and 75%.
ꢀ
ꢀ
sealed together at 393 K. Electrolytes were I /I (I 100 mM in
2
3
distilled 3-methoxypropionitrile) for N719 and the literature
41
electrolyte for SQ1. Electrolyte fill holes in the counter elec-
trode were then sealed using Surlynꢀ. For the fast dyeing
method, dye solutions were pumped through the device cavity
individually, sequentially, or as a dye solution mixture through
a hole in the counter electrode for between 5 and 10 minutes at
Sample and reference (MnTPPS) PAC signals were analysed
after subtraction of the (almost negligible) background signal.
The reference is defined as a dye that converts all absorbed
energy into heat on a time-scale much shorter than the time-
ꢂ
24
5
0 C using the method described previously.
Current–voltage characteristics were measured with either
26
resolution of acoustic detection. The energy of the charge-
a Universal Photovoltaic Testing System (Dyesol) or an ABET
Solar Simulator with a Xe arc lamp and a Keithley 2400 at
separated (CS) state was then calculated from eqn (1) where 4 is
1
the fraction of energy released by the sample compared with the
ꢀ
2
1
00 mW cm or 1 Sun between 0 and 1 V. Spectral response
heat released by the reference, Ehn ¼ 2.50 eV at 495 nm, and F is
cs
measurements were made from 300–800 nm either on an IPCE
Measurement Apparatus (Dyesol Ltd) at a resolution of 5 nm or
on a QEX10 Quantum Efficiency Measurement System in DC
mode at a resolution of 10 nm. Light soaking experiments were
carried out using a Dyesol light soaking chamber fitted with
a 1000 W metal halide high-intensity discharge (HID) bulb using
a 420 nm cut-off filter. Lamps were calibrated to 1 Sun (100 mW
the quantum yield of the CS state which is assumed to be close to
unity (0.99) based on previous experiments:
ð1 ꢀ 4 ÞEhn
1
E
cs
¼
(1)
F
cs
Electron lifetime measurements
ꢀ2
cm ) using a certified (Oriel 91150V) monocrystalline silicon
reference cell traceable to the National Renewable Energy
Laboratory (NREL).
Impedance spectroscopy measurements were carried out on
a Gamry Reference 600 potentiostat–galvanostat–ZRA. Devices
were measured in a 2 electrode set-up and illuminated with
a white LED at varying intensities. The bias potential was set
equal to the Voc obtained at a given LED light intensity with the
measurement taking place under illumination at that intensity.
AC amplitude was set at ꢃ10 mV with a frequency range of
50 kHz to 0.1 Hz. Equivalent circuit modelling was achieved
using a transmission line model.
Dye loadings
Dye loadings were measured for fast dyed devices which had
been dyed for different lengths of time by pumping 40 mM tert-
Bu NOH
4
solution through the device cavity to desorb the
(aq)
2
dyes from the TiO surface followed by UV-visible spectroscopic
analysis of the dye solutions. Dye uptake versus time for passive
dyed devices was measured using a method adapted from Hol-
Conclusions
42
liman et al. where TiO
varying amounts of time (0–18 h) before dye desorption using
0 mM tert-Bu NOH(aq) and colorimetric measurement by
2
films were immersed in dye solutions for
Two key goals for DSCs are to increase device efficiency and to
improve device processing towards manufacturing scale-up. We
have shown that ultra-fast co-sensitization is possible in a period
of minutes using several dye combinations leading to increased
4
4
UV-visible spectroscopy.
Jsc with EQE data confirming greater photon harvesting across
Photoacoustic (PAC) measurements
the solar spectrum. We have also reported the first example of
ultra-fast tri-sensitization. As part of this approach, we have
synthesized a new, yellow triarylamine dye (6) to harvest more
photons at the blue end of the spectrum where N719 is less
efficient. The high 3 of (6) is crucial for this because a lower dye
loading of (6) is required which does not impair the N719
loading. Light soaking data for (6) have also been measured
showing excellent stability over 1800 h compared to N719. Our
data also show that successful co-sensitization is not as simple as
mixing dyes with coordinating spectral responses. Instead, to
2 2
TiO films were prepared by doctor blading one layer of TiO
colloid (DSL18NRT, Dyesol) onto glass microscope slides and
calcining at 723 K for 30 min. A reference film was dyed in
MnTPPS for several hours to give an absorbance of 0.2 at the
excitation wavelength (480 nm for N719 and 420 nm for (6))
before subtraction of the TiO background signal. Separate films
2
were partially saturated with (6), N719 or (6) + N719 in ethanol
for 2–5 min to match the reference absorbance and to avoid
multi-layer or multi-photonic effects. Pulsed laser excitation was
1
3326 | J. Mater. Chem., 2012, 22, 13318–13327
This journal is ª The Royal Society of Chemistry 2012

View Online
achieve increased device efficiency, the sensitization procedure
must be carefully controlled to optimize dye loading and inter-
dye interactions on the metal oxide surface to minimize recom-
bination processes. In addition, each partner dye within the
device must not interfere with the successful operation of any
other. We have used PAC to measure HOMO, LUMO and lmax
energy levels of (6) and N719 and our data suggest that Voc is set
by the lowest LUMO level of the co-sensitizing dyes used. Thus,
for N719, any additional dye LUMO levels must be above that
12 S. Pack, H. Choi, C. Kim, N. Cho, S. So, K. Song, M. K. Nazeeruddin
and J. Ko, Chem. Commun., 2011, 47, 2874.
1
3 J. N. Clifford, E. Palomares, M. K. Nazeeruddin, R. Thampi,
M. Gr €a tzel and J. R. Durrant, J. Am. Chem. Soc., 2004, 126, 5670.
14 H. Choi, S. Kim, S. O. Kang, J. Ko, M.-S. Kang, J. N. Clifford,
A. Forneli, E. Palomares, M. K. Nazeeruddin and M. Gratzel,
€
Angew. Chem., Int. Ed., 2008, 47, 8259.
1
5 F. Inakazu, Y. Noma, Y. Ogomi and S. Hayase, Appl. Phys. Lett.,
008, 93, 093304.
16 R. Y. Ogura, S. Nakane, M. Morooka, M. Orihashi, Y. Suzuki and
2
K. Noda, Appl. Phys. Lett., 2009, 94, 073308.
7 K. Lee, S. Woong Park, M. Jae Ko, K. Kim and N. Park, Nat. Mater.,
009, 8, 665.
1
2
for N719 or the TiO Fermi level and Voc will drop. In summary,
2
the approach of using ultra-fast multiple dyeing to extend the
spectral response represents huge promise for DSCs because it is
rapid (5 min), reproducible and scalable and, by using different
dyes, it is possible to tune device response to particular spectral
regions.
1
8 J.-J. Cid, J.-H. Yum, S.-R. Jang, M. K. Nazeeruddin, E. Martinez-
Ferrero, E. Palomares, J. Ko, M. Gr €a tzel and T. Torres, Angew.
Chem., Int. Ed., 2007, 46, 8358.
9 D. Kuang, P. Walter, F. N u} esch, S. Kim, J. Ko, P. Comte,
S. K. Zakeeruddin, M. K. Nazeeruddin and M. Gr €a tzel, Langmuir,
2007, 23, 10906.
1
2
0 X.-F. Wang, Y. Koyama, O. Kitao, Y. Wada, S. Sasaki, H. Tamiaki
and H. Zhou, Biosens. Bioelectron., 2010, 25, 1970.
Acknowledgements
21 J. N. Clifford, A. Forneli, H. Chen, T. Torres, S. Tan and
E. Palomares, J. Mater. Chem., 2011, 21, 1693.
We gratefully acknowledge EPSRC (EP/E03585X/1) support for
AC, ERDF-WAG LCRI funding for SPARC (MLD and AC),
Iraqi Govt support for MM and Swansea University EPSRC
National Mass Spectometry Service Centre. We thank EPSRC
for funding the National Crystallography Service (Southampton
and Newcastle), and Diamond Light Source for access to
synchrotron facilities. We thank Tata Steel Colors Ltd and
Dyesol Ltd for help with light soaking and EQE measurements
and Dr Piers Barnes (ICL) for assistance with EIS transmission
line modelling. CS and OHF acknowledge support for project
PTDC/QUI/QUI/099730/2008 from Funda c¸ a˜ o para a Ci ^e ncia e
a Tecnologia (Portugal) through the COMPETE programme
and from FEDER (European Union).
22 C.-M. Lan, H.-P. Wu, T.-Y. Pan, C.-W. Chang, W.-S. Chao,
C.-T. Chen, C.-L. Wang, C.-Y. Lin and E. W.-G. Diau, Energy
Environ. Sci., 2011, in press.
2
3 A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran,
M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin
and M. Gr €a tzel, Science, 2011, 334, 629.
2
4 P. J. Holliman, M. L. Davies, A. Connell, B. Vaca Velasco and
T. M. Watson, Chem. Commun., 2010, 46, 7256.
2
5 S. Hwang, J.-H. Lee, Cha. Park, H. Lee, Cha. Kim, Chi. Park,
M.-H. Lee, W. Lee, J. Lee, K. Kim and Chu. Kim, Chem.
Commun., 2007, 4887.
2
6 C. Serpa, J. Schabauer, A. P. Piedade, C. J. P. Monteiro,
M. M. Pereira, P. Douglas, H. D. Burrows and L. G. Arnaut, J.
Am. Chem. Soc., 2008, 130, 8876.
2
7 L. Alibabaei, J.-H. Kim, M. Wang, N. Pootrakulchote, J. Teuscher,
D. Di Censo, R. Humphry-Baker, J.-E. Moser, Y.-J. Yu,
K.-Y. Kay, S. M. Zakeeruddin and M. Gr €a tzel, Energy Environ.
Sci., 2010, 3(11), 1757.
2
8 K.-F. Chen, C.-W. Chang, J.-L. Lin, Y.-C. Hsu, M.-C. P. Yeh,
C.-P. Hsu and S.-S. Sun., Chem.–Eur. J., 2010, 16, 12873.
9 S. Tatay, S. A. Haque, B. O’Regan, J. R. Durrant, W. J. H. Verhess,
J. M. Kroon, A. Vidal-Ferran, P. Gavi n~ a and E. Palomares, J. Mater.
Chem., 2007, 17, 3037.
Notes and references
2
1
K. Hara and H. Arakawa, Dye-Sensitized Solar Cells, in Handbook of
Photovoltaic Science and Engineering, ed. A. Luque and S. S. Hegedus,
WILEY, Chichester, UK, 2003.
3
3
0 P. J. Holliman, M. L. Davies, unpublished work.
2
3
B. O’Regan and M. Gr €a tzel, Nature, 1991, 353, 737.
M. K. Nazeeruddin, F. De Angelis, S. Fantacci, A. Selloni,
G. Viscardi, P. Liska, S. Ito, B. Takeru and M. Gr €a tzel, J. Am.
Chem. Soc., 2005, 127, 16835.
1 S. Ito, S. M. Zakeeruddin, R. Humphry-Baker, P. Liska, R. Charvet,
P. Comte, M. K. Nazeeruddin, P. P ꢀe chy, M. Takata, H. Miura,
S. Uchida and M. Gr €a tzel, Adv. Mater., 2006, 18, 1202.
2 Y. Chen, Z. Zang, C. Li, W. Wang, X. Wang and B. Wang, New J.
Chem., 2005, 29, 773.
3
3
4
5
6
Y. Cao, Y. Bai, Q. Yu, Y. Cheng, S. Liu, D. Shi, F. Gao and P. Wang,
J. Phys. Chem. C, 2009, 113, 6290.
M. K. Nazeeruddin, P. P ꢀe chy and M. Gr €a tzel, Chem. Commun., 1997,
3 X. Chen, J. Guo, X. Peng, M. Guo, Y. Xu, L. Shi, C. Liang, L. Wang,
Y. Gao, S. Sun and S. Cai, J. Photochem. Photobiol., A, 2005, 171,
1705.
2
31.
K. Hara, M. Kurashige, Y. Dan-oh, C. Kasada, A. Shinpo, S. Suga,
K. Sayama and H. Arakawa, New J. Chem., 2003, 27, 783.
3
4 (a) M. K. Nazeeruddin, S. M. Zakeeruddin, R. Humphry-Baker,
M. Jirousek, P. Liska, N. Vlachopoulos, V. Shklover, C.-H. Fischer
and M. Gr €a tzel, Inorg. Chem., 1999, 38, 6298; (b) F. Lenzmann,
J. Krueger, S. Burnside, K. Brooks, M. Gr €a tzel, D. Gal, S. R u€ hle
and D. Cahen, J. Phys. Chem. B, 2001, 105, 6347.
5 F. De Angelis, S. Fantacci, E. Mosconi, M. K. Nazeeruddin and
M. Gr €a tzel, J. Phys. Chem. C, 2011, 115, 8825.
6 CrystalClear-SM Expert 2.0 r5 (Rigaku, 2010); APEX2 (Bruker AXS,
7
8
T. Horiuchi, H. Miura and S. Uchida, Chem. Commun., 2003, 3036.
(a) G. Zhang, H. Bala, Y. Cheng, D. Shi, X. Lv, Q. Yu and P. Wang,
Chem. Commun., 2009, 2198; (b) H. Im, S. Kim, C. Park, S.-H. Jang,
C.-J. Kim, K. Kim, N.-G. Park and C. Kim, Chem. Commun., 2010,
3
46, 1335.
9
(a) J.-H. Yum, S. Jang, R. Humphry-Baker, M. Gr €a tzel, J.-J. Cid,
T. Torres and M. K. Nazeeruddin, Langmuir, 2008, 24, 5636; (b)
T. Bessho, S. M. Zakeeruddin, C.-Y. Yeh, E. Wei-Guang Diau and
M. Gr €a tzel, Angew. Chem., Int. Ed., 2010, 49, 6646.
3
2
010).
3
3
7 R. Hooft, Collect: Data Collection Software, Nonius B.V., 1998.
8 G. M. Sheldrick, SADABS, Version 2.10, Bruker AXS Inc., Madison,
Wisconsin, USA, 2003.
1
0 (a) J.-H. Yum, P. Walter, S. Huber, D. Rentsch, T. Geiger, F. N u€ esch,
F. De Angelis, M. Gr €a tzel and M. K. Nazeeruddin, J. Am. Chem.
Soc., 2007, 129(24), 10320; (b) H. Choi, J.-J. Kim, K. Song, J. Ko,
Md. N. Nazeeruddin and M. Gr €a tzel, J. Mater. Chem., 2010, 20,
3
9 G. M. Sheldrick, Acta Crystallogr., Sect. A: Found. Crystallogr., 2008,
6
0 L. J. Farrugia, J. Appl. Crystallogr., 1997, 30, 565.
4, 112.
4
4
3
280; (c) T. Geiger, S. Kuster, J.-H. Yum, S.-J. Moon,
M. K. Nazeeruddin, M. Gr €a tzel and F. N u€ esch, Adv. Funct. Mater.,
009, 19, 2720.
1 A. Burke, L. Schmidt-Mende, S. Ito and M. Gr €a tzel, Chem. Commun.,
2007, 234.
2 P. J. Holliman, B. V. Velasco, I. Butler, M. Wijdekop and D. Worsley,
Int. J. Photoenergy, 2008, 2008, 1–8.
2
4
11 T. Ono, T. Yamaguchi and H. Arakawa, Sol. Energy Mater. Sol.
Cells, 2009, 93, 831.
This journal is ª The Royal Society of Chemistry 2012
J. Mater. Chem., 2012, 22, 13318–13327 | 13327