Factors Affecting the Performance of Champion Silyl-Anchor Carbazole Dye Revealed in the Femtosecond to Second Studies of Complete ADEKA-1 Sensitized Solar Cells

Article abstract of DOI:10.1002/chem.201603059

Record laboratory efficiencies of dye-sensitized solar cells have been recently reported using an alkoxysilyl-anchor dye, ADEKA-1 (over 14 %). In this work we use time-resolved techniques to study the impact of key preparation factors (dye synthesis route, addition of co-adsorbent, use of cobalt-based electrolytes of different redox potential, creation of insulating Al₂O₃layers and molecule capping passivation of the electrode) on the partial charge separation efficiencies in ADEKA-1 solar cells. We have observed that unwanted fast recombination of electrons from titania to the dye, probably associated with the orientation of the dyes on the titania surface, plays a crucial role in the performance of the cells. This recombination, taking place on the sub-ns and ns time scales, is suppressed in the optimized dye synthesis methods and upon addition of the co-adsorbent. Capping treatment significantly reduces the charge recombination between titania and electrolyte, improving the electron lifetime from tens of ms to hundreds of ms, or even to single seconds. Similar increase in electron lifetime is observed for homogenous Al₂O₃over-layers on titania nanoparticles, however, in this case the total solar cells photocurrent is decreased due to smaller electron injection yield from the dye. Our studies should be important for a broader use of very promising silyl-anchor dyes and the further optimization and development of dye-sensitized solar cells.

Full text of DOI:10.1002/chem.201603059

DOI: 10.1002/chem.201603059
Full Paper
&
Physical Chemistry
Factors Affecting the Performance of Champion Silyl-Anchor
Carbazole Dye Revealed in the Femtosecond to Second Studies of
Complete ADEKA-1 Sensitized Solar Cells
[a, b]
[c]
[b]
[a]
´
´
˙
Błazej Gierczyk, Gotard Burdzinski, Mariusz Jancelewicz,
Jan Sobus,
Enrico Polanski,[d, e] Anders Hagfeldt,[d] and Marcin Ziꢀłek*[b]
Abstract: Record laboratory efficiencies of dye-sensitized
solar cells have been recently reported using an alkoxysilyl-
anchor dye, ADEKA-1 (over 14%). In this work we use time-
resolved techniques to study the impact of key preparation
factors (dye synthesis route, addition of co-adsorbent, use of
cobalt-based electrolytes of different redox potential, crea-
tion of insulating Al2O3 layers and molecule capping passiva-
tion of the electrode) on the partial charge separation effi-
ciencies in ADEKA-1 solar cells. We have observed that un-
wanted fast recombination of electrons from titania to the
dye, probably associated with the orientation of the dyes on
the titania surface, plays a crucial role in the performance of
the cells. This recombination, taking place on the sub-ns and
ns time scales, is suppressed in the optimized dye synthesis
methods and upon addition of the co-adsorbent. Capping
treatment significantly reduces the charge recombination
between titania and electrolyte, improving the electron life-
time from tens of ms to hundreds of ms, or even to single
seconds. Similar increase in electron lifetime is observed for
homogenous Al2O3 over-layers on titania nanoparticles, how-
ever, in this case the total solar cells photocurrent is de-
creased due to smaller electron injection yield from the dye.
Our studies should be important for a broader use of very
promising silyl-anchor dyes and the further optimization and
development of dye-sensitized solar cells.
Introduction
ed by inorganic silicon cells and thin film cells (mainly CdTe
and CIGS), used so far. Thanks to the specific properties of
DSSC such as good sunlight conversion efficiency under mod-
erate illumination intensity, possible construction of semi-trans-
parent systems and selection of dyes with different absorption
bands, they can be used in indoor environments, in decora-
tions or on building faÅades (i.e., in the so-called BIPV: build-
ing-integrated photovoltaics). However, their widespread use
requires improvement of the DSSC efficiency (the highest certi-
fied one is 11.9%[5]). Another important problem to be solved
is the long-term stability of DSSC devices.
Dye-sensitized solar cells (DSSC)[1–4] represent one of the most
important emerging photovoltaic technologies, which—thanks
to potential inexpensiveness—have a real chance to become
an alternative to the relatively expensive solar panels, dominat-
´
[a] Dr. J. Sobus, Dr. M. Jancelewicz
´
NanoBioMedical Centre, Adam Mickiewicz University in Poznan
Umultowska 85, 61-614 Poznan (Poland)
´
´
´
[b] Dr. J. Sobus, Dr. G. Burdzinski, Dr. M. Ziꢀłek
Quantum Electronics Laboratory, Faculty of Physics
´
Adam Mickiewicz University in Poznan, Umultowska 85
Total charge separation in DSSC, the quantum yield of which
determines the maximum current flowing in the solar cell, in-
volves a few partial processes: electron injection from the dye
to the semiconductor (usually titania) nanostructure, the re-
generation of the oxidized dye, and charge collection from the
nanostructured photoanode.[1,3,4] To ensure quantum yields of
charge separation close to 100%, some excess energy is
needed for partial separation processes that have to compete
with the undesired pathways of charge transfer. However, the
greater the excess energy (energy loss) the lower the voltage
of the open circuit (VOC). The value of this voltage, along with
the photocurrent intensity, determines the actual efficiency of
the solar cell.[1,6] For many years the dominant DSSC systems
were based on ruthenium dyes and an iodide electrolyte (con-
taining the pairs Iꢀ/I3ꢀ, with multi-electron redox reaction
mechanism), in which an effective process of dye regeneration
´
61-614 Poznan (Poland)
E-mail: marziol@amu.edu.pl
[c] Dr. B. Gierczyk
´
Faculty of Chemistry, Adam Mickiewicz University in Poznan
Umultowska 89b, 61-614, Poznan (Poland)
´
[d] Dr. E. Polanski, Prof. A. Hagfeldt
Laboratory of Photomolecular Science
Institute of Chemical Sciences and Engineering
ꢁcole Polytechnique Fꢂdꢂrale de Lausanne
CH-1015, Lausanne (Switzerland)
[e] Dr. E. Polanski
Present address: Department of Physical Chemistry
University of Tor Vergata–Rome, Via della Ricerca Scientifica 1
00133 Rome (Italy)
Supporting information and the ORCID identification number for the
author of this article can be found under
http://dx.doi.org/10.1002/chem.201603059 including Figures S1–S11, Ta-
bles S1–S4 and the details of electrochemical impedance studies.
Chem. Eur. J. 2016, 22, 1 – 13
1
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
required particularly large driving force.[7–9] Recently, alternative
electrolytes based on cobalt ions (with single-electron redox
pair Co2+/Co3+) have been developed.[10–15] Their redox level is
energetically lower (has more positive potential).[10–12] The use
of these cobalt electrolytes in combination with organic dyes
without ruthenium content, has led to higher VOC values of the
solar cells at similar values of photocurrent to the ones report-
ed before. This in turn has brought laboratory efficiencies of
DSSC using porphyrin dyes as high as up to 13%.[16,17] The
ways to use cobalt complexes in quasi-solid electrolyte for
stable DSSC have been also proposed.[18,19] Calculations sug-
gest that the use of cobalt-based electrolytes and optimization
of all parameters should theoretically bring the DSSC efficiency
to almost 18% (at the total energy loss reduced to 0.4 eV in
contrast to 0.75 eV for iodide electrolytes).[6] However, in the
systems with cobalt electrolytes, the common problem is a fast
process of undesirable recombination of electrons from the
conduction band. It enforces the use of special methods for in-
time-resolved studies have been performed for the dyes with
silyl anchoring groups so far.
Therefore, in this work we employ time-resolved laser spec-
troscopy and electrochemical impedance spectroscopy to
study the dynamics of charge transfer in complete solar cells
sensitized with ADEKA-1 dye on the time scale from 200 fs to
several seconds. We investigate the influence of several optimi-
zation procedures (employed for DSSC sensitized with ADEKA-
1[21]) on the partial charge separation efficiencies. These addi-
tional treatments are: the incorporation of co-adsorbent, the
use of cobalt-based electrolytes of different redox potentials,
creation of insulating Al2O3 layers and passivation of the elec-
trode by molecule capping. Moreover, we report, in more
detail, the route for the most efficient (and free from polymeri-
zation problems) synthesis of ADEKA-1 and related alkoxysilyl-
anchor dyes.
sulation of metal oxide nanoparticles from the electrolyte for Results and Discussion
the majority of applied dyes.
Effect of the synthesis route
A novel family of dyes with silyl anchoring groups have
been proposed by Japanese researchers for employment as
sensitizers in DSSC.[20–24] It has been shown that the covalent
bond made by the alkoxysilyl groups on the surface of the
metal oxide nanoparticles (SiꢀOꢀTi) is much stronger than the
one made by the traditional carboxylic groups (CꢀOꢀTi).
Owing to this property, it was possible to apply the passivation
of the metal oxide surface by few rather strong acids (called
multi-capping). Their molecules fill the voids between the dye
molecules and block the access of the cobalt electrolyte to the
metal oxide nanoparticles. This method could not be applied
for the systems with dyes comprising carboxylic groups be-
cause it resulted in the detachment of dye molecules from the
nanoparticles’ surface.[21] Moreover, silyl-anchor dyes, which
can be used with the multi-capping technique, exhibit remark-
able stability and resistance to water in DSSC devices.[21]
The synthetic procedure for ADEKA-1 published previously[21] is
complex in several aspects. First, the authors have reported
the use of water during work-up of the reaction mixture, that
is, extraction of polar byproducts with this solvent. It is well
known, that alkoxysilane derivatives are highly unstable in hy-
drolytic conditions.[29,30] In the presence of water they undergo
polycondensation (via silanol intermediate). This process is
faster in the presence of ionic substances in water. As a conse-
quence, the use of this procedure results in a mixture of
ADEKA-1 and polymeric products formed during its polycon-
densation. Depending on the time of shaking with water and
on its amount, the organic layer contains a mixture of mono-
meric and polymeric form of the dye or even only polymer
(Figure 1). Moreover, the authors of reference [21] applied
column chromatography on silica to purify the product. The
second aspect of alkoxysilane chemistry is the tendency of
these compounds to react with surface of silica (and other
oxide materials).[31–33] The presence of silanol groups on the
SiO2 surface results in immobilization of SiOR containing mole-
cules linked through SiOSi bond formation. This interaction
strongly decreases the yield of the product. If the elution of
the dye is carried out slowly, almost 100% of the monomer is
covalently bonded to the silica. The fast elution allows separa-
tion of the expected fraction however it reduces its purity.
Work-up of the reaction mixture according to the published
procedure results in isolation of ADEKA-1 (5–10 mg) in low
yields. The composition of the eluent is also important. The au-
thors have used a mixture of chloroform/hexane (1:1 v/v). Both
ADEKA-1 and MK2 dyes are basically immobile on silica in this
solvent system. The use of pure diethyl ether, acetonitrile,
chloroform/methanol (1:1) or chloroform/acetone (1:2) is
needed to elute ADEKA-1. Therefore, in the present work we
describe an upgraded synthetic procedure in the Experimental
Section.
The best dye of this new family so far is ADEKA-1, a silyl
modification of the popular carbazole sensitizer MK2.[25–28] For
ADEKA-1 used alone, the best efficiency of DSSC was 12.5%,[21]
and in combination with SFD-5 (another silyl-anchor dye, ab-
sorbing in more short-wavelength range of the solar spectrum)
the efficiency increased to 12.8%,[23] while in co-sensitization
with LEG4 (a carboxy-anchor dye with triphenylamine electron
donor group) the best efficiency exceeded 14%.[24] According
to our knowledge, this is a record reported laboratory efficien-
cy of DSSC. Therefore, it is quite likely that the dyes with silyl
anchoring groups in combination with cobalt electrolytes can
instigate revival of the DSSC technology.
On the other hand, however, studies of DSSC employing
silyl-anchor dyes have not been undertaken in other laborato-
ries so far. One of the reasons might be difficulties with the
proper dye synthesis. Moreover, no fundamental characteriza-
tion of the dyes with silyl anchoring groups is available. This
problem needs thorough fundamental research that would
bring information about the interaction of the dyes with metal
oxide nanoparticles so that the design of new DSSC would be
more effective. In particular, to the best of our knowledge, no
During the process of synthesis optimization several batches
of ADEKA-1 were obtained, with different content of the poly-
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
2
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
1
Figure 1. ADEKA-1 dye structure (top) and H NMR spectra of: MK2, ADEKA-1, and partially polymerized ADEKA-1 samples (ca. 30 and 5% of monomer, re-
spectively) in CDCl3 (bottom).
merized version of the dye (ranging from purely monomer,
through 50%/50% monomer/polymer mixture, to the 5%/95%
monomer/polymer batch). The polymer content was verified
by NMR spectroscopy in all cases. For simplicity, the samples
will be named (here and in the rest of the text) according to
the monomer content, for example, “60%m” means the
sample consisting of 60% of monomer and 40% of polymer.
The cells made with the following samples were investigated:
100%m, 100%m_old (the same sensitizing solution kept for
more than 2 months), 60, 50 and 5%m. The differences in the
monomer content can be also observed in the normalized sta-
tionary absorption spectra in toluene by examination of the
local absorption minimum around 423 nm (Figure S3). A clear
minimum is present for pure monomer synthesis (both in fresh
and aged solutions), while the spectrum of polymerized dye
(5%m) does not exhibit it. Moreover, it turned out that the
polymerized dye has its absorption maximum moved slightly
to the shorter wavelength, while the absorption tail is extend-
ed to the red. For an equal mixture of monomer and polymer,
the absorption spectrum is more similar to 100%m, but the
minimum at 423 nm is flattened.
General trends observed in transient absorption
Femtosecond transient absorption spectroscopy was used to
measure the relative efficiency of the charge separation occur-
ring up to 3 ns (time window of the ultrafast transient absorp-
tion setup). A simple way to test this is a comparison of the
normalized kinetics probed at 750 nm. As previously reported
by us for the reference MK2 dye,[34] the combined spectrum of
dye radical cation and electrons in titania has a maximum
close to this wavelength. On the other hand, the initial signal
(just after excitation) is proportional to the population of the
singlet excited state of the dye. Therefore, the magnitude of
the signal left after the 3 ns (ratio of the final to initial ampli-
tude) shows the amount of the charge carriers that were suc-
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
3
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
cessfully separated and did not recombine with the dye mole-
cule itself. In principle, the extinction coefficient of radical
cation of monomer and polymer can be different, but assum-
ing that the proportional difference is also maintained for the
excited state of monomer and polymer (initial signal), the nor-
malized kinetics give proper estimation of the charge separa-
tion. The selected, representative kinetics for different ADEKA-
1 cells (as well as the reference kinetics for MK2 cell) are pre-
sented in Figure 2A. As can be seen, there is a large variety of
the residual amplitude for different ADEKA-1 cells, and only
the best configurations match the efficiency obtained for MK2
cell.
tion of stationary absorption spectrum of the sensitized elec-
trode (with subtracted spectrum of an unloaded electrode)
and by taking into account the photon flux spectrum from
AM1.5G data.
As shown in Figure 2B and in Table 1, the amplitude of the
residual signal at 750 nm corresponds well to the total APCE of
the device, independently of the dye series, co-adsorbent addi-
tion and surface treatments (described later). The MK2 dye was
used as a reference here due to its high efficiency on this time-
scale. If it can be matched by ADEKA-1 on the ultrafast time-
scale, the alkoxysilane bonding provides ways to prevent trans-
port recombination on the longer timescales and improve volt-
age of the device (as will be shown later). Judging from
Figure 2 and Table 1 one can see that the polymerized ADEKA-
1 dye (5%m) gives the lowest efficiency. The cells from the
aged synthesis (100%m_old) are marginally better. The best
efficiency on the ultrafast timescale is obtained for 100%m
sample.
More detailed information was obtained from the global
analysis of the transient absorption spectra. The analysis was
performed in a similar way to that recently reported by us for
MK2 cells.[34] The transient absorption spectra are decomposed
by using four time constants plus an offset constant in the
time window of an experiment. The pre-exponential factor
spectra of the fitted components are shown in Figure 3 for se-
lected samples.
The first three components were similar for all cells, so for
a better comparison between different cells their average
values (0.2, 3 and 40 ps) were used in the final fitting as fixed
values. The fourth, sub-ns component (in the range of several
hundred ps) was left as a variable one. The fastest component
(0.2 ps) is comparable to the IRF of the setup and it contains
significant contribution of coherent artifacts that occur during
pump-probe overlap,[35] so for the sake of clarity of the presen-
tation it is absent in the graphs. The two shortest components
(0.2 and 3 ps) are attributed to the electron-injection process
from the excited dyes. As can be seen in Figure 3, the pre-ex-
ponential factor spectra of the 3 ps component are usually
negative in the range from 600 to 750 nm, which probably in-
dicates the rise of the population of dye radical cation and
trapped electrons in titania,[34] as well as a possible contribu-
tion of the decay of the stimulated emission from the excited
state of the dye.
Figure 2. A) Kinetic of transient absorption traces at 750 nm of various cells
sensitized with ADEKA-1 dye. Time zero was shifted to 1 ps in order to pres-
ent the time axis in logarithmic scale. B) Total APCE of the various cells plot-
ted versus the relative amplitude of residual signal at 750 nm in the transi-
ent absorption experiment measured after 3 ns after sample excitation. The
detailed parameters of the cells represented by filled squares (various ver-
sions of ADEKA-1 dye, surface treatments and addition or lack of OTMS co-
adsorbent) are presented in Table 1. The open circle was obtained for MK2
cell.
Longer components (40 ps and sub-ns) should probably be
assigned to the fast recombination between the injected elec-
trons and the dyes.[34,36–41] We think that the factors driving
such recombination can be similar to those recently reported
for DSSCs with porphyrin dyes.[42,43] As has been shown, the
fast recombination rates are directly dependent on the dye
molecule orientation (tilt angle) on the surface, with perpen-
dicular molecules having longest recombination time con-
stants.[43] Similarly to our previous assignment for MK2,[34] the
40 ps component represents the recombination between the
electrons injected to the trap states (yet to be successfully sep-
arated) and the dye. For all ADEKA-1 cells the amplitude of the
40 ps component mirrors the 3 ps one, and their relative ratio
is independent from the cell efficiency. On the contrary, the
It should be noted that the above analysis reflects the initial
charge-separation efficiency with respect to the number of ab-
sorbed photons, so it cannot be directly correlated with short-
circuit current density (JSC) of the cells having different number
of adsorbed dyes. The global solar cell parameter that ac-
counts for the differences in light absorption is total absorbed
photon to current efficiency (APCE). This parameter was esti-
mated by dividing the photocurrent of the full working device
by the number of the photons absorbed by the dye. The
number of the photons absorbed was obtained from integra-
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
4
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Table 1. Photovoltaic parameters of representative cells used for femtosecond transient absorption studies.[a]
Cell
VOC [V]
FF
JSC [mAcmꢀ2
]
Eff. [%]
Nph [1020
s
ꢀ1 mꢀ2
]
Amax
Total APCE
Sres
100%m_old+OTMS
100%m_old+OTMS+ALD“short”
100%m_old+OTMS+multi-capping
100%m_old+OTMS +ALD“long”
5%m
5%m+OTMS
50%m
50%m+OTMS
50%m+OTMS+multi-capping
50%m+OTMS+ALD“short”
50%m+OTMS+ALD“long”
60%m
60%m+OTMS
60%m+OTMS+multi-capping
100%m
0.70
0.62
0.67
0.85
0.61
0.61
0.74
0.72
0.70
0.73
0.92
0.72
0.70
0.69
0.71
0.72
0.46
0.37
0.61
0.59
0.52
0.66
0.54
0.50
0.59
0.48
0.80
0.50
0.55
0.66
0.54
0.45
3.60
2.16
2.95
2.28
1.17
1.27
7.35
6.57
4.22
8.59
0.16
6.87
7.43
4.83
6.98
11.24
1.16
0.50
1.19
1.14
0.37
0.51
2.92
2.39
1.75
3.02
0.12
2.44
2.87
2.20
2.70
3.64
5.4
5.0
3.8
7.2
8.5
6.8
9.3
8.1
6.0
7.8
7.6
9.1
8.4
6.4
8.3
8.7
0.71
0.58
0.48
1.22
1.56
1.04
1.70
1.43
0.85
1.40
1.54
1.56
1.50
0.90
1.32
1.51
0.42
0.27
0.48
0.20
0.09
0.12
0.49
0.51
0.44
0.69
0.01
0.47
0.55
0.47
0.53
0.81
0.32
0.27
0.40
0.19
0.22
0.27
0.46
0.69
0.54
0.59
0.14
0.48
0.64
0.60
0.67
0.70
MK2
[a] Photocurrent density (JSC), open circuit voltage (VOC), fill factor (FF), efficiency (Eff.), corrected photoconversion efficiency (total APCE), absorbance of the
cells close to the maximum (at 500 nm, Amax), number of absorbed photons (Nph), and the amplitude of the normalized residual signal of kinetic at
750 nm (Sres, ratio of the amplitude after 3 ns to the initial one).
“sub-ns” and constant offset components vary a lot for the
cells of different performance. Therefore, we think that the
sub-ns component represents the recombination of the inject-
ed electrons with the dye between the free electrons in titania
and the donor part of the dye molecule. Since it competes
with the constant offset signal of the successfully separated
oxidized dyes and charge carriers, the low amplitude of sub-ns
component and its red shift is mostly beneficial for the suc-
cessful charge separation.
that in better cells the orientation of molecules is more per-
pendicular to the surface, which makes the direction of the
ADEKA-1 dipole moment better aligned with electric field di-
rection, and in turn enhances the Stark shift effect as well.[44–46]
Selected fully assembled cells were also probed with transi-
ent absorption measurements in ns–ms timescale. Resulting ki-
netics were analyzed by using global fitting multi-exponential
function. The resultant time constants with the wavelength-de-
pendent amplitudes (associated with each time constant) are
shown in Figure 4A–C. Besides the three components shown
in the figures, an additional 20 ns component (IRF of the
setup) was also used in the fit to account for the scattered
laser excitation light. Two shorter components (around 200 ns
and 3.5–4 ms) represent the decay of the oxidized dye mole-
cules (radical cation) due to electron-dye recombination and
dye regeneration (by electrolyte). Their spectra can be divided
into two parts. The positive part corresponds to the absorption
of oxidized dye. The negative one shows disappearing deple-
tion of the ground state of dye molecules (bleach signal) and
a possible contribution from the Stark shift effect. The longest
time, having positive amplitude over all wavelengths, repre-
sents absorption of electrons left in titania.[47,48]
It is clearly visible from Figure 3 that the polymerized version
(5%m) has the highest amplitude of the sub-ns signal com-
pared to the blue residual one. The aged dyes also exhibit
higher sub-ns signal than the constant offset one. This means
that in both cases the unwanted recombination has the high-
est contribution. At the moment, we do not have a clear ex-
planation why the aged 100%m solution gives a much worse
performance then the fresh one, since both stationary absorp-
tion (Figure S3) and NMR analysis do not reveal any differences
in the signals of ADEKA-1. One explanation might be that the
dye can suffer from isomerization over time which facilitates
recombination due to closer orientation towards the nanopar-
ticle surface. Another possibility might be that older solution
enhanced formation of dye aggregates on titania surface.
Moving to the 50%m and 60%m synthesis, the recombination
signal is suppressed successfully. Increasing the monomer con-
tent by 10% makes the recombination signal even lower and
extends it characteristic time by 100 ps. The optimized synthe-
sis process (100%m) results in the lowest amplitude of the
sub-ns component and the position where its spectrum
changes the sign from negative to positive (intersection with
DA=0 line) shifts as far as 670 nm. The red shift of the nega-
tive amplitude in the sub-ns component in efficient cells might
be due to the Stark shift effect.[44–46] As more electrons are fully
separated in better cells, they create larger electric field on the
nanoparticle surface. It leads to enhanced negative signal due
to the Stark shift of dye absorption band. Another possibility is
The results for reference MK2 cells (Figure 4A) are similar to
the ones in our previous paper,[34] but in that report we re-
solved one averaged short component (about 1 ms), assigned
to dye regeneration dynamics. Here we find that this compo-
nent should rather be divided into two contributions (200 ns
and 4 ms) of slightly different pre-exponential factor spectra.
Most probably, the faster one contains a significant part of
electron-dye recombination, which might be a continuation of
the sub-ns recombination observed in ultrafast transient ex-
periments (lowering the performance of the solar cell). Indeed,
for the ADEKA-1 cell with the poor efficiency (Figure 4B, old
solution) the contribution of the 200 ns component is higher
than that in the better cell (Figure 4C). At the same time, the
amplitude of the final component (due to the electrons left in
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
5
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Figure 3. Wavelength dependent amplitudes of the indicated time constants obtained from global analysis of transient absorption spectra of full cells sensi-
tized with various versions of ADEKA-1: A) 5%m, B) 100%m_old, C) 50%m, D) 60%m, E) 100%m, F) data obtained for the reference MK2.
titania) is lower in the former than in the latter cell. Moreover,
the effect of the pump pulse intensity investigated for the nor-
malized kinetics at 750 nm (Figure 4D) confirms this assign-
ment. As a result of higher pump intensity (thus higher popu-
lation of injected electrons and oxidized dyes per nanoparticle)
the contribution of the fast component increases (as expected
for the recombination process), while the dynamics of the
middle, microsecond component is unchanged. Therefore, the
latter can be assigned exclusively to the dye-regeneration pro-
cess.
cells corrected for the number of adsorbed dyes (total APCE).
The cells with higher monomer content give smaller recombi-
nation and better performance. Enhanced recombination in
ADEKA-1 compared to the reference MK2 dye can probably be
explained by higher electronic coupling between dye and tita-
nia (due to different anchoring groups) and/or additional mole-
cule rotation possibilities present in the extended anchoring
group of ADEKA-1. In the next sections, the influence of other
factors on the recombination dynamics will be presented. They
include the effect of different levels of the redox couple poten-
tial, the addition of co-adsorbent, capping treatment and the
effect of alumina protection layer. Besides the fs–ms laser spec-
troscopies, electrochemical impedance spectroscopy will also
be employed to test the charge-separation dynamics occurring
on longer time scales.
To summarize this section, we have observed that fast re-
combination plays a crucial role in the performance of the
solar cells with ADEKA-1 dye. The unwanted dye recombina-
tion, probably associated with the orientation of the dyes on
the titania surface, takes place on the sub-ns and ns time
scales. Its amount is correlated with the photocurrent of the
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
6
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Figure 4. A–C) Wavelength-dependent amplitudes with the indicated time constants obtained from the global fit to the flash photolysis results of the MK2
and ADEKA-1 cells. D) Kinetics at 750 nm obtained in flash photolysis experiments for ADEKA-1 cell and the indicated pump-pulse energies. Inset: the initial
signal over a shorter time scale.
Effects of cobalt complex, co-adsorbent and surface treat-
difference in the dye regeneration time constant (Figure S6).
ment
For MK2 cells, the middle component in the global analysis
For selected MK2 and ADEKA-1 cells we checked the effect of
changing the electrolyte redox pair from cobalt-phenenatroline
(Co-phen) to cobalt-bypyridine (Co-bpy). The reason behind it
was to check the effect of shifting the position of electrolyte
redox potential (from 0.62 V vs. NHE for Co-phen to 0.56 V vs.
NHE for Co-bpy) on the observed charge separation dynamics.
The macroscopic parameters were affected in the way analo-
gous to that frequently reported:[1,3,4] for more positive redox
potential the VOC of the cell increases, but at some point the
JSC start to decrease due to not sufficiently fast dye regenera-
tion. Photovoltaic parameters of an example set of MK2 and
ADEKA-1 cells are presented in Table S1. As can be seen, use of
Co-phen instead of Co-bpy yields about 20 mV higher voltage
at the expense of about 1 mAcmꢀ2 lower photocurrent. Similar
effects were observed for ADEKA-1 cells (in a few configura-
tions for which the redox couple effect was measured): lower
(see previous section and Figure 4) was equal to 4 ms for Co-
phen and 3 ms for Co-bpy. Similar shortening of the regenera-
tion kinetics was observed for ADEKA-1 cells—the averaged
time constants decreased from 3.5 ms for Co-phen to 2.5 ms for
Co-bpy. For both dyes this effect can be attributed to the
larger driving force for the dye-regeneration process when bi-
pyridine redox complex is used with respect to the phenena-
troline one. On the other hand, the differences in the fitted
time constants between both Co complexes are not large.
However, bearing in mind the discovered fast recombination
(with the time scales shorter than regeneration dynamics),
even small improvements in the shortening of regeneration
time constant might be responsible for the evident increase in
JSC of the cells.
A common method used to improve the efficiency of the
device is the addition of a co-adsorbent to the dye solution. It
decreases dye agglomeration and increases average distance
between dye molecules, which in turn reduces the energy
transfer between them.[49,50] It can also help in arranging the
dye molecules perpendicularly to the oxide surface, which re-
duces the recombination of electrons from titania to the oxi-
dized dye, in a way that was previously reported for porphyrin
dyes.[42,43] The absorbance was compared for the photoanodes
sensitized with the dye solutions with and without addition of
V
OC by about 20–30 mV and higher JSC by about 10–20%, when
Co-phen is replaced with Co-bpy.
Within experimental error, no differences were observed be-
tween the ultrafast transient absorption results of the cells
filled with the two cobalt electrolytes, both for MK2 and
ADEKA-1. This could be expected as the position of the elec-
trolyte redox potential should not influence either electron in-
jection or electron-dye recombination dynamics. On the con-
trary, nanosecond flash photolysis measurements revealed the
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
7
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
the OTMS (Figure S4). The number of absorbed photons is re-
duced upon addition of OTMS by 10–20% (Table 1).
HOMOꢀ1 to LUMO orbitals (11%).[24] HOMO orbital is localized
on the three thiophene groups closest to the carbazole group,
while the HOMOꢀ1 orbital is localized mainly on the carbazole
group.[24] Therefore, these groups are the electron-donor parts
of ADEKA-1, and their distance from titania surface is probably
crucial for the occurrence of sub-ns recombination. Unlike
MK2, ADEKA-1 can suffer from the rotation around C31ꢀN
amide bond (Figure 1A) of a partial double bond character. It
can result in the twisted structure with carbazole and next
thiophene groups laying closer to the injected electrons on
TiO2 surface and enhance back electron transfer. The above ro-
tation takes place easier in the absence of co-adsorbent mole-
cules and for more polymerized content of ADEKA-1, in agree-
ment with the findings of transient absorption experiments.
Femtosecond transient absorption spectroscopy was also
used to determine the impact of the capping process (single-
and multi-capping was analyzed) and alumina shell created by
the ALD process. Multi-capping sometimes increases or de-
creases the residual signal at 750 nm compared to cells with
just OTMS present (Table 1). The positive action for the early
time behavior is observed in “bad” performing cells, while the
opposite is found for “good” cells (sensitized with fresh solu-
tions of high monomer-content and with OMTS). As described
earlier, OTMS probably acts to straighten the dye molecules
perpendicularly to the surface and capping may act in a similar
way. In the cells with a suboptimal dye orientation, it helps to
set dye molecules more perpendicular to the surface, thus re-
ducing recombination and improving the final signal. In the
cells with optimal dye orientation, capping treatment may
cause slight unbalance in the system, marginally lowering the
final signal.
Figure 5A shows the global analysis spectrum of the 50%m
cell after addition of OTMS as a co-adsorbent (analogous plot
for and 60%m cell is shown in Figure S7A). Indeed, the sub-ns
recombination component is heavily reduced and the residual
one (constant offset) is improved in both cases (to be com-
pared with Figure 3 within of the same synthesis series, but
without OTMS). Moreover, the sub-ns spectrum is moved
a little to the red and the characteristic time is vastly increased.
This most likely indicates that adsorption of OTMS molecules
on titania surface takes up empty spaces between dye mole-
cules and prevents their “tiling” on the surface, consequently
reducing recombination. The overall positive impact of the ad-
dition of OTMS on the sub-ns processes is also reflected by the
higher residual signal of the kinetics at 750 nm for the samples
containing OTMS, compared to their counterparts without it
(e.g., Figure 2A for 60%m).
With all the above results in hand, the differences in initial
charge separation between MK2 and ADEKA-1 molecules can
be compared based on the analysis of their structure and the
results of previously reported molecular modeling. LUMO orbi-
tal is localized close to cyano and amide groups. The optical
transition of ADEKA-1 in the visible region (S0!S1) consists
mainly of the transition from HOMO to LUMO (77%) and
As for the ALD utilized in the photoanode preparation, we
observed crucial importance of the deposition settings. The
impact of these parameters has recently been investigated for
planar structures.[51] In order to investigate the effect of partial
ALD coating, cells were prepared with longer and shorter dep-
osition and purging times. The details of the comparison are
presented in Supporting Information. Only the “long” method
was able to form uniform monolayer of alumina. For the cells
with “long” ALD treatment a very profound recombination
(sub-ns component) signal and low residual (constant offset
component) signal was observed in transient absorption ex-
periments (Figure 5B). This can be explained by the blocking
of electron injection from dye to titania material by 1 nm Al2O3
layer, probably causing significant recombination from the
electrons trapped in surface states of Al2O3 localized close to
the oxidized dyes. These observations are further supported by
a residual signal in the kinetics from Figure 2A (a huge drop in
final kinetic amplitude for the “long” ALD treatment). On the
contrary, “short” ALD method, similarly to the capping treat-
ment, has no significant effect on the ultrafast processes (Fig-
ure S7B).
Electrochemical impedance spectroscopy was used to study
the effects of co-adsorbent and various kinds of surface treat-
ment on the longer-time-scale charge separation. A detailed
presentation of the results is provided in the Supporting Infor-
mation. Briefly, slightly increased transport resistance and small
Figure 5. Wavelength dependent amplitudes of the indicated time constants
obtained from global analysis of transient absorption spectra of full cells
sensitized with ADEKA-1 dye and addition of OTMS: A) 50%m, B) 50%m
with additional “long” ALD treatment. Panel A can be compared to the cor-
responding samples without OTMS presented in Figure 3C.
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
8
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
conduction band shift occur after addition of OTMS (Fig-
ure S9B). As a result, the decrease of the photocurrent for the
cells with OTMS (Table 1 and Table S2) can be explained by the
combined action of a smaller number of absorbed photons
and increased transport resistance. However, in the full opti-
mized cell (when light absorption is improved due to increased
TiO2 thickness and/or additional scattering layers) the overall
impact of OTMS should be positive due to decreased sub-ns
recombination.
tude of sub-ns recombination component significantly in-
creased.
Secondly, flash photolysis measurements have shown that
there is also a connection between performance of the device
and relative amplitude of the observed time components—the
shortest one (assigned to sub-ms electron recombination) and
the longest one (assigned to electron population left in TiO2).
The dominance of the latter indicates a well performing
device. It has also been confirmed that a Co-phenenatroline
redox shuttle yields higher voltages than a Co-bipyridine one,
but at the expense of a photocurrent drop, which is the result
of slightly slower (3.5 vs. 2.5 ms) dye regeneration.
Both molecular capping and creation of insulating Al2O3
layers were found to increase the lifetime of electrons signifi-
cantly increasing the charge-collection efficiency beyond 90%.
Moreover, the capping treatment causes slight shift in conduc-
tion band in the opposite direction compared to OTMS. The
Al2O3 layer on the other hand causes voltage and fill-factor im-
provements, but at the expense of photocurrent drop. Cap-
ping treatment significantly reduces the charge recombination
between titania and electrolyte: the electron lifetime is about
10 times longer for the cells with a single-capping (with re-
spect to the cells with no capping), and further 3–5 times
longer for the cells with a multi-capping (Figure S11C). Similar
increase in electron lifetime is observed for homogenous Al2O3
over-layers on titania nanoparticles (“long” ALD), about 2 times
per 0.2 nm thickness of the layer (Figure S10C). However, the
positive effect of insulating layer on the titania-electrolyte re-
combination is offset by much lower initial charge-separation
efficiency, which results in significantly worse global per-
formance of the cells with homogenous Al2O3 over-layers.
Next, electrochemical impedance spectroscopy has provided
information about the impact of the surface treatments on the
electron transport and recombination between titania and
electrolyte. Both molecular capping and creation of insulating
Al2O3 layers were found to increase the lifetime of electrons.
Moreover, the capping treatment causes slight (up to 55 mV)
shift in conduction band. Finally, we have underlined the im-
portance of the parameters used during the atomic layer dep-
osition of alumina shell, which may lead to disparate results af-
fecting the properties of the device in an unexpected way.
We believe that these results will enable better understand-
ing of alkoxysilyl dyes and lead to development of better
DSSCs in general.
Experimental Section
Dye synthesis
Chemical compounds and apparatus used
Conclusions
All solvents were dried with molecular sieves (3 ꢂ, 20% m/v). Reac-
tions were conducted in flame-dried glassware under argon atmos-
phere. Dimethylformamide (DMF) was allowed to stand over CaH2
for 24 h and distilled under reduced pressure immediately prior to
use.
The dynamics of charge transfer in complete solar cells sensi-
tized with very efficient silyl-anchor ADEKA-1 dye (with differ-
ent preparation parameters) were studied on the time scale
from 200 fs to several seconds. We have also devised a new,
optimized synthesis process for ADEKA-1 dye providing pure
monomer compound with high yield.
1H and 13C NMR spectra were recorded at 298 K on an Agilent DD2
800 spectrometer, operating at frequencies 799.926 and
201.162 MHz, respectively, equipped with 1H[13C/15N] triple reso-
The most important conclusions arise from the transient ab-
sorption spectroscopy results in the fs–ps regime. In this work
we have shown that there is a time component in the global
analysis of these spectra that can be directly connected with
the photon to current conversion efficiency. It is associated
with the undesired ultrafast recombination between dye mole-
cule and oxide surface with characteristic time of hundreds of
picoseconds (sub-ns). With its increase in amplitude, the
number of successfully injected electrons falls down, directly
lowering the performance of the whole device. We have
proved that in the case of the ADEKA-1 dye an increase in the
degree of polymerization of the dye leads to increase of this
unwanted recombination. Therefore, pure monomers in the
dye solution are desirable for the best performing devices. Ad-
dition of the co-adsorbent (OTMS) was found to decrease the
unwanted sub-ns recombination component as well, probably
by reorientation of the tilt of dye molecules on the surface of
the oxide. Creation of the uniform alumina shell led to de-
creased efficiency of initial charge separation, with the ampli-
1
nance probe (908 pulse width: H: 13.6 ms; 13C: 6.9 ms). The sample
concentrations were 0.01m. The 1D spectra were recorded by
using standard one-pulse sequence (s2pul), 2D experiments were
made using standard sequences with gradient pulses (gCOSY,
gHMQCAD, gHMBCAD). The 13C NMR spectra were recorded with
1
broadband H decoupling. Elementary analyses were performed on
a Vario EL III analyzer. FT-IR spectra were recorded on a Bruker IFS
66 s spectrometer in KBr disc (1 mg in 200 mg KBr).
Synthesis process
To a stirred solution of 100 mg of MK2 dye (0.105 mmol) in 2 mL of
chloroform, a solution of 14.6 mg oxalyl chloride (0.115 mmol) in
the same solvent (200 mL) was added, followed by addition of
100 mL of DMF. After 1 h the mixture was cooled to 08C and the
solutions of 4-(trimethoxysilyl)aniline (22.3 mg, 0.105 mmol) and
ethyldiisopropylamine (27.1 mg, 0.210 mmol) in chloroform
(200 mL) were added. The mixture was then mixed for 2 h at room
temperature. After the solvent was evaporated, the solid residue
was dissolved in 1 mL of chloroform and 50 mL of methanol was
added. The solid, blackish product was filtered off, washed with
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
9
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
methanol and dried in vacuum. Yield is 110 mg (91%). 1H NMR
(CDCl3): d=8.59 (1H, s; H-29); 8.31 (1H, d, J=1.7 Hz; H-8); 8.14
(1H, d, J=7.4 Hz; H-5); 7.96 (1H, s; H-46); 7.72 (1H, dd, J=8.4,
1.7 Hz; H-10); 7.68 (4H, s; H-33 & 34); 7.49 (1H, dd, J=8.1, 7.5 Hz;
H-3); 7.42 (1H, d, J=8.1 Hz; H-2); 7.41 (1H, d, J=8.4 Hz; H-11);
7.26 (1H, t, J=7.4 Hz; H-4); 7.20 (1H, s; H-14); 7.11 (1H, s; H-26);
7.03 (1H, s; H-18 or H-22); 7.02 (1H, s; H-18 or H-22); 4.39 (2H, q,
J=7.3 Hz; H-44); 3.64 (9H, s; H-36); 2.86 (8H, m; H-38); 1.71 (8H,
m; H-39); 1.46 (16H, m; H-40 & 41); 1.46 (3H, t, J=7.3 Hz; H-45);
1.34 (8H, m; H-42); 0.91 ppm (12H, m; H-43). 13C NMR (CDCl3): d=
155.54 (C-31); 151.4 (C-30); 144.8 (C-25); 143.6 (C-13); 142.7 (C-29);
140.9 & 143.3 (C-17 & 21); 140.8 (C-16); 140.4 (C-1); 139.6 (C-12);
135.9 (C-34); 135.4 & 136.8 (C-20 & 24); 129.8 (C-27); 129.2 (C-28);
128.9 & 129.1 (C-19 & 23); 129.0 & 128.3 (C-18 & 22); 128.5 (C-15);
127.4 (C-26); 126.0 (C-3); 125.7 (C-32 & 35); 125.4 (C-9); 125.1 (C-
14); 123.9 (C-10); 123.4 (C-7); 122.8 (C-6); 120.6 (C-5); 119.5 (C-33);
119.1 (C-4); 117.9 (C-37); 117.6 (C-29); 108.8 & 108.7 (H-2 & H-11);
50.9 (C-36); 37.7 (C-44); 28.0–32.0 (C-38, 39, 40 & 41); 22.6 (C-42);
14.1 (C-43); 13.6 ppm (C-45). IR (KBr): n˜ =3440, 3415, 2928, 2852,
2195, 1683, 1569, 1513, 1491, 1418, 1315, 1196, 1085, 825, 801,
724 cmꢀ1. Elemental analysis: calculated: C 69.93, H 7.27, N 3.65, S
11.15; found: C 70.02, H 7.35, N 3.59, S 11.05. ADEKA-1 structure
and atom numbering are shown in Figure 1. 1H-13C HSQC and IR
spectra are shown in Figures S1 and S2 (in the Supporting Informa-
tion).
Dye deposition
Prior to dye deposition, the photoanodes were dried in 4008C for
30 min to remove any residual water or other contaminants.
Unless otherwise indicated the dye solution had a concentration of
0.2 mm, the solvent used was toluene (dyes without co-adsorbent)
or 9:1 mixture of toluene and acetonitrile (dyes with co-adsorbent).
The dyes used were MK2 (Sigma) and ADEKA-1 (synthesis route de-
scribed above). The co-adsorbent used was isooctyltrimethoxysi-
lane (OTMS, Sigma–Aldrich) with concentration of 0.1 or 0.2 mm.
Working electrodes were placed in the dye solution and stored at
58C for 16–20 h to enable efficient adsorption of the dye.
Cell assembly
Both electrodes were connected together by using a polymer seal
(25 mm Surlyn, Meltronix, Solaronix SA) with conducting surfaces
facing inwards. They were then filled with electrolyte through the
holes in the counter electrode and sealed with the cover glass on
top. The electrolyte consisted of 0.25m Co2+ bis(trifluoromethane)-
sulfonimide (TFSI), 0.035m of Co3+TFSI, 0.1m of LiTFSI and 0.5m
tert-butylpyridine (TBP). The cobalt redox couple used was cobalt-
phenantroline (Co-phen) in most cases (if not stated otherwise).
For comparison studies, some cells were prepared with cobalt-bi-
pyridine complexes (Co-bpy).
ALD and capping treatment
Cell preparation
In some of the cells additional preparation steps were involved in
order to make core-shell structures or incorporate the capping
treatment. Alumina shell was deposited on the photoanodes
before submerging them in the dye by using the atomic layer dep-
osition (ALD) technique. It was performed using R-200 reactor (Pi-
cosun) with deionized water and trimethylaluminium as oxygen
and aluminium sources, respectively. The duration of the deposi-
tion (0.1 or 2 s) and purge (3 or 8 s) steps of the cycle were highly
important. The capping treatment was performed by sequentially
dipping the photoanodes in the appropriate solutions and rinsing
them with toluene, just after taking them out from the dye solu-
tion and right before cell assembly. All capping solutions had
1 mm concentration in 1:1 toluene/acetonitrile solvent. For
a single-capping treatment the electrode was submerged in hepta-
noic acid solutions for 30 min. For multi-capping treatment it was
dipped in n-octadecyl succinic acid for 10 min, n-hexadecyl malonic
acid for 10 min, tetradecylphosphonic acid for 5 min, octylphos-
phonic acid for 5 min, heptanoic acid for 10 min and finally ethyl-
phosphonic acid for 5 min.
Preparation of the photoanode
Glass plates of the size 16ꢃ14 mm2 were cut from the FTO glass
sheet (solar 4 mm thickness, 10 Wsqꢀ1, Nippon Sheet Glass, Japan).
They were cleaned by using a solution of 2 g Degonex detergent
in 1 L of water in an ultrasonic bath for 45 min. Afterwards, they
were further purified by treatment in a UV-O3 system (Model No.
256–220, Jelight Company, Inc.) for 15 min. Next they were sub-
merged in a 40 mm aqueous solution of TiCl4 for 40 min in 708C,
and then rinsed in water and ethanol. Subsequently, an approxi-
mately 8 mm thick layer of mesoporous titania was deposited by
screen printing twice (54T, Estal Mono, Schweiz, Seidengazefabrik,
AG, Thal) using the commercially available screen-printing pastes
(30NR-D or 18NR-T, Dyesol) with 8 min paste settling on 1258C hot-
plate after the first print. No scattering layer was used, since it
would prohibit transient absorption studies and complicate the in-
terpretation of impedance analysis. After printing and verifying
thickness by the needle profilometry, the photoanodes were again
subjected to TiCl4 treatment described before with the concentra-
tion lowered to 20 mm. Finally, the electrodes with the TiO2 pastes
were gradually heated under an airflow at 1258C for 5 min, 3258C
for 5 min, at 3758C for 5 min, at 4508C for 15 min, and at 5008C
for 15 min.
Cell characterization
Current–voltage characterization of the solar cell was performed
with a potentiostat (model M101 with a frequency response ana-
lyzer FRA32M module, Autolab) coupled to a photoelectric spec-
trometer equipped with the option of solar simulator (Photon Insti-
tute, Poland). The sunlight conditions were simulated with a Xe
lamp with AM 1.5 G spectral filter and intensity adjusted to
100 mWcmꢀ2 using a calibrated silicon cell (RR-74, Rera Systems). It
was used to obtain characteristics of the cells subjected to transi-
ent absorption measurements. Part of the current voltage charac-
terization was done on a solar simulator equipped with a 450 W
xenon lamp (Model No. 81172, Oriel) calibrated by using a reference
Si photodiode equipped with an IR-cutoff filter (KG-3, Schott) as
well (cells to be later analyzed by EIS). Electrochemical impedance
spectroscopy (EIS) measurements were performed using SP-300
Counter electrode preparation
Counter electrodes were prepared by cutting 18ꢃ14 mm2 plates
from the FTO-covered glass sheet (LOF Industries, TEC 15 Wsqꢀ1
,
2.2 mm thickness) and had 1 mm conical holes sand-blasted in
them. Afterwards they were cleaned in water and acetone and
dried. Heating to 4008C for 30 min in order to remove any organic
contamination left was followed by drop-wise deposition of 80 mL
of graphene nanoplatelets solution (1 mg per 2 mL of acetone).
Upon drying in low humidity conditions it resulted in a uniform
carbon layer on the glass surface.
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
10
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
[5] www.nrel.gov/ncpv, National Renewable Energy Laboratory (NREL).
potentiostat (BioLogic) over the range from 0.1 Hz to 7 MHz
spread in 80 measuring points distributed evenly on the logarith-
mic frequency scale. The obtained curves were analyzed by using
the ZView software using the transmission line element.[52] More in-
formation about the voltage range and equivalent circuits is pre-
sented in the Supporting Information.
´
[6] J. Sobus, M. Ziꢀłek, Phys. Chem. Chem. Phys. 2014, 16, 14116–141126.
[7] B. O’Regan, M. Grꢄtzel, Nature 1991, 353, 737–740.
[8] M. K. Nazeeruddin, F. De Angelis, S. Fantacci, A. Selloni, G. Viscardi, P.
Liska, S. Ito, B. Takeru, M. Grꢄtzel, J. Am. Chem. Soc. 2005, 127, 16835–
16847.
[9] Q. Yu, Y. Wang, Z. Yi, N. Zu, J. Zhang, M. Zhang, P. Wang, ACS Nano
2010, 4, 6032–6038.
[10] S. M. Feldt, G. Wang, G. Boschloo, A. Hagfeldt, J. Phys. Chem. C 2011,
115, 21500–21507.
[11] D. Zhou, Q. Yu, N. Cai, Y. Bai, Y. Wang, P. Wang, Energy Environ. Sci. 2011,
4, 2030–2034.
[12] M. Pazoki, P. W. Lohse, N. Taghavinia, A. Hagfeldt, G. Boschloo, Phys.
Chem. Chem. Phys. 2014, 16, 8503–8508.
[13] L. Giribabu, R. Bolligarla, M. Panigrahi, Chem. Rec. 2015, 15, 760–788.
[14] K. Lim, M. J. Ju, J. Na, H. Choi, M. Y. Song, B. Kim, K. Song, J.-S. Yu, E.
Kim, J. Ko, Chem. Eur. J. 2013, 19, 9442–9446.
[15] K. D. Seo, I. T. Choi, H. K. Kim, Chem. Eur. J. 2015, 21, 14804–14811.
[16] 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, M. Grꢄtzel, Science 2011, 334,
629–634.
[17] S. Mathew, A. Yella, P. Gao, R. Humphry-Baker, B. F. E. Curchod, N.
Ashari-Astani, I. Tavernelli, U. Rothlisberger, M. K. Nazeeruddin, M. Grꢄt-
zel, Nat. Chem. 2014, 6, 242–247.
All transient absorption studies were performed for fully assem-
bled, working solar cells (the same as those characterized in cur-
rent–voltage measurements). The setup for ultrafast broadband
transient absorption has been described before (Helios spectrome-
ter, Ultrafast Systems, and Spectra Physics laser system).[53] The
pump pulses were set at 500 nm and the IRF (pump-probe cross
correlation function) was about 250 fs (FWHM). The pump pulse
energy of 60 nJ corresponds to energy density of about 30 mJcmꢀ2
.
The transient absorption measurements were performed in the
spectral range of 450–850 nm and in the time range of up to 3 ns.
The nanosecond flash photolysis setup was based on Q-switched
Nd:YAG laser and a 150 W xenon arc lamp as the excitation and
the probing light sources, respectively.[46] The pump pulse wave-
length was set at 532 nm and the energy was set at 0.15 mJ
(energy density of about 300 mJcmꢀ2).
The global analysis of the transient absorption data was performed
by using Surface Explorer software (Ultrafast Systems) and Asufit
program[54] for ultrafast transient absorption and nanosecond flash
photolysis, respectively. Both programs fit a multi-exponential func-
tion (convoluted with IRF) to either the kinetic vectors of a selected
number of singular values (Surface Explorer) or to the kinetic
traces for all analyzed wavelengths (Asufit). As a result of the analy-
sis, the characteristic time constants were obtained as well as the
wavelength-dependent amplitudes associated with them (also
called decay associated difference spectra or pre-exponential factor
spectra).
[18] F. Bella, N. Vlachopoulos, K. Nonomura, S. M. Zakeeruddin, M. Grꢄtzel, C.
Gerbaldia, A. Hagfeldt, Chem. Commun. 2015, 51, 16308–16311.
[19] F. Bella, S. Galliano, C. Gerbaldi, G. Viscardi, Energies 2016, 9, 384.
[20] K. Kakiage, T. Tokutome, S. Iwamoto, T. Kyomen, M. Hanaya, Chem.
Commun. 2013, 49, 179–180.
[21] K. Kakiage, Y. Aoyama, T. Yano, T. Otsuka, T. Kyomen, M. Unno, M.
Hanaya, Chem. Commun. 2014, 50, 6379–6381.
[22] S. K. Matta, K. Kakiage, S. Makuta, A. Veamatahau, Y. Aoyama, T. Yano,
M. Hanaya, Y. Tachibana, J. Phys. Chem. C 2014, 118, 28425–28434.
[23] K. Kakiage, Y. Aoyama, T. Yano, K. Oya, T. Kyomen, M. Hanaya, Chem.
Commun. 2015, 51, 6315–6317.
[24] K. Kakiage, Y. Aoyama, T. Yano, K. Oya, J.-i. Fujisawa, M. Hanaya, Chem.
Commun. 2015, 51, 15894–15897.
[25] N. Koumura, Z.-S. Wang, S. Mori, M. Miyashita, E. Suzuki, K. Hara, J. Am.
Chem. Soc. 2006, 128, 14256–14257.
[26] T. N. Murakami, N. Koumura, T. Uchiyama, Y. Uemura, K. Obuchi, N.
Masaki, M. Kimurac, S. Mori, J. Mater. Chem. A 2013, 1, 792–798.
[27] M. K. Kashif, J. Axelson, N. W. Duffy, C. M. Forsyth, C. J. Chang, J. R. Long,
L. Spiccia, U. Bach, J. Am. Chem. Soc. 2012, 134, 16646–16653.
[28] K. Hara, Z.-S. Wang, Y. Cui, A. Furube, N. Koumura, Energy Environ. Sci.
2009, 2, 1109–1114.
It should be noted that the cells measured by transient absorption
and electrochemical impedance methods were assembled in two
different laboratories with varying conditions. Therefore, it is diffi-
cult to directly compare the absolute efficiencies of the cells used
for transient absorption with those measured by impedance spec-
troscopy.
[29] M.-C. Brochier Salon, M. Naceur Belgacem, Phosphorus Sulfur Silicon
2011, 186, 240–254.
Acknowledgements
[30] K. J. McNeil, J. A. DiCaprio, D. A. Walsh, R. F. Pratt, J. Am. Chem. Soc.
1980, 102, 1859–1865.
[31] P. Van Der Voorta, E. F. Vansanta, J. Liq. Chromatogr. Relat. Technol. 1996,
19, 2723–2752.
This work was supported by NCN (National Science Centre,
Poland) under project 2015/18/E/ST4/00196. J.S. is a holder of
a scholarship funded within Human Capital Operational Pro-
[32] J. W. Goodwin, R. S. Harbron, P. A. Reynolds, Colloid Polym. Sci. 1990,
268, 766–777.
[33] T. Deschner, Y. Liang, R. Anwander, J. Phys. Chem. C 2010, 114, 22603–
22609.
´
gramme, European Social Fund. Katarzyna Pydzinska is kindly
acknowledged for help with the preparation and characteriza-
tion of one series of ADEKA-1 cells. We would also like to
thank Dr. Juan-Pablo Correa-Baena for help with preparation of
some ALD layers and Prof. Shaik Mohammed Zakeeruddin for
help with interpretation of the results.
´
´
[34] J. Sobus, J. Kubicki, G. Burdzinski, M. Ziꢀłek, ChemSusChem 2015, 8,
3118–3128.
[35] M. Lorenc, M. Ziꢀłek, R. Naskre˛cki, J. Karolczak, J. Kubicki, A. Maciejew-
ski, Appl. Phys. B 2002, 74, 19.
[36] X.-H. Zhang, J. Ogawa, K. Sunahara, Y. Cui, Y. Uemura, T. Miyasaka, A.
Furube, N. Koumura, K. Hara, S. Mori, J. Phys. Chem. C 2013, 117, 2024–
2031.
[37] J. Wiberg, T. Marinado, D. P. Hagberg, L. Sun, A. Hagfeldt, B. Albinsson,
J. Phys. Chem. B 2010, 114, 14358–14363.
Keywords: dyes/pigments · cobalt-based electrolytes · dye-
sensitized solar cells · time-resolved laser spectroscopy
[1] A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo, H. Pettersson, Chem. Rev. 2010,
110, 6595–6663.
[2] M. Grꢄtzel, Nature 2001, 414, 338–344.
[3] M. Grꢄtzel, J. R. Durrant, in Nanostructured and Photoelectrochemical Sys-
tems for Solar Photon Conversion (Eds.: M. D. Archer, A. J. Nozik), Imperi-
al College Press, 2008, pp. 503–536.
[38] P. Piatkowski, C. Martin, M. R. di Nunzio, B. Cohen, S. Pandey, S. Hayse,
A. Douhal, J. Phys. Chem. C 2014, 118, 29674–29687.
[39] W.-K. Huang, C.-W. Cheng, S.-M. Chang, Y.-P. Lee, E. W.-G. Diau, Chem.
Commun. 2010, 46, 8992–8994.
[40] B.-G. Kim, K. Chung, J. Kim, Chem. Eur. J. 2013, 19, 5220–5230.
[41] T. Debnath, P. Maity, H. Lobo, B. Singh, G. S. Shankarling, H. N. Ghosh,
Chem. Eur. J. 2014, 20, 3510–3519.
[4] K. Kalyanasundaram, Dye-Sensitized Solar Cells, EPFL Press, 2010.
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
11
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
´
[42] H. Imahori, S. Kang, H. Hayashi, M. Haruta, H. Kurata, S. Isoda, S. E.
Canton, Y. Infahsaeng, A. Kathiravan, T. Pascher, P. Chꢅbera, A. P. Yartsev,
V. Sundstrçm, J. Phys. Chem. A 2011, 115, 3679–3690.
[43] S. Ye, A. Kathiravan, H. Hayashi, Y. Tong, Y. Infahsaeng, P. Chabera, T.
Pascher, A. P. Yartsev, S. Isoda, H. Imahori, V. Sundstrçm, J. Phys. Chem. C
2013, 117, 6066–6080.
[49] J. Sobus, J. Karolczak, D. Komar, J. A. Anta, M. Ziꢀłek, Dyes Pigm. 2015,
113, 692–701.
[50] C. Martꢆn, M. Ziꢀłek, A. Douhal, J. Photochem. Photobiol. C 2016, 26, 1–
30.
[51] I. Iatsunskyi, E. Coy, R. Viter, G. Nowaczyk, M. Jancelewicz, I. Baleviciute,
K. Załe˛ski, S. Jurga, J. Phys. Chem. C 2015, 119, 20591–20599.
[52] F. Fabregat-Santiago, G. Garcia-Belmonte, I. Mora-Serꢀ, J. Bisquert, Phys.
Chem. Chem. Phys. 2011, 13, 9083–9118.
[44] S. Ardo, Y. Sun, A. Staniszewski, F. N. Castellano, G. J. Meyer, J. Am.
Chem. Soc. 2010, 132, 6696–6709.
´
[45] U. B. Cappel, S. M. Feldt, J. Schçneboom, A. Hagfeldt, G. Boschloo, J.
Am. Chem. Soc. 2010, 132, 9096–9101.
[53] J. Idꢆgoras, G. Burdzinski, J. Karolczak, J. Kubicki, G. Oskam, J. A. Anta, M.
Ziꢀłek, J. Phys. Chem. C 2015, 119, 3931–3944.
´
[46] G. Burdzinski, J. Karolczak, M. Ziꢀłek, Phys. Chem. Chem. Phys. 2013, 15,
[54] E. Katilius, J. Hindorff, N. Woodbury, p. ASUFIT program available at
www.public.asu.edu/~laserweb/asufit/asufit.html.
3889–3896.
[47] T. Yoshihara, R. Katoh, A. Furube, Y. Tamaki, M. Murai, K. Hara, S. Murata,
H. Arakawa, M. Tachiya, J. Phys. Chem. B 2004, 108, 3817–3823.
[48] Y. Tamaki, K. Hara, R. Katoh, M. Tachiya, A. Furube, J. Phys. Chem. C
2009, 113, 11741–11746.
Received: June 27, 2016
Published online on && &&, 0000
&
&
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
12
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
Pressing charges: Time-resolved tech-
niques were employed to study the
impact of key preparation factors on
the partial charge separation efficiencies
in ADEKA-1 solar cells. Unwanted re-
combination of electrons from titania to
this alkoxysilyl-anchor dye results in the
lowering of residual signal at long
delays.
&
Physical Chemistry
´
´
J. Sobus, B. Gierczyk, G. Burdzinski,
M. Jancelewicz, E. Polanski, A. Hagfeldt,
M. Ziꢀłek*
&& – &&
Factors Affecting the Performance of
Champion Silyl-Anchor Carbazole Dye
Revealed in the Femtosecond to
Second Studies of Complete ADEKA-
1 Sensitized Solar Cells
Chem. Eur. J. 2016, 22, 1 – 13
www.chemeurj.org
13
ꢁ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ