Dye-sensitized solar cells for efficient power
generation under ambient lighting
Marina Freitag
1
†
, Joël Teuscher
2
, Yasemin Saygili
1
, Xiaoyu Zhang
3
, Fabrizio Giordano
4
, Paul Liska
4
,
Jianli Hua
3
, Shaik M. Zakeeruddin
4
, Jacques-E. Moser
2
, Michael Grätzel
4
*
and Anders Hagfeldt
1
*
Solar cells that operate efficiently under indoor lighting are of great practical interest as they can serve as electric power
sources for portable electronics and devices for wireless sensor networks or the Internet of Things. Here, we demonstrate
a dye-sensitized solar cell (DSC) that achieves very high power-conversion efficiencies (PCEs) under ambient light
conditions. Our photosystem combines two judiciously designed sensitizers, coded D35 and XY1, with the copper complex
Cu(
II
/
I
)(tmby) as a redox shuttle (tmby, 4,4′,6,6′-tetramethyl-2,2′-bipyridine), and features a high open-circuit
photovoltage of 1.1 V. The DSC achieves an external quantum efficiency for photocurrent generation that exceeds 90%
across the whole visible domain from 400 to 650 nm, and achieves power outputs of 15.6 and 88.5 μWcm
–2
at 200 and
1,000 lux, respectively, under illumination from a model Osram 930 warm-white fluorescent light tube. This translates
into a PCE of 28.9%.
S
ince the industrial revolution, humans have contributed more
carbon dioxide to the atmosphere than Earth’s plants can
recycle, which has resulted in a global temperature rise. In
2016, CO
2
concentration in the atmosphere passed the mark of
400 ppm (refs 1,2) To realize a ‘low–carbon society’
3
, photovoltaics
will play a key role in energy harvesting
4
. Next to the widely com-
mercialized semiconductor technologies based on crystalline and
thin-film Si solar cells, alternative photovoltaics are emerging
5
.
Apart from thin-film systems such as CuInGaSe
2
(refs 6,7) or
CdTe (ref. 6) cells, there has been a rapid development of perovskite
solar cells
8–10
during the past five years. The latter evolved from dye-
sensitized solar cells (DSCs), which themselves have recently under-
gone major advances as part of novel environmentally friendly
photovoltaic technologies since their first report by Grätzel and
co-workers
11,12
.
Currently, the market for solar cells can be divided into large
module installations for terrestrial power generation and smaller
modules to power portable electronics
13
. DSCs can be used in
both areas, but they hold particular promise within the second cat-
egory. They show an outstanding performance under indoor, con-
ditions with an artificial light source in comparison with other
solar cell technologies
14
. The unique properties of DSCs make
them the best alternative to wired and battery energy sources, as
DSCs are capable of maintaining a high photovoltage even in
diffuse light conditions
15–20
.
DSCs are known to perform well in ambient light; however, very
few studies have been published regarding the performance under
such conditions. In addition, the few existing reports all use iodide-
based electrolytes in combination with a ruthenium-based inorganic
dye
21–23
. However, in 2005, Fukuzumi and co-workers reported that
copper complexes worked well as redox mediators at reduced light
intensities (
∼20 mW cm
–2
) (ref. 24). The bis(2,9-dimethyl-1,10-phe-
nanthroline) copper-based DSCs were further improved by Wang
and co-workers by combining them with an organic sensitizer,
which led to an increase in the power conversion efficiencies
(PCEs) from 7.0% at 100 mW cm
–2
to 8.3% at ∼23 mW cm
–2
air
mass 1.5 global (AM 1.5G) light
25
. With further improvements of
the DSCs and the introduction of new copper-based redox shuttles,
Freitag and co-workers recently surpassed the 10.0% efficiency
mark at 100 mW cm
–2
AM 1.5G light for this family of alternative
redox mediators
26–28
.
Herein, for the first time, we introduce a DSC design that
outperforms other photovoltaic technologies, including GaAs
thin-film solar cells
29
, in terms of efficiency and cost under
ambient light conditions. By judiciously combining two previously
reported chromophores
25
, that is, the donor-π-acceptor (D-π-A)
dye coded D35 (refs 30,31) and the recently discovered benzothia-
diazole-based D-A-π-A sensitizer XY1 (ref. 32), we achieved
highly effective light harvesting in the visible region extending
from 400 to 650 nm and converted the absorbed photons to elec-
trons with an external quantum efficiency (EQE) of 90% (ref. 33).
Using these sensitizers in conjunction with the copper redox
shuttle Cu(
II/I)(tmby)
2
TFSI
2/1
(tmby, 4,4′,6,6′-tetramethyl-2,2′-
bipyridine; TFSI, bistrifluoromethane sulfonimidate) we achieved
a PCE of 28.9% under indoor conditions at 1,000 lux, where lux is
the unit of light illuminance. Importantly, the cells maintain their
high performance over a large domain of light intensities and
spectral distributions, and yield a PCE of 11.3% in AM1.5G
sunlight.
Results and discussion
Photovoltaic performance of co-sensitized DSCs with Cu(II/I)
(tmby)
2
as redox shuttle. Here we exploit fully the advantages of
the Cu(
II/I)(tmby)
2
complexes, which are distinguished by a low
reorganization energy that allows the regeneration of the sensitizer
with a very small driving force of 0.1 V (ref. 25). The ‘blend’ of
1
Laboratory of Photomolecular Science, Institute of Chemical Sciences and Engineering, École Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne,
Switzerland.
2
Photochemical Dynamics Group, École Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland.
3
Key Laboratory for Advanced
Materials and Institute of Fine Chemicals, School of Chemistry and Molecular Engineering, East China University of Science and Technology, 130 Meilong
Road, Shanghai 200237, China.
4
Laboratory for Photonics and Interfaces, Institute of Chemical Sciences, Engineering École Polytechnique Fédérale de
Lausanne (EPFL), 1015 Lausanne, Switzerland.
†
Present address: Department of Chemistry, Ångström Laboratory, Uppsala University, 75126 Uppsala,
Sweden.
*
e-mail: michael.graetzel@epfl.ch; anders.hagfeldt@epfl.ch
ARTICLES
PUBLISHED ONLINE: 1 MAY 2017 | DOI: 10.1038/NPHOTON.2017.60
NATURE PHOTONICS | VOL 11 | JUNE 2017 | www.nature.com/naturephotonics372
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
organic chromophores (Fig. 1a) brings out the outstanding
attributes of both dyes. The recently introduced XY1 sensitizer, a
D-A-π-A dye, has a very high molar extinction coefficient and
spectral response that extends beyond 700 nm, whereas D35 is an
organic D-π-A dye endowed with a judiciously designed
arylamine donor structure, which suppresses electron recapture by
the Cu(
II) complex from the TiO
2
conduction band and generates
a high open-circuit potential
30,32
. Even during the early stage of
this study we found that the XY1 dye in combination with D35
showed impressive photovoltaic metrics, specifically under indoor
and ambient light illumination. A schematic representation of the
energy levels in the device, and molecular structures of the D35
and XY1 dyes and the Cu(
II/I) (tmby)
2
complexes are given in Fig. 1.
The solar cells were fabricated using a sandwich structure
(Fig. 1e,f ). The sensitized photoanode consisted of 8 µm of TiO
2
(4 µm 30NRD + 4 µm scattering layer) on a fluorine-doped tin
oxide (FTO) substrate. Sensitization of the mesoporous TiO
2
layer
was achieved by immersion into a dye solution with different dye
ratios by varying the concentration of the sensitizers. The sensitizing
solutions employed an equimolar mixture of acetonitrile and tert-
butanol as solvent and contained the organic dyes in the molar
ratios listed in Supplementary Table 1.
Using a thermoplastic spacer, we contacted the sensitized film
with a poly(3,4-ethylenedioxythiophene) (PEDOT)-covered FTO
conducting glass that served as the counter electrode
34
. The
copper redox mediator that contained the electrolyte was introduced
through a hole in the back electrode. The electrolyte employed a sol-
ution of 0.2 M Cu(
I) as well as 0.04 M Cu(II) complexes, 0.1 M
LiTFSI (lithium bis(trifluoromethylsulfonyl)imide) and 0.6 M tri-
butyl phosphate (TBP) in acetonitrile. Supplementary Table 1
gives an overview of the photocurrent density–voltage character-
istics of the DSCs co-sensitization series between D35 and XY1.
We calculated the PCE values using equation (1), where J
sc
is the
short-circuit photocurrent density, V
oc
the open-circuit voltage,
FF the fill factor and I
o
the intensity of incident light:
PCE = J
sc
× V
oc
×
FF
I
o
(1)
For standard full AM 1.5G sunlight, we obtained the best conversion
efficiency of 11.3% at a D35:XY1 ratio of 4:1 in the staining solution.
We measured a J
sc
of 16.2 mA cm
−2
and a FF of 68%. V
oc
values over
1.0 V were reached for the whole series, with a maximum of 1.1 V for
pure D35. At a lower solar light intensity of 12 mW cm
−2
,the
b
c
N
N
N
N
Cu
N
O
O
O
O
N
S
N
S S
CN
COOH
N
O
O
O
O
S
COOH
CN
XY1
D35
Cu(tmby)
2
d
0.99
−0.98
E (V) vs NHE
E
F
CB
E*
Excited-state
potential
1.04
−1.35
TiO
2
(Li
+
doped)
XY1
D35
Cu(tmby)
2
2+/1+
0.87 V
a
Cu
2+/1+
(redox couple)
E
ox
Oxidation
potential
e
1 cm
f
FTO
Cu(tmby)
2
electrolyte
Dye/TiO
2
Thermoplastic
spacer
The
PEDOT on FTO
Figure 1 | Molecular photovoltaics with Cu(II/I)(tmby)
2
redox mediator and XY1 and D35 sensitizers. a, Schematic representa tion of the energy levels of
the sensitizers XY1 and D35 with the Cu(tmby)
2
redox couple and possible electron-transfer processes. b–d, Molecular structures of the XY1 dye (b), D35
dye (c) and redox mediator Cu(tmby)
2
(d); the counterion is TFSI. e, Photograph of the DSC photoanode with the co-sensitized mesoporous TiO
2
films
(EPFL logo) and PEDOT counter electrode. f, Schematic of a DSC.
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.60
ARTICLES
NATURE PHOTONICS | VOL 11 | JUNE 2017 | www.nature.com/naturephotonics 373
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
maximum PCE increased to 13% for the same staining solution; these
are exceptionally high values for a solar cell system under these low
light conditions (Fig. 2c). Here it is evident that both dyes comp-
lement each other in terms of photovoltaic performance—XY1 con-
tributes to the large photocurrents because of its high molar
extinction coefficient in the 500–650 nm wavelength region, and
D35 supports the high V
oc
values as it is endowed with groups that
block sterically the access of the Cu(
II)(tmby)
2
complex to the TiO
2
surface. This reduces the ability of the Cu(II) complex to recapture
the photoinjected electrons from the TiO
2
conduction band.
Figure 2d shows the photocurrent dynamics as a function at
various light intensities for the best-performing DSC. The J
sc
depends linearly on the light intensity and hence it is not limited by
mass transport up to full sun illumination.
The incident photon to electron conversion efficiency (IPCE)
spectra for the DSC a t the various co-sensitization ratios, recorded at
a 10% light-emitting diode bias light intensity, are shown in Fig. 3.
For the best-performing co-sensitized sy s tem (4:1, D35:XY1), the
IPCE rea ches its highest values within the series of 91% at 540 nm.
Considering the optical loss of the FT O substr a te, the internal
quantum efficiency ranges between 90% and pra ct ically 100%
(380–600 nm). For comparison, DSCs that incorpora te the single dye
D35 show a lower IPCE of 80% at 540 nm and a narrower spectral
response of the photocurrent, which is limited to the 380–550 nm wave-
length domain. DSCs with only the XY1 sensitizer sho w a larger spectral
coverage up to 640 nm, but a slightly lower IPCE of 88% at 540 nm.
The charge-extraction measurements as a function of V
oc
,pre-
sented in Supplementary Fig. 1a, show a downshift of 70 mV of
the conduction band edge for the best-performing co-sensitization
ratio of the D35:XY1 dye (4:1). More charge is extracted at a
lower V
oc
especially in comparison with a DSC with only one of
the two sensitizers. This is probably related to the optimal dye cover-
age between the two sensitizers on the TiO
2
layer. We employed the
copper(
II/I) complex-based electrolyte to measure electron lifetimes
as a function of V
oc
of the complete DSCs for the co-sensitized
series. These are displayed in Supplementary Fig. 1b as a semiloga-
rithmic plot of electron lifetime as a function of V
oc
. The electron
lifetimes of DSCs made at higher ratios of the D35 dye are longer
than those obtained from the XY1-rich staining solutions. Given
that the same electrolyte composition and redox mediators were
used for all the studied solar cells, this indicates a higher rate of
recombination for the XY1 dye. The curved shape of the slopes
can be attributed to electron recombination from the FTO substrate
(with a thin TiO
2
blocking layer) to the redox mediator. It is prob-
able that the blocking layer contains pinholes, which become more
apparent at lower light intensities
35
.
Transient absorption and steady-state absorption spectroscopy.
To gain better insight into the electron-transfer dynamics of the
D35/XY1 co-sensitized mesoscopic TiO
2
films, time-resolved
nanosecond transient and steady-state absorbance measurements
were performed. In total, we examined seven dye-sensitized TiO
2
electrodes with different D35/XY1 loading ratios of 100% D35,
9% XY1, 20% XY1, 50% XY1, 80% XY1, 90% XY1 and 100%
XY1. The relative surface coverage by the two sensitizers will
differ from their concentration ratios in the staining solution. This
J
sc
(mA cm
−2
)J
sc
(mA cm
−2
)
J
sc
(mA cm
−2
)J
sc
(mA cm
−2
)
0.0
0.0 0
−1
−2
−3
−4
−5
−6
−7
−8
−9
−10
−11
−12
−13
−14
−15
−16
−17
−0.2
−0.4
−0.6
−0.8
−1.0
−1.2
−1.4
−1.6
−1.8
−2.0
−2.2
0.1 sun
ab
cd
100% sun
1.0 sun
1.0 sun
0.7 sun
0.5 sun
0.4 sun
0.1 sun
0.05 sun
50% sun
12% sun
Dark current
At 1 sun
D35:XY1
1:0
4:1
10:1
1:1
1:10
1:4
0:1
D35:XY1
1:0
4:1
10:1
1:1
1:10
1:4
0:1
−2.4
0.1 0.2 0.3 0.4 0.5
V (V)
V (V)
V (V)
0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.70 0.8 0.9 1.0
0.0
0
−5
−10
−15
0.2 0.4 0.6 0.8 1.0 0
0
5
5
10
10
15
15
Time (s)
20
20
25
1.1
Figure 2 | Photovoltaic characteristics of D35/XY1 co-sensitized systems with Cu(tmby)
2
as the redox mediator. a,b,MeasuredJ–V curves at
12 mW cm
−2
(10% sun) (a) and 100 mW cm
−2
(100% sun) (b) for various ra tios of D35 and XY1. c, J–V curves of the champion solar cell at the ratio of
4:1 (D35:XY1) at 100, 50 and 12% sun (solid lines) and the corresponding dark currents (dashed line). d, Dependence of the photocurrent dynamics on the
solar light intensity for the champion solar cell.
ARTICLES
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.60
NATURE PHOTONICS | VOL 11 | JUNE 2017 | www.nature.com/naturephotonics374
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
is confirmed by the absorption spectra of the sensitized TiO
2
(Fig. 4b). Photoinduced absorption spectroscopy (PIA) was
performed to resolve the oxidized dyes spectra on TiO
2
along
with bleaching of their ground states. We use this measurement to
extract two wavelengths to monitor preferentially one oxidized dye
or the other. Figure 4a shows that at 780 nm we mainly observe
oxidized XY1 (oxidized D35 at its minimum absorbance and
oxidized XY1 close to its maximum absorbance), whereas the
1,200 nm probing targets oxidized D35 (oxidized XY1 at its
minimum absorbance and oxidized D35 close to its maximum
absorbance). In between, a smooth transition is observed, which
indicates that we effectively load the films with different
concentration ratios. No significant lateral hole transfer takes
place between both dyes as none of the two oxidized species is
preferentially observed. From the PIA spectra, transient
absorption spectroscopy (TAS) was performed at both
wavelengths (780 and 1,200 nm) in the micro- to millisecond
timescale to follow the transient decay of the dyes’ oxidized
species, and therefore recombination of the oxidized dyes with
TiO
2
electrons or regeneration by the copper-based electrolyte.
The time evolution of the transient absorption at 780 nm, which
reflects the lifetime of oxidized XY1, does not appear to depend
on the dye-concentration ratio. In an inert electrolyte, we observe
similar lifetimes for the oxidized state of both dyes, that is, 2.10
and 1.85 ms for D35 and XY1, respectively (Fig. 4c). On the
addition of the redox electrolyte that contains Cu(tmby)
2
, both
sensitizers are regenerated efficiently. From monitoring the
transient absorption at 1,200 nm, we infer that the regeneration of
D35 occurs with time constants that range from 2 to 10.8 µs and
without any identifiable trend, whereas at 780 nm we observe the
regeneration of XY1 oxidized molecules with time constants that
range from 1.1 to 5.2 µs (Fig. 4d). Overall dye-regeneration yields
are only marginally affected by these variations and are all above
99.5%, using pseudo-first-order rate constants for regeneration
and first-order rate constants for recombination reactions.
DSCs performance under indoor-light conditions. Indoor-light
conditions are very different compared with the solar irradiance
outdoors, because the light intensity is orders of magnitude lower
and the spectra of the indoor-light sources differs greatly from
that of solar emission. Standard indoor illumination has an
intensity between 200 and 2,000 lux. Illuminance is analogous to
100
80
IPCE (%)
60
40
20
0
400 500 600
Wavelength (nm)
700 800
1:0
10:1
4:1
1:1
1:4
1:10
0:1
D35
XY1
D35:XY1
Figure 3 | IPCE spectra of DSCs co-sensitized at various dye ratios.
cd
ab
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Absorbance change
10
−5
2 3 4 5 6 7 89
10
−4
2 3 4 5 6 7 89
10
−3
2 3 4 5 6 7 89
10
−2
Time (s)
At 780 nm (for XY1)
inert electrolyte
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Absorbance change
9
10
−6
2 3 4 5 6 7 8 9
10
−5
2 3 4
Time (s)
At 780 nm (for XY1)
Cu(tmby)
2+/1+
electrolyte
1:0 1.1 μs 1:4 5.2 μs
10:1 1.8 μs
1:10 4.7 μs
4:1 2.6 μs 0:1
3.4 μs
1:1 3.8 μs
500 600 700
500400
Absorbance (a.u.)
0.0
0.2
0.4
0.6
0.8
1.0
XY1
1.2
600 700
800 900
At 780 nm (XY1)
At 1,200 nm (D35)
1,000 1,100
Wavelength (nm) Wavelength (nm)
−2
−4
−ΔT/T ×10
−3
2
0
1:0
10:1
4:1
1:1
1:4
1:10
0:1
D35:XY1
D35:XY1
D35:XY1
D35:XY1
1:0
10:1
4:1
1:1
1:4
1:10
0:1
1:0
10:1
4:1
1:1
1:4
1:10
0:1
Figure 4 | Absorption spectroscopy and time-resolved laser spectroscopy of interfacial electron transfer that involve the D35 and XY1 sensitizers.
a, Photoinduced absorption (PIA) spectra of dye-sensitized TiO
2
electrodes with different dye solution ratios of D35 and XY1 organic dyes. b,UV–vis absorption
spectra of the corresponding sensitized TiO
2
substrates. c,d, Transient absorption decays of the sensitized TiO
2
films after 532 nm excitation with an inert (c)and
Cu(tmby)
2+/1+
-based (d) electrolyte. The red lines in both transient absorption decays represent the fits of the signal a t the 4:1 D35:XY1 dye ratio. The curves of
the transient absorption decays have an offset of 0.3 for better visualization.
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.60
ARTICLES
NATURE PHOTONICS | VOL 11 | JUNE 2017 | www.nature.com/naturephotonics 375
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
irradiance (watts per square metre) commonly used for outdoor
devices, but relates the intensity to how it is perceived by humans
instead of power
36
. As a comparison between the two units, full
solar irradiation, that is, 100 mW cm
–2
AM 1.5G, corresponds to
about 100,000–110,000 lux. Therefore, indoor light at 1,000 lux
corresponds to
∼1% of the standard AM 1.5G solar irradiation
37–39
.
As relatively few reports exist that focus on the use of DSCs for
indoor-light harvesting, no standard indoor-light source has been
established so far. We chose a typical light tube, Osram Warm
White 930 (representative spectra of the light source are given in
Supplementary Figs 5 and 6), to test the DSCs under realistic
indoor conditions. We measured the light intensity that strikes the
DSC by using of a lux meter (TES 1334).
The DSCs fabricated for the study of indoor-light conditions
have a larger area of 2.8 cm
2
compared with the normally used lab-
oratory test DSCs (0.22 cm
2
). The purpose behind this is to have a
smaller discrepancy between ‘real-life application’ and laboratory-
produced solar cells, which tend to have a better efficiency attributed
to their smaller size. Further, the DSCs with a less volatile electrolyte
solvent, propionitrile, were compared with the DSCs using the stan-
dard electrolyte composition that contained acetonitrile
23
.
Otherwise, the same composition for the Cu(
II/I)(tmby)
2
was
applied. The mesoscopic TiO
2
films were sensitized with D35 and
XY1 dyes in a staining solution that contained 0.08 mM D35 and
0.02 mM XY1 (ratio of 4:1).
The PV metrics and power output (P
out
) of the DSCs and GaAs
solar cells (flexi-GaAs (Alta Devices)) were compared under similar
indoor-light conditions (Table 1). The efficiencies of the solar cells
at indoor conditions were calculated with equation (2), where P
out
(W cm
–2
) is the output power of the solar cell and P
in
(W cm
–2
)
is the incident power of the light source, measured by a calibrated
Si-diode or the lux meter:
η =
P
out
P
in
(2)
Given that the spectral region in which the DSCs absorb light is
limited to the visible part of the solar spectrum, it is expected that
their efficiency in the visible region (400–700 nm) is twice the full-
spectrum value
20
. Taking into consideration that the indoor-light
sources emit mostly visible light, a DSC with an 11.3% PCE under
full solar illumination is expected to have an efficiency over 20%
in its particular spectrum of operation indoors. When an indoor-
light source was used, the DSCs outperformed the GaAs solar cells
under similar conditions. The illumination intensity of the indoor-
light sources was varied between 200 and 1,000 lux and, in both
cases, the power output of the DSCs was higher. At 200 lux, the
best-performing DSC yields a power output of 15.6 µW cm
–2
(the
average of 19 samples was 13.5 µW cm
–2
), which is substantially
higher than the 13.1 µW cm
–2
obtained for GaAs. At 1,000 lux,
the best DSC gives 88.5 µW cm
–2
(the average of seven samples
was 80.0 µW cm
–2
), again clearly above the 74.5 µW cm
–2
produced
by GaAs (Fig. 5 and Table 1). This translates into a PCE of 28.9% for
the DSC that operates at 1,000 lux. Low-power electronic devices,
such as wireless sensor nodes or low-power microcontroller units,
typically consume about 100 µW in sleeping mode
40–42
. Roundy
et al. illustrate that at this average power consumption, a 1 cm
3
primary battery (0.8 W h cm
–3
) lasts for about 11 months, before
the node goes into an idle state
43
.A2cm
2
DSC would therefore
be able to render the low-power device fully autonomous
43
.
We advance the following rationale to explain the outstanding
ambient light performance of dye-sensitized solar cells that
employ a combination of the two sensitizers with the Cu(
II/I)
redox-elect electrolyte:
(1) The co-sensitization of mesoscopic titania scaffolds by the two
sensitizers with complementary absorption spectra extends the
light-harvesting ability of the device over a wider spectral
domain, which increases its short-circuit photocurrent. In the
present case, the D35 absorbs mainly blue and green light,
whereas the XY1 covers the yellow and red spectral region.
(2) The sensitizers are judiciously engineered on the molecular scale
such that their electron-acceptor moiety, the cyanoacrylate
group, is attached by coordinative bonding to the titanium
ions surface of the TiO
2
, whereas their arylamine donor group
is positioned away from the interface at the opposite end of
the dye molecule. In the chemical design of the sensitizer struc-
ture their lowest unoccupied molecular orbital is adjusted to
match the energy of the conduction band edge of TiO
2
formed by the Ti (3d) orbitals. In this way, the energy loss
associated with the interfacial electron-injection process is mini-
mized. Optical excitation promotes electrons from the donor to
Table 1 | Photovoltaic metrics for DSCs and GaAs solar cells for indoor-light sources at 200 lux and 1,000 lux.
Solar cell Light source Light intensity (lux) J
sc
(μAcm
–2
) V
oc
(mV) FF (%) P
in
(μWcm
–2
) P
out
(μWcm
–2
) PCE (%)
DSC* Osram Warm White 930 200 27.2 732.0 0.79 61.3 15.6 25.5
DSC
†
200 24.8 700.0 0.79 61.3 13.7 22.3
DSC
*
1,000 138.0 797.0 0.80 306.6 88.5 28.9
DSC
†
1,000 137.2 766.0 0.80 306.6 84.1 27.4
Flexi-GaAs (Alta) Osram Warm White 827 200 20.1 870.0 0.75 70.6 13.1 18.6
Flexi-GaAs (Alta) 1,000 99.0 940.0 0.80 354.0 74.5 21.0
*Acetonitrile-based electrolyte.
†
Propionitrile-based electrolyte. The PCEs for the solar cells are determined from equation (2). Flexi-GaAs solar cells are from Alta Devices measured at GCell with a configuration
of six cells of area 8.33 cm
2
in parallel and in series connected to a mini-module of size 50 cm
2
.
−120
1,000 lux
DSC with Cu(tmby)
2
/acetonitrile
DSC with Cu(tmby)
2
/propionitrile
−140
−100
J
sc
(μA cm
−2
)
−80
−60
−40
200 lux
−20
0
0.0 0.1 0.2 0.3 0.4
V (V)
0.5 0.6 0.7 0.8
Figure 5 | Photovoltaic characteristics of co-sensitized DSCs measured
under indoor-light conditions. J–V curves at 200 lux and 1,000 lux of DSCs
fortheco-sensitizeddyesD35:XY1(4:1)withtheCu(tmby)
2
acetonitrile-
based electrolyte (solid line) and the Cu(tmby)
2
propionitrile-based
electrolyte (dashed line).
ARTICLES
NATURE PHOTONICS DOI: 10.1038/NPHOTON.2017.60
NATURE PHOTONICS | VOL 11 | JUNE 2017 | www.nature.com/naturephotonics376
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
