Solar Energy Materials & Solar Cells 73 (2002) 51–58

A 5% efficient photoelectrochemical solar cell

based on nanostructured ZnO electrodes
.
Karin Keis, Eva Magnusson, Henrik Lindstrom,
Sten-Eric Lindquist, Anders Hagfeldt*
Department of Physical Chemistry, University of Uppsala, Box 532, S-75121 Uppsala, Sweden
Received 5 February 2001; received in revised form 30 May 2001

Abstract

Nanoporous ZnO electrodes, dye-sensitized with a ruthenium bipyridyl complex, were used
as photoanodes in photoelectrochemical solar cells. By improving the interfacial contact
between dyes and ZnO particles in the film, overall solar-to-electric energy conversion
efficiencies of up to 5% were obtained. The solar cell performance was studied as a function of
film thickness, residence time of the electrodes in the dye solution, electrolyte composition and
light intensity. r 2002 Elsevier Science B.V. All rights reserved.

Keywords: Nanostructured; ZnO; Dye; Solar cell

1. Introduction

Nanostructured metal oxide materials is an intensive research area today with

many potential applications. Dye-sensitized nanoporous materials are especially of
interest for solar cells. In those so-called Gr.atzel-type of solar cells the nanometer
sized semiconductor particles are sintered together to form a three-dimensional
network. Upon optical excitation, the chemically adsorbed dye molecules on the
surface inject electrons into the conduction band of the semiconductor. Electrons
then diffuse to the back-contact, and redox species present in the electrolyte solution
regenerate the dye. A closed regenerative cell is achieved by regeneration of redox
species at a counter electrode, draining electrons from the outer circuit. Titanium
dioxide is so far the most studied nanostructured electrode material giving the

*Corresponding author. Fax: +46-18-508-542.

E-mail address: [email protected] (A. Hagfeldt).

0927-0248/02/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 0 2 4 8 ( 0 1 ) 0 0 1 1 0 - 6
52 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58

highest solar-to-electric conversion efficiency [1–4]. Work on other types of

materials, however, is important to get a better understanding of the nanostructured
oxide properties of the photoelectrode, as well as looking for other high efficiency
systems.
In this paper, nanostructured ZnO electrodes with one order of magnitude larger
particles (150 nm) than normally used in these type of systems (5–30 nm) are
investigated. High quality ZnO films are prepared by a novel method for
manufacturing nanostructured semiconductor layers on the conducting substrates.
It is observed that the crucial parameters for optimizing the efficiency of the
photoelectrochemical solar cells based on zinc oxide are the dye-sensitization process
and the preparation of the nanostructured zinc oxide film.

2. Experimental section

The ZnO powder was synthesized by mixing 0.1 M zinc nitrate water solution with
0.1 M triethanolamine [5]. The solution was then allowed to age in an oven at 1001C
for 24 h. The resulting powder was washed with water, centrifuged and finally dried.
The ZnO films were prepared on conducting glass sheets (Hartford Glass Co. Inc.,
fluorine-doped SnO2 glass with a sheet resistance of 8 O/square) by using a new
method for manufacturing nanoporous semiconductor layers developed in our
laboratory [6]. This method involves compression of the deposited particle layer onto
the conducting substrate at room temperature. A colloidal solution was obtained by
mixing ZnO powder with ethanol and stirring the resulting suspension overnight.
Ethanolic solutions with ZnO contents of 20, 33 and 40 percent by weight were
prepared. The suspension was applied onto the conducting substrate by doctor
blading using adhesive tape (Scotch brand) as a frame and spacer. The films were
then allowed to dry in air. A pressure of 1000 kg/cm2 was applied by a 150 ton strain
rod press on the deposited suspension layer placed between two planar steel plates. A
thin plastic film (PolyethyleneFlow density, 0.03 mm, GoodFellow) was placed
between the zinc oxide film and the upper steel plate to prevent adhesion between
them. Resulting films can be directly used in photoelectrochemical measurements or
for dye-sensitization.
The film thickness of the prepared films was determined by profilometry (Tencor
Alpha Step). The surface morphology was investigated by scanning electron
microscopy (SEM, LEO, 1530 Gemini). Surface area and the average pore diameter
of 100 mg samples scraped off the substrate were determined by a gas adsorption
method using a BET adsorption–desorption isotherm (N2 as adsorbate, ASAP 2010,
Micromeritics).
The results on TiO2 films prepared by the compression method have shown that
the firing step is not necessary, since there are no organic additives to burn up and
the electrical contact between particles is achieved by the compression [6]. In the
present study, the ZnO films were sintered shortly after deposition in an airflow at
4001C for 5 min in order to use similar conditions for dye-sensitization as in earlier
studies. The ZnO electrodes were then immersed into 0.5 mM ethanolic solution of
 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58 53

cis-bis(isothiocyanato)bis(2,20 bipyridy1-4,40 dicarboxylato) ruthenium (II) bis-tetra-

butylammonium. This dye is commercially available from Solaronix S.A. (Lausanne,
Switzerland), and was used without further purification. The dye is denoted by N719
below. The amount of adsorbed dye was determined by desorbing the dye from a
14 mm thick ZnO film (dye sensitized for 30 min) into a 1 mM KOH water solution
and measuring the absorption spectrum.
Series of I–V measurements were performed in air and at room temperature using
a sandwich type of solar cell where the dye-sensitized ZnO electrodes were
illuminated through the substrate. The counter electrode was a thermally platinized
conducting glass (5 mM H2PtCl6 in dry isopropanol, heated at 3801C on a
conducting glass substrate for 10 min). As a standard electrolyte we used 0.5 M
LiI/50 mM I2 in 3-methoxypropionitrile. For comparison, 0.5 M 4-tertbutylpyridine
was added to the standard electrolyte solution. The I–V curves were monitored and
recorded using a computerized Keithley Model 2400 source measure unit. The light
source (Light Drive 1000, type 1400-E2/1, Fusion Lighting, Inc.) was calibrated to
give equivalent I–V curves as under direct sunlight. The active electrode area was
typically 0.5 cm2. The results are not corrected with respect to transmission losses in
the conducting substrate.

3. Results and discussion

All results from the I–V measurements presented here have been performed in
several series under a three months period. The photoelectrochemical output of the
solar cells are highly reproducible when repeating the preparation of ZnO electrodes.
The SEM picture of the prepared ZnO film is presented in Fig. 1. The film is
composed of particles with an average diameter of 150 nm. The resulting films are
light-scattering in the visible region due to the large particle size. The solutions with
different content of ZnO resulted in electrodes with thicknesses of 4, 8 and 14 mm.
The BET pore size distribution in the compressed film ranged from 20 to 90 nm with
a maximum at 55 nm. A BET surface area of 16.6 m2/g and a porosity of 49% were
measured on a 14 mm thick electrode. This is in good agreement with a gravimetric
porosity estimation of 53%. From these values the roughness factor is estimated to
be around 600. Thus, morphology, particle size and porosity of the films are similar
to ZnO samples used earlier in our group [5]. However, samples used in Ref. [5] were
prepared by applying ZnO paste with organic additives onto the glass substrate by
doctor blading, followed by a sintering step at 4001C for at least 30 min [5]. For the
latter method, an overall solar-to-electric energy conversion efficiency of about 2%
was obtained as reported earlier [7].
The concentration of dye in the 14 mm thick ZnO film from dye-desorption
measurements was found to be 5.2  105 mol cm3, which is lower but comparable
with the obtained value of 6.9  105 mol cm3 for a 10 mm thick nanoporous TiO2
film. These films were sensitized for 30 min. From optical measurements using a
spectrophotometer with an integrating sphere, an absorptance (A=1TR)
maximum of 0.86 at 530 nm was found for an 8 mm thick ZnO film, dye sensitized
54 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58

Fig. 1. SEM picture of a ZnO electrode.

for 30 min in 0.5 mM N719.1 This indicates that the light harvesting efficiency is
similar in TiO2 and ZnO films.
Recent results from our laboratory showed that the photoelectrochemical output
of the solar cells based on ZnO electrodes are dependent on the residence time in the
dye solution and on the pH of the solution [7]. I–V measurements were therefore
performed varying the electrode residence time in the dye solution. For three
different thicknesses of the ZnO electrodes the results under 100 W/m2 illumination
in terms of open circuit voltage (Voc), short circuit current (Isc), fill factor (FF) and
overall solar-to-electric energy conversion efficiencies (Z) are presented in Table 1.
When increasing the film thickness from 4 to 8 mm, the Isc, at similar residence time in
dye solution, increases by a factor of 2.6 whereas the Voc is nearly the same for all
three thicknesses. For the studied film thicknesses of 4, 8, and 14 mm, the low Isc
values at short residence time in the dye solution (2–10 min) are due to inefficient
light harvesting efficiency. These films are still uncolored close to the back-contact
region. For each film thickness a maximum in Isc was observed at a specific residence
time. Leaving the samples for a very long time in the dye solution (hours) results,
thus, in a decrease of Isc. As discussed in Ref. [7], the possible reason for this
observation is the formation of aggregates between dye molecules and dissolved
Zn2+-ions originating from the ZnO surface. It is important to note that depending
on the rates of the diffusion of the dye into the nanostructure, adsorption of the dye,
dissolution of Zn surface ions and formation of aggregates in the pores of the film,
the outer part of the electrode may be in the process of forming aggregates when the

1
The measurement is uncertain since part of the light may be scattered through the side of the substrate
glass making the value an upper limit of the absorptance maximum. That the light harvesting efficiency is
similar in TiO2 and ZnO film can, however, be concluded (also by visual inspection).
 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58 55

Table 1
Results from I–V characteristics of ZnO electrodes with different film thicknesses under 100 W/m2
illumination using a standard electrolyte solution

Film thickness (mm) Residence time in dye Voc (V) Isc (mA/cm2) FF Z (%)

4 10 min 0.49 0.52 0.47 1.2

4 20 min 0.51 0.55 0.50 1.4
4 1h 0.54 0.46 0.49 1.2
4 24 h 0.55 0.44 0.44 1.1
8 2 min 0.42 0.77 0.61 2.0
8 5 min 0.45 1.1 0.57 2.9
8 10 min 0.48 1.4 0.54 3.5
8 20 min 0.51 1.4 0.55 4.0
8 30 min 0.50 1.4 0.50 3.4
8 1h 0.50 1.2 0.54 3.3
8 3h 0.47 0.93 0.52 2.3
8 24 h 0.50 0.65 0.43 1.4
14 20 min 0.50 1.3 0.50 2.9
14 30 min 0.46 1.3 0.55 3.4
14 1h 0.47 1.0 0.59 2.9
14 24 h 0.46 0.54 0.45 1.1
14 24 ha 0.45 1.3 0.58 3.4
a
Experiment performed with potassium hydroxide in the dye solution.

two first mentioned processes occur close to the back-contact. Thus, it is difficult to
avoid dye-aggregation completely and the residence time has to be optimized for a
given film thickness in order to obtain the highest possible solar cell efficiency.
Beside the possibility in minimizing aggregation during the dye-sensitization
process by careful control of sensitization time, it was proposed that the formation of
Zn2+/dye aggregates is suppressed when an unprotonated dye or a more basic dye
solution is used [7]. This idea is supported in the present work by the following
experiment. By adding an equivalent amount of potassium hydroxide to the dye
solution, a short circuit current of 1.3 mA/cm2, open circuit potential of 0.45 V and
overall efficiency of 3.4% was observed after 24 h in the dye-bath for a 14 mm thick
electrode. Corresponding values after 24 h sensitization when using the usual
sensitization procedure without adding the base were 0.54 mA/cm2, 0.46 V and 1.1%
(see Table 1, 14 mm thick electrode). Thus, for an efficient dye-sensitization process
of ZnO electrodes, dyes with no acidic protons are preferred in order to suppress the
dissolution of Zn surface ions and formation of Zn2+/dye aggregates.
In Fig. 2, we present the effect of light intensity on the short circuit current. The
relationship between light intensity and Isc for all measured thicknesses of ZnO films
is reasonably linear using light intensities up to 400 W/m2. Further increase in light
intensities leads to a flattening out of the curve probably due to mass transport
limitations in the electrolyte.
The effect of adding 4-tertbutylpyridine is displayed in Fig. 3, showing the
current–voltage characteristics obtained with the 14 mm thick ZnO electrode under
100 W/m2 illumination. Conversion efficiencies of 3.4% and 5.0%, fill factors of 0.55
56 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58

Fig. 2. Short circuit current versus light intensity for 4, 8 and 14 mm thick dye-sensitized ZnO electrodes
dye-sensitized for 10, 20, and 30 min, respectively.

Fig. 3. I–V characteristics for 14 mm thick ZnO electrodes (0.42 cm2) sensitized in 0.5 mM N719 dye
solution for 30 min. Illumination intensity100 W/m2. (A). Electrolyte solution: 0.5 M LiI/50 mM I2 in 3-
methoxypropionitrile (B). Electrolyte solution: same as A except that 0.5 M 4-tertbutylpyridine was added
to the electrolyte solution.

and 0.68, Voc of 0.46 and 0.56 V, Isc of 1.3 and 1.3 mA/cm2 were obtained using the
standard electrolyte and an electrolyte with 4-tertbutylpyridine, respectively. Thus,
the addition of 4-tertbutylpyridine increases the fill factor and Voc while keeping Isc
unchanged. The photocurrent action spectrum of this film has a maximum incident
photon-to-current conversion efficiency of 65% at 540 nm (not corrected for
 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58 57

Table 2
Results from I–V characteristics of ZnO electrodes with different film thicknesses under 100 W/m2
illumination using a 4-tertbutylpyridine in electrolyte solution.

Film thickness (mm) Residence time in dye Voc (V) Isc (mA/cm2) FF Z (%)

4 10 min 0.51 0.55 0.49 1.4

8 20 min 0.55 1.4 0.62 4.8
14 30 min 0.56 1.3 0.68 5.0

transmission losses in the conducting substrate). Additional data are presented in

Tables 1 and 2.
To understand why the new compression method gives higher efficiencies of dye-
sensitized solar cells based on ZnO, we have made preliminary studies of the electron
transport properties by time resolved photocurrent transient measurements on bare
ZnO electrodes. For a detailed description of the method see Refs. [8,9]. Two ZnO
electrodes having similar macroscopical properties (film thickness, particle size and
porosity) were prepared by the different methods, i.e. compression and temperature
sintering. The results show that the average traveling time for electrons from the
outer layer of the film to the conducting substrate (electrolyte/electrode illumination)
is similar for the two ZnO films. Thus, the new compression method and the
previously used method give similar results. Therefore, we suggest that the novel
preparation method does not influence the charge transport properties but rather
leads to improved interfacial kinetics between dye molecules and ZnO (e.g. the
exclusion of organic additives in the film preparation process may be of relevance).
The compression method may thus result in more efficient electron injection into the
conduction band of the zinc oxide, and/or suppressed back reactions from the zinc
oxide to the oxidized dye.
In summary, the ZnO films used in this study show suitable properties for solar
energy conversion applications. The improvement in the overall solar cell efficiency
from 2% to 5% in the present work originates from a controlled dye-sensitization
procedure (avoiding dye aggregation) and from the new compression method (the
exclusion of organic additives probably leads to better interfacial kinetics). In a
previous work, we have observed that ZnO films exhibit fast and efficient electron
transport with small recombination losses [5]. In comparison with TiO2, the ZnO
electrodes showed actually better charge transport properties. Also, the light
harvesting efficiency is high and comparable with TiO2 based solar cells. For further
improvement of dye-sensitized ZnO solar cells a much better understanding of the
electron transfer reactions at the ZnO/dye/electrolyte interface is required for
systems prepared under optimal dye-sensitization conditions.

Acknowledgements

This work was supported by the Foundation for Strategic Environmental

Research (MISTRA) and the Swedish National Energy Administration.
58 K. Keis et al. / Solar Energy Materials & Solar Cells 73 (2002) 51–58

References

[1] B. O’Regan, M. Gr.atzel, Nature 353 (1991) 737.

[2] M.K. Nazeerudin, A. Kay, I. Rodicio, R. Humphry-Baker, E. Mueller, N. Vlachopoulos, M. Gr.atzel,
J. Am. Chem. Soc. 115 (1993) 6382.
.
[3] A. Hagfeldt, B. Didriksson, T. Palmqvist, H. Lindstrom, .
S. Sodergren, H. Rensmo, S.-E. Lindquist,
Sol. Energy Mater. Sol. Cells 31 (1994) 481.
[4] A. Hagfeldt, M. Gr.atzel, Acc. Chem. Res. 33 (2000) 269.
[5] K. Keis, L. Vayssieres, H. Rensmo, S.-E. Lindquist, A. Hagfeldt, J. Electrochem. Soc. 148 (2001)
A149.
. E. Magnusson, A. Holmberg, S.-E. Lindquist, A. Hagfeldt, Sol. Energy Mat. Sol. Cells,
[6] H. Lindstrom,
forthcoming.
[7] K. Keis, J. Lindgren, S.-E. Lindquist, A. Hagfeldt, Langmuir 16 (2000) 4688.
[8] P. Hoyer, H. Weller, J. Phys. Chem. 99 (1995) 14096.
. H. Rensmo, A. Hagfeldt, S.-E. Lindquist, S. Sodergren,
[9] A. Solbrand, H. Lindstrom, . J. Phys. Chem. B
101 (1997) 2514.