4688 Langmuir 2000, 16, 4688-4694

Studies of the Adsorption Process of Ru Complexes in

Nanoporous ZnO Electrodes
Karin Keis,† Jan Lindgren,‡ Sten-Eric Lindquist,† and Anders Hagfeldt*,†
Department of Physical Chemistry, Uppsala University, Box 532, S-751 21 Uppsala, Sweden,
and Department of Inorganic Chemistry, Uppsala University, Ångström Laboratory, Box 538,
S-751 21 Uppsala, Sweden

Received September 27, 1999. In Final Form: January 31, 2000

Nanostructured ZnO films consisting of 150-nm sized particles were dye-sensitized with different
ruthenium complexes. Incident photon-to-current conversion efficiencies (IPCE) were found to be dependent
on dye concentration and on residence time of the ZnO film in the dye solution. Results obtained using
infrared and Raman spectroscopies together with traditional two-electrode measurements using a sandwich-
type cell gave evidence of dye agglomeration in the nanostructured ZnO film. Possible reasons for the
agglomeration process in ZnO films are discussed.

Introduction 0.4%.16 They concluded that the low efficiency was due to
Photon-to-current conversion properties of numerous a reduced uptake of the dye and low quantum yield for
semiconducting nanostructured thin film electrodes have charge injection.
been reported; for examples, TiO2,1 SnO2,2 Fe2O3,3 ZnO,4 Recent results from our laboratory showed that bare
CdS, and CdSe.5 Systems based on TiO2 are the most nanostructured ZnO films with 150-nm sized spherical
extensively studied and, at present, the most promising particles have excellent electron transport properties,
for photoelectrochemical systems.6 However, other metal yielding an IPCE value of 90% at 360 nm.17 Thus, it was
oxides, like ZnO with a band gap similar to that of TiO2, shown that ZnO is a possible candidate for highly efficient
could be an alternative material for solar cell applications. dye-sensitized solar cells. However, compared with dye-
Several attempts have been made to use ZnO in photo- sensitized TiO2 solar cells with overall light-to-electric
electrochemical solar cells. Dye-sensitized semiconducting energy conversion efficiencies of up to 10%, moderate
single-crystal ZnO electrodes were studied by Gerischer efficiencies of 2% were obtained for ZnO samples sensitized
and Tributsch as early as 1969.7,8 In 1980, a 2.5% in 0.5 mM ethanolic Ru-based dye solution.18
monochromatic energy conversion efficiency at 562 nm In this paper, the ZnO/dye systems are studied in more
was reported by Matsumura et al.9 In the following years detail. Information about binding modes is obtained from
many different dyes were tried for photosensitization of infrared (IR) and Raman spectroscopies. These results
ZnO electrodes e.g.; chlorophyll,10 phthalocyanides,11 rose- are then used to interpret the photoelectrochemical output
bengal,9,12 and Cu(I)13,14 and Ru(II)15-complexes. In 1994, from solar cell experiments.
Redmond et al. achieved an incident photon-to-current
Experimental Section
conversion efficiency (IPCE) of 13% at 520 nm and an
overall light-to-electric energy conversion efficiency of The synthesis of ZnO powder and preparation of colloidal
dispersion are described elsewhere.17 The electrodes were
* To whom correspondence should be addressed. prepared by spreading out the viscous dispersion of ZnO onto
† Department of Physical Chemistry. transparent conducting glass sheets (Libbey Owens Ford,
‡ Department of Inorganic Chemistry. fluorine-doped SnO2 glass with a sheet resistance of 8 Ω/square).
(1) Hagfeldt, A.; Björksten, U.; Lindquist, S.-E. Sol. Energy Mater. Using adhesive tape (Scotch brand) as a frame and spacer, a
Sol. Cells 1992, 27, 293. layer was formed by raking off the excess of the colloidal solution
(2) Bedja, I.; Hotchandani, S.; Kamat, P. V. J. Phys. Chem. 1994, 98, with a glass rod. Based on preliminary sintering temperature
4133. studies, the samples were sintered at 400 °C in air for 30 min
(3) Björksten, U.; Moser, J.; Grätzel, M. Chem. Mater. 1994, 6, 858.
(4) Holler, P.; Weller, H. J. Phys. Chem. 1995, 99, 14096. to form a nanostructured film electrode. ZnO films used in our
(5) Hodes, G.; Howell, D. J.; Peter, L. J. Electrochem. Soc. 1992, 139, experiments consist of 150-nm sized particles forming a highly
3136. porous structure, shown in Figure 1. Thickness of films were
(6) O’Regan, B.; Grätzel, M. Nature 1991, 353, 737. determined by profilometry (Tencor Alpha Step) and were
(7) Gerischer, H.; Tributsch, H. Ber. Bunsen-Ges. Phys. Chem. 1969, typically 8-10 µm.
73, 251. Four different dyes were used for the sensitization of ZnO
(8) Gerischer, H.; Tributsch, H. Ber. Bunsen-Ges. Phys. Chem. 1969,
73, 850. electrodes. Two dyes were purchased from Solaronix (Lausanne,
(9) Matsumura, M.; Matsudaira, S.; Tsubomura, H. Ind. Eng. Chem. Switzerland): cis-bis(isothiocyanato)-bis(2,2′bipyridyl-4,4′dicar-
Prod. Res. Dev. 1980, 19, 415. boxylato) ruthenium (II) complex (hereafter denoted by N3) and
(10) Miyasaka, T.; Watanabe, T.; Fujishima, A.; Honda, K. J. Am. cis-bis(isothiocyanato)bis(2,2′bipyridyl-4,4′dicarboxylato) ruthe-
Chem. Soc. 1978, 100, 6657. nium (II) bis-tetrabutylammonium (hereafter denoted by N719).
(11) Minami, N.; Watanabe, T.; Fujishima, A.; Honda, K. Ber. The synthesis of the third dye, [bis-(2,2′bipyridyl)] [4-(2-lithium
Bunsesn-Ges. Phys. Chem. 1979, 83, 476.
(12) Kokado, H.; Nayama, T.; Inoue, E. J. Phys. Chem. Solids 1974, carboxylate ethyl)-4′methyl-2,2′-bipyridine] ruthenium bis-
35, 1169.
(13) Alonso-Vante, N.; Ern., V.; Chartier, P. Nouveau Journal de (16) Redmond, G.; Fitzmaurice, D.; Grätzel, M. Chem. Mater. 1994,
Chemie 1983, 7, 1, 3. 6, 686.
(14) Alonso-Vante, N.; Nierengarten, J. F.; Sauvage, J. P. J. Chem. (17) Keis, K.; Vayssieres, L.; Rensmo, H.; Lindquist, S.-E.; Hagfeldt,
Soc., Dalton Trans. 1994, 1649. A. J. Electrochem. Soc., submitted for publication.
(15) Bedja, I.; Kamat, P.; Hua, X.; Lappin, G.; Hotchandani, S. (18) Rensmo, H.; Keis, K.; Lindström, H.; Södergren, S.; Solbrand,
Langmuir 1997, 13, 2398. A.; Hagfeldt, A.; Lindquist S.-E. J. Phys. Chem. B. 1997, 101, 14, 2598.

10.1021/la9912702 CCC: $19.00 © 2000 American Chemical Society

Published on Web 04/07/2000
Absorption of Ru Complexes in ZnO Electrodes Langmuir, Vol. 16, No. 10, 2000 4689

Figure 1. SEM picture of nanostructured ZnO electrode. The

bar corresponds to 1 µm.

hexafluorophosphate (hereafter denoted by C1), is described by Figure 2. IPCE at 520 nm versus sensitizing time obtained
Aranyos et al.19 The fourth dye was trans-[2,6-bis(methyl- from 2-electrode measurement with nanostructured ZnO
benzimidazol-2-yl)-pyridine](2,2′-bipyridyl-4-phosphonic acid) electrodes with thicknesses of 10 µm. Dependence was studied
ruthenium(II) chloride and is denoted as P1 below. The electrodes in three different N3 concentrations: 0.5, 0.25, and 0.125 mM.
were immersed in a room-temperature ethanolic dye solution
while still being warm at ∼80 °C. The concentrations of used dye whereas strong bands using the 514-nm laser were obtained.
solutions are indicated later, and the “sensitizing time” is defined This difference is due to the resonance Raman effect (i.e., an
as the residence time of the electrode in the dye solution. After electronic transfer at 520 nm). Therefore, vibrational transitions
the dye-bath, the excess solution on the electrode surface was are expected mainly from the aromatic system, which is directly
gently dried with tissue paper (Kleenex). The films were then bonded to the ruthenium ion.
dried in air before measurement.
The TiO2 colloidal solution was prepared by hydrolysis of Results
titanium tetraisopropoxide.6 The electrode preparation and dye-
sensitization procedure (0.5 mM N3 dye in ethanol, 24 h in dye-
In Figure 1 we show a scanning electron microscopy
bath) were similar to those for ZnO electrodes. (SEM) picture of the nanostructured ZnO electrode used
In the two-electrode measurements, a sandwich-type solar in our measurements. The ∼150-nm sized particles form
cell was constructed by gently squeezing together the dye- a highly porous structure of the film. The structural and
sensitized ZnO electrode with a platinized conducting glass using photoelectrochemical characterization of this material is
a spring. The electrolyte solution, 0.5 M LiI and 50 mM I2 in described elsewhere.17
3-metoxypropionitrile, was attracted into the cavities of the ZnO Two-Electrode Measurements with Sandwich-
electrode by capillary forces. The working electrode was il- Type Cells. Figure 2 shows IPCE values at 520 nm for
luminated through its substrate by a 450 W xenon lamp together ZnO electrodes with a film thickness of 10 µm plotted
with an 80-mm water filter. The IPCE values were determined against sensitizing time in three different N3 concentra-
at 20-nm intervals between 400 and 800 nm. The results were
not corrected for intensity losses due to light absorption and
tions. Contrary to the very slow dye uptake of ZnO
reflection by the glass support. electrodes (> 24h) reported in the previous literature,16
The overall light-to-electric energy conversion efficiency we observed a very fast dye uptake (minutes to a few
measurements were performed with a Light Drive 1000 as a hours) that is comparable to that of nanostructured films
light source (embodying a microwave-powered sulfur plasma). of TiO2. The strong dark red color indicates that the dye
The lamp simulates the visible part of the solar spectrum rather is adsorbed homogeneously through the ZnO nanoporous
well. There are, however, very low intensities in the ultraviolet film in a few minutes using dye solutions with 0.5 mM
(UV) and IR ranges. The light intensity was typically 60 W/m2, concentration. However, after the initial rapid increase,
measured by a pyranometer. The active electrode area was the IPCE values decreased when electrodes were left in
typically 0.5 cm2. the dye-bath for longer times. Corresponding changes in
Reflection measurements were performed with an integrating
the action spectra are shown in Figure 3. To follow the
sphere using a Cary 2000 spectrophotometer.
IPCE dependence on sensitizing time in more detail, we
Fourier transform infrared (FT-IR) spectral measurements
were performed on dye-sensitized ZnO and TiO2 samples (1.5 × used solutions with lower dye concentrations. Measuring
2 cm2) and on powdered Zn2+/dye complex in Nujol mull between the IPCE for ZnO electrodes sensitized for only a few hours
CaF2 windows. A Digilab-Biorad FTS-45 FT-IR spectrometer in 0.25 and 0.125 mM dye solutions, efficiencies of ∼30%
covering the range 400-4400 cm-1 with a resolution of 2 cm-1 were obtained. Because it was observed that these films
was used for the measurements. Reflection-absorption spectra were still white close to the conducting glass substrate,
were recorded using a reflectance unit (Seagull, Harrick Scientific we concluded that the dye molecules were not fully
Corp.) with incidence and reflectance angles of 45°. Sixteen scans penetrating the films in this time range. Using the same
were averaged in all cases. Baseline corrections were made, and concentrations but longer sensitization times, we observed
if necessary, water vapor bands were subtracted. The Renishaw
that the films were colored all the way to the back contact.
System 1000 micro-Raman spectrometer equipped with a 514-
nm Argon laser (25 mW) and a 780-nm diode laser (20 mW) with All used dye concentrations gave a maximum IPCE value
2 cm-1 resolution was used in recording the Raman spectra. No after a certain sensitizing time (Figure 2). We noticed
useful Raman spectra from the 780-nm laser could be registered, that by decreasing the dye concentration, it took a longer
time to reach the maximum efficiency. The maximum
(19) Aranyos, V.; Grennberg, H.; Tingry, S.; Lindquist, S.-E.; Hagfeldt, IPCE of ∼50% (at 520 nm) was obtained by using a 0.125
A. Sol. Energy Mater. Sol. Cells, in press. mM ethanolic N3 solution for 7 h. It is important to note
4690 Langmuir, Vol. 16, No. 10, 2000 Keis et al.

Figure 3. IPCE action spectra for ZnO electrodes with different

sensitizing times in 0.5 mM N3 solution. Note that the values
for 2 h sensitization time refer to the right-hand axis, whereas
the other curves refer to the left-hand axis.

that different series of measurements gave slightly

different IPCE values, whereas the IPCE dependence on
sensitizing time was always observed (similar to Figure
2).
In Figure 4A, the IPCE action spectra of ZnO/N719 is
shown. The effect of sensitizing time on efficiency is similar
to that of the N3. The IPCE decreased with longer times
in the dye solution. Also, the maximum around 520 nm
flattened out for both dyes (Figures 3, 4A). The situation
was different when the 0.5 mM Pl and C1 solutions were
used. In the case of P1, there was no or little effect of the
sensitizing time on the IPCE values (Figure 4B). In Figure
4C, a slight increase in IPCE with sensitizing time was
observed for C1. Because this increase was slow and did
not reach a maximum within the time of the experiment
(24 h), we suggest that for this dye (with only one carboxylic
group) it takes a comparatively longer time to fill all sites
with a monolayer of the dye. The IPCE values at short
wavelengths (λ < 420 nm, Figures 4A, B, C) are mainly
due to band-gap excitation of ZnO.
Current-Voltage Characteristics. When ZnO elec-
trodes were sensitized in 0.5 mM N719, the short circuit
current (Isc) decreased with longer sensitization time
(Figure 5). A similar behavior was observed with the N3
(not presented here). The highest Isc (6.7‚10-4 A/cm2) was
obtained with shortest sensitization time (2 min) and the
lowest Isc (2.2‚10-4 A/cm2) was obtained with the longest
sensitization time (24 h). The open-circuit voltage (Voc)
varied in the range of the experimental error and could
be considered constant with an average value of 0.48 V
(Figure 6). A highest overall light-to-electric energy
conversion efficiency of 2.1% was obtained with the
electrodes sensitized with N719 for the shortest time (2
min in dye-bath).
In the case of the P1 and C1 dyes, both the short-circuit
currents and open-circuit voltages were unaffected by the
sensitizing time, varying within the experimental error.
The Isc varied between 2.0‚10-4 and 2.5‚10-4 A/cm2, with
Voc being 0.2 V, for P1. For C1, the Isc was between 1.2‚10-4 Figure 4. IPCE action spectra for ZnO electrodes. Effect of
and 1.8‚10-4 A/cm2 and Voc was ∼0.26 V (Figures 5 and sensitizing time in following solutions: (A) 0.5 mM N719; (B)
6). 0.5 mM P1; and (C) 0.5 mM C1.
Optical Measurements. In the optical measurements with increasing sensitizing time. The absorption curves
we observed an overall decreased reflection from electrodes in Figure 7 give the approximate upper limit for absorption
Absorption of Ru Complexes in ZnO Electrodes Langmuir, Vol. 16, No. 10, 2000 4691

Figure 5. The short circuit current density versus sensitizing Figure 7. Absorption spectra for ZnO electrodes with different
time for ZnO sensitized with 0.5 mM N719, C1, and P1 solutions. sensitizing times in 0.5 mM N3.
Light intensity of 60 W/m2.
Moreover, in the case of ZnO, the absorption bands
increased with increasing sensitizing time, except for the
carbonyl band, which was practically unchanged. This
behavior was observed for both the N3 and the N719 dyes
anchored to ZnO films (Figures 8A and 8B). For ZnO/
N719, we observed no carbonyl absorption bands (Figure
8B). In Figure 8C we show the IR spectra for ZnO
sensitized with C1. Only the absorption bands from the
carboxylate group appear, broadened and slightly shifted
compared with the other dyes. There was no change in
the spectra with increasing sensitizing time, except for a
slight increase in absorbance, indicating a somewhat
larger amount of the dye on the surface. In Figure 9, the
IR spectrum of a red precipitate formed by mixing a low
concentration of Zn2+ ions (zinc nitrate) in ethanol with
a N3 solution is shown (see Discussion).
In Figure 10A, Raman spectra for the ZnO/N3 system
are presented. At longer wavenumbers, the vibration
bands from mainly aromatic groups of the dye are
observed. We notice that for longer sensitizing time, the
ZnO vibrational band at 450 cm-1 disappears. In the case
Figure 6. The open circuit voltage versus sensitizing time for of the ZnO/P1 system, the ZnO vibrational band is seen
ZnO sensitized with 0.5 mM N719, C1, and P1 solutions. Light even with the longest (24 h) sensitizing time (Figure 10B).
intensity of 60 W/m2.
Discussion
and were calculated by neglecting the transmission. The The theoretical incident monochromatic photon-to-
absorption maximum at ∼510 nm widened to shorter current conversion efficiency is given by the expression:
wavelengths when the sensitizing time was increased.
Additionally, we observed a change in the shape of the φ(λ) ) LHE(λ) φinj ηc
curve; that is, the absorption increased more quickly
around 650 and 450 nm with increased sensitizing time where LHE(λ) is the monochromatic light harvesting
compared with other parts of the spectrum. efficiency, φinj is the quantum yield for charge carrier
IR and Raman Spectra. IR spectra for a ZnO/N3 injection, and ηc is the efficiency of collecting the injected
system are shown in Figure 8A. The thiocyanate ligand charge at the back-contact. To get the correct φ(λ) values,
at 2100 cm-1 was N-bounded in agreement with the corrections for losses of light in the glass, back-contact,
literature.20,21 Absorptions at 1728, 1610, and at 1378 cm-1 and different interfaces must be made. To understand
correspond to a carbonyl stretching vibration from carboxyl why φ(λ) values of the studied dye-sensitized ZnO solar
groups and an asymmetric and symmetric C-O stretching cells are significantly smaller than those of systems based
vibration from carboxylate groups, respectively. In the on nanostructured TiO2 , the differences in those three
case of TiO2, these three absorption bands have about terms for the two materials need to be considered.
equal intensities, whereas in the case of ZnO, the In a previous work, we found the ηc to be as efficient in
absorption bands from the carboxylate groups dominate. ZnO as for TiO2.17 Several authors have concluded that
(20) Murakoshi, K.; Kano, G.; Yanagida, S.; Miyazaki, H.; Matsumoto,
low-light-harvesting efficiency is due to poor porosity or
M.; Murasawa, S. J. Electroanal. Chem. 1995, 396, 27. thin ZnO films.15,16 Bedja et al. concluded that the weak
(21) Kohle, O.; Ruile, S.; Grätzel M. Inorg. Chem. 1996, 35, 16, 4779. interaction in the ZnO/dye surface results in a poor uptake
4692 Langmuir, Vol. 16, No. 10, 2000 Keis et al.

Figure 9. IR spectrum of the Zn2+/N3 complex in Nujol mull

between CaF2 windows. Nujol bands have been subtracted.

of dye.15 However, from our data we can conclude that the

light-harvesting efficiency for ZnO is similar to that of
systems based on TiO2. The last parameter to consider is
φinj, which may be assumed to be equal for ZnO and TiO2
because the energetics are similar. For the TiO2/N3 system,
φinj is normally observed to be unity. Redmond et al.
explained the low efficiencies for the ZnO/N3 system to
be due to adsorption of dye molecules in a manner that
they do not efficiently inject an electron into the semi-
conductor upon being photoexcited.16 Recent studies of
electron injection dynamics in dye-sensitized ZnO films
showed the presence of multiexponential injection kinetics
with much slower injection times, thus being different
from TiO2.22 Therefore, it seems that the lower efficiencies
in case of ZnO compared with TiO2 can be attributed to
the parameter φinj. Possible reasons for this will now be
presented on the bases of our experimental observations.
The discussion will proceed by considering the following
processes: (a) diffusion of dye into the ZnO nanostructure,
(b) adsorption of the dye to the ZnO surface, and (c)
formation of dye agglomerates in the pores of the ZnO
film.
If it is assumed that only dye molecules in direct contact
with the ZnO surface can inject electrons and contribute
to the photocurrent, we see from our experiment that the
transport and adsorption of the dye to the ZnO -
depending on the concentration - proceeds over minutes
to hours (Figure 2). We noted that the film at the early
stages of the process remained uncolored in a zone close
to the back-contact. The initial increase in IPCE in all
curves in Figure 2 is clearly due to an increase of the
interfacial concentration of effectively injecting dye mol-
ecules. The dye molecules are first adsorbed in the outer
part of the film and, as they penetrate deeper into the
film, the IPCE increases. The flux of dye molecules toward
the back-contact is driven by the concentration gradient
of dye in the solution. If the adsorption is followed by a
desorption of the dye, we have to consider the equilibrium
between dye molecules in solution and dyes adsorbed at
the surface.
Figure 8. IR spectra of ZnO electrodes sensitized in (A) 0.5 When the first monolayer of N3 or N719 is formed in
mM N3, (B) 0.5 mM N719, and (C) 0.5 mM C1 solutions for the outer part of the film, it is ready to proceed to step (c)
different times. The spectrum for TiO2 electrodes sensitized in
0.5 mM N3 solution for 24 h is shown in Figure 8A. Spectra are (22) Asbury J. B.; Wang Y. Q.; Lian T. J. Phys. Chem. B. 1999, 103,
offset from zero absorbance for clarity. 32, 6643.
Absorption of Ru Complexes in ZnO Electrodes Langmuir, Vol. 16, No. 10, 2000 4693

Depending on the rate of the different processes (a)-

(c), the outer part on the electrode may well be in the
process of accumulating dye agglomerates in the pores
when the first dye molecules are reaching the interface
between the back-contact and the ZnO film. Thus, the
latter part is on the way to increasing the efficiency,
whereas the former is in a phase of decreasing the
efficiency. The results in Figures 2 and 3 are consistent
with such processes. The IPCE at 520 nm goes through
an optimum. The increase is due to the increasing number
of effectively injecting dye molecules. The decrease is due
to the light absorbed in the noninjecting agglomerated
dye molecules. The time at which the optimal efficiency
appears at each concentration will depend on the dynamics
of the different competing processes.
In the following we will discuss the adsorption and
accumulation of the dye in the nanostructured ZnO film
in more detail. We propose that the mechanism of dye
agglomeration is due to dissolved Zn2+ ions binding to dye
molecules, forming a complex that stays inside the matrix
of the particles. There is much evidence supporting this
mechanism. First, in a beaker, a solution with a low
concentration of Zn2+ ions in ethanol was mixed with a
N3 solution. After 24 h, a colorless ethanolic solution with
a red precipitate in the bottom was formed. An IR spectrum
of the red precipitate (shown in Figure 9) shows two broad
bands, centered at 1630 and 1370 cm-1 corresponding to
an asymmetric and symmetric C-O stretching vibration
from a carboxylate ion, indicating that all protons in the
dye molecule have been exchanged for Zn2+. Another
indication of the existence of such a complex is that in the
Raman spectra the ZnO band disappears after longer
sensitization times due to covering of the complexes on
the ZnO surface (Figure 10A). The large increase of the
carboxylate bands in the IR data of the sensitized ZnO
electrodes also indicates a situation with all four car-
boxylate groups bonded to Zn2+ ions. The detailed structure
of the formed Zn2+/dye agglomerate can, however, not be
deduced from our data. All these observations indicate
the presence of dye agglomerates in the ZnO/dye system.
From the literature we also note reports on complex
formation between ZnO and different sensitizers due to
the dissolution of zinc ions from the surface. In 1973,
Figure 10. Raman spectra of ZnO electrodes sensitized in (A) Baumeler et al. studied ZnO/dye complexes using 1-(2′hy-
0.5 mM N3 and (B) 0.5 mM P1 solutions for different times. The droxy-4′-X-5′-Y-phenylazo)-2-naphthol, where X and Y
spectrum for P1 powder is presented in Figure 10B. stands for different ligands.24 It was shown that the formed
complexes contain zinc that was originally on the particle
in our aforementioned scheme. After longer sensitizing surfaces. Furthermore, the formation of tail-like complexes
times, in particular at high dye concentrations, the amount have been shown by electron microscopy by Brändli et
of dye in the ZnO films increases tremendously, resulting al.25
in a broad absorption spectrum with high absorbances Interestingly we notice that in the ZnO/C1 and ZnO/P1
even for wavelengths with relatively low extinction systems there is negligible or no formation of the Zn2+/
coefficients (see Figure 7). This result means that after dye complex. The initial increase in IPCE with time
a complete surface coverage, extra dye molecules are corresponds to the filling of adsorption sites with dye
accumulated in the film. These extra dye molecules do molecules (Figure 4B and 4C). At equilibrium conditions,
not inject electrons but are rather shielding the dye there is no additional dye uptake. In the corresponding
molecules in direct contact with ZnO from absorbing light. IR and Raman spectra, there are no observable changes
This shielding lowers the overall φinj and explains the present. The results indicate that protons play an im-
decrease in IPCE values (see Figure 3). The accumulation portant role in the exchange reaction with Zn2+ in ZnO.
of dye molecules in the pores of ZnO also explains the The fewer protons in the dye molecule, the more sup-
observed spectral changes presented in Figure 3. The pressed is the dissolution of Zn2+ ions from the surface of
maximum in the action spectra widens because the filter- the ZnO.
effect is larger to wavelengths with relatively high The formation of dye agglomerates does not take place
extinction coefficients. We also note that Khazraji et al. in the case of TiO2 electrodes but are apparently extremely
have observed similar changes in the photocurrent action important in the case of ZnO. There are many possible
spectra and absorption spectra when sensitizing TiO2 with
monomeric and aggregate forms of Merocyanine 540.23 (24) Baumeler, R.; Rys, P.; Zollinger, H. Helv. Chimica Acta 1973,
56, 251, 2451.
(23) Khazraji, A. C.; Hotchandani, S.; Das, S.; Kamat, P. V. J. Phys. (25) Brändli, R.; Rys, P.; Zollinger, H.; Oswald, H. R.; Schweizer, F.
Chem. B 1999, 103, 4693-4700. Helv. Chim. Acta 1970, 53, 130, 1133.
4694 Langmuir, Vol. 16, No. 10, 2000 Keis et al.

reasons why the dissolution process is not observed in the of formic acid on the ZnO(101h 0) surface, the bridging
case of the TiO2/dye system. One reason is the difference structure has been found to be energetically the most
in surface chemical properties of these two metal oxides. favorable and stable geometry.29,30 Additionally, the
The pH of point of zero charge (pzc) lies at ∼5 for TiO2 and abstraction of acidic hydrogen by the basic surface oxygen
at ∼9 for ZnO. The acid-base properties of colloidal ZnO to form surface hydroxyl groups was observed.
and TiO2 particles have been studied by Bahnemann.26 It
was shown that titration with acid leads to the dissolution Conclusions
of the ZnO colloid already below pH 7.4 and the titration
with base gave a maximum coagulation rate at the pHpzc In this paper we show the importance of the dye
due to surface electroneutrality. A difference in interaction sensitization process in nanostructured ZnO electrodes.
between dye molecules and these two metal oxides can be The sensitization process depends on parameters such as
demonstrated using a simple experiment. By mixing 0.5 dye composition and concentration and sensitizing time.
g of ZnO powder with 10 mL of 0.5 mM N3 ethanolic Our results give evidence for dye agglomeration in the
solution, we observed, after 24 h, a colorless solution with ZnO nanostructured electrodes. The process is irreversible,
all dye molecules forming a precipitate on the surface of and the formed complex is not easily removed from the
the ZnO powder. The corresponding experiment with film. The agglomeration is due to the dissolution of Zn2+
mixing the TiO2 powder with N3 solution resulted in a ions from ZnO in which protons from the carboxylic groups
red-colored solution with no formation of dye precipitate. of the dye are active. The dissolution of the Zn2+ ions are
Some additional information about the nature of the explained by low local pH at the surface of ZnO. In a second
surface binding between the dye and ZnO can be obtained step, the Zn2+ ions initiate a formation of Zn2+/dye
from IR spectral data. In an investigation of ruthenium- complexes that agglomerate in the pores of the ZnO matrix.
(II)bipyridyl dicarboxylic acid as a photosensitizer to TiO2 Optimization of the sensitization process to obtain efficient
it was concluded that carboxylate groups are attached via dye-sensitized ZnO mesoscopic electrodes for solar energy
bridged coordination to the titanium ions.27,28 This con- conversion should therefore be performed in basic solutions
clusion was based on a comparison of the spectra of or with dyes that have no acid protons in their structure
adsorbed species with carboxylate ions in solution. In but still will attach to the ZnO particles so that effective
particular, the wavenumber difference between the an- electron injection can take place. Further work in these
tisymmetric and symmetric carboxylate stretching bands directions is in progress in our laboratory.
of 230 and 244 cm-1 were observed for the TiO2/N3 and
the TiO2/N719 systems, respectively.28 A chelation situ-
Acknowledgment. This work has been supported
ation would result in a substantially smaller wavenumber
by the Swedish Research Council for Engineering
difference and unidentade complexes in a larger wave-
Sciences (TFR), Swedish Natural Science Research
number difference compared with the corresponding
carboxylate species in solution. In the present IR spectra Council (NFR), Strategic Foundation for Environmental
of the ZnO/N3 system, the spacing between the carboxylate Research (MISTRA), and Swedish National Energy Ad-
stretching bands was 228 and 248 cm-1 for the ZnO/N719 ministration. We thank Viviane Aranyos and Helena
system, which may indicate a similar binding to that of Grennberg for kindly supplying us with the C1-dye and
TiO2. Additionally, the IR spectra for ZnO/N719 indicate Oliver Kohle for the P1-dye. Mårten Edwards is gratefully
that the binding to the ZnO surface involves two proto- acknowledged for his help with optical measurements.
nated groups of the dye, which become deprotonated. From We also thank Petter Persson, Gunnar Westin, and Karin
theoretical investigations of the adsorption and reaction Westermark for helpful discussions.
LA9912702
(26) Bahnemann, D. W. Israel J. Chem. 1993, 33, 115.
(27) Duffy, N. W.; Dobson, K. D.; Gordon, K. C.; Robinson, B. H.;
McQuillan, A. J. Chem. Phys. Lett. 1997, 226, 451. (29) Nakatsuji, H.; Yoshimoto, M.; Umemura, Y., Takagi, S.; Hada,
(28) Finnie, K. S.; Bartlett, J. R.; Woolfrey, J. L. Langmuir 1998, 14, M. J. Phys. Chem. 1996, 100, 694.
2744. (30) Persson, P.; Ojamäe, L. Chem. Phys. Lett., in press.