Solar Energy Materials & Solar Cells 61 (2000) 427}441

Mesoporous electrodes having tight

agglomeration of single-phase anatase TiO
2
nanocrystallites: Application to dye-sensitized
solar cells
Shingo Kambe!, Kei Murakoshi!, Takayuki Kitamura!,
Yuji Wada!, Shozo Yanagida!,*, Hiroshi Kominami",
Yoshiya Kera"
!Material and Life Science, Graduate School of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
"Department of Applied Chemistry, Faculty of Science and Engineering, Kinki University, Kowakae 3-4-1,
Higashi-Osaka 577-0818, Japan
Received 2 August 1999; received in revised form 10 October 1999; accepted 10 October 1999

Abstract

Single-phase anatase nanocrystalline HyCOM-TiO (Hydrothermal Crystallization in Or-

2
ganic Media) to label this method was synthesized by high-temperature hydrolysis of titanium
tetrabutoxide in toluene. The resulting HyCOM-TiO nanocrystallites were found to be
2
covered by n-butoxide, yielding mesoporous, transparent anatase "lms with a narrow pore size
distribution and good electron transport characteristic when sintered at 350}5503C on optically
transparent conducting glass. Dye-sensitized solar cells made of the Ru-dye-adsorbed mesopor-
ous HyCOM "lms as photoanodes achieved better photo-energy conversion e$ciency as
compared to those prepared using commercially available Degussa P25 "lms. ( 2000 Elsevier
Science B.V. All rights reserved.

Keywords: Mesoporous; TiO electrode; Anatase; Single phase; Dye-sensitization

2

1. Introduction

Dye-sensitized TiO solar cells using reactive ruthenium dyes (N3 dye, cis-
2
di(thiocyanato)-N,N@-bis(2,2@-bipyridyl-4,4@-dicarboxylic acid)ruthenium(II) complex)

* Corresponding author. Tel.: #81-6-879-7924; fax: #81-6-877-9067.

E-mail address: [email protected] (S. Yanagida)

0927-0248/00/$ - see front matter ( 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 0 2 4 8 ( 9 9 ) 0 0 1 6 6 - X
428 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

as sensitizers, porous TiO these types of solar cells, photo-excited electrons are
2
injected e$ciently from "lms as electron transport layers and organic liquid electro-
lytes containing a redox couple (e.g. I~/I~) as hole transport layer are currently
3
attracting much interest. This is because of their low cost with quantum e$ciencies
comparable to that of amorphous silicon solar cells [1}3]. In these types of solar cell,
photo-exited electrons are injected e$ciently from excited dye molecules into the
conduction band of the TiO electrode. Mesoporous TiO electrodes with high
2 2
surface area are indispensable (roughness factor '1000) in this system because of
the use of monolayer of dye which is anchored on the TiO electrode that can
2
absorb solar light e$ciently, much like the thylakoid membrane structures in green
plants [3,4].
Another important characteristic of the dye-sensitized solar cells is an excellent
electron #ow within a porous nanostuctured TiO network [5}12]. The e$cient
2
electron transport of the injected electrons from dye molecules is essential to obtain
respectable e$ciency in such dye-sensitized solar cells. To explain the electron
transport across the nanocrystalline "lms where macroscopic electrical "elds are
absent, a di!usion model due to electron scattering or hopping phenomena (trapping
and detrapping phenomena), and possibility of tunneling through a potential barrier
of the nanoparticles has been proposed [6]. Recently, the negative charge caused by
electrons trapped in interfacial band-gap states was proposed to enhance the driving
force for the electron transport in the mesoporous TiO electrode. Recently, Motto
2
transition model has been proposed for the sharp rise in conductivity of the mesopor-
ous anatase TiO "lms; an insulator}metal transition should occur at the initial
2
charging of the "lms [12]. These mechanistic models for the enhanced conductivity of
mesoporous TiO "lms suggest that the crystalline structure and the grain boundaries
2
of the mesoporous TiO "lms should be another important factor which determines
2
the overall e$ciency of the solar cells [13].
In order to obtain higher photoconversion e$ciency in the TiO solar cells,
2
synthesis of anatase TiO nanocrystallites and preparation of mesoporous TiO
2 2
electrodes using their aqueous suspensions are currently being studied worldwide.
Generally, TiO nanocrystallites are prepared by hydrolysis of titanium alkoxide
2
followed by autoclaving for several hours above 2003C. The resulting slurry of anatase
TiO nanocrystallites have been used to prepare mesoporous photoanodes for dye-
2
sensitized solar cells as well as aqueous suspension of commercially available P25
TiO powder (P25-TiO ) [5,13}15]. In the procedure, preparation of TiO seeds as
2 2 2
precursors before autoclaving is known to be a key step for the preparation of
well-de"ned single-phase anatase nanocrystallites. The condensation reaction
and the morphology of "nal TiO nanocrystallites are in#uenced by structures
2
of the precursors; the substitution of the isopropoxide groups in titanium isopropox-
ide with ethoxy (more hydrolyzable), n-butoxy group (less hydrolyzable), or
acetyl groups lead to more monodispersed and smaller sols due to di!erences in
dispersibility and condensation reaction rates [13]. These observations and facts
suggest that organic groups that may locate at surface of the nanocrystallites should
also play important roles during sintering for preparation of the porous "lms as
electrodes.
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 429

Recently, direct hydrolysis of titanium tetra-n-butoxide in toluene under quasi-

supercritical conditions, i.e., at high pressure (1}3 MPa) and temperature (150}3003C),
was found to give directly single-phase anatase TiO nanocrystallites (ca. 13 nm)
2
[16,17]. The authors have used the acronym HyCOM (hydrothermal crystallization
in organic media) to label this method. Such TiO nanocrystallites (HyCOM-TiO )
2 2
are obtained as a powder, not sol, and have single-phase anatase structure with high
surface activity. The authors also reported that HyCOM-TiO showed remarkably
2
higher photocatalytic activity than P25-TiO , as a mixture of anatase and rutile (8 : 2)
2
phases. The high activity was attributed to long-lived photoformed electron}hole
pairs on HyCOM-TiO [18,19].
2
In this study, we prepared nano-structural TiO electrodes using HyCOM-TiO
2 2
powder, and examined electrochemical characteristics of N3 dye sensitized HyCOM-
TiO electrodes as a function of the preparation conditions (e.g. "lm thickness and
2
sintering temperatures). We have intended to explore the following points: "rst, what
kinds of organic molecules are located on HyCOM-TiO powder; second what are
2
the di!erences in morphology (crystalline phase, crystallite size, TiO "lm microstruc-
2
tures) and electron transferability when compared with those of P25-TiO "lms
2
prepared under comparable conditions; and third, whether the TiCl post treatment
4
in#uences the photoenergy conversion e$ciency of the solar cells using dye-anchored
HyCOM-TiO electrodes.
2

2. Experimental procedure

2.1. Apparatus

UV-Vis absorption spectra were measured on a Hitachi U-3300 spectro-

photometer. Powder X-ray di!raction (XRD) spectra were recorded on a Rigaku
RU-200B using Cu Ka, (50 kV - 150 mA, j"1.54056 As ) irradiation. Di!erential
scanning calorimetry (DSC) measurements were carried out on Shimadzu DSC-60.
GC-mass analysis was done on a Shimadzu, GCMS-QP2000A instrument combined
with GC-14A [GC/column: Shimadzu capillary column Hi-Cap CBP-1 (equivalent to
OV-1, f"0.2 mm diameter]25 m); carrier gas: He; #ow pressure: 1.0 kg cm~2; col-
umn temperature: 803C; injection temperature: 2503C, MS/temperature: 2503C; ioniz-
ation: EI (70 eV)]. The surface areas and the pore size distributions of TiO "lms were
2
measured using the nitrogen adsorption method (Gemini 2375, Micro-metrics Instru-
ment Corp.). The surface areas were determined by the BET method. The roughness
factors were calculated as a ratio of the BET surface area of "lms to the apparent
projection areas.

2.2. Preparation of nanocrystalline TiO2 xlms

Porous TiO "lms were prepared using TiO nanocrystallites synthesized by direct
2 2
hydrolysis of titanium n-butoxide in organic solvent under high pressure at high
temperature (150}3003C) (the HyCOM method) [16]. Two kinds of HyCOM-TiO ,
2
430 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

HyCOM1 and HyCOM2, were synthesized as HyCOM-TiO from titanium n-

2
butoxide using 25 and 10 ml of water in 70 ml of toluene, respectively. Details of the
synthesis were reported previously [16]. The BET surface area was determined to be
112 m2 g~1 for HyCOM1 and 140 m2 g~1 for HyCOM2, which were much larger
than that of P25-TiO (55 m2 g~1) and the HyCOM-TiO have single anatase phase
2 2
with diameter of ca. 11}13 nm. The HyCOM1 was analyzed by DSC and GC-mass.
The eliminated compound was identi"ed as n-butanol (M/z (%), 39 (25), 40 (17), 41
(100), 42 (43), 43 (77), 44 (71), 45 (31), 55 (18), 56 (93)). TiO electrodes were prepared as
2
follows; 0.07 g of HyCOM-TiO was mixed with 0.01 ml of conc. nitric acid, and was
2
ground mechanically. During vigorous mixing, 0.32 ml of water was added dropwise
into the mixture. Neat Triton X-100 (1 ml) was "nally added to facilitate spreading of
the paste onto the substrate. The resulting colloidal paste was deposited onto
a #uorine-doped SnO glass (OTE glass, sheet resistance of 10 )/square, Nippon
2
Sheet Glass, Japan). After drying slowly for 1 h at room temperature, the "lm was
heated at 350}5503C at heating rate of 11}173C/min and left at 350}5503C for 30 min
before cooling to room temperature. The thickness of "lms prepared by the above
procedure was ca. 1 lm. Thicker "lms were prepared by repeating the above proced-
ure. Commercially available TiO particle (P25-TiO , Nippon Aerosil) was also used
2 2
for preparation of porous TiO "lms for comparison [1]. The N3 dye, cis-
2
di(thiocyanato)-N,N@-bis(2,2@-bipyridyl-4,4@-dicarboxylic acid)ruthenium(II) complex
([RuL (NCS) ]), was anchored onto the surface of the resulting TiO electrodes in an
2 2 2
ethanol solution of the dye (3]10~4 M) by re#uxing under nitrogen atmosphere at
803C for 1 h [20]. The quantity of the anchored dye was determined by measuring the
quantity of the dye desorbed after treatment with a 0.1 M NaOH solution [2].
As TiCl post-treatment, a cold solution of TiCl (0.2 M) was applied onto the
4 4
TiO electrode and the electrode was kept standing for 10 h before resintering at
2
4503C as reported [13]. The dye was then applied as described above.

2.3. Photoelectrochemical measurements

Photoelectrochemical measurements were performed using a three-electrode cell

with a Pt counter electrode. Electrode potential was referred to an Ag/0.01 M AgNO
3
reference electrode. A mixture of ethylene carbonate and acetonitrile (1 : 1) containing
0.5 M tetrapropylammonium iodide and 0.04 M I was employed as an electrolyte.
2
A potentiostat (Toho Technical Research, 2001) and a programmed function gener-
ator (Toho Technical Research, FG-02) were used to measure photoelectrochemical
response of the electrochemical cell. A 500 W Xe lamp was employed as a light source.
Optical "lters were used to remove ultraviolet (j(400 nm) and near-infrared
(j'1000 nm) radiation. Light intensity was 22 mW cm~2. For the measurements of
photocurrent } voltage characteristics, the focused light from light source was irra-
diated directly to the cell.
The photoenergy conversion e$ciency was determined using a sandwich type cell.
The dye anchored TiO electrode was "xed onto a platinized conducting glass
2
electrode as a photocathode using thermal adhesive "lms (HIMILAN 1652, thickness,
50 lm, Mitsui-Dupont Polychemical, Japan). An electrolyte consisting 0.1 M LiI,
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 431

Fig. 1. DSC curves of HyCOM "lm before (a) and after (b) sintering, and that of P25 "lm before (c) and after
(d) sintering at 4503C.

0.6 M 1,2-dimethyl-3-propylimidazolium iodide, 0.1 M I , 1 M 4-tert-butylpyridine

2
and 3-methoxy propionitrile as a solvent was introduced into the gap of the sand-
wiched electrodes. Photoelectrochemical measurements were performed using a solar
simulator (YSS-80: Yamashita Denso Co.; irradiance was adjusted to 10 mW cm~2
and 100 mW cm~2 with AM 1.5 spectral distribution) as a light source and a computer-
controlled digital multimeter (model 166; Keithley, model HP3478A; Hewlett-Packard)
to measure photocurrent and voltage. All measurements were carried out at 203C.

3. Results and discussion

3.1. Characterization of HyCOM-TiO2 and HyCOM-TiO2 xlms

As reported recently, we synthesized single-phase anatase nanocrystalline TiO

2
(HyCOM-TiO ) by high-temperature hydrolysis of titanium tetra-n-butoxide with
2
a limited amount of water dissolved in toluene. Thermal analysis of HyCOM-TiO
2
showed one exothermic peak at 2203C with a little enhanced rate in weight loss in its
thermal gravimetric analysis (TGA) (Fig. 1).1 No peak was observed when the sample

1 We reported previously that HyCOM-TiO showed 4.9% weight loss during thermalysis starting from
2
1303C to 10003C and no exothermic peaks was detected in the di!erential thermal analysis [17].
432 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

Fig. 2. Formation of the chemical bond at the HyCOM-TiO surface during sintering.
2

was once sintered at 4503C. The eliminated compound was identi"ed as n-butanol by
GC-mass analysis. HyCOM1 was found 1.30 wt% of carbon content by elemental
analysis. Assumed that HyCOM1 consists of 13 nm sized particles, the number of
n-butanol on the one particle was calculated to be more than 620. These facts suggest
that the surface of HyCOM-TiO should be bound by n-butoxy groups to apprecial
2
extent. In the case of P25-TiO , no exothermic peak was observed either before or
2
after sintering. The gradual weight loss observed in TGA analysis of HyCOM-TiO
2
and P25-TiO was ascribed to interparticle neck growth by dehydration during the
2
sintering as is depicted in Fig. 2.
Fig. 3 shows images of the TiO "lms obtained by scanning electron micrograph at
2
two di!erent magni"cations, suggesting that the HyCOM "lm is smoother than the
P25 "lm and appears to have smaller crystallites.
X-ray di!raction patterns of the both "lms are shown in Fig. 4. Di!raction signal
assigned to anatase (1 0 1) structure at 25.33 were clearly observed in both HyCOM
and P25 "lms. The di!raction signal at 27.43 due to rutile phase (1 1 0) was not
observed in HyCOM "lm. The intensity ratio of the di!raction re#ects ca. 20%
content of rutile phase in the P25 "lms. Absence of the di!raction due to rutile phase
in the HyCOM "lm indicates that they consist of phase pure anatase as the starting
HyCOM-TiO , [17] and that the sintering process in the "lm fabrication does not
2
e!ect the crystalline structure.
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 433

Fig. 3. Scanning electron micrograph of HyCOM "lm (a, b) and P25-TiO "lm (c, d). Left panels (a, c) are
2
low resolution and right panels (b, d) are of high resolution.

Fig. 4. X-ray powder di!raction spectra of HyCOM "lm (above), P25-TiO "lm (below).
2
434 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

Fig. 5. Absorption spectra of TiO "lms; thickness of the HyCOM "lms are 2.9 lm (a), 6.3 lm (b), 12.1 lm
2
(c), and 16.0 lm (d). That of P25-"lm is 11.3 lm (e).

Table 1
Surface area (R. F.) of the "lms

R.F. of "lms Nano-crystallites BET

area (m~2 g~1)
d"3 lm d"10 lm

HyCOM 780 1900 112

P25 380 770 55

Fig. 5 shows the absorption spectra of HyCOM "lms with di!erent thicknesses and
a P25 "lm of 10 lm thickness. The HyCOM "lms were found more transparent than
the P25 "lm. The absorbance of the HyCOM "lms was less than the P25 "lm, even
when the thickness of HyCOM "lms was larger than the P25 "lm. Steep absorption
onset and lower absorption at longer wavelength region (j'450 nm) of HyCOM
"lms may be due to the absence of light scattering due to the aggregation of
mono-dispersed nanocrystallites of which crystallite size is far less than the
wavelength of visible light.
The roughness factors of the "lms are listed on Table 1 with the surface areas of the
nanocrystalline TiO powders. The "lms made from HyCOM gave two times higher
2
roughness factors than those made from P25. The high roughness factor of HyCOM
"lm should be attributed to the smaller particle size of HyCOM than P25 as clearly
shown in the di!erence of the BET surface areas.
The HyCOM "lm and the P25 "lm showed the remarkable di!erence in their pore
size distribution (Fig. 6). Both "lms had the very small pores at 5 nm however, only the
HyCOM "lm possessed large amount of mesopores at around 10 nm. On the other
hand, the P25 "lm showed the very broad pore size distribution even larger than
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 435

Fig. 6. Pore-size distribution of HyCOM "lm (L) and P25 "lm (]).

Fig. 7. Dependence of the amount of chemisorbed dye on the thickness of HyCOM1 "lms (L), HyCOM2
"lm (n), and P25-TiO "lm (]).
2

20 nm. The HyCOM "lm is characterized by the predominant presence of the

mesopores. The smaller particles size and the homogeneous particle size distribution
of HyCOM should result in the characteristic pore size distribution of the HyCOM
"lm.

3.2. Dye chemisorption onto the xlms and ewect of sintering temperature

The quantity of chemisorbed dye on the "lm was determined by measuring the
quantity of the dye desorbed into a NaOH solution (0.1 M). Fig. 7 shows that the
quantity of the dye on HyCOM "lms was more than twice as much as on P25 "lm at
comparable thickness and that the dye quantity is proportional to the thickness of
the HyCOM "lms. The HyCOM1 "lm comprised by the smaller size (11 nm) of
436 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

Fig. 8. Photocurrent}potential characteristics of the dye-adsorbed HyCOM "lm with thickness of 13.4 lm
(a), 8.5 lm (b), 2.7 lm (c), and under dark (d), determined using three electrodes photo electrochemical cell
(above). Photocurrent}potential characteristics of P25-TiO "lm with thickness of 15.0 lm under illumina-
2
tion and dark (below).

HyCOM-TiO chemisorbed slightly more dye molecules than the HyCOM2 "lm
2
(13 nm).
The e!ect of sintering temperature on the dye chemisorption was also examined.
The quantity of the dye on HyCOM1 "lms with the thickness of 4 lm were 0.54, 0.49,
and 0.47]10~7 mol cm~2, when sintered at 350, 450 and 5503C, respectively. Slight
decrease in the dye chemisorption with increase of sintering temperature re#ected the
decrease in the surface area of the "lms owing to necking between the particles.

3.3. Photoelectrochemical response of the dye-chemisorbed TiO2 xlms

Photoelectrochemical response of the dye-chemisorbed TiO "lms was evaluated

2
using a three-electrode electrochemical cell. Fig. 8 shows current-voltage character-
istics of the TiO "lms under illuminated conditions. Maxima of photocurrent of
2
HyCOM "lms depended on the thickness of the "lms. As the thickness of HyCOM
"lms increased from 2.7 to 13.4 lm, photocurrent increased from 1.0 to 1.9 mA cm~2.
Recently, similar photoelectrochemical behavior was reported for nanostructured
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 437

Fig. 9. Dependence of photocurrent maxima on the amount of chemisorbed dye on HyCOM1 "lms (L),
HyCOM2 "lm (n), and P25-TiO "lm (]).
2

TiO "lms with thicknesses ranging from ca. 0.6 to 45 lm; under band gap UV
2
illumination in the presence of formic acid as electron donor, the photocurrent
increased with increase of the "lm thickness [12].
It should be noted that the thin HyCOM "lm (2.7 lm) showed higher photocurrent
(1.0 mA cm~2) than the thick P25 "lm (15 lm thickness, 0.5 mA cm~2), although the
former chemisorbed about one-half of the dye molecules on the latter (Fig. 7). To
evaluate the e$ciency of the dye molecules which generate photocurrent, maximum
values of photocurrent (I ) were plotted as a function of the quantity of chemisorbed
.!9
dye molecules per unit area of the "lms in Fig. 9. Photocurrent increased almost
linearly as the amount of the dye molecules increased, although slight saturation of
the photocurrent was observed at large values of the dye-chemisorption
('0.7]10~7 mol cm~2). Higher photocurrent was observed for HyCOM "lms than
for P25 "lms. The HyCOM2 "lm yielded a larger photocurrent than the HyCOM1
"lm, while the chemisorbed dye on the HyCOM1 "lms was comparable or slightly
larger than that of the HyCOM2 "lm.
Importance of light scattering in electrodes was pointed out and better photovoltaic
response was con"rmed especially for the red light by introducing large scattering
particles in the TiO electrodes. The scattering of light does not play a role in the
2
photovoltaic response of the HyCOM systems, because the HyCOM-TiO "lms are
2
fully transparent while P25-TiO "lm is translucent or opaque. Generally, lattice
2
imperfections in crystal such as kink, vacancy, lattice mismatch, interfere with electron
transport in semiconducting materials. Taking into account that P25-TiO is a mix-
2
ture of ca. 80% anatase and ca.20% rutile nanocrystallites, the higher photocurrent
generation in the dye-chemisorbed HyCOM "lms suggested that electron transport of
photo-injected electron should become e!ective in the single phase anatase HyCOM-
TiO agglomeration.
2
Grain boundaries between the particles should a!ect the electron transport of
injected electrons as well as the crystallinity of the particles. The present systems of
HyCOM electrodes consist of a few tens nm crystallites, and thus contain a signi"cant
438 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

amount of grain boundaries. The higher e$ciency of HyCOM electrodes implies that
not only the single phase anatase crystalline structure but also the smooth neck
growth at the grain boundaries should contribute to the e!ective electron transport.
Linear increment of photocurrent depending on the thickness of the HyCOM "lms
(Fig. 9) also supports the formation of the e!ective crystalline neck growth in the
agglomeration with thickness up to 13 lm. Taking into account the n-butoxide layer
on HyCOM nanocrystallites, the gradual elimination of the butoxide might contrib-
ute to the e!ective crystalline neck growth during sintering of the agglomeration (Fig.
2). Improvement of the e$ciency at the HyCOM "lms sintered at higher temperature
(5503C) was observed, which should be also attributed to the e!ective crystalline
necking at higher temperature.
Photocurrent}voltage characteristics of the dye-chemisorbed HyCOM electrodes
showed slightly positive onset potential of photocurrent (; "0) compared with that
i
of P25 "lm. In the system of dye-sensitization, values of ; "0, where net photocur-
i
rent was not observed, re#ect injected electron density of the conduction band under
illumination [21}23]. Although positive shift of ; "0 is generally caused by in-
i
crement of recombination process of photoinjected electron in conduction band and
electron acceptors at the interface. Such as oxidized dye or I~ in the redox electrolytes,
3
the observed shift in the system of the HyCOM electrodes should not be attributed to
this e!ect, because such recombination also decreases the photocurrent e$ciency.
Positive ; "0 of the HyCOM-TiO system should be ascribed to intrinsic surface
i 2
properties a!ecting the #at band potential of the electrode. Fig. 8 also shows cur-
rent}potential characteristics of the HyCOM electrode system under dark conditions.
When compared to those of P25 electrode (below), onset potential of the reduction
current under dark was as positive as under illumination. Thus, the positive shift of
; "0 can be attributed to the intrinsic properties of the HyCOM electrode. It is
i
well-known that proton-rich surface of TiO "lms causes positive shift of the #at-band
2
potential [14]. Thus, the lower #at band potential, i.e., the more positive shift of ; "0
i
for the HyCOM electrodes might re#ect more protonated surface, i.e. more acidic
surface of the dye-chemisorbed HyCOM electrodes than that of the P25 electrodes.

3.4. Characteristics in photoenergy conversion device

Sandwich-type solar cells were fabricated to determine the photoenergy conversion

e$ciency for cells using HyCOM and P25 "lms. Though open-circuit voltage (< ) of
0#
the HyCOM cells is lower than that of the P25 cells, short-circuit current (I ) of the
4#
HyCOM cells is higher under low and high irradiation (10 and 100 mW cm~2). This
result is consistent with the tendency of #at band potential shift in ; . The post-
i/0
treatment of the porous "lms with aqueous TiCl solution followed by sintering was
4
conducted to improve the performance of the solar cells as reported Barbe [13]. Fig.
10 shows photocurrent}voltage characteristics of the cells using HyCOM and P25
"lms with and without the post-treatment. The TiCl treatment enhanced the di!er-
4
ence between the performance of the cell using HyCOM "lms and P25 "lms. Short-
circuit photocurrent (I ) and open-circuit voltage (< ) of the cell were improved from
4# 0#
0.75 mA cm~2 and 0.58 V to 0.95 mA cm~2 and 0.60 V, respectively. The similar
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 439

Fig. 10. Photocurrent}voltage characteristic of the cells using; HyCOM "lm with the treatment (bold solid
line), HyCOM "lm without the treatment (bold dashed line), P25 "lm with the treatment (solid line), P25
"lm without the treatment (dashed line), determined by the sandwich-type cell. Illuminated light intensity
was 10 mW cm~2 with AM 1.5 spectral irradiance.

Table 2
Photovoltaic properties of dye-sensitized solar cells!

Intensity (mW cm~2) < (V) I (mA cm~2) Fill factor g! (%)
0# 4#
HyCOM "lm 10 0.56 0.75 0.72 3.1
HyCOM "lm/TiCl 10 0.60 0.95 0.73 4.3
4
treatment
HyCOM "lm/TiCl 100 0.62 8.8 0.52 2.8
4
treatment
P25 "lm 10 0.63 0.68 0.69 3.1
P25 "lm/TiCl treatment 10 0.63 0.66 0.66 2.7
4
P25 "lm/TiCl treatment 100 0.70 6.8 0.55 2.9
4
!Overall light-to-electric energy conversion e$ciency.

treatment of P25 "lms, however, led to the slight decline of the cell performance. The
post-treatment should result in the formation of fresh TiO layer onto the surface of
2
porous "lms. The formation of fresh surface should improve inter-connection between
the particles as in the case of the necking during the sintering. The e!ective neck
growth would facilitate the electron transport between the particles, leading to the
improvement of the cell performance. Furthermore, the improved surface layer might
contribute to e!ective adsorption of dye molecules on the TiO "lms [20]. The
2
e!ective chemisorption should assist e!ective electron injection at the dye-sensitiza-
tion. Such surface activation should occur more e!ectively in the HyCOM "lms than
P25 "lms. Under the AM1.5 conditions (100 mW cm~2), the HyCOM TiO solar cell
2
(TiCl treatment) gave a less conversion e$ciency when compared with the case of the
4
440 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441

P25 solar cell (TiCl treatment) (Table 2). The di!erence is explained as due to
4
di!erence in pore size distribution. The HyCOM "lms had the pore size distribution
that ranged from 4 to 20 nm, while the P25 "lms from 4 to 50 nm (Fig. 6). The narrow
pore size distribution is not suitable for di!usion of the hole transporting I~/I~
3
species in the electrolyte under high light intensity.

4. Conclusions

Nanocrystalline titanium dioxide (HyCOM-TiO ) with high surface area and

2
single-phase anatase structure was synthesized under quasi-supercritical conditions.
Porous TiO "lms with narrow pore-size distribution and high optical transmittance
2
were successfully prepared using two kinds of HyCOM-TiO (d"13 and 11 nm).
2
Transparency of the fabricated "lms (16 lm thickness) was superior to that of
P25-TiO "lms (11 lm thickness). Electrochemical characteristics of dye-anchored
2
"lms depend on surface and boundary structures of the TiO nanocrystallites in the
2
"lms, which are in#uenced by fabrication conditions such as source of TiO , sintering
2
temperature, thickness, dye-adsorption, and the TiCl post-treatment of the "lms. The
4
characteristics of the HyCOM "lms are attributed to their surface conditions and
crystallinity of the starting HyCOM nanocrystallites, and the e!ective neck growth
due to thermolysis of the surface n-butoxide during sintering.

Acknowledgements

This work was supported in part by Grants-in-Aid for Scienti"c Research on

Priority Areas, `Electrochemistry of Ordered Interfacesa (No. 10131244) from the
Ministry of Education, Science, Sports, and Culture, Japan, and a grand by a `Re-
search for the Futurea Program of JSPS.

References

[1] B. O'Regan, M. GraK tzel, Nature 353 (1991) 737.

[2] M.K. Nazeeruddin, A. Kay, I. Rodicio, B.R. Humphry, E. Mueller, P. Liska, N. Vlachopoulos, M.
GraK tzel, J. Am. Chem. Soc. 115 (1993) 6382.
[3] C. J. Brinker, G. W. Scherer, Sol-Gel Science, Academic Press, San Diego, 1990.
[4] Y. Boukennous, B. Benyahia, M.R. Charif, A. Chikousche, J. Phys. III France 5 (1995) 1297.
[5] A. Solbrand, H. Lindstrom, H. Rensmo, S. Lindquist, S. Sodergren, J. Phys. Chem. B 101 (1997) 2514.
[6] S. Sodergren, A. Hagfeldt, J. Olsson, S. Lindquist, J. Phys. Chem. 98 (1994) 5552.
[7] H. Lindstrom, H. Rensmo, S. Sodergren, A. Solbrand, S. Lindquist, J. Phys. Chem. 100 (1996) 3084.
[8] A. Kay, R. Humphry-Baker, M. GraK tzel, J. Phys. Chem. 98 (1994) 952.
[9] G. Redmond, D. Fitzmaurice, M. GraK tzel, J. Phys. Chem. 97 (1993) 6951.
[10] R. KoK nenkamp, R. Henninger, P. Hoyer, J. Phys. Chem. 97 (1993) 7328.
[11] P. Hoyer, H. Weller, J. Phys. Chem. 99 (1995) 14096.
[12] A. Wahl, J. Augustynski, J. Phys. Chem. B 102 (1998) 7820.
 S. Kambe et al. / Solar Energy Materials & Solar Cells 61 (2000) 427}441 441

[13] C.J. Barbe, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, M. GraK tzel, J. Am. Ceram.
Soc. 80 (1997) 3157.
[14] B. O'Regan, J. Moser, M. Anderson, M. GraK tzel, J. Phys. Chem. 94 (1990) 8720.
[15] N. Papageorgiou, C. Barbe, M. GraK tzel, J. Phys. Chem. B 102 (1998) 4156.
[16] H. Kominami, T. Matsuura, K. Iwai, B. Ohtani, S. Nishimono, Y. Kera, Chem. Lett. (1995) 693.
[17] H. Kominami, H. Takada, Y. Kera, M. Inoue, T. Inui, J. Mater. Sci. Lett. 15 (1996) 197.
[18] B. Ohtani, K. Iwai, H. Kominami, T. Matsuura, Y. Kera, S. Nishimoto, Chem. Phys. Lett. 242 (1995)
315.
[19] B. Ohtani, R.M. Bowman, D.P. Colombo Jr., H. Kominami, H. Noguchi, K. Uosaki, Chem. Lett.
(1998) 579.
[20] K. Murakoshi, G. Kano, Y. Wada, S. Yanagida, H. Miyazaki, M. Matsumoto, S. Murasawa, J.
Electroanal. Chem. 396 (1995) 27.
[21] A. Hagfeldt, M. GraK tzel, Chem. Rev. 95 (1995) 49.
[22] S.Y. Huang, G. SchlichthoK rl, A.J. Nozik, M. GraK tzel, A.J. Frank, J. Phys. Chem. B 101 (1997) 2576.
[23] G. SchlichthoK rl, S.Y. Huang, J. Sprague, A.J. Frank, J. Phys. Chem. B 101 (1997) 8141.