Photocatalytic activity of transition-metal-loaded titanium(IV) oxide powders suspended in aqueous solutions: Correlation with electron-hole recombination kinetics

Article abstract of DOI:10.1039/b008028o

Photocatalytic reactions by transition-metal (V, Cr, Fe, Co, Cu, Mo, or W) loaded TiO₂ (M-TiO₂) powders suspended in aqueous solutions of methanol, (S)-lysine (Lys), or acetic acid were investigated. The photoactivities of various samples were compared with the rate constant (k_r) of recombination of photoexcited electrons and positive holes determined by femtosecond pump-probe diffuse reflection spectroscopy (PP-DRS). As a general trend, increased loading decreased the rate of formation of the main products (H₂, pipecolinic acid (PCA), and CO₂) under UV (> 300 nm) irradiation, and the effect became more intense on increasing the loading. In PP-DRS, these M-TiO₂ gave similar decays of absorption at 620 nm arising from excitation by a 310 nm pulse (< 100 fs). The second-order rate constant (k_r) markedly increased with loading, even at a low level (0.3%) and further increased with an increase in loading up to 5%. The photocatalytic activity of platinized M-TiO₂ for H₂ and PCA production under deaerated conditions depended strongly on k_r, but the relation between k_r and the rate of CO₂ production by unplatinized M-TiO₂ under aerated conditions was ambiguous; other factor(s) might control the rate of the latter. These different k_r dependences of photoactivity on the reaction kinetics governed by e^--h⁺ recombination were attributed to the presence of O₂ and Pt deposits. A simple kinetic model to explain the overall rate of these photocatalytic reactions is proposed, and the effect of recombination kinetics on photoactivity is discussed.

Full text of DOI:10.1039/b008028o

View Article Online / Journal Homepage / Table of Contents for this issue
Photocatalytic activity of transition-metal-loaded titanium(IV) oxide
powders suspended in aqueous solutions: Correlation with
electron–hole recombination kinetics
Shigeru Ikeda,ab Noboru Sugiyama,b Bonamali Pal,a Giuseppe Marc•,c Leonardo Palmisano,c
Hidenori Noguchi,d Kohei Uosakid and Bunsho Ohtaniab
a Catalysis Research Center, Hokkaido University, Sapporo 060-0811, Japan.
E-mail: ohtani=cat.hokudai.ac.jp
b Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0810,
Japan
c Dipartimento di Ingegneria Chimica dei Processi e dei Materiali, Universita di Palermo,
V iale delle Scienze, 90128, Palermo, Italy
d Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan
Received 4th October 2000, Accepted 28th November 2000
First published as an Advance Article on the web 21st December 2000
Photocatalytic reactions by transition-metal (V, Cr, Fe, Co, Cu, Mo, or W) loaded TiO (M-TiO ) powders
2
2
suspended in aqueous solutions of methanol, (S)-lysine (Lys), or acetic acid were investigated. The
photoactivities of various samples were compared with the rate constant (k ) of recombination of photoexcited
r
electrons and positive holes determined by femtosecond pumpÈprobe di†use reÑection spectroscopy (PP-DRS).
As a general trend, increased loading decreased the rate of formation of the main products (H , pipecolinic
2
acid (PCA), and CO ) under UV ([300 nm) irradiation, and the e†ect became more intense on increasing the
2
loading. In PP-DRS, these M-TiO gave similar decays of absorption at 620 nm arising from excitation by a
310 nm pulse (\100 fs). The second-order rate constant (k ) markedly increased with loading, even at a low
level (0.3%) and further increased with an increase in loading up to 5%. The photocatalytic activity of
2
r
platinized M-TiO for H and PCA production under deaerated conditions depended strongly on k , but the
2
2
r
relation between k and the rate of CO production by unplatinized M-TiO under aerated conditions was
r
2
2
ambiguous; other factor(s) might control the rate of the latter. These di†erent k dependences of photoactivity
r
on the reaction kinetics governed by e~Èh` recombination were attributed to the presence of O and Pt
2
deposits. A simple kinetic model to explain the overall rate of these photocatalytic reactions is proposed, and
the e†ect of recombination kinetics on photoactivity is discussed.
on visible-light response. As a possible explanation, some
Introduction
investigators have concluded that the doping or loading of
metals, acting as or leading to the formation of defective sites
in the crystal lattice, accelerates recombination of photoexcit-
ed electrons and positive holes.10,12 However, this is still only
speculation, and no clear evidence of an e†ect of transition
Titanium(IV) oxide (TiO ) has been investigated extensively as
one of the most promising candidates for a semiconductor
2
photocatalyst,1 and has been found to be suitable for a wide
range of processes, including solar energy conversion and
storage,2,3 reductive Ðxation of carbon dioxide,4 organic syn-
thesis,5,6 and mineralization and/or detoxiÐcation of organic
metal on the photocatalytic activity of TiO due to its band-
2
gap excitation has yet been obtained. Furthermore, to the best
compounds.7,8 A disadvantage of the use of TiO as a photo-
of our knowledge, few detailed examinations on the corre-
lation between the recombination rate and photocatalytic
activity have been performed.
2
catalyst is that it absorbs only UV light at j \ 400 nm, and
therefore a signiÐcant fraction of the solar radiation is not
used in TiO photocatalytic reactions, though this photoab-
In the primary stage of the photocatalytic reaction, the
2
sorption property, as TiO is practically white or colorless,
band-gap photoabsorption of TiO generates photoexcited
2
2
enables us to apply photocatalytic coatings without changing
electrons and holes, which can migrate to the surface to drive
the appearance of the underlying materials. In order to extend
the range of photoabsorption to visible light matching the
solar radiation, doping of metals, especially transition metals
ions, into the TiO crystal lattice or loading them on the TiO
surface has been shown to be successful in several studies.9h14
The absorption spectrum has been extended to the visible
region, and photoresponses, i.e., visible light-induced photo-
redox reactions with adsorbed substrates, competing with
their disappearance due to mutual recombination. Although
the mechanism and/or kinetics of such fast primary events in
heterogeneous systems have been poorly understood so far,
mainly due to the limitation of time resolution in conventional
techniques, recent progress in laser technology has enabled
direct measurement of the dynamics of heterogeneous electron
transfer with a time resolution even faster than 100 fs.15h26
2
2
catalytic reactions, on modiÐed TiO have been reported.
2
However, it should be noted that in most cases the fundamen-
For TiO particles and colloids,20h26 femtosecond PP tran-
2
tal photocatalytic activity of TiO , UV light-induced band-
sient absorption spectroscopy has shown that certain surface
2
gap excitation seemed to be reduced, in spite of the emphasis
sites trap electrons within a few tens of fs after the photoexci-
DOI: 10.1039/b008028o
Phys. Chem. Chem. Phys., 2001, 3, 267È273
This journal is ( The Owner Societies 2001
267

View Article Online
tation, giving a broad absorption at around 500È650 nm,27,28
and their recombination with holes induces decay of absorp-
tion within 100 ps with second-order kinetics.23,24 Analysis of
The di†usely reÑected probe beam was detected by a photo-
diode, and the output from the photodiodes was collected by
two separate gated integrators (Stanford Research Systems,
SR250) and digitized with an 8-bit A/D converter (Stanford
Research Systems, SR245). ReÑectance data were accumulated
the ultrafast absorption decay kinetics of opaque TiO
2
powders measured in a di†use reÑection mode has shown that
the observed lifetime of trapped electrons is closely related to
the photocatalytic activity;24h26 the longer the lifetime of
trapped electrons, the higher the photocatalytic activity of
and recorded by a personal computer as absorption, 1-R/R ,
0
where R and R represent the intensity at a given delay and in
0
the absence of pump pulse, respectively. Both R and R were
0
commercial and synthesized TiO powders. These facts
corrected by the reference beam intensity of each probe pulse
2
suggest that kinetic parameters, e.g., the rate constant of the
and only the absorption for the probe pulse with intensity
within a given, typically ^5%, Ñuctuation. The delay between
the pump and probe pulses was controlled by an optical delay
line including a computer-controlled translation stage with 1
lm (\3.3 fs) resolution and 1 ns maximum delay (Sigma
Koki, AS NET-3/STM-160). Before each measurement, the
zero point of delay, i.e., the timing of the coincidence of pump
and probe beams and spatial overlapping of two beams, was
adjusted by measuring the transient absorption of 1,1@-bin-
aphthyl (Tokyo Kasei Kogyo) dissolved in methanol. All the
measurements were carried out in air at ambient temperature.
absorption decay, could be a signiÐcant measure of the photo-
catalytic activity.24
The purpose of this study was to apply PP-DR spectros-
copy to transition-metal-loaded TiO powders to clarify the
2
inÑuence of loaded metal on the rate of recombination of elec-
trons and holes under UV photoirradiation. A series of TiO
photocatalysts with varying amounts of loaded transition
2
metals (V, Cr, Fe, Co, Cu, Mo, or W) were used for several
photocatalytic reactions: methanol dehydrogenation, conver-
sion of (S)-lysine (Lys) into (S)-pipecolinic acid (PCA), and oxi-
dation of acetic acid. The initial reaction rates were compared
with the relative rate constants of electronÈhole recombi-
nation determined using PP-DR spectroscopy.
Photoirradiation and product analysis
The photocatalyst (50 mg) was suspended in an aqueous solu-
tion (5 cm3) containing (a) methanol (50 vol.%), (b) Lys (100
lmol), or (c) acetic acid (5 vol.%), and the suspensions were
photoirradiated in Ar (a) and (b) or air (c) with magnetic stir-
ring (1000 rpm). In reaction (a), the catalyst was platinized in
situ by adding H PtCl É 6H O (Wako Pure Chemical), the
Experimental
Materials
Bare and metal-loaded TiO powders were prepared as
follows: An aqueous solution of ammonia (Carlo Erba RPE,
25%) was added dropwise at 298 K to an aqueous solution of
2
2
6
2
amount of which corresponded to 2 wt.% loading of Pt. The
platinized TiO powders were recovered and used in reaction
TiCl (Carlo Erba RPE, 15%) to obtain a white precipitate,
2
3
(b). Decomposition of acetic acid (reaction (c)) was also exam-
which was then dried in air at 373 K followed by calcination
ined in air with platinized M-TiO recovered from reaction (a)
at 773 K for 24 h in air. The as-prepared bare TiO was
suspended in an aqueous solution of metal nitrates or
ammonium salts of hetero polyacids (Cr(NO ) É 9H O,
2
2
(c1) or in Ar with unplatinized photocatalysts (c2). Irradiation
with light of wavelength j [ 300 nm from a high-pressure
mercury arc (Eiko-sha, 400 W) was performed through a
cylindrical Pyrex glass Ðlter and a glass reaction tube (18 mm
in diameter and 180 mm in length), so that light at j [ 300
nm reached the suspension. The temperatures of suspensions
under photoirradiation were kept at 298 ^ 0.5 K in a ther-
3 3
2
2
Fe(NO ) É 9H O,
3 3
NH VO , (NH ) Mo O , and (NH ) W
Co(NO ) É 6H O,
Cu(NO ) É 3H O,
3 2
), and then
2
3 2
2
O
4
3
4 6
7
24
4 6 12 39
the slurry was dried and calcined at 773 K for 24 h in air. In
all catalyst preparation processes, the content of metal was
0.3, 1, 2, or 5%, calculated as the molar ratio of loaded metal
to the total metal, including Ti. Measurements of powder
X-ray di†raction and nitrogen adsorption at 77 K showed
mostatted water bath. Gaseous products, such as H and
2
CO , were analyzed by gas chromatography (GC) and pro-
2
ducts in the liquid-phase were analysed by high-performance
that all the TiO powders consisted mainly of well-crystallized
2
liquid chromatography (HPLC). Details of procedures for the
photocatalytic reactions investigated and product analyses
have been described previously (for methanol,30 Lys,31h33 and
acetic acid34,35).
anatase with a similar BET surface area of ca. 35È60 m2 g~1.
The detailed characterization of these samples will be
published elsewhere.29
Femtosecond PP-DRS
The experimental setup is summarized here. A train of ca.
100-fs pulses with 1 mJ energy at a repetition rate of 1 kHz
was generated by a regenerative ampliÐer system (Quantronix,
4812RGA/4823S/C), which was synchronously pumped by a
mode-locked YLF laser (Quantronix, 527DP-H). Seed pulses
of 82 MHz centered at 790 nm were supplied from a mode-
locked Ti:Sapphire laser (Spectra-Physics, Tsunami: 3960-
L2S) pumped by an argon ion laser (Spectra-Physics,
BeamLok: 2580C). The Ðnal output was split into two beams
with almost comparable intensities in order to pump two
identical optical parametric generation (OPG)/optical para-
metric ampliÐcation (OPA) systems (Light Conversion,
TOPAS). The 620 nm output beam from one OPG/OPA was
frequency-doubled in a BBO crystal and used as a pump
beam (0.1 D 0.03 mJ (pulse)~1), and the 620-nm beam from
another OPG/OPA was used as a probe beam (1.5 mJ
(pulse)~1). The probe beam was split into two beams; a minor
part, directly detected by a photodiode (Hamamatsu Photon-
ics, Co. Ltd., S2387-66R), was used as a reference. The pump
and a major part of the probe beams were collinearly focused
with a focal lens ( f \ 10 cm) and overlapped at the sample.
Results and discussion
PP measurements of bare and metal-loaded TiO powders
2
Fig. 1 shows representative time-proÐles of the transient
absorption of bare and Fe-loaded (5%) TiO powders in
PP-DRS. As a common feature, a very rapid rise of absorp-
tion at 620 nm, within or comparable to the limit of time
resolution of ca. 500 fs, followed by decay over several
hundreds ps, was observed. The rapid rise is thought to be due
to capture of photoexcited electrons at a trapping site and
broad-band absorption of the trapped electrons in the visible
and near-IR regions. Following the pioneering work with
2
PP-DRS of TiO powders and suspensions by Colombo and
2
Bowman,23 most of the reported PP-DRS data have been
analyzed with absorption (1-R/R ), and the decay proÐles
0
could be best Ðtted to a function of the sum of a second-order
rate expression with a constant k and very slow decay com-
ponent, as follows:24
r
(absorption) \ a([e] /(1 ] k [e] t) ] BL),
(1)
0
r
0
268
Phys. Chem. Chem. Phys., 2001, 3, 267È273

View Article Online
for each loaded metal, the loading enhanced the recombi-
nation of electrons and holes.
E†ect of transition-metal loading on photocatalytic activity
Fig. 2 shows the rates of formation of (a) H , (b) PCA, and (c)
2
CO as functions of the amount of transition-metal loading.
2
Previous reports ((a),30 (b),31h33 and (c)34,35) indicated that
the stoichiometry of the main reactions occurring in our
system is
CH OH \ HCHO ] H
(2)
3
2
H NCH (CH ) CH(COOH)NH
2
2
2 3
2
\ HN(CH ) CHCOOH ] NH (3)
2 4
3
Â
Â
CH COOH ] 2O \ 2CO ] 2H O.
(4)
3
2
2
2
Consequently, we could estimate the overall rate of the redox
reactions from the rate of main-product formation. The rate of
formation of gas phase products, H and CO , was deter-
2
2
mined from the slope of the plots relative to the amount of
products vs. irradiation time by considering the initial 1.5 h of
irradiation; the rate of PCA formation was determined, by
considering only 1.0 h of irradiation. For each reaction,
loading of any kind of transition metal gave rise to a decrease
in photocatalytic activity and the e†ect was more signiÐcant
as the amount of loaded metal increased. Since the high-
pressure mercury arc used in this study emits light in the
visible region, in addition to UV light, which excites electrons
Fig. 1 Representative decay proÐle of M-TiO powder induced by
2
ultrafast pump (ca. 100 fs). Results of bare and Fe (5%)-loaded TiO
2
powders are shown. Solid line indicates the Ðtting with eqn. (1).
where [e] , k , t, and BL represent, respectively, the initial
0
r
concentration of a part of the trapped electrons giving a rela-
tively fast decay just after the pump pulse, the second-order
rate constant (cm3 ps~1), the time delay of the probe pulse
(ps), and a component of the long-lived baseline. The param-
eter a (cm3) is related to the photoabsorption cross section of
the components and the depth of penetration of the pump
pulse. This rate equation can be rationalized with the assump-
tion that some of the trapped electrons recombine with posi-
tive holes and that the others remain unreacted (BL) in this
time region. It is clear by comparing the two proÐles in Fig. 1
that the Fe loading gave rise to faster initial decay in the time
region of \100 ps, which corresponds to larger k . Table 1
r
summarizes the k values of bare and M-TiO samples deter-
r
2
mined by using eqn. (1), assuming a \ 1 for every sample. In
contrast to Ðrst-order decay kinetics, evaluation of the second-
order rate constant requires the absolute value of the reaction
rate and thereby k depends on a. The actual value of a need
r
not be unity and must vary with the measurement conditions.
Therefore, the determined value of k was a relative measure.
r
Some deeply colored samples containing, for example, 5 mol%
of V, Cr, Co, Cu, or Mo absorbed a large part of the probe
pulse, leading to much reduced di†use reÑection, and did not
allow us to evaluate k with high accuracy and reproducibility.
r
It is clear from Table 1 that k increased with metal loading
r
Table 1 Second-order rate constant (k )a of bare and transition-
r
metal-loaded TiO powders
2
k /cm3 ps~1
r
Loaded
metal
0%
1.4
0.3%
1%
2%
5%
Bareb
V
Cr
1.9
2.8
2.6
2.3
2.2
1.8
2.3
3.1
2.3
4.1
2.5
2.3
2.1
1.9
3.7
3.4
4.6
3.0
2.5
5.2
2.3
È
È
4.8
È
È
È
2.2
Fe
Co
Cu
Mo
W
Fig. 2 Rates of main product formation on bare and transition
metal-loaded TiO powders vs. amount of loaded metal. Platinized
2
M-TiO powders, prepared by in situ photodeposition from aqueous
a Estimated from the decay proÐle of PP-DRS measurement on the
assumption that the relatively faster component obeys second-order
2
H PtCl , were used in reactions (a) and (b). Bare TiO (L); V-TiO
2
6
2
2
(+); Cr-TiO (|); Fe-TiO (K); Co-TiO ()); Cu-TiO (…); Mo-
rate equation. b A TiO powder without transition metal.
2
2
2
2
2
TiO (=); W-TiO (È).
2
2
Phys. Chem. Chem. Phys., 2001, 3, 267È273
269

View Article Online
beyond the band gap, it was expected that loading would be
beneÐcial to the photoactivity. Indeed, absorption of visible
light can occur in the presence of a loaded metal. However,
the results shown in Fig. 2 clearly indicate that visible light-
induced activity had little or no e†ect and that there was an
appreciable reduction in activity by band-gap excitation.
Therefore, we can assume that the photoactivity shown in Fig.
2 is essentially due to band-gap excitation.
An exception to the metal-loading e†ect was Cu in reaction
(c); the rate of CO evolution increased with Cu-loading up to
2
1%, but higher loading gave rise to slightly slower rates com-
pared with the observed maximum value. Possible explana-
tions are: (1) Loaded Cu itself induces photochemical reaction
for the decomposition of acetic acid to CO . (2) Loaded Cu
2
was dissolved as cations in the aqueous phase (It has been
reported that the oxidation of several organic substrates can
be considerably accelerated by addition of Cu2` to an
aqueous suspension.36h39) (3) Loaded Cu is reduced into a
metallic state (Cu0) to act as a co-catalyst, enhancing the
photocatalytic activity. The Ðrst hypothesis seems unlikely
because photoirradiation of a Cu (1%)ÈSiO (OX50, Nippon
2
Aerosil, 50 m2 g~1) sample, prepared by a procedure similar
to that used for the Cu-TiO samples, resulted in no evolution
2
of CO . The second possibility, i.e., acceleration by dissolved
Fig. 3 Relation between k values and rate of main product forma-
2
r
Cu species (presumably Cu2`) is even less probable. In the
tion. Upper, H evolution (reaction (a)); lower, PCA formation
2
literature, indeed, it has been reported that the presence of
dissolved Cu2` ions is beneÐcial because they can act as elec-
tron traps, improving the availability of the holes for pro-
duction of active hydroxyl radicals;36,37 if so, the rate for
reactions (a) and (b) would be increased, as for reaction (c). On
the basis of these considerations, the hypothesis of the pres-
ence of deposited metallic Cu could justify the enhancement of
the reaction rate. The redox potential of Cu2`/Cu (]0.1 V vs.
SCE) is more positive than the conduction-band edge of
TiO . In fact, an ivory-colored suspension of Cu-TiO in
(reaction (b)). The meanings of the symbols same as those in Fig. 2.
The relation between the rate of reaction (c) and k (Fig. 4)
r
was rather ambiguous, even if the data obtained using Cu-
TiO samples were not considered. The reciprocal relation
2
between k and the reaction rate cannot be applied to the
r
results shown in Fig. 4, as the reaction rate in this case
appears to be independent of k . This tendency appears
r
clearer if the plots of the results of Cr- and V-TiO samples
2
2
2
are removed from both Ðgures, as these powders showed
aqueous acetic acid solution turned black during the photoir-
radiation. The deposit may act as a co-catalyst for CO evolu-
tion in reaction (c). Similar deposition might occur in
reactions (a) and (b), but we noticed only a negligible change
in color, due to the presence of pre-deposited Pt. Pt acts as a
co-catalyst and consequently might obscure the e†ect of Cu.
much less signiÐcant photocatalytic activity in all the reaction
systems. Although no experimental evidence has so far been
2
reported, the activity of M-TiO for reaction (c) may be gov-
2
erned by factors other than the recombination rate, such as
redox reactions consuming e~ and h`, and/or metal surface
coverage inhibiting e~ (or h`) transfer to the reacting
substrate(s) or the adsorption of the substrate(s). On the basis
The photocatalytic activity for H and PCA production of the
2
unplatinized Cu-TiO sample was appreciable but smaller
2
of these considerations, the di†erence in k dependence,
than that of the corresponding platinized sample.
r
reciprocal or independent, will be discussed and rationalised
in the following sections.
Apparently, there were two signiÐcant di†erences in the
experimental conditions under which the reactions (a) and (b)
Dependence of photocatalytic activity on k
r
Fig. 3 shows the correlations of the rates of reaction (a)
and (c) were carried out: (1) the absence and presence of O
(upper) and (b) (lower) with k , determined by PP-DRS and
2
r
summarized in Table 1. As has been previously reported,24
there was an almost linear relation between k values mea-
r
sured in air and in suspension for several TiO samples. Fur-
2
thermore, a negligible di†erence was obtained with and
without O . Consequently, the k value obtained in the
2
r
powder di†use reÑection system was used to examine the
correlation with photocatalytic activity in aqueous suspen-
sions. Although both plots were fairly scattered, it was clear
that the rate tended to decrease with increase in k . We
r
assumed that the activity for these reactions was inversely
proportional to the increase in k , i.e., the net activity was
r
dominated by the electronÈhole recombination rate. We
attempted to measure k for the platinized M-TiO samples
r
2
which were used in reaction (a) or (b) but failed due to the lack
of reÑection of the probe beam on their highly colored sur-
faces. However, in separate PP-DR analyses using several
commercial and synthesised TiO powders without transition-
2
metal-loading, an almost linear relation between the k values
r
of platinized and unplatinized samples was observed.40 There-
fore, the k values of unplatinized TiO powders were used for
convenience throughout this study.
Fig. 4 Relation between k values (cm~3 ps~1) rate of CO evolution
r
2
r 2
in reaction (c). The meanings of the symbols are as in Fig. 2.
270 Phys. Chem. Chem. Phys., 2001, 3, 267È273

View Article Online
and (2) the presence and absence of Pt deposits for (a) and (b)
and (c), respectively. In order to examine the inÑuences of
these factors, the following two additional experiments were
performed: decomposition of acetic acid under air with platin-
ized M-TiO powders (c1) and under an Ar atmosphere with
2
unplatinized M-TiO (c2). In these experiments, M-TiO
2
2
samples showing photoactivity similar to that of bare TiO in
reaction (c) were used to evaluate the dependence of photo-
2
catalytic activity on k as a sole parameter. Fig. 5È7 show the
r
results. For reaction (c1), the rate increased ca. three-fold for
bare TiO , but the ratio between reaction rates observed with
2
platinized and unplatinized samples vs. the k s decreased to 1
r
when k was [4 (Fig. 5). Thus, a k dependence of the (c1)
reaction similar to those of reactions (a) and (b) was obtained.
r
r
Fig. 7 Rate of CO evolution as a function of k values. Upper,
2
r
results in Ar (open symbols) and in air (closed symbols) for the
unplatinized samples; lower, ratio of these rates (Ar/air). The mean-
ings of the symbols are as those in Fig. 5.
It is known that the photocatalytic activity on platinized
TiO depends strongly on the procedure of Pt deposition, e.g.,
2
the di†erence in photocatalytic activity could be explained in
terms of the distribution of Pt deposited on the TiO particles,
2
not on the total amount of Pt deposit.41 Depending on the
intrinsic photocatalytic activity of M-TiO , the in situ photo-
2
platinization method used in this study might lead to a di†er-
ent Pt distribution. Although the di†erence, if it actually
exists, may be a measure of the intrinsic photocatalytic activ-
ity, this possibility could be excluded by performing the fol-
lowing experiment. The photocatalytic reaction (c1) was
Fig. 5 Rate of CO evolution vs. k values. Upper, platinized (open
2
r
symbols) and unplatinized (Ðlled symbols) M-TiO in the presence
2
of O ; lower, ratio of these rates, (platinized/unplatinized). Pt was
2
deposited by photoplatinization in reaction (a). Circles, triangles,
carried out by using platinized M-TiO prepared by deposi-
squares, diamonds, and inverted triangles denote bare TiO , Cr-TiO ,
2
2
2
tion from colloidal Pt. Since the platinization was achieved
Fe-TiO , Co-TiO , and W-TiO samples, respectively.
2
2
2
without any chemical reaction, similar distributions of Pt par-
ticles could be expected.32,42,43 Fig. 6 shows the results. A
similar relation between activity and k was also observed in
r
this case, indicating that the inÑuence of di†erent Pt distribu-
tions among the M-TiO powders could be neglected, i.e., the
2
observed k dependence of the photocatalytic activity is
r
caused by the inÑuence of the loaded transition-metal. In
other words, the platinization made the k dependence notice-
r
able.
In reaction (c2), the photocatalytic activity also became k -
r
dependent (Fig. 7). The activity decreased to ca. 2/3 of that of
bare TiO due to the transition-metal loading, though the
2
activity of the bare TiO was almost half of that under air.
2
The stoichiometry of this photocatalytic reaction was not
clear, although it is clear that acetic acid was decomposed to
liberate CO , and that the rate was much slower than in the
2
other systems. However, it should be noted that the presence
of O enhances the photocatalytic reaction and, at the same
2
time, obscures the inÑuence of e~Èh` recombination, as was
evaluated with k . Thus, the k dependence disappears in the
r
r
presence of O , while it is intensiÐed by platinization.
2
A model of the reaction kinetics
As described above, we observed that transition-metal loading
accelerates the e~Èh` recombination, resulting in the
reduction of photocatalytic activity of, in particular, the plat-
inized M-TiO , while the activity of the unplatinized M-TiO
Fig. 6 Rate of CO evolution vs. k values. Upper, platinized and
2
r
unplatinized M-TiO in the presence of O ; lower, ratio of these rates,
2
2
(platinized/unplatinized). Pt was deposited from colloidal Pt. The
meanings of the symbols are as in Fig. 5.
2
2
in the presence of O seems to be almost independent of the
2
Phys. Chem. Chem. Phys., 2001, 3, 267È273
271

View Article Online
loading. Since (1) photocatalytic reactions proceed via pho-
toexcitation of electrons beyond the band gap to yield e~ and
h` and (2) these redox active species react with surface-
adsorbed substrates (or otherwise recombine leading to no
chemical reaction), the rate of recombination must be included
in the rate expression. The following discussion on the model
of the reaction kinetics is based on the assumptions that only
there is no, or only negligible, migration of e~ (or h`) essential
for the occurrence of photocatalytic reactions.
Summary and conclusion
Several studies have suggested acceleration of e~Èh` recombi-
nation by transition metal doping and/or loading of
TiO ,9h14 but there is insufficient experimental evidence
the band-gap excitation of TiO participates, even though M-
2
2
because the recombination dynamics is much faster than the
TiO had an extended absorption, and also that the rate of
2
detection limit of conventional time-resolving measurements.
The present study on PP-DR spectroscopy of transition
e~Èh` formation in each M-TiO photocatalyst is the same.
When Ðrst-order rates are assumed for convenience for both
2
metal-loaded TiO powders directly revealed acceleration of
recombination of e~Èh` (mutual recombination) and surface
2
e~Èh` recombination by metal loading. Furthermore, two dif-
reaction of e~ and h` with the substrates, the rate of e~Èh`
ferent relations between photocatalytic activity and recombi-
nation rate were found, depending on the photocatalytic
system studied and on the experimental conditions. Reduction
of photocatalytic activity by metal loading for dehydroge-
nation of methanol and conversion of Lys into PCA (typical
photo-reactions carried out in deaerated conditions and in the
presence of Pt deposits on the surface of the catalyst) indi-
cated some difficulty in improving the photocatalytic activity,
e.g., an extension of the response in the visible-light region by
loading other metals or metal oxides on semiconducting
materials. In contrast, the results of photocatalytic oxidation
of acetic acid suggested that it might be possible to induce a
visible light response by metal loading for photocatalytic oxi-
dation systems in aerated conditions, such as decomposition
and/or mineralisation of organic compounds, without
reducing the primary photocatalytic activity in the UV region.
Thus, we have shown the e†ect of transition-metal loading of
escaping from their recombination (r ) is given, using a
esc
steady-state approximation for e~Èh` pairs, as follows:
r
\ I/kC/(kC ] k ),
(5)
esc
r
where I and / are the light Ñux of excitation and the probabil-
ity of light absorption to produce e~ and h`. A rate constant
(k) and surface concentration of substrate (C) were also
deÐned tentatively. In this ““homogeneousÏÏ model, an e~Èh`
pair behaves as a photoexcited molecule in a solution contain-
ing a reaction substrate of concentration C. When k is com-
r
parable to or larger than kC, r
should be inversely
esc
proportional to k , as observed in reactions (a) and (b). Thus,
it is expected that e~Èh` recombination is detrimental to the
photocatalytic activity unless the pairs react with the surface-
r
adsorbed substrate(s). On the other hand, if kC is much larger
than k , i.e., the surface reaction(s) proceeds efficiently before
r
the mutual recombination, r becomes independent of k , as
esc
r
TiO powders on photocatalytic activity and correlated the
was observed for reaction (c). It seems reasonable to expect
2
relation between photocatalytic activity and recombination
that the surface-adsorbed O reacts with e~. By assuming that
2
rate. These results should be useful for the design of further
functionalized photocatalytic materials for di†erent kinds of
reactions.
the overall photocatalytic reaction rate (r
) is proportional
overall
to r and that r is governed by eqn. (5), r
for reaction
esc
esc
overall
(c) must be larger than that for reaction (a) or (b). If k is
r
negligible, r is equal to I/ i.e., the quantum efficiency is
esc
Acknowledgements
100%. However, almost comparable rates, corresponding at
most to only a few% of quantum efficiency, were determined,
indicating that other factors might be responsible for the
decrease in r . For example, the reverse reaction of inter-
overall
mediate species formed via the surface reaction may cause the
The present work was partly supported by a Grant-in-Aid for
ScientiÐc Research in Priority Area of ““Electrochemistry of
Ordered InterfacesÏÏ (No. 11118208) from the Ministry of Edu-
cation, Science, Sports, and Culture, Japan.
reduction in r
.
overall
Another explanation for the di†erent k dependences is
based on the assumption that the distance between the place
r
References
where e~Èh` pairs form and react with the substrate(s)
depends on the type of photocatalytic reaction and on the
experimental conditions used. It has been reported that platin-
1
2
3
4
A. Mills and P. Sawunyama, J. Photochem. Photobiol. A, 1997,
108, 1.
K. Sayama and H. Arakawa, J. Chem. Soc., Chem. Commun.,
1992, 150.
ization of TiO efficiently enhances activity for photocatalytic
2
S. Tabata, H. Nishida, Y. Masaki and K. Tabata, Catal. L ett.,
1995, 34.
dehydrogenation of propan-2-ol and deaminocyclization of
Lys (reaction (b)).41 For these reactions, it was proved that
M. Anpo and H. Yamashita, in Heteroreneous Photcatalysis, ed.
M. Schiavello, Wiley, New York, 1997, p. 531.
M. A. Fox and M. T. Dulay, Chem. Rev., 1993, 93, 341.
B. Ohtani, in T rends in Photochemistry and Photobiology, ed.
Council of ScientiÐc Information, Research Trends, Trivandrum
(India), 1994, p. 531.
TiO particles loaded with at least one Pt deposit show suffi-
2
cient ability for propan-2-ol dehydrogenation and that further
5
6
Pt loading has a negligible or, rather, a negative e†ect. This
indicates that e~ formed in a relatively small (20È50 nm; cor-
responding to ca. 50 m2 g~1) TiO particle can migrate within
7
8
9
M. R. Ho†mann, S. T. Martin, W. Choi and D. W. Bahnemann,
Chem. Rev., 1995, 95, 69.
2
the whole volume of the particle. Since reaction (b) proceeds
via a redox combined mechanism including oxidation and
reduction by h` and e~, respectively, a larger number of Pt
deposits is needed to decrease the distance between the site for
oxidation of Lys and for reduction of the reaction interme-
diate. Thus, redox reactions occur at separate sites also in the
reactions (a) and (b). On the other hand, reaction (c) does not
need Pt loading. It has been presumed that oxidation of
A. L. Linsebigler, G. Lu and J. T. Yates, Chem. Rev., 1995, 95,
735.
E. Borgarello, J. Kiwi, M. Gratzel, E. Pelizzetti and M. Visca, J.
Am. Chem. Soc., 1982, 104, 2996.
10 Z. Luo and Q. H. Gao, J. Photochem. Photobiol. A, 1992, 63, 367.
11 K. E. Karakitsou and X. E. Verykios, J. Phys. Chem., 1993, 97,
1184.
12 N. Serpone and D. Lawless, L angmuir, 1994, 10, 643.
13 T. Ohno, F. Tanigawa, K. Fujihara, S. Izumi and M. Matsumura,
J. Photochem. Photobiol. A, 1999, 127, 107.
surface-adsorbed acetic acid by h` and reduction of O by e~
2
occur at neighbouring sites. On the basis of these consider-
14 M. Anpo, Y. Ichihashi, M. Takeuchi and H. Yamashita, Res.
Chem. Intermed., 1998, 24, 143.
ations, it seems reasonable to assume that the rates of reac-
tions (a) and (b) depend on k since e~ (or h`) must migrate in
15 J. Z. Zhang, R. H. OÏNeil and T. W. Roberti, Appl. Phys. L ett.,
1994, 64, 1989.
r
the bulk or on the surface of a TiO particle unless they
2
recombine at the recombination centres. The rate of reaction
16 J. Z. Zhang, R. H. OÏNeil, T. W. Roberti, J. L. McGowen and
J. E. Evans, Chem. Phys. L ett., 1994, 218, 479.
(c), on the other hand, is almost independent of k because
r
272
Phys. Chem. Chem. Phys., 2001, 3, 267È273

View Article Online
17 J. Z. Zhang, R. H. OÏNeil and T. W. Roberti, J. Phys. Chem.,
1994, 98, 3859.
30 S.-i. Nishimoto, B. Ohtani and T. Kagiya, J. Chem. Soc., Faraday
T rans. 1, 1985, 81, 2467.
18 J. J. Cavaleri, D. E. Skinner, J. D. P. Colombo and R. M.
Bowman, J. Chem. Phys., 1995, 103, 5378.
31 B. Ohtani, S. Tsuru, S.-i. Nishimoto, T. Kagiya and K. Izawa, J.
Org. Chem., 1990, 55, 5551.
19 J. J. Cavaleri, D. P. Colombo, Jr. and R. M. Bowman, J. Phys.
Chem. B, 1998, 102, 1341.
32 B. Ohtani, J.-i. Kawaguchi, M. Kozawa, S.-i. Nishimoto, T. Inui
and K. Izawa, J. Chem. Soc., Faraday T rans., 1995, 91, 1103.
33 B. Ohtani, J.-i. Kawaguchi, M. Kozawa, Y. Nakaoka, Y. Nosaka
and S.-i. Nishimoto, J. Photochem. Photobiol. A, 1995, 90, 75.
34 H. Kominami, J.-i. Kato, Y. Takada, Y. Doushi, B. Ohtani, S.-i.
Nishimoto, M. Inoue, T. Inui and Y. Kera, Catal. L ett., 1997, 46,
235.
20 D. E. Skinner, J. D. P. Colombo, J. J. Cavaleri and R. M.
Bowman, J. Phys. Chem., 1995, 99, 7853.
21 D. P. Colombo, Jr. and R. M. Bowman, J. Phys. Chem., 1995, 99,
11752.
22 D. P. Colombo, Jr., K. A. Roussel, J. Saeh, D. E. Skinner, J. J.
Cavaleri and R. M. Bowman, Chem. Phys. L ett., 1995, 232, 207.
23 D. P. Colombo, Jr. and R. M. Bowman, J. Phys. Chem., 1996,
100, 18445.
24 B. Ohtani, H. Kominami, R. M. Bowman, P. Colombo, Jr., H.
Noguchi and K. Uosaki, Chem. L ett., 1998, 579.
25 A. Furube, T. Asahi, H. Masuhara, H. Yamashita and M. Anpo,
Chem. L ett., 1997, 735.
35 H. Kominami, J.-i. Kato, M. Kohno, Y. Kera and B. Ohtani,
Chem. L ett., 1996, 1051.
36 A. Fujihira, Y. Satoh and T. Osa, Chem. L ett., 1981, 1053.
37 A. Sclafani, L. Palmisano and M. Schiavello, Res. Chem. Inter-
med., 1992, 18, 211.
38 M. Bideau, B. Claudel, L. Faure and H. Kazouan, J. Photochem.
Photobiol. A, 1991, 61, 269.
26 A. Furube, T. Asahi, H. Masuhara, H. Yamashita and M. Anpo,
J. Phys. Chem. B, 1999, 103, 3120.
39 M. I. Litter, Appl. Catal. B, 1999, 23, 89.
40 S. Ikeda, H. Kominami, Y. Kera, H. Noguchi, K. Uosaki and B.
Ohtani, J. Phys. Chem., to be submitted.
27 G. Rothenberger, J. Moser, M. Grazel, N. Serpone and D. K.
Sharma, J. Am. Chem. Soc., 1985, 107, 8054.
41 B. Ohtani, K. Iwai, S.-i. Nishimoto and S. Sato, J. Phys. Chem. B,
1997, 101, 3349.
42 A. Mills, J. Chem. Soc., Chem. Commun., 1982, 367.
43 P.-A. Brugger, P. Cuendet and M. Grazel, J. Am. Chem. Soc.,
1981, 103, 2923.
28 N. Serpone, D. Lawless and R. Khairutdinov, J. Phys. Chem.,
1995, 99, 16655.
29 A. D. Paola, S. Ikeda, G. Marc•, B. Ohtani, L. Palmisano and M.
Schiavello, to be published.
Phys. Chem. Chem. Phys., 2001, 3, 267È273
273