Kinetic study on photocatalytic hydrogenation of acetophenone derivatives on titanium dioxide

Article abstract of DOI:10.1039/c3cy00879g

Acetophenone (AP) derivatives were photocatalytically hydrogenated to afford the corresponding secondary alcohols with excellent chemical efficiencies on titanium dioxide (Degussa P25, TiO₂) under UV light irradiation. Maximum reaction rates (k_max) and apparent adsorption constants (K_LH) under irradiation were obtained from the Langmuir-Hinshelwood kinetic analysis. The k_max values showed a tendency to decrease with the decreasing reduction potentials (E_red) of the AP derivatives, while the K_LH values were distributed in the range of 280-780 L mol^-1. Among these, simple AP exhibited the greatest adsorptivity upon the UV irradiated TiO₂ surface. Additionally, it was demonstrated that the electrons trapped at surface defect Ti (Ti_st) sites on the TiO₂ actually hydrogenated the AP derivatives. The amount of reacted electrons also showed a tendency to decrease with decreasing E_red values, in accord with the dependence on k_max. These results indicate that the electrons accumulated at shallow Ti_st states easily participate in the hydrogenation of AP derivatives, whereas those trapped at deeper states hardly react with the substrates. The results strongly support the electron transfer reaction model via the Ti_st sites in the photocatalytic hydrogenation on TiO₂. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c3cy00879g

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Kinetic study on photocatalytic hydrogenation of
acetophenone derivatives on titanium dioxide†
Cite this: Catal. Sci. Technol., 2014,
4, 1084
*
*
Shigeru Kohtani, Yuna Kamoi, Eito Yoshioka and Hideto Miyabe
Acetophenone (AP) derivatives were photocatalytically hydrogenated to afford the corresponding
secondary alcohols with excellent chemical efficiencies on titanium dioxide (Degussa P25, TiO2) under
UV light irradiation. Maximum reaction rates (kmax) and apparent adsorption constants (KLH) under
irradiation were obtained from the Langmuir–Hinshelwood kinetic analysis. The kmax values showed a
tendency to decrease with the decreasing reduction potentials (Ered) of the AP derivatives, while the KLH
values were distributed in the range of 280–780 L mol−1. Among these, simple AP exhibited the greatest
adsorptivity upon the UV irradiated TiO2 surface. Additionally, it was demonstrated that the electrons
trapped at surface defect Ti (Tist) sites on the TiO2 actually hydrogenated the AP derivatives. The amount
of reacted electrons also showed a tendency to decrease with decreasing Ered values, in accord with the
dependence on kmax. These results indicate that the electrons accumulated at shallow Tist states easily
participate in the hydrogenation of AP derivatives, whereas those trapped at deeper states hardly react
with the substrates. The results strongly support the electron transfer reaction model via the Tist sites in
the photocatalytic hydrogenation on TiO2.
Received 3rd November 2013,
Accepted 7th January 2014
DOI: 10.1039/c3cy00879g
www.rsc.org/catalysis
are conventional solvents, such as water or alcohols, which
concurrently act as h+ scavengers, as mentioned above.
Introduction
Some metal oxides (e.g. titanium dioxide (TiO2)) are regarded
as semiconductors, in which electrons (e−) photogenerated in
the conduction band (CB) and holes (h+) simultaneously gen-
erated in the valence band (VB) play important roles in chem-
ical transformations as e− and h+ can induce redox reactions
on the surface. Photocatalytic hydrogenation on such semicon-
ductor particles proceeds via electron transfer into a substrate
followed by protonation. In general, photo-hydrogenation can
take place in the presence of a large excess amount of
electron donors, such as water, alcohols or amines, and in
the absence of molecular oxygen (O2). The aim of using an
electron donor is to scavenge h+ generated in the VB, thereby
diminishing the degree of recombination between e− and h+
within the particles.
Photocatalytic hydrogenation has received increasing
attention as a new method for the synthetically useful reduc-
tion of organic compounds having various double or triple
bonds.1–6 This methodology has some advantages compared
to conventional reduction methods: (1) the most important
merit is that particular reducing agents (e.g. H2 gas, NaBH4,
LiAlH4 etc.) are not necessary. In most cases, the reductants
Therefore, this method allows us to avoid both the use of
harmful and dangerous chemical reagents and the emission
of harmful waste. (2) The reactions mostly proceed under
mild conditions, e.g. under ordinary temperature and pres-
sure, and therefore are safe. (3) In the case of TiO2 or other
stable metal oxide photocatalysts, the materials are chemi-
cally stable, easily removable, and reusable. These significant
advantages imply that this methodology holds great promise
to become an alternative “green” synthetic method for reduc-
tive chemical transformations.
We have recently reported that acetophenone (AP) deriva-
tives can be photocatalytically hydrogenated to afford the
corresponding secondary alcohols on Degussa P25 TiO2
powders under UV light irradiation (Scheme 1).2 The desired
secondary alcohols were obtained with excellent chemical
efficiency, almost 100% yields, by choosing ethanol as a
sacrificial h+ scavenger, which was oxidized to acetaldehyde.
More recently, we have demonstrated that some AP deriva-
tives can be hydrogenated on P25 TiO2 powder modified with
organic dyes under visible light irradiation.3 In the study,
Department of Pharmacy, School of Pharmacy, Hyogo University of Health Sciences,
1-3-6, Minatojima, Chuo-ku, Kobe, 650-8530, Japan. E-mail: kohtani@huhs.ac.jp,
miyabe@huhs.ac.jp; Fax: +81 78 304 2858; Tel: +81 78 304 3158
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c3cy00879g
Scheme
1 Photocatalytic hydrogenation of AP derivatives: Ar =
aromatic ring, R = H, Me, Et, i-Pr, or CF3.
1084 | Catal. Sci. Technol., 2014, 4, 1084–1091
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
without further purification: 1-phenyl-2,2,2-trifluoroethanol
(Aldrich,
98%),
1-(2′,3′,4′,5′,6′-pentafluorophenyl)ethanol
(Aldrich, 97%), 1-(2′-fluorophenyl)ethanol (Aldrich, 97%),
1-(3′-fluorophenyl)ethanol (Avocado, 97%), 1-phenylethanol
(TCI, >98.0%), 1-(4′-fluorophenyl)ethanol (Avocado, 99%), and
1-phenyl-1-propanol (TCI, >97.0%).
2.2 Prolonged UV irradiation experiment
Irradiation experiments were carried out for a mixture of the
AP derivatives (initial concentration range: 1–20 mmol L−1)
and TiO2 (0.10 g) in a deaerated ethanol solution (30 mL)
under irradiation with UV light (>350 nm, light intensity:
790 mW cm−2) at 305 K. The details of the irradiation experi-
ment and GC analysis have been described in our previous
reports.2,4
Fig. 1 Reduction potential of AP derivatives vs. SHE in acetonitrile.2
we found that a suitable combination of dye-TiO2 and
triethylamine as
a sacrificial electron donor successfully
2.3 Pre-UV irradiation experiment
extended the UV response of TiO2 towards the visible light
region. We have further examined the adsorptive and kinetic
behavior in the photocatalytic hydrogenation of AP 5 and
2,2,2-trifluoroacetophenone (TFAP) 1 upon the UV irradiated
TiO2 surface.4 The study showed that the trapped electrons at
surface defect (Tist) sites7–9 actually take part in the hydroge-
nation of AP 5 and TFAP 1. Therefore, it is important for us to
prove the generality of the electron transfer reaction via the
Tist sites for other AP derivatives in photocatalytic hydrogenation.
For this purpose, we selected seven AP derivatives, 1–7,
with different reduction potentials (Ered vs. Standard Hydro-
gen Electrode, SHE), depicted in Fig. 1, and investigated the
dependence of Ered on the hydrogenation efficiency. Here,
maximum reaction rates (kmax) and apparent adsorption con-
stants (KLH) under UV irradiation were determined by the
Langmuir–Hinshelwood (LH) kinetic analysis. Furthermore,
the molar amount of reacted electrons trapped at Tist with
the AP derivatives was also estimated using a pre-irradiated
TiO2 system. On the basis of these data, a reasonable electron
transfer model via the Tist sites will be proposed in the
TiO2-catalyzed photo-hydrogenation of AP derivatives.
The TiO2 powder (0.10 g) was dispersed in ethanol (30 mL)
and this suspension was degassed with argon bubbling for at
least 30 min. After the degassed solution was irradiated with
UV light (>350 nm, light intensity: 1280 mW cm−2) for 2 h,
the white color of the TiO2 powder changed into blue–gray.
After confirming the color change, 300 μmol of the AP deriva-
tives was injected into this TiO2 suspension in the dark.
Then, electron transfer from the Tist sites to the adsorbed
AP derivatives took place and afforded the corresponding
secondary alcohols. The amounts of the products were quan-
titatively analyzed by GC-MS.4
2.4 Quantum chemical calculations
Density functional theory (DFT) calculations were performed
with Spartan'10 at the B3LYP/6-31+G* level of theory. Electro-
static potential energy maps of the AP derivatives were
obtained when the geometry optimizations were achieved,
which were evaluated by the heat of formation.
Results and discussion
3.1 Photocatalytic hydrogenation during the prolonged
UV irradiation
Materials and methods
Fig. 2 shows the time profiles of the photocatalytic hydro-
genation at several initial concentrations of o-F-AP 3 as a
representative AP derivative. The time dependence on the
decays of 3 (Fig. 2(a)) and the formation of the corresponding
secondary alcohol (Fig. 2(b)) indicate very similar features to
those reported for AP:4 (1) the hydrogenation reaction pro-
ceeds almost quantitatively, (2) the time dependence shows a
linear relationship between the concentration and irradiation
time in the high concentration region, suggesting zero-order
rate dependence at high concentration, and (3) the slopes of
the time profiles become smaller with a decreasing concen-
tration of 3, implying that the kinetic behavior varies from
zero- to first-order depending on the concentration of 3.
Other substrates 2, 4, 6, and 7 also exhibited similar time
dependence except for 1 (see: Fig. S1–S4 in the ESI†).
2.1 Materials
The TiO2 powder (Degussa P25, specific surface area:
35–65 m2 g−1) was used as received. HPLC grade ethanol
was purchased from Nacalai Tesque and used without
further purification. The following reagents were used as
substrates and their quantification on gas chromatography
(GC) as received: 2,2,2-trifluoroacetophenone (1, TFAP,
Aldrich, 99%), 2′,3′,4′,5′,6′-pentafluoroacetophenone (2, PFAP,
Aldrich, 97%), 2′-fluoroacetophenone (3, o-F-AP, Aldrich, 97%),
3′-fluoroacetophenone (4, m-F-AP, Aldrich, 97%), acetophen-
one (5, AP, Nacalai Tesque, 98.5%), 4′-fluoroacetophenone
(6, p-F-AP, Aldrich, 99%), propiophenone (7, PP, Aldrich,
99%). The following reagents were used for the identi-
fication and quantification of the desired products
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1084–1091 | 1085

View Article Online
Paper
Catalysis Science & Technology
ketal nor hemiketal species was observed for substrates 2–7.
Accordingly, common kinetic behavior was confirmed for the
AP derivatives 2–7.
The results listed in Table 1 indicate that Ered seems to
correlate with kmax but not with KLH. The KLH values are dis-
tributed in the range of 280–780 L mol−1, which are associ-
ated with adsorptivity onto the TiO2 surface under UV
irradiation. Henderson proposed the major adsorption state
of acetone on a rutile TiO2 (110) single crystal on the basis of
high-resolution electron energy-loss spectroscopy, in which
acetone presumably binds at Ti4+ sites via the lone pair on
the oxygen atom of acetone (η1 geometry).9,10 Later, he and
his co-workers calculated the adsorption energy for various
carbonyl compounds, including AP, using the DFT method
and obtained a sufficient negative adsorption energy for AP
(−77 kJ mol−1),11 suggesting the presence of η1 species on the
rutile TiO2 (110) surface. A similar adsorption model can also
be adapted to our system. We calculated the electrostatic
potential energy map of the AP derivatives 2–7 using the DFT
method (see Fig. S5 in the ESI†). The calculated electrostatic
potential energy on the oxygen atom seems to be well corre-
lated with the KLH values (Table 1), meaning that the AP
derivatives would most likely adsorb onto the surface Ti4+
sites with similar η1 geometry. The KLH value for AP 5 is
larger than that for PP 7, while the electrostatic potential
energy for AP 5 and PP 7 shows almost the same value
(ca. −180 kJ mol−1). This discrepancy may be explained in
terms of the UV-induced super-hydrophilic TiO2 surface.8,9,12
The hydrophobicity of organic compounds can be evaluated
by the octanol–water partition coefficient (P). The found values
of log P for AP 5 and PP 7 are reported to be 1.58 and 2.19,
respectively.13 This inversely implies that the hydrophilicity
of AP 5 is greater than that of PP 7, which therefore should
become a great advantage for the adsorption of AP 5 onto the
super-hydrophilic TiO2 surface.
Fig. 2 Time profiles of (a) the decay of o-F-AP 3 and (b) the formation
of the corresponding secondary alcohol in ethanol under the irradia-
tion conditions (>350 nm, 790 mW cm−2) at 305 K.
Fig. 3 depicts the initial reaction rates, v0, (the slopes in
Fig. 2(a)) vs. the initial concentration of the substrates in
ethanol at 305 K. The rates, v0, increase to attain maxi-
mum values asymptotically with increasing concentration in
solution. These data are well analyzed by the LH kinetic
expression (1):
kmaxKLHC0
v0 
(1)
1 KLHC0
where v0 is the initial reaction rate, kmax is the maximum
value of the reaction rate, KLH is the apparent adsorption
constant under irradiation, and C0 is the initial concentration
of the AP derivatives in equilibrium. The best fitting parame-
ters are listed in Table 1. The kmax values for 2–7 obtained
from the LH kinetic analysis show a tendency to decrease
with decreasing Ered of the AP derivatives. On the other hand,
the photo-hydrogenation of 1 followed the first-order rate law
because of the predominant formation of ketal or hemiketal
species in ethanol, i.e., only a few percent of 1 remained in
the original keto form.4 Thus, the kmax value for 1 is not eval-
uated in this study. On the contrary, neither the formation of
3.2 Electron transfer efficiency on the pre-irradiated
TiO2 surface
The electron transfer efficiency from the surface defect Tist to
the adsorbed AP derivatives on TiO2 was evaluated by the
injection experiment for a pre-irradiated TiO2 suspension.4 It
is well known that there are various Tist sites, for example,
the five coordinate Ti and the O vacancies on the TiO2 sur-
face and so on.7–9 The electronic energy of the Tist states is
almost located just below the CB edge of TiO2 in the range of
ca. 1 eV.9 Therefore, electrons excited in the CB band should
4+
relax to the Tist sites to yield the accumulated electrons
(Tist3+), which are indicated by the blue–gray color with an
absorption band from the visible to the IR region.14–19 After
being confirmed by the color change from white to blue–gray
in the 2 h pre-irradiation of the TiO2 suspension, a sufficient
amount of each AP derivative was injected into this suspen-
sion in the dark so that the surface electron transfer from
Fig. 3 Dependencies of the initial rates (v0) of hydrogenation at the
initial concentration of the six AP derivatives (2–7) under UV irradiation
(>350 nm, light intensity: 790 mW cm−2) at 305 K. These data were
3+
Tist to the adsorbed AP derivatives took place to afford the
fitted by eqn (1): v0
= kmaxKLHC0/(1 + KLHC0). The best fitting
parameters are listed in Table 1.
corresponding secondary alcohols.4 Fig. 4 shows plots of the
1086 | Catal. Sci. Technol., 2014, 4, 1084–1091
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Table 1 Reduction potentials, best fitting parameters for the LH kinetic expression (1), and the electrostatic potential energy on the oxygen atom
of the AP derivatives
Substrate
Ereda/V
kmax/10−3 mol L−1 h−1
KLH/L mol−1
Potential energyb/kJ mol−1
2
3
4
5
6
7
−1.59
−1.62
−1.80
−1.89
−1.92
−1.94
3.4 0.2
2.2 0.2
2.0 0.1
1.9 0.1
1.2 0.1
0.75 0.05
280 50
420 100
330 90
780 90
510 90
560 110
−147.4
−173.2
−168.4
−180.5
−172.4
−178.4
a
Reduction potential vs. SHE determined by cyclic voltammetry in CH3CN containing a Bu4NClO4 supporting electrolyte, in which the
standard potential of the Ag wire reference electrode used was compensated by the potential of the Fc+/Fc couple vs. SHE.2 b Calculated using
the DFT method (B3LYP/6-31+G*).
3+
Tist were consumed for the reductive hydrogenation of 1
because the color change of TiO2 from blue–gray to white
completed within 3 h after the injection of 1.4 Therefore, the
3+
total amount of Tist generated on the TiO2 powder was esti-
mated to be ca. 100 μmol g−1 after 5 h (the plots for 1 in
Fig. 4). This value is roughly consistent with the reported one
(50 μmol g−1), determined using the surface reaction of Ti3+
with methylviologen to afford its cation radical on the P25
TiO2 powder in a deaerated aqueous solution containing
methanol as a sacrificial reagent.20 On the other hand, in the
case of the other AP derivatives 2–7, part of the blue–gray spe-
cies on TiO2 remained even 5 h after the AP derivatives were
injected, which must be due to the deeply trapped electrons
at Tist remaining on the P25 TiO2 powder. This means that
Fig. 4 Time evolutions of the molar amount of reacted electrons after
the injection of substrate 1–7 (300 μmol) into the 2 h pre-irradiated
TiO2 suspension at 305 K.
3+
the electrons accumulated at the shallow defect states easily
participate in the reduction of the AP derivatives 2–7, whereas
those trapped at deeper states hardly transfer on the TiO2
surface. Assuming that all of the accumulated electrons react
with 1 on the TiO2 surface, the percentages of reacted
electrons were estimated for the other AP derivatives, as sum-
marized in Table 2. The values roughly depend on Ered except
for AP 5. The upward deviation for AP 5 may be due to the
greater adsorptivity, as mentioned in the previous section.
molar amount of reacted electrons after the injection of each
of the seven substrates. The amount of reacted electrons first
grew up and almost reached a constant value after 3 h. Just
for information, the amount of reacted electrons is two times
larger than that of the secondary alcohols, because the single
electron transfer takes place twice in the hydrogenation.
Although TFAP 1 gives the ketal or hemiketal form in
ethanol in equilibrium, a sufficient concentration of the keto
form of 1 (ca. 10 mmol L−1) near the TiO2 surface can be
attained immediately after the injection of 1, because it sur-
vives as the keto form for about ten minutes at 305 K.4 In
addition, we found that all of the accumulated electrons at
3.3 Reaction model on the TiO2 surface
Fig. 5 indicates the dependencies of the amount of reacted
electrons (several μmol) and the maximum reaction rates
(kmax) on Ered for the seven AP derivatives 1–7. The amount of
Table 2 Reduction potentials, the amount of reacted electrons at 5 h, and percentages of reacted electronsa
Substrate
Ered/Vb
The amount of reacted electrons/μmol
Percentagec/%
1
2
3
4
5
6
7
−1.35
−1.59
−1.62
−1.80
−1.89
−1.92
−1.94
10.2
8.22
6.32
6.09
7.38
5.70
4.76
100
81
62
60
72
56
47
a
The molar amount of reacted electrons was estimated at 5 h after the injection of substrate 1–7 (300 μmol) into the 2 h pre-irradiated TiO2
b
c
suspension at 305 K. Reduction potential vs. SHE (see the footnote in Table 1). Percentage of the reacted electrons per the total amount of
Tist (10.2 μmol) generated on 0.10 g of the TiO2.
3+
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1084–1091 | 1087

View Article Online
Paper
Catalysis Science & Technology
Fig. 5 The dependencies of the amount of reacted electrons (■) and
k
max (●) on Ered for the seven AP derivatives 1–7.
Fig. 6 Schematic illustration of the electron transfer reaction from the
Tist sites to the adsorbed AP derivatives, where Ered is the reduction
potential of the AP derivative,
λ is the reorganization energy
reacted electrons showed a tendency in decrease on decreas-
ing the Ered values, in accord with the dependence on kmax
(ca. 0.7 eV),25 and E0 (= qEred − λ) is the energy at the top of the curve
for the acceptor level (solid line). The dotted line indicates the donor
energy level for the anionic species of the AP derivatives.
.
Thus, the kmax values of the AP derivatives show a strong
correlation with the amount of reacted electrons on the pre-
irradiated TiO2 surface. This implies that the rates of photo-
catalytic hydrogenation of the AP derivatives are governed by
the electron transfer efficiency from the Tist sites to the
adsorbed AP ones.


2 

(E  qEred  )
4kBT
k (E)  k (E )exp
(3)
et
et
0




Let us consider the dependence of kmax on Ered and the
relationship between kmax and the amount of reacted
electrons on the basis of the Marcus theory at the semicon-
ductor/liquid interface.21–24 From the result of the good cor-
relation between kmax and the amount of reacted electrons
(Fig. 5), the reaction rate for AP derivatives possessing differ-
ent Ered is supposed to be closely associated with the
electronic energy of Tist states and their distribution within
the band gap. However, it is difficult to obtain information
about the energy distribution of Tist states on the actual TiO2
surface. Ondersma and Hamann have recently applied this
theory to the model of electron transfer recombination
where λ is the reorganization energy of the acceptor species
near the TiO2 surface, which is the sum of the inner-sphere
and outer-sphere components,21–24 q is the charge of an
electron, kB is the Boltzmann's constant, and T is the temper-
ature. ket(E) also contains the electronic coupling term, which
may inherently differ between types of Tist states and the
adsorbed AP derivatives but is assumed here to be indepen-
dent of the energy over the range of interest, as previously
reported.22–24 When E is E0 (= qEred − λ) at the top of the
curve of the acceptor level (solid line), ket(E0) exhibits the
peak rate constant, which is expected to be a weak function
of λ (ket(E0) ∝ λ−1/2). The reorganization energy, λ, is found to
be more or less 0.7 eV for all the AP derivatives because the
molecular radius of the AP derivatives is almost the same size
(ca. 0.35 nm, see Fig. S5 in the ESI†).25
The dependence of kmax on Ered, depicted in Fig. 5, should
be closely related to eqn (2) and (3). Therefore, it is important
for us to consider the relative energy position between the CB
edge and Ered (Fig. 6). There is a discrepancy between the CB
edge of TiO2 in ethanol (−1.15 V vs. SHE)26 and Ered of the AP
derivatives. The reduction potential of AP in an ethanol–H2O
(1 : 1) mixture was reported to be −1.42 V,27 the value of
which is considerably shifted toward the positive direction
(+0.47 V) compared to that in acetonitrile (−1.89 V) listed in
Table 1 and 2. Thus, Ered for other AP derivatives would also
shift in the same direction in ethanol. However, even though
the positive shift is taken into consideration, the Ered of AP
in ethanol (−1.42 V) is located at a more negative potential
than the flatband potential in the dark (−1.15 V vs. SHE),26
which is almost comparable with the CB edge of TiO2. Even
between a nanoparticle TiO2 electrode and redox shuttle.24
A
similar model can be illustrated for our electron transfer
reaction, as shown in Fig. 6, where the fluctuating energy
levels for the adsorbed acceptor (solid line: neutral AP deriva-
tives) and donor (broken line: their anionic species) are indi-
cated in accordance with the Marcus theory. In this model,
when the AP derivatives fully adsorb onto the Tist sites and
show the maximum coverage, the contributions to the reac-
tion rate from all the occupied Tist states must be included
in the net electron transfer rate, kmax, which is therefore inte-
grated from the bottom energy of the Tist states (Eb) to the
band edge of the CB (Ecb) as follows:
Ecb
kmax

g(E) f (E  EF)k et (E)dE
(2)
E
b
where g(E) is the density of surface trap states, f(E − EF)
is the Fermi–Dirac function indicating the trap state occu-
pancy at a given Fermi level EF, and ket(E) is the electron trans-
fer rate constant from a surface Tist state at energy E given by
1088 | Catal. Sci. Technol., 2014, 4, 1084–1091
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
if the CB edge of TiO2 under UV irradiation slightly shifts
towards the negative direction because of the electron
accumulation in the CB and at the Tist sites, the Ered of AP
(−1.42 V) is still negatively positioned compared to the CB
edge of TiO2. This implies that some AP derivatives with neg-
ative Ered values (e.g. 4–7) cannot be reduced on UV irradiated
TiO2, which is, of course, not true. An interesting question is
how the AP derivatives, especially substrates 4–7 possessing
negative Ered values, can be hydrogenated on the TiO2 sur-
face. The adsorption model of the η1 geometry may offer an
reasonable explanation for the further positive shift of Ered of
the AP derivatives because the interaction between the car-
bonyl oxygen atom and the surface Ti4+ sites induces an
electron deficiency in the π-electron system and lowers the π*
LUMO level of the AP derivatives. Consequently, the adsorbed
AP derivatives can be reduced by the electrons trapped at the
Tist sites on the TiO2 surface. The effect of adsorption on the
positive shift of Ered would be greater for AP 5 because of its
strongest adsorption property (see Table 1). Thus, the effect of
adsorption on the positive shift of Ered can lead to the enhance-
ment of the electron transfer efficiency on the TiO2 surface.
On the basis of the above consideration, the relative
energy position between the CB edge and Ered can be reason-
ably illustrated in Fig. 6, where Ered varies depending on the
AP derivatives whereas λ can be fixed to be ca. 0.7 eV.25
Therefore, the Gaussian curves of the acceptor and donor
levels shown in Fig. 6 move upward and downward with the
change in Ered but without a significant change in the shape.
This means that a large number of Tist sites are accessible to
the electron transfer for the AP derivatives with a positive Ered
(e.g. 1–3) while the number of available Tist sites is limited
for those possessing a negative Ered (e.g. 6, 7). The proportion
of accessible Tist sites should be related to the percentages
listed in Table 2. Here, the fluctuating acceptor level
(the Gaussian curve) of TFAP 1 would entirely cover the whole
range of the electronic energy of Tist states within the band
gap, thereby attaining a percentage of 100%. Thus, this
model can satisfactorily explain the kmax dependence on Ered
as well as the relationship between kmax and the amount of
reacted electrons shown in Fig. 5, though the numerical anal-
ysis of this model seems to be difficult since the individual
contributions of ket(E) (eqn (3)) are convoluted in eqn (2).
The Marcus inverted region23 was not observed in our reac-
tion system because of two reasons: 1) the considerably nega-
tive Ered of the AP derivatives, i.e., an insufficient driving
force for this reaction and 2) the widely distributed Tist states
within the band gap (1 eV more or less).9
In the photocatalytic hydrogenation of AP, the maximum
reaction rate, kmax, increased with increasing incident light
intensity, I.4 Therefore, the photo-hydrogenation of AP pro-
ceeds in a light-limited controlled manner. Since the excited
electrons in the CB rapidly migrate and distribute to the Tist
sites (within 500 ps even for the deep trap sites),28,29 the con-
centration of excited electrons accessible to the reaction
should be related to the light intensity. The intensity depen-
dence on kmax can be formulated in kmax ∝ In,30–32 where n is
usually taken from 0.5 to 1.0 depending on the light inten-
sity. At a high light intensity, n is 0.5 because the Tist sites
are saturated by the excited electrons and the band to band
e−–h+ recombination becomes dominant (second order). At a
very low light intensity, n is 1.0 because the Tist sites are not
saturated and therefore the e−–h+ recombination via the Tist
sites becomes first order. The intensity dependence on kmax
for AP was in proportion to I0.68 (n = 0.68), indicating that
both the band to band recombination and the recombination
via the Tist sites occur simultaneously.
Finally, we propose a reasonable reaction mechanism,
depicted in Fig. 7. First, the adsorbed AP species (most likely
Fig. 7 A proposed reaction mechanism for the photo-hydrogenation of AP derivatives on the TiO2 surface.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1084–1091 | 1089

View Article Online
Paper
Catalysis Science & Technology
η1 geometry) are reduced by the accumulated electrons to the
anionic species through the first electron transfer step, which
is strongly affected by the relative position of Ered. In this
step, the trapped electrons seem to be transferred from the
Ti 3d orbital to the π* orbital on the carbonyl moiety. The
next protonation step should be prior to the back electron
transfer from the anionic species to the deep Tist sites
because the trapped electrons at the deep Tist sites mostly
remain, or even if the trapped electrons are consumed by the
reaction, another electron can be rapidly supplied from the
accumulated CB electrons within 500 ps (Fig. 6).28,29 This
protonation step may be associated with the ethanol oxida-
tion by h+ on the TiO2 surface. Recently, Morris et al. pro-
posed the photo-oxidation mechanism of methanol on rutile
TiO2 nanoparticles by means of FT-IR spectroscopy, in which
the photogenerated holes play a crucial role in the methanol
oxidation and produce H+ and a new surface hydroxyl group,
HObr−, during the reaction.33 In a similar manner, H+ and/or
2 S. Kohtani, E. Yoshioka, K. Saito, A. Kudo and H. Miyabe,
Catal. Commun., 2010, 11, 1049–1053.
3 S. Kohtani, S. Nishioka, E. Yoshioka and H. Miyabe,
Catal. Commun., 2014, 43, 61–65.
4 S. Kohtani, E. Yoshioka, K. Saito, A. Kudo and H. Miyabe,
J. Phys. Chem. C, 2012, 116, 17705–17713.
5 S. Füldner, R. Mild, H. I. Siegmund, J. A. Schroeder,
M. Gruber and B. König, Green Chem., 2010, 12, 400–406.
6 Y. Shiraishi, Y. Togawa, D. Tsukamoto, S. Tanaka and
T. Hirai, ACS Catal., 2012, 2, 2475–2481.
7 U. Diebold, Surf. Sci. Rep., 2003, 48, 53–229, and references therein.
8 T. L. Thompson and J. T. Yates, Chem. Rev., 2006, 106,
4428–4453, and references therein.
9 M. A. Henderson, Surf. Sci. Rep., 2011, 66, 185–297, and
references therein.
10 M. A. Henderson, J. Phys. Chem. B, 2004, 108, 18932–18941.
11 M. A. Henderson, N. A. Deskins, R. T. Zehr and M. Dupuis,
J. Catal., 2011, 279, 205–212.
12 R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni,
E. Kojima, A. Kitamura, M. Shimohigashi and T. Watanabe,
Nature, 1997, 388, 431–432.
−
HObr might be generated during the photo-oxidation of
ethanol on the P25 TiO2 surface and associated with the pro-
tonation reaction. The second electron transfer should be
faster than the first one because the predicted reduction
potential of the acetophenone ketyl radical (−1.59 V vs. SHE
in CH3CN calculated by a quantum chemical calculation)34
is +0.3 V more positive than that of AP 5 (−1.89 V). At the
end of the sequential reaction, the final secondary alcohol
product would be formed via the rearrangement and the sec-
ond protonation.
13 J. Li, Chromatographia, 2004, 60, 63–71.
14 T. Beager, M. Sterrer, O. Diwald, E. Knozinger, D. Panayotov,
T. L. Thompson and J. T. Yates, J. Phys. Chem. B, 2005, 109,
6061–6068.
15 S. H. Szczepankiewicz, A. J. Colussi and M. R. Hoffmann,
J. Phys. Chem. B, 2000, 104, 9842–9850.
16 S. H. Szczepankiewicz, J. A. Moss and M. R. Hoffmann,
J. Phys. Chem. B, 2002, 106, 2922–2927.
17 S. H. Szczepankiewicz, J. A. Moss and M. R. Hoffmann,
J. Phys. Chem. B, 2002, 106, 7654–7658.
Conclusions
We have demonstrated that the photocatalytic hydrogenation
of AP derivatives proceeds via the surface defect Ti (Tist) sites
on the TiO2 surface, where they are not only adsorption
sites but also electron trap sites. Under UV irradiation, the
electron transfer event at the Tist sites initiates the hydroge-
nation of AP derivatives, in which the reaction rate is strongly
affected by the reduction potential (Ered) of the substrates.
A reasonable electron transfer reaction model via the Tist
sites (Fig. 6) is proposed in this study. The results provide
good insight into valuable information about the reaction
mechanism of the photo-hydrogenation of AP derivatives as
well as the physicochemical properties of the Tist sites on
the TiO2 surface.
18 M. Takeuchi, G. Martra, S. Coluccia and M. Anpo, J. Phys.
Chem. C, 2007, 111, 9811–9817.
19 M. Takeuchi, J. Deguchi, S. Sakai and M. Anpo, Appl. Catal., B,
2010, 96, 218–223.
20 S. Ikeda, N. Sugiyama, S. Murakami, H. Kominami, Y. Kera,
H. Noguchi, K. Uosaki, T. Torimoto and B. Ohtani,
Phys. Chem. Chem. Phys., 2003, 5, 778–783.
21 Y. Q. Gao, Y. Georgievskii and R. A. Marcus, J. Chem. Phys.,
2000, 112, 3358–3369.
22 N. S. Lewis, J. Phys. Chem. B, 1998, 102, 4843–4855.
23 T. W. Hamann, F. Gstrein, B. S. Brunschwig and N. S. Lewis,
J. Am. Chem. Soc., 2005, 127, 7815–7824.
24 J. W. Ondersma and T. W. Hamann, J. Am. Chem. Soc., 2011,
133, 8264–8271.
25 The outer-sphere reorganization energy for the AP derivatives
at the TiO2/ethanol interface can be estimated by the
following equation:22–24
Acknowledgements
This work was supported by Grant-in-Aid for Scientific
Research (no. 22590026 and 24590067) from the Japan
Society for the Promotion of Science.
2
2
2
2









2










n
 n




(zq)
8
1
1
1
1
1
1
TiO2
s
TiO2
TiO2
s
s



 




 




2
2
 




a

2R

s
n
n
n
 n
s
0
s



s

s
TiO2





Notes and references
where Δ z is the change in the charge of the AP derivative, a
is the radius of the AP derivative (ca. 0.35 nm estimated
1 S. Kohtani, E. Yoshioka and H. Miyabe, in Hydrogenation,
ed. Karamé, InTech, Rijeka, 2012, ch.12, pp. 291–308, (This is an
open access chapter, DOI: 10.5772/45732), and references therein.
using the DFT calculation), ns and nTiO are the refractive
2
1090 | Catal. Sci. Technol., 2014, 4, 1084–1091
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
index of ethanol (1.36)27 and anatase TiO2 (2.5),24 respec-
28 Y. Tamaki, A. Furube, M. Murai, K. Hara, R. Katoh and
M. Tachiya, Phys. Chem. Chem. Phys., 2007, 9, 1453–1460.
29 Y. Tamaki, K. Hara, R. Katoh, M. Tachiya and A. Furube,
J. Phys. Chem. C, 2009, 113, 11741–11746.
30 U. Stafford, K. A. Gray and P. V. Kamat, J. Catal., 1997, 167,
25–32.
31 A. Mills, J. Wang and D. F. Ollis, J. Phys. Chem. B, 2006, 110,
14386–14390.
32 A. Mills, J. Wang and D. F. Ollis, J. Catal., 2006, 243, 1–6.
33 D. A. Panayotov, S. P. Burrows and J. R. Morris, J. Phys.
Chem. C, 2012, 116, 6623–6635.
tively, εs and εTiO are the static dielectric constant of ethanol
2
(24.6)27 and anatase TiO2 (114),24 respectively, and R is the
distance from the adsorbed AP derivatives to the TiO2 sur-
face. Thus, we obtained the outer-sphere reorganization
energy as 0.7 eV by assuming R = a = 0.35 nm. Although λ is
the sum of the inner-sphere and outer-sphere components, λ
should be more or less 0.7 eV because the calculated inner-
sphere reorganization energy for organic dyes adsorbed on
TiO2 nanoparticles is reported to be negligibly small.35
26 G. Redmond and D. Fitzmaurice, J. Phys. Chem., 1993, 97,
1426–1430.
34 Y. Fu, L. Liu, H.-Z. Yu, Y.-M. Wang and Q.-X. Guo, J. Am.
Chem. Soc., 2005, 127, 7227–7234.
35 E. Maggio, N. Martsinovich and A. Troisi, J. Phys. Chem. C,
2012, 116, 7638–7649.
27 M. Montalti, A. Credi, L. Prodi and M. T. Gandolfi, in
Handbook of Photochemistry, CRC, Boca Raton, 3rd edn,
2006, ch. 7, pp. 493–527.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1084–1091 | 1091