Role of water and carbonates in photocatalytic transformation of CO 2 to CH4 on titania

Article abstract of DOI:10.1021/ja108791u

Using the electron paramagnetic resonance technique, we have elucidated the multiple roles of water and carbonates in the overall photocatalytic reduction of carbon dioxide to methane over titania nanoparticles. The formation of H atoms (reduction product) and ?OH radicals (oxidation product) from water, and CO₃^- radical anions (oxidation product) from carbonates, was detected in CO₂-saturated titania aqueous dispersion under UV illumination. Additionally, methoxyl, ?OCH₃, and methyl, ?CH₃, radicals were identified as reaction intermediates. The two-electron, one-proton reaction proposed as an initial step in the reduction of CO₂ on the surface of TiO₂ is supported by the results of first-principles calculations.

Full text of DOI:10.1021/ja108791u

ARTICLE
pubs.acs.org/JACS
Role of Water and Carbonates in Photocatalytic Transformation
of CO to CH on Titania
2
4
,
†,‡
#
†
‡
#
Nada M. Dimitrijevic,* Baiju K. Vijayan, Oleg G. Poluektov, Tijana Rajh, Kimberly A. Gray,
§
§,†
Haiying He, and Peter Zapol
†
‡
§
Chemical Sciences and Engineering Division, Center for Nanoscale Materials, and Materials Science Division,
Argonne National Laboratory, Argonne, Illinois 60439, United States
#
Department of Civil & Environmental Engineering, Northwestern University, Evanston, Illinois 60208, United States
S Supporting Information
b
ABSTRACT: Using the electron paramagnetic resonance technique, we have elucidated the multiple roles
of water and carbonates in the overall photocatalytic reduction of carbon dioxide to methane over titania
•
nanoparticles. The formation of H atoms (reduction product) and OH radicals (oxidation product) from
-
water, and CO3 radical anions (oxidation product) from carbonates, was detected in CO -saturated
2
•
•
titania aqueous dispersion under UV illumination. Additionally, methoxyl, OCH , and methyl, CH ,
3
3
radicals were identified as reaction intermediates. The two-electron, one-proton reaction proposed as an
initial step in the reduction of CO on the surface of TiO is supported by the results of first-principles
2
2
calculations.
’
INTRODUCTION
potentials are presented in reference to NHE at pH 7.
þ
-
Photocatalytic reduction of CO to formaldehyde, formic acid,
2
CO þ 2H þ 2e
f HCOOH
2
CB
methanol, and methane as main products over semiconductor
Eꢀ ¼ - 0:61 V
ð1Þ
ð2Þ
1
particles was demonstrated three decades ago. Since then, there
þ
-
has been an effort to design an efficient and selective photo-
HCOOH þ 2H þ 2eCB f HCHO þ H2O
Eꢀ ¼ - 0:48 V
catalytic system for the reduction of CO in water that can work
2
2
-4
without any loss of the light energy through chemical storage.
þ
-
The focus has been on TiO , an efficient photocatalyst for
HCOH þ 2H þ 2e
f CH OH
2
CB
3
environmental applications, because it is highly stable, nontoxic,
and cheap and absorbs part of the solar spectrum. The use of
titania and water (without sacrificial hole scavengers) ultimately
provides a “green chemistry” approach to the transformation of
Eꢀ ¼ - 0:38 V
ð3Þ
ð4Þ
ð5Þ
ð6Þ
ð7Þ
þ
-
CH OH þ 2H þ 2e
f CH þ H O
3
CB
4
2
Eꢀ ¼ - 0:24 V
CO to fuel. However, the photoefficiency of CO reduction to
2
2
H2O þ hVB f OH þ Hþ
þ
methane over titania is very low, <0.1%, using only water as an
electron donor. The majority of research is aimed at improving
photocatalytic efficiency and/or harvesting visible light. The
Eꢀ ¼ þ 2:32 V
þ
þ
2
H O þ 2h
f H O þ 2H
2
VB
2
2
improved efficiency of CO reduction was demonstrated for
2
Eꢀ ¼ þ 1:35 V
the synthesized novel titania-based nanomaterials having specific
5
,6
7,8
9
2H2O þ 4hVB f O2 þ 4Hþ
þ
surface structures, for coupling titania to dyes, enzymes,
metallic catalysts,
1
0,11
12
or quantum dots. Still, a better under-
Eꢀ ¼ þ 0:82 V
standing of processes that occur on the surface of TiO during
2
Although conduction band electrons of TiO , E = -0.50 V at
2
CB
carbon dioxide reduction, including adsorption/desorption of
17
pH 7, are sufficiently energetic to drive CO reduction to
2
CO and intermediate products, as well as the role of adsorbed
2
methane, Eꢀ(CO /CH ) = -0.24 V, the detailed mechanism is
not well understood, and it is proposed to involve shared
intermediates and multiple reaction pathways starting with the
formation of CO2 radical anion bound to the oxide surface.
To our knowledge, there is very little experimental data on the
2
4
water is needed in order to elucidate the reaction mechanism
which, in synergy with targeted synthesis, can improve overall
efficiency.
-
18
The reduction of CO to methane is a multistep process,
2
13-16
generally presented by the set of eqs 1-4.
dispersions/solutions, electrons are provided by photoexcitation
In aqueous
reaction intermediates. The formation of surface-adsorbed
of TiO (conduction band electrons), while water acts as both a
proton donor (eqs 1-4) and an electron donor (eqs 5-7). All
Received: September 29, 2010
Published: February 24, 2011
2
r 2011 American Chemical Society
3964
dx.doi.org/10.1021/ja108791u
|J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
CO2- on MgO in the absence of water and in the presence of
’ EXPERIMENTAL SECTION
excess electrons has been confirmed by electron paramagnetic
1
9-21
-
Materials. Sodium carbonate (Aldrich), acetonitrile (Sigma), and
the spin trap 5,5-dimethyl-1-pyrrolyne N-oxide (DMPO, Sigma) were
all analytical grade and used as received, without further purification.
Research-grade carbon dioxide, 99.999% (Airgas), was passed consecu-
tively through two hydrocarbon traps (Supelco) to remove even a trace
resonance (EPR),
while formation of CO2 at the surface
of TiO under UV illumination was detected by infrared (IR)
2
2
2
spectroscopy. Carbonate, bicarbonate, formic acid, formalde-
hyde, and methoxy have been observed on the oxide surfaces
11,20
during CO reduction.
2
amount of impurities. Powdered TiO (Degussa P25 Aeroxide) was
2
In general, a low yield of methane production over titania is
attributed to two major factors: (i) the strong oxidizing power of
valence band holes (EVB = þ2.70 V vs NHE at pH 7) that can
result in the reaction of holes (or OH radicals) with intermediate
molecular products (formate, formaldehyde, methanol), and (ii)
the thermodynamically unfavorable one-electron reduction of
used as received; no heat treatment was used to remove bound water.
Milli-Q water was used immediately upon production to reduce disso-
lution of carbon dioxide from air. TiO
2 2
was left in D O (Cambridge
Isotope Laboratories) for 3 months in order to exchange bound water.
-
5
Samples were degassed at 77 K using a 10 bar vacuum system, and a
controlled amount of CO gas (in the range 300-700 mbar) was
2
CO , based on a very negative redox potential for this process in
2
-
introduced into suprasil EPR tubes (Wilmad LabGlass).
an aqueous homogeneous system, Eꢀ(CO2aq/CO2 ) = -1.9 V
vs NHE. However, the presence of proton donors together with
aq
Instrumentation. X-band continuous-wave EPR experiments were
conducted on a Bruker Elexsys E580 spectrometer equipped with an
Oxford CF935 helium flow cryostat with an ITC-5025 temperature
controller. EPR spectra of photogenerated charges were recorded at
cryogenic temperatures, from 4.5 to 77 K, while measurements using
DMPO spin trap were performed at room temperature. The g factors
were calibrated for homogeneity and accuracy by comparison to a coal
standard, g = 2.00285 ( 0.00005. The weak pitch reference sample, in
combination with a known concentration of TEMPO radicals
(2,2,6,6-tetramethylpiperidine-1-oxyl, Aldrich), was applied for quan-
titative analysis. Double integration of the EPR spectra was carried out
using the Bruker data analysis package after baseline correction.
Samples were excited either with 355 nm photons from Nd:YAG
the binding of CO to the oxide surface could diminish redox
2
23,24
potential by almost an order of magnitude.
Anpo and co-
workers probed the properties of highly reactive titania-based
oxide materials as related to CO reduction and showed that the
2
product yields depend strongly on the coordination of surface Ti
atoms, the H O:CO ratio (maximum for 5:1), and the reaction
2
2
5
temperature. They demonstrated that the yield of methane
production increases with the concentration of undercoordi-
5
nated Ti at the surface. Carbon dioxide adsorbs preferentially on
25
undercoordinated Ti surface sites, both five- and four-
5
coordinated. The proximity/binding of CO to the surface
2
not only affects its redox properties but competes with the
recombination of charges, a thermodynamically favored process.
In contrast to CO , water dissociates on the surface of TiO ;
laser (power 9 mJ) or by using a 300 W Xe lamp (ILC) with a photon
-2 -1
flux of 1 mmol m
s .
2
2
An HP 5890 gas chromatograph equipped with a flame ionization
detector (GC-FID) was used to monitor photocatalytic production of
i.e., water oxygen forms a dative bond with a surface Ti, whereas
the hydrogens form two weak H-bonds with bridging oxy-
methane. Production of CH was measured in a closed reactor at room
2
6-28
4
gens,
giving rise to the formation of both terminal and
temperature containing 10 mg of TiO
atmospheric pressure) under continuous illumination with 365 nm
photons from a mercury vapor UV lamp (100 W), energy density of
2 2
(P25) and CO /water vapor
bridging OH groups at the surface. Molecular H O is further
2
(
bound to these OH surface groups, and more than three layers
29
constitute a bulk water. This strong binding/dissociation trans-
2
1
10 W/m . The sample holder was a circular pot with a diameter of
lates to the lowering of the barrier for a one-electron-transfer
1
.8 cm (total area for irradiation). The details of catalytic test activity are
process. Furthermore, water solvates CO , and at room tem-
30
2
presented elsewhere.
perature 0.2-1% of solvated/dissolved CO is in a form of
2
Computational Methods. A periodic slab model was used to
calculate the interactions on the surface. Our model system represents
an anatase (101) surface, which is the most abundant surface in anatase
featuring two types of under coordinated atoms: five-fold Ti atoms and
two-fold O atoms. A 2ꢂ1 supercell along the [010] and [101] directions
carbonic acid, eq 8. Interaction of carbonic acid with an OH-
functionalized surface could lead to dehydration and binding of
bicarbonate, or bicarbonate ions can be present in TiO /water
2
double layer, in both cases participating in a photocatalytic
process.
consisting of six TiO trilayers with four Ti atoms per layer was used.
2
Atoms in the bottom trilayer were fixed to their bulk positions. A vacuum
½
H2CO3ꢀ
-3
layer of about 11 Å was placed between slabs along the z direction.
Keq ¼
ꢁ 1:7 ꢂ 10
ð8Þ
31
½
CO2ꢀ
Calculations were done using PBE exchange correlation functional,
plane wave basis sets, and PAW potentials as implemented in the VASP
aq
3
2
In this paper we address the role of water and dissolved CO in
the form of carbonates/bicarbonates in the overall mechanism/
efficiency of CO reduction on titania, in the absence of sacrificial
2
program. The PBE functional has previously proved to reproduce
experimental results of the structural and energetic properties of TiO
2
3
3
2
bulk and surfaces. A Monkhorst-Pack grid of 2ꢂ2ꢂ1 was used to
-5
hole scavengers, applying EPR spectroscopy. EPR techniques
enable studies of photogenerated charges localized on titania and
charge-transfer processes leading to formation of paramagnetic
radicals directly or indirectly via a spin-trap method. As a
photocatalyst we have chosen Degussa P25 Aeroxide to avoid
discrepancies in materials synthesized in different laboratories,
and because it is widely used as a reference material for
demonstrating the efficiency of various photocatalysts. In this
work we compare data for partially hydrated and fully hydrated
titania surfaces in the absence and in the presence of adsorbed
CO2.
sample the first Brillouin zone. The total energy is converged to 10 eV,
and the force on each atom is converged to 0.03 eV/Å. A hydrogen atom
is introduced on the surface-bridging two-fold O atom to become a
proton with a formal charge state of þ1. The electron from this H
redistributes over Ti atoms to populate the bottom of the TiO2
conduction band within self-consistent calculations mimicking a photo-
excited electron, while the proton mimics a localized hole in terms of its
3
4
Coulomb potential. Charge distribution (see Supporting Information)
is analyzed using Bader charge analysis. Transition states along the
reaction pathways were located with the Climbing Image Nudged Elastic
3
5
3
6
Band method.
3
965
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
Scheme 1. Salient Presentation of the Energetics of Photo-
generated Charges, As Observed by EPR under Illumination
of TiO at Different Temperatures
2
Figure 1. Production of CH
0 mg of TiO (P25) and CO
under continuous illumination with 365 nm photons. For calculating
4
measured in a closed reactor containing
1
2
2
/water vapor (atmospheric pressure)
1
5
2
2
TON, the values of 10 active sites/m and surface area of 55 m /g for
P25 were taken into account.
4þ
a strong tridentate complex with Ti ions on the surface of
anatase nanoparticles, resulting in a one-electron reduction of
PQQ to semiquinone, to some extent even at extremely low
temperatures, below 30 K. Thus, our strategy was to illuminate
’
RESULTS AND DISCUSSION
43
The photocatalytic reduction of CO to CH was demon-
a reaction mixture of TiO /CO /H O at a higher temperature,
2
4
2
2
2
strated for TiO P25, a mixed-phase anatase/rutile titania.
77 K, in order to increase the yield of reducing reactions, and to
record EPR spectra at 4.5 K, in order to increase the sensitivity of
the measurements.
2
Figure 1 presents the turnover number (TON), expressed as a
number of CH molecules formed per number of surface active
4
1
0
12
2 37
sites (typically 10 -10 per cm ), under bandgap excitation
of TiO at room temperature in a system containing only carbon
The EPR spectrum recorded in the dark at 4.5 K after 1 h
illumination of TiO in the presence of CO and H O at 77 K is
2
2
2
2
dioxide and water vapor at atmospheric pressure. The quantum
presented in Figure 2. The spectrum consists of multiple para-
magnetic species having different g-tensor values and hyperfine
-
3
yield of the order of 10 (Supporting Information) is compar-
able to the ones previously reported for various titania nano-
composites, including pure anatse nanoparticles, mixed-phase or
splittings. The EPR signals with g = 1.991 and g = 1.975 are
^
^
attributed to electrons trapped in the lattice of anatase and rutile,
3
þ
38
nitrogen-doped titania, and TiO in the presence of metallic
(Ti )latt, respectively. The signal from holes localized on the
2
5
4þ •-
cocatalysts.
surface of titania, (Ti O )surf, is also denoted in Figure 2. The
EPR spectroscopy has been widely used to examine paramag-
two-line signal with 50.8 mT hyperfine splitting corresponds to
44
•
netic species formed upon bandgap excitation of TiO , including
H atoms, while CH radicals are identified by a four-line
2
3
3
8
P25. Excitation of nanocrystalline TiO with photon energies
spectrum with intensity ratios of 1:3:3:1, having g-tensor of
2
4
5
larger than its bandgap results in the femtosecond formation of
conduction band electrons and valence band holes. Once pro-
duced, the charge carriers become trapped into lower energy
states (shallow and deep traps), accompanied by their recombi-
nation. The strategy of using temperatures below 80 K in EPR
measurements allows for the detection of localized charges in
titania, since the rate of electron-hole recombination is reduced
2.002 and hyperfine splitting constant a = 2.3 mT. Similar
H
spectra were observed previously upon illumination of anatase-
5
type TiO . These results suggest that photogenerated electrons
2
react not only with CO but also with protons from H O bound/
2
2
dissociated on the TiO surface. Therefore, in the overall
2
photocatalytic process of CO reduction, water acts as an electron
2
acceptor. Competitive formation of H atoms with transfer of
39
5
at these temperatures. It was demonstrated by EPR that, when
electrons to carbon dioxide was previously suggested by Anpo
11
TiO is illuminated at temperatures <10 K, photogenerated
and Wu. When H O was exchaned with D O, no signals from
H atoms or CH radicals were observed upon illumination
2
2 2
•
electrons localize in the interior of nanoparticles, while holes,
3
40
due to their higher mobility, localize on surface oxygen,
under the same conditions; instead, a seven-line spectrum with
3
þ
9,41
4þ •-
•
forming (Ti
)
and (Ti O )
paramagnetic species,
hyperfine splitting of 0.36 mT, corresponding to CD radicals,
latt
surf
3
3
respectively.
trapped holes react with adsorbed molecules even at these
Due to their strong oxidation power, surface-
was detected (Figure 3). Due to the low sensitivity of EPR
•
signals, spectra of D and CD radicals were greatly masked by
3
42
extremely low temperatures. On the other hand, for electrons
to react with adsorbed molecules, an increase in temperature is
required, usually >30 K, because it allows for detrapping of
electrons from lattice to conduction band and their further
migration to the surface (the activation energy for detrapping
intense signals from holes and electrons on titania.
Methyl radicals, the last intermediate species in the overall
reduction process, were the only radicals observed in the EPR
spectra. Their concentration increases with a decrease in H O:
2
CO molar ratio and with the time of illumination. The yield of
2
4
3
•
from lattice shallow traps is ∼20 meV). Once photogenerated
CH radicals is directly correlated to the decrease in the concen-
3
electrons reach the surface of TiO , they can react with adsorbed
tration of electrons localized on TiO (Supporting Information).
2
2
molecules or localize on surface trapping sites (Scheme 1). The
exception arises when electron-accepting molecules make strong
complex with surface titanium ions, which allows molecules to
compete for conduction band electrons with their thermaliza-
tion/localization on lattice trapping sites. This was shown
previously for pyrroloquinoline quinone (PQQ), which forms
Multiple Roles of Water. In order to further reveal the role of
water in a complex photocatalytic process, we have examined
charge separation in TiO having partially and fully hydrated
2
surfaces in the absence of CO . A “dry” powder TiO P25, as was
2
2
used in these studies, is partially hydrated having 1.7-4.5 OH
2
46
groups/nm , while dispersing P25 in water results in fully
3
966
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
2 2
Figure 2. EPR spectrum recorded at 4.5 K after 1 h of illumination at 77 K. Light source, 300-W Xe lamp; sample, 50 mg of TiO , 2.8 mmol of H O, and
2
2.8 mmol of CO . Instrument conditions: power, 2.0 mW; modulation amplitude, 0.5 mT. For both the inset and detailed detection of H atom signals,
power was reduced to 0.2 mW and modulation amplitude to 0.2 mT.
Figure 3. EPR spectrum recorded at 4.5 K after 1 h of illumination at 77
K of 50 mg of TiO , 2.8 mmol of D O, and 2.8 mmol of CO . Instrument
2 2 2
conditions and light source as in Figure 2. Inset: experimental EPR
spectrum after baseline correction, recorded at 20 K, 6 mW power, and
0
.1 mT modulation amplitude, and simulated spectrum of CD radical.
3
hydrated surfaces. When a vacuum-degassed aqueous dispersion
of TiO was illuminated at 4.5 K, the signals from lattice-trapped
2
electrons in both anatase (g = 1.991) and rutile (g = 1.975) and
^
^
surface-trapped holes were observed (Figure 4). The oxygen-
centered radical, due to the trapping of holes on the surface of
titania, is characterized by g = 2.002, g = 2.017, and g = 2.026 in
Figure 4. EPR spectra recorded under illumination at 4.5 K of degassed
“dry” TiO (black line) and aqueous TiO (red line). Power, 0.2 mW;
modulation amplitude, 0.5 mT; light source, 355 nm photons from a Nd:
YAG laser (power 9 mJ).
2
2
x
y
z
the aqueous dispersion and can be assigned according to the
4
þ
•-
39,41,47
previous studies to a (Ti O ) paramagnetic species.
surf
At the same time, for partially hydrated “dry” TiO , a slight shift
of OH radicals formed upon excitation of TiO in an aqueous
2
2
(
g = 2.023) and an increase in intensity of the g component can
dispersion. The formation of OH radicals from water acting as an
electron donor is well established (eq 5), and an EPR spin-trap
z
z
48
be observed. The differences suggest that, for partially hydrated
surfaces, holes localize on lattice oxygen atoms located in the
subsurface layer with a structure of [Ti O Ti OH ], as
technique is commonly used for their detection (see further
text). If any, the contribution of OH radicals to the overall EPR
4
þ
•- 4þ
-
41
49
the magnitude of the g component is largely controlled by the
spectrum presented in Figure 4 is extremely low, as no changes in
z
local electric field gradient.
the values or intensities of g and g components could be obser-
x
y
Another possibility to explain the differences between EPR
spectra of partially and fully hydrated surfaces is the contribution
ved. It was demonstrated that OH radicals on the surface of metal
oxides not only lead to the broadening and decrease in intensity
3
967
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
signals associated with holes/oxidation products differ from
those in water. We propose that, for dissolved CO in aqueous
2
TiO dispersions, where some of the CO is transformed into
2
2
carbonic acid, the observed changes in the EPR spectrum are due
to the contribution of a signal associated with the formation of
-
CO3 anions. The carbonate/bicarbonate ions act as hole scaven-
gers, as shown in eq 9a.
hVB þ CO3 f CO3-
þ
2-
ð9aÞ
ð9bÞ
hVB þ HCO3 f CO - þ Hþ
þ
-
3
As a proof of the role of carbonates, the spectrum of
illuminated, degassed, aqueous 0.5 M Na CO /TiO is pre-
2
3
2
-
sented in Figure 5, showing formation of CO3 radical anions.
The principal g-tensor values, g = 2.006, g = 2.012, and g =
x
y
z
2
.019, observed for the Na CO /TiO reaction mixture are in
2 3 2
-
agreement with literature values for orthorhombic CO radical
anions in proximity of K , Na , or Ca cations and suggest
binding/adsorption of CO3 to Ti
CO3 results in suppressed charge recombination, seen as an
increased intensity of signals associated with electrons localized in
anatase and rutile lattices, as charges become spatially separated.
The experimental EPR spectrum of TiO /H O/CO is over-
3
þ
þ
2þ
4þ/3þ
51
-
ions. The formation of
-
2
2
2
lapped and superimposed with spectra corresponding to holes on
-
titania and to CO3 radical anions. The simulated spectrum
Figure 5. EPR spectra recorded under illumination at 4.5 K of 10 mg of
presented in Figure 5 is a sum of the spectra of these two
TiO in the presence of (a) 5 mmol of CO and 2 mmol of H O (blue
2
2
2
4þ
•-
paramagnetic species (with 60% contribution from (Ti O )
and 40% from CO3 radicals).
line), (b) 2 mmol of H
2
O (red line), and (c) 0.5 M aqueous Na
2
CO
3
surf
-
(black line). The simulated spectrum is presented as a gray line. Power,
0
.2 mW; modulation amplitude, 0.5 mT. Inset: part of the EPR spectra
When temperature was increased to 50 K, lower yields of
lattice- and surface-trapped electrons for the TiO /H O/CO
2
that corresponds to photogenerated electrons of the same samples
illuminated and measured at 50 K (power, 2.0 mW). The light source
was 355 nm photons from a Nd:YAG laser (power, 9 mJ).
2
2
reaction mixture as compared with TiO /H O were observed
2
2
(
Figure 5, inset). When the temperature is increased, the
contribution of reaction of photogenerated electrons with ad-
-
of the g component compared to O radicals but also exhibit
sorbed CO molecules on titania increases and competes with
z
2
50
different values of all three g-components.
More importantly, the presence of bulk water and/or a fully
hydrated surface enables better charge separation. We have
the charge recombination process, as discussed previously and
presented in Scheme 1. A slight decrease in the concentration
of photogenerated electrons was also observed for a 0.5 M
Na CO /TiO mixture as compared to TiO /H O; however,
15
found that the number of spins increases from 0.8 ꢂ 10
2
3
2
2
2
3
15
3
-
spins/cm for “dry” TiO to 1.0 ꢂ 10 spins/cm for aqueous
the mechanism is not clear. Although reduction of HCO3 to
2
TiO dispersion. As can be seen from Figure 4, the intensities of
formate over Pd/TiO in the presence of the sacrificial hole
2
2
52
signals associated with trapped electrons in rutile and anatase are
noticeably higher for fully hydrated titania. For better visualiza-
tion, the signals corresponding to holes/oxidation products are
normalized for different spectra presented in Figures 4 and 5, and
lines are drawn to indicate relative intensities of photogenerated
electrons. As the EPR spectra were recorded under illumination,
the intensities of the signals correspond to the steady-state
concentrations of products, in our case ruled by the formation
and recombination of electrons and holes and products of their
subsequent reactions.
scavenger oxalate was proposed previously, we cannot rule out
a contribution of the reaction of adsorbed CO3 anions with
-
photogenerated electrons at elevated temperatures under steady-
-
state conditions, CO3 being a strong, one-electron oxidation agent.
EPR Spin Trapping. To further examine the role of water and
carbonates, and to identify possible radicals formed upon illumi-
nation, the spin-trapping technique was applied to titania-CO
2
systems. Figure 6 presents EPR spectra recorded after 1 min of
illumination of titania dispersions containing DMPO as a spin
trap, all containing CO . The formation of DMPO-OH (a =
2
N
In the presence of bulk water, charges are stabilized either due
to the dipolar interaction with water or due to the band bending
at the TiO /water interface, which results in suppressed charge
a = 1.48 mT) and DMPO-CH adducts (a = 1.57 mT, a =
H
3
N
H
2.085 mT) is observed both in the absence (Figure 6a) and in the
presence (Figure 6b) of sodium carbonate. For these measure-
ments, the pH of water was adjusted to 11 to match that of the
sodium carbonate solution. The band edges change with pH
2
recombination.
Role of Carbonates. When CO was added to an aqueous
2
TiO dispersion, further changes in the EPR spectra of photo-
generated charges were observed upon illumination at 4.5 K
according to the Nernst equation (ECB = -0.74 V, and E
=
2
VB
þ2.46 V at pH 11) and could affect charge transfer to adsorbed
molecules, and thus the yield of products. A higher concentration
of methyl radicals is formed in the presence of 0.5 M Na CO , as
(
Figure 5). First, an even higher concentration of spins was
observed compared to that in pure water, revealing better charge
2
3
separation, and second, the line shape and g-tensor values of
compared to the concentration of OH radicals, proving that
3
968
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
Scheme 2. Proposed Mechanism of Photocatalytic Trans-
formation of CO to Methoxyl Radical over TiO in the
2
2
Presence of Dissociated/Bound Water
53
ues. The generation of any type of hydrocarbon radical from “dry”
TiO P25 suggests that H atoms are formed from dissociated/
2
submonolayer water rather than from bulk water. Bulk water
leads to detachment of radicals from the surface, and the reaction
of methoxyl radical with H O can lead to formation of methanol.
2
Under our experimental conditions (the maximum concentra-
tion of DMPO used was 0.1 M), we did not observe DMPO-H
-
or DMPO-CO adducts, which suggests a fast rate of hydro-
2
carbon formation on the surface of TiO at room temperature.
2
Initial Steps in CO Reduction. The direct detection of H
2
atoms and CH radicals present simultaneously upon illumina-
3
tion of P25 with CO /H O (1:1 molar ratio) suggests compe-
2
2
titive electron transfer to adsorbed/bound carbon dioxide and
adsorbed/bound protons on the surface of TiO (Scheme 2).
2
The initial electron transfer is accompanied by breaking of the
double OdCdO bond and attachment of a H atom, resulting in
the formation of formate. Consecutive electron/proton transfer
leads to formation of methoxyl radical.
The initial step in the proposed mechanism (competitive
electron transfer) thus corresponds to two-electron, one-proton
transfer, as shown in eq 11.
CO þ 2e þ H f HCOO-
-
þ
ð11Þ
2
In order to estimate the energetics of the simultaneous two-
electron process (eq 11) on the surface of TiO , theoretical
2
modeling in the gas phase has been carried out. For this purpose
we have considered the case where a reactant hydrogen is initially
adsorbed on the two-fold bridging oxygen site and its electron is
Figure 6. EPR spectra recorded at room temperature after 1 min of
illumination of TiO -CO and DMPO in (a) H O, pH 11, (b) 0.5 M
redistributed to the conduction band of TiO . It has been shown
2
2
2
2
that the energies from localized and delocalized states of an extra
aqueous Na
amplitude, 0.1 mT; light source, 300-W Xe lamp. A capillary tube was
filled with a solution of TiO (20 mg/mL) and DMPO (60 mM) and
placed in a vacuum, 4 mm diameter Suprasil tube. After degassing at 77
K, 6.2 mmol of CO was introduced into the tube, and the whole setup
2 3
CO , and (c) acetonitrile. Power, 33.2 mW; modulation
5
4
electron on anatase (101) surface are nearly degenerate. The
extent of localization of the initial unpaired electron, although a
2
55
56
topic of interest, has little effect on the energies of surfaces.
Hence, the initial spatial distribution of the electron has little
effect on the adsorption and reaction energies of the surfaces with
an adsorbate in the case of charge transfer. The calculated
reaction pathway is plotted in Figure 7.
2
was brought to room temperature.
carbonate ions act as electron donors, competing with water for
photogenerated holes. Some of the OH radicals can be formed
not only from direct reaction of valence band holes with water
We have chosen the sum of energies of an isolated CO
2
þ
molecule and two protons (2H ) adsorbed on the negatively
(
eq 5) but also from reaction of H O with surface-trapped
-
2
2
charged (2e ) anatase (101) surface as the energy reference (see
3
þ
electrons, (Ti )surf, a Fenton-type reaction (eq 10). Due to the
increased local concentration of OH radicals on the surface of
TiO , the yield of H O (formed from combination reaction of
34
Figure 7). As reported previously, CO prefers a linear vertical
2
adsorption (labeled as A1 configuration) at the five-fold Ti site
on the anatase (101) surface. It binds to the surface relatively
weakly, with a binding energy of 0.21 eV, which is not affected by
the presence of one adsorbed hydrogen in its vicinity
2
2 2
two OH radicals) is higher than in homogeneous systems.
3
þ
-
4þ
H2O2 þ ðTi Þ f OH þ OH þ ðTi Þ
ð10Þ
surf
surf
þ
-
[
CO (A1) H (a) þ e ]. Co-adsorption of another proton
2
3 3 3
þ
-
When acetonitrile was used as a dispersant, allowing for
nearby [CO (A1) 2H (a) þ 2e ] results in a metastable
2
3 3 3
illumination of partially hydrated TiO , the formation of an
state, which is unfavorable energetically by 0.26 eV relative to the
2
•
þ
oxygen-centered methoxyl radical, OCH , was detected
proton farther away. In the presence of two H adsorbed on the
3
(
1
Figure 6c). The signal with hyperfine splitting constants a =
.35 mT, aH = 0.775 mT, and aH = 0.17 mT is attributed to a
surface next to a CO molecule and two extra electrons in the
conduction band (representing photoexcited electrons), CO2
can be activated and converted to formate, involving a concerted
N
2
β
γ
DMPO-OCH adduct, in accordance with the literature val-
3
3
969
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
ARTICLE
PBE calculations. This material is available free of charge via the
Internet at http://pubs.acs.org.
’
AUTHOR INFORMATION
Corresponding Author
dimitrijevic@anl.gov
’
ACKNOWLEDGMENT
This work was performed under the auspices of the U.S.
Department of Energy under Contract DE-AC02-06CH11357.
Use of computational resources at Argonne National Laboratory
Computing Resource Center is gratefully acknowledged.
Figure 7. First-principles calculations of the reaction pathway from
CO and hydrogen to formate on the anatase (101) surface. Important
2
bond lengths (Å) and angles (degrees) are marked in all geometries with
atom colors showing Ti in gray, O in red, C in blue, and H in white. The
species along the reaction pathway are labeled at the bottom, where “g”
denotes gas phase and “a” denotes adsorbed species on the anatase (101)
’ REFERENCES
(1) Inoue, T.; Fujishima, S.; Konishi, S.; Honda, K. Nature 1979, 277, 637.
(2) Usubharatana, P.; McMartin, D.; Veawab, A.; Tontiwachwuthikul,
P. Ind. Eng. Chem. Res. 2006, 45, 2558.
3) Kitano, M.; Matsuoka, M.; Ueshima, M.; Anpo, M. Appl. Catal., A:
Gen. 2007, 325, 1.
2
surface; A1 is a linear vertical adsorption configuration of CO on the
(
surface. The “þ” sign indicates non-interacting species, while “...”
indicates two species in proximity.
(4) Centi, G.; Perathoner, S. Catal. Today 2009, 148, 191.
(
5) (a) Anpo, M.; Yamashita, H.; Ichihashi, Y.; Ehara, S. J. Electro-
two-electron, one-proton transfer from the TiO substrate. The
2
anal. Chem. 1995, 396, 21. (b) Anpo, M.; Takeuchi, T. J. Catal. 2003,
16, 505. (c) Anpo, M.; Thomas, J. M. Chem. Commun. 2006, 3273.
6) Chen, L.; Graham, M. E.; Li, G.; Gentner, D. R.; Dimitrijevic,
N. M.; Gray, K. A. Thin Solid Films 2009, 517, 5641.
7) Nguyen, T. V.; Wu, J. C. S.; Chiou, C. H. Catal. Commun. 2008,
9, 2073.
(8) Ozcan, O.; Yukruk, F.; Akkaya, E. U.; Uner, D. Top. Catal. 2007,
44, 523.
transition state (TS) for this process in Figure 7 is represented by
2
-
CO2 (one electron transferred already) strongly bound to the
(
five-fold Ti site with a Ti-O distance of 2.05 Å and bent
(
—OCO ∼140ꢀ) toward a proton adsorbed on the surface,
(
forming a hydrogen bond 2.1 Å in length. The effective barrier for
the formation of formate via a two-electron, one-proton process
is calculated to be 0.82 eV, a value lower than the activation
(
9) Woolerton, T. W.; Sheard, S.; Reisner, E.; Pirece, E.; Ragsdale,
energy for one-electron transfer to CO alone. The relatively
2
S. W.; Armstrong, F. A. J. Am. Chem. Soc. 2010, 132, 2132.
(
7, 37.
(
high activation barrier of 0.82 eV would account for the relatively
low efficiency determined experimentally at room temperature,
together with the complex nature of the processes on the surface
10) Tseng, I. H.; Chang, W. C.; Wu, J. C. S. Appl. Catal., B 2002,
3
11) Wu, J. C. S. Catal. Surv. Asia 2009, 13, 30.
12) Wang, C.; Thompson, R. L.; Baltrus, J.; Matranga, C. J. Phys.
of TiO , considered here only for water and carbonates.
2
(
Chem. Lett. 2010, 1, 48.
13) Kaneco, S.; Shimizu, Y.; Ohta, K.; Mizuno, T. J. Photochem.
Photobiol. A 1998, 115, 223.
14) Tseng, I. H.; Chang, W. C.; Wu, J. C. S. Appl. Catal. B: Environ.
002, 37, 37.
15) Belapurkar, A. D.; Kishore, K. J. Photochem. Photobiol. A 2004,
63, 503.
’
CONCLUSION
(
Water, both dissociated on the surface of TiO and in
2
(
subsequent molecular layers, has a three-fold role: (i) stabilization
of charges (preventing electron-hole recombination), (ii) as an
electron donor (reaction of water with photogenerated holes to
give OH radicals), and (iii) as an electron acceptor (formation of H
atoms in a reaction of photogenerated electrons with protons on
the surface, -OH ). Dissolved CO in the form of carbonates/
bicarbonates competes with water for photogenerated holes and
thus can act as a hole scavenger. CO3 radicals are relatively strong
one-electron oxidation agents. Finally, observation of such reaction
intermediates as H atoms, OCH radicals, and CH radicals, and
first-principles calculations, suggest a concerted two-electron, one-
proton transfer to adsorbed carbon dioxide molecules, demonstrat-
ing the importance of hydrogen/water management on the
photocatalyst surface.
2
(
1
(
(
16) Dey, G. R.; Pushpa, K. K. Res. Chem. Intermed. 2007, 33, 631.
17) (a) Duonghong, D.; Ramsden, J. M.; Gratzel, M. J. Am. Chem.
þ
2
2
Soc. 1982, 104, 2977. (b) Dimitrijevic, N. M.; Savic, D.; Micic, O. I.;
Nozik, A. J. J. Phys. Chem. 1984, 88, 4278.
(18) Centi, G.; Perathoner, S.; Wine, G.; Gangeria, M. Green Chem.
2007, 9, 671 and references cited therein.
(19) Lunsford, J. H.; Jayne, J. P. J. Phys. Chem. 1965, 69, 2182.
-
•
•
3
3
(
20) (a) Tanaka, T.; Kohno, Y.; S. Yoshida, S. Res. Chem. Intermed.
000, 26, 93. (b) Teramura, K.; Tanaka, T.; Ishikawa, H.; Kohno, Y.;
Funabiki, T. J. Phys. Chem. B 2004, 108, 346.
21) (a) Chiesa, M.; Giamello, E. Chem.—Eur. J. 2007, 13, 1261. (b)
Preda, G.; Pacchioni, G.; Chiesa, M.; Giamello, E. J. Phys. Chem. C 2008,
12, 19568.
2
(
1
’
ASSOCIATED CONTENT
(22) J. Rasko, J.; Solymosi, F. J. Phys. Chem. 1994, 98, 7147.
(
23) Yoneyama, H. Catal. Today 1997, 39, 169.
S
Supporting Information. Determination of quantum
b
(24) Hemminger, J. C.; Carr, R.; Somorjai, G. A. Chem. Phys. Lett.
1978, 57, 100.
yield for CH formation; evolution of EPR spectra with time of
4
illumination and CO concentration, together with EPR spectra
of reference samples (blank experiments), for both direct and
spin-trap EPR measurements; electron density distribution from
(25) Green, J.; Carter, E.; Murphy, D. M. Chem. Phys. Lett. 2009,
477, 340.
(26) Henderson, M A. Langmuir 1996, 12, 5093.
2
3
970
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971

Journal of the American Chemical Society
27) Ketteler, G.; Yamamoto, S.; Bluhm, H.; Andersson, K.; Starr,
ARTICLE
(
D. E.; Ogletree, D. F.; Ogasawara, H.; Nilsson, A.; Salmeron, M. J. Phys.
Chem. C 2007, 111, 8278.
(
28) He, Y.; Tilocca, A.; Dulub, O.; Selloni, A.; Diebold, U. Nat.
Mater. 2009, 8, 858.
29) Levchenko, A. A.; Li, G.; Boerio-Goates, J.; Woodfield, B. F.;
Navrotsky, A. Chem. Mater. 2006, 18, 6324.
30) Vijayan, B.; Dimitrijevic, N. M.; Rajh, T.; Gray, K. A. J. Phys.
Chem. C 2010, 114, 12994.
31) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996,
7, 3865.
(
(
(
7
(
(
32) Kresse, G.; Furthmuller, J. Phys. Rev. B 1996, 54, 11169.
33) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. Rev. B 2001,
63, 155409.
(
34) He, H.; Zapol, P.; Curtiss, L. J. Phys. Chem. C 2010, 114, 21474.
(35) Sanville, E.; Kenny, S. D.; Smith, R.; Henkelman, G. J. Comput.
Chem. 2007, 28, 899.
(
36) Henkelman, G.; Jonsson, H. J. Chem. Phys. 2000, 113, 9978.
(37) Parmon, V.; Emeline, A. V.; Serpone, N. Int. J. Photoenergy
2
002, 4, 91.
38) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer,
M. C. J. Phys. Chem. B 2003, 107, 4545.
39) (a) Berger, T.; Sterrer, M.; Diwald, O.; Knozinger, E.; Panayotov,
(
(
D.; Thompson, T. L.; Yates, J. T., Jr. J. Phys. Chem. B 2005, 109, 6061. (b)
Berger, T.; Sterrer, M.; Diwald, O.; Knozinger, E. Chem. Phys. Chem. 2005,
6, 2104.
(
(
40) Enright, B.; Fitzmaurice, D. J. Phys. Chem. 1996, 100, 1027.
41) (a) Kumar, C. P.; Gopal, N. O.; Wang, T.-C.; Wong, M.-S.; Ke,
S.-C. J. Phys. Chem. B 2006, 110, 5223. (b) Ke, S.-C.; Wang, T.-C.; Wong,
M.-S.; Gopal, N. O. J. Phys. Chem. B 2006, 110, 11628.
(42) See for example: Rajh, T.; Makarova, O. V.; Thurnauer, M. C.;
Cropek, D. In Synthesis, Functionalization and Surface Treatment of
Nanoparticles; Baraton, M.-I., Ed.; American Science Publishers: Valencia,
CA, 2003; pp 147-171.
(
43) Dimitrijevic, N. M.; Poluektov, O. G.; Saponjic, Z. V.; Rajh, T.
J. Phys. Chem. B 2006, 110, 25392.
44) Bielski, B.; Gebicki, J. M. Atlas of Electron Spin Resonance
Spectra; Academic Press: New York/London, 1967.
45) (a) Florin, R. E.; Brown, H. W.; Wall, L. A. J. Phys. Chem. 1962,
(
(
6
3
6, 2672. (b) Fessenden, R. W.; Schuler, R. H. J. Chem. Phys. 1963,
9, 2147.
(
46) (a) Rodriguez, R.; Blesa, M. A.; Regazzoni, A. E. J. Colloid Interface
Sci. 1996, 177, 122. (b) Erdem, B.; Hunsicker, R. A.; Simmons, G. W.;
Sudol, E. D.; Dimonie, V. L .; El-Aasser, M. S. Langmuir 2001, 17, 2664.
(
47) Dimitrijevic, N. M.; Saponjic, Z. V.; Rabatic, B. M.; Poluektov,
O. G.; Rajh, T. J. Phys. Chem. C 2007, 111, 14597.
48) See for example: (a) Hoffmann, M. R.; Martin, S. T.; Choi, W.;
(
Bahnemann, D. W. Chem. Rev. 1995, 95, 69. (b) Mills, A.; Le Hunte, S.
J. Photochem. Photobiol. A Chem. 1997, 108, 1.
(49) Jaeger, C. D.; Bard, A. J. J. Phys. Chem. 1979, 83, 3146.
(50) Giamello, E.; Calosso, L.; Fubini, B.; Geobaldo, F. J. Phys. Chem.
1
993, 97, 5735.
51) (a) Chantry, G. W.; Horsfield, A.; Morton, J. R.; Whiffen, D. H.
Mol. Phys. 1962, 5, 589. (b) Serway, R. A.; Marshal, S. A. J. Chem. Phys.
(
1
967, 47, 868. (c) Koksal, F.; Koseoglu, R. Radiat. Phys. Chem. 2000,
57, 59.
(
52) Goren, Z.; Willner, I.; Nelson, A. J.; Frank, A. J. J. Phys. Chem.
990, 94, 3784.
53) (a) Li, A. S. W.; Cummings, K. B.; Roethling, H. P.; Buettner,
1
(
G. R.; Chignell, C. F. J. Magn. Reson. 1988, 79, 140. (b) Izakovic, M.;
Sima, J.; Brezova, V. J. Photochem. Photobiol. A 2004, 167, 81.
(
54) Di Valentin, C.; Pacchioni, G.; Selloni, A. J. Phys. Chem. C 2009,
13, 20543.
55) Minato, T.; Sainoo, Y.; Kim, Y.; Kato, H. S.; Aika, K.; Kawai, M.;
1
(
Zhao, J.; Petek, H.; Huang, T.; He, W.; Wang, B.; Wang, Z.; Zhao, Y.;
Yang, J.; Hou, G J. Chem. Phys. 2009, 130, 124502.
(56) Deskins, N. A.; Rousseau, R.; Dupuis, M. J. Phys. Chem. C 2010,
114, 5891.
3
971
dx.doi.org/10.1021/ja108791u |J. Am. Chem. Soc. 2011, 133, 3964–3971