Identification and reactivity of a surface intermediate in the photoreduction of CO2 with H2 over ZrO2

Article abstract of DOI:10.1039/a801055b

Zirconium oxide is active for photoreduction of gaseous carbon dioxide to carbon monoxide with hydrogen. A stable surface species arises under the photoreduction of CO₂ on zirconium oxide and is identified as surface formate by IR spectroscopy. Adsorbed CO₂ is converted to formate by photoreaction with hydrogen. The surface formate is a true reaction intermediate since CO is formed by the photoreaction of formate and CO₂; surface formate works as a reductant of carbon dioxide to yield carbon monoxide. The dependence on the wavelength of irradiation light shows that bulk ZrO₂ is not the photoactive species. When ZrO₂ adsorbs CO₂ a new band appears in the photoluminescence excitation spectrum. The photoactive species in the reaction between CO₂ + H₂ which yields HCOO^- is presumably formed by the adsorption of CO₂ on the ZrO₂ surface.

Full text of DOI:10.1039/a801055b

View Article Online / Journal Homepage / Table of Contents for this issue
IdentiÐcation and reactivity of a surface intermediate in the
photoreduction of CO with H over ZrO
2
2
2
Yoshiumi Kohno, Tsunehiro Tanaka, Takuzo Funabiki and Satohiro Yoshida
Department of Molecular Engineering, Graduate School of Engineering, Kyoto University,
Kyoto 606-8501, Japan
Zirconium oxide is active for photoreduction of gaseous carbon dioxide to carbon monoxide with hydrogen. A stable surface
species arises under the photoreduction of CO on zirconium oxide and is identiÐed as surface formate by IR spectroscopy.
2
Adsorbed CO is converted to formate by photoreaction with hydrogen. The surface formate is a true reaction intermediate since
2
CO is formed by the photoreaction of formate and CO ; surface formate works as a reductant of carbon dioxide to yield carbon
monoxide. The dependence on the wavelength of irradiation light shows that bulk ZrO is not the photoactive species. When
ZrO adsorbs CO a new band appears in the photoluminescence excitation spectrum. The photoactive species in the reaction
between CO ] H which yields HCOO~ is presumably formed by the adsorption of CO on the ZrO surface.
2
2
2
2
2
2
2
2
It is of urgent necessity to decrease emissions of carbon
dioxide to protect the environment from the greenhouse e†ect.
Thus the development of methods to convert carbon dioxide
to more valuable compounds inexpensively is desirable. From
this standpoint, catalytic reduction of carbon dioxide with a
reductant such as hydrogen1,2 or methane3,4 has been studied.
However, carbon dioxide is a very stable and inert compound,
and CO2 reduction scarcely occurs at room temperature
under ambient pressure. On the other hand, irradiation of a
reaction was also examined. Since we expected that the Ðrst
step of the photoreaction was an interaction of an excited
species in a triplet state with a reactant, we tried to detect the
photoactive species by photoluminescence spectroscopy.
Experimental
Zirconium oxide used in this study was prepared from an
aqueous solution of zirconium oxychloride (Nacalai Tesque,
photoactive catalyst often enables reduction of CO at a sig-
niÐcant reaction rate under mild conditions of room tem-
GR) by precipitation with 25 mass% NH (aq). The precipitate
obtained was washed with distilled water followed by Ðl-
2
3
perature and low pressure because the energy of the light
assists the reaction. In fact, many studies have reported the
tration until the Ðltrate was negative towards AgNO . The
sample was dried at 373 K overnight, ground into a powder,
3
possibility of CO photoreduction over various heterogeneous
photocatalysts.5h16
and calcined at 773 K for 5 h in a dry air stream. The XRD
pattern indicated that the resulting powder was zirconium
oxide in a mixture of monoclinic and tetragonal phases.
Hydrogen was puriÐed through a liquid nitrogen trap.
Formic acid (Nacalai Tesque, GR), oxygen and carbon
dioxide were puriÐed by vacuum distillation at low tem-
perature. 13C-labeled carbon dioxide (99 atom% 13C) was
obtained from ICON and used without further puriÐcation.
IR spectra were recorded with a Perkin-Elmer PARAGON
1000 PC Fourier transform IR spectrometer. A zirconium
2
Among these reports, titanium oxide or metal loaded tita-
nium oxide systems have mainly been used as photo-
catalysts.5h9 On the other hand, zirconium oxide, which is
also one of the oxides of group 4A transition metals, has
rarely been employed as a photocatalyst. However, Sayama
and Arakawa recently reported that zirconium oxide showed
an excellent photocatalytic activity for photodecomposition of
water and photoreduction of aqueous carbonate.17,18 They
reported that zirconium oxide did not require any metal
oxide sample (ca. 80 mg) was pressed into
a disk
loading to exhibit photocatalytic activity, in contrast to TiO .
The observation of surface species to study the reaction
pathway would be easier without metal loading, because
loaded metals often complicate the surface structure. From
this viewpoint, zirconium oxide is expected to be useful for the
investigation of mechanisms of reactions on the surface.
(diameter \ 10 mm) at a pressure of 4 MPa and suspended
2
with platinum wire in the same cell as mentioned elsewhere.21
The cell allowed us to perform heating, O treatment, intro-
2
duction of substrates, photoirradiation and measurements of
spectra in situ. Before a measurement, the sample disk was
activated by evacuation at 673 K for 60 min, treatment with 8
We have previously reported the application of ZrO to
kPa O for 120 min and evacuation for 60 min at the same
2
2
photoreduction of gaseous carbon dioxide,19,20 and found
temperature. Introduction of reaction substrates (CO and
2
that ZrO has sufficient activity to reduce CO to CO selec-
H ) was carried out as follows. CO (3.3 kPa) was introduced
2
2
2
2
tively with hydrogen under photoirradiation at room tem-
perature.19 We have also found that heat treatment at 673 K
after photoreaction yields additional CO evolution, which
indicates the existence of stable surface species decomposed to
CO by heat. We expect that the surface species is a reaction
intermediate.
to the cell at room temperature followed by evacuation and
H (4.7 kPa) was then introduced. Formic acid was adsorbed
2
at 573 K followed by evacuation, since too large amounts of
formic acid were adsorbed at room temperature. A 500 W
ultrahigh-pressure mercury lamp (Ushio Denki USH-500D)
was used as the light source for photoirradiation of the disk.
For each spectrum, data from 200 scans were accumulated at
a resolution of 4 cm~1.
The photoreaction was carried out in a closed static system
connected to a vacuum line. A zirconium oxide sample (0.3 g)
was spread on the Ñat bottom of a quartz reactor. Before pho-
toreaction, the sample was heated at 673 K for 30 min in air
and evacuated for 30 min at the same temperature, followed
In this study, we attempted to detect and identify any
surface intermediate by IR spectroscopy. The detection and
identiÐcation of surface intermediates is important for the
clariÐcation of reaction mechanisms.
The results of reactions under various conditions allowed us
to discuss how the intermediate was formed, and how it gave
the Ðnal product. The nature of the photoactive species in the
J. Chem. Soc., Faraday T rans., 1998, 94(13), 1875È1880
1875

View Article Online
by treatment with 8 kPa O for 75 min and evacuation for 30
2
min at 673 K. Unless otherwise noted, a mixture of CO (150
2
lmol) and H (50 lmol) was admitted to the reactor, and the
2
total pressure in the reactor was ca. 25 kPa. A 500 W
ultrahigh-pressure mercury lamp as described above was used
as the light source, and the reactor was illuminated from the
bottom. The area subjected to illumination was 12 cm2. When
the short wavelength part of irradiation light was required to
be cut o†, a series of glass Ðlters (Toshiba UV-29, UV-31,
UV-33 and UV-37, which permitted 50% transmission of light
at wavelengths of 290, 310, 330 and 370 nm, respectively) was
placed between the mercury lamp and the reactor. After pho-
toreaction for 6 h, the gaseous products were analyzed, and
after 2 min evacuation at room temperature the sample was
heated at 673 K for 20 min, and the desorbed gases were also
analyzed. Analysis of products was performed with an on-line
TCD gas chromatograph (Shimadzu GC-8A) equipped with
Molecular Sieve 5A packed column using argon as carrier gas,
which could detect H , CH and CO. In temperature pro-
Fig. 1 IR spectra of ZrO (a) after pretreatment, (b) after intro-
2
duction of CO and H to (a), and (c) after photoirradiation of (b).
2
2
The inset is the di†erence spectrum between (a) and (b), indicating the
spectrum of adsorbate.
2
4
grammed desorption (TPD) experiments, the desorbed gases
were detected by a quadrupole-type mass spectrometer
bicarbonate (hydrogen carbonate) species according to pre-
vious studies.22h24 Two intense bands at 1562 and 1310 cm~1
were assigned to surface carbonate species. Bands at 1452 and
1224 cm~1 and a weak shoulder band at around 1610 cm~1
were assigned to bicarbonate species. A band at 1054 cm~1
was due to the superposition of these two species. These two
species were both formed by introduction of carbon dioxide
even in the dark, and they were not removed from the surface
by evacuation at room temperature since the absorption
bands did not disappear upon evacuation.
(
ULVAC Massmate-100). In order to avoid the interference of
nitrogen to the signal of mass number \ 28, 13C-labeled CO
was used to detect the desorption of carbon monoxide.
2
The photoluminescence spectra were recorded at room tem-
perature with a Hitachi F-3010 Ñuorescence spectrometer and
an in situ cell. Before a measurement, a zirconium oxide
sample placed in the reactor was heated at 673 K for 30 min
in air and evacuated for 30 min at the same temperature, fol-
lowed by treatment with 8 kPa O for 120 min and evac-
The introduction of hydrogen to the zirconium oxide pre-
2
uation for 60 min at 673 K. The e†ect of CO adsorption on
2
adsorbing CO led to almost no spectral change. In particu-
2
the photoluminescence was investigated by recording the
lar, no new peak at around 1560 cm~1 was observed. A band
spectra under the equilibrium adsorption of CO
temperature.
2 at room
at around 1560 cm~1 has been assigned to ZrwH species
formed by dissociative adsorption of hydrogen on ZrO , and
2
this species has been reported to be formed when hydrogen is
contacted with ZrO at room temperature.25h27 The absence
2
Results and Discussion
of ZrwH is a consequence of the variation of pretreatment
temperature. Zirconium oxide pretreated at 973 K25,27
adsorbs hydrogen dissociatively to exhibit the band at 1560
cm~1, whereas a sample pretreated at 700 K28 or at below
IdentiÐcation of the surface intermediate
In the previous study,19 we have demonstrated the existence
of surface species on zirconium oxide used for photoreduction
of carbon dioxide with hydrogen. Carbon dioxide is reduced
to carbon monoxide with hydrogen over zirconium oxide by
photoirradiation. When the sample is evacuated after the pho-
toreaction to remove any gaseous substrates or products, and
the evacuated sample is heated at 673 K, further CO evolution
is observed. This result shows the existence of some species
decomposed to CO by heat on the surface. This surface
species is not adsorbed CO produced during photoreaction
since the desorption temperature of photo-adsorbed CO is dif-
ferent from the observed decomposition temperature of the
species. From these results, we can conclude that a surface
species is formed on the surface of zirconium oxide during
photoreaction, and that the heat treatment at 673 K after
photoreaction decomposes this surface species and leads to
additional CO evolution. We expected that the surface species
is a reaction intermediate. In the following, we show the
results of detection and identiÐcation of surface species by IR
spectroscopy, and provide evidence that the surface species is
a reaction intermediate.
8
73 K29 does not. Since the ZrO used in this study was
2
pretreated at 673 K, it does not adsorb detectable amounts of
dissociated hydrogen. Dissociative adsorption of hydrogen
implies the activation of hydrogen molecules on the surface.
Therefore we conclude that hydrogen is scarcely activated on
ZrO at room temperature at least in the dark.
2
The spectral change caused by photoirradiation was so
small that it was hardly observed in Fig. 1. To detect any
noticeable change, di†erence spectra between before and after
photoirradiation were obtained. Fig. 2(a) and (b) show such
di†erence spectra in the ranges from 3000È2800 and 1800È
1000 cm~1, respectively. When the sample was kept in the
dark, no spectral change was observed between 3000 and 2800
cm~1 [thin curve in Fig. 2(a)]. This was also the case in the
region 1800È1000 cm~1 [thin curve in Fig. 2(b)]. On the other
hand, photoirradiation of the sample caused a signiÐcant
growth of two peaks at 2970 and 2886 cm~1, and the growth
was enhanced as the irradiation time was extended as is
shown by the thick curves in Fig. 2(a). These two bands may
be assigned to the surface species arising during photoirradia-
tion, since no reaction proceeds without photoirradiation.19
In the wavenumber region 1800È1000 cm~1, however, less
noticeable changes were observed since the strong absorption
bands of surface carbonate and bicarbonate overlapped [thick
curves in Fig. 2(b)]. As spectral information was insufficient to
identify the surface species, a method of removing adsorbed
carbonates without desorbing the surface species was
required.
Detection of surface species. Fig. 1 shows the IR spectra of
ZrO (a) before and (b) after introduction of CO and H , and
2 2 2
c) after photoirradiation for 21 h in CO and H . The inset of
(
2
2
Fig. 1 is the subtraction (b) [ (a), indicating the spectrum of
adsorbate.
Introduction of CO leads to the appearance of six bands in
2
the range 1800È1000 cm~1, as shown in the inset of Fig. 1. All
To remove surface carbonates only, we utilized the di†er-
ence of desorption temperature between adsorbed carbon
of these bands were assigned to bidentate surface carbonate or
1876
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
Fig. 4 Di†erence IR spectrum of the ZrO sample evacuated at 573
2
K and 673 K after photoirradiation in CO and H
2
2
to yield carbon monoxide at a temperature of 673 K (see Fig.
). From these results, we conclude that these peaks were
3
derived from the surface species. The surface species were
found to be formed during photoirradiation on ZrO2 by the
reaction between hydrogen and adsorbed carbon dioxide.
IdentiÐcation of the surface species. We searched for an
adsorbate giving a spectrum close to that of the surface
species, and formic acid was found to reveal a similar spec-
trum when adsorbed on ZrO . The spectrum of adsorbed
2
formic acid is shown in Fig. 5. Absorption bands were
observed at 2968, 2882, 1568, 1382 and 1368 cm~1 with posi-
tions quite similar to those of bands due to the surface species
(2968, 2886, 1566, 1390 and 1372 cm~1). These bands were not
assigned to molecular formic acid but to surface formate
ion.24,30 Therefore, we have identiÐed the surface species as
surface formate. This conclusion is not unexpected considering
reports23,24 in which surface formate has been produced
during the thermal reaction of adsorbed carbon dioxide and
Fig. 2 Di†erence IR spectra of ZrO with CO and H before and
2
2
2
after photoirradiation. Fig. 2(a) and 2(b) show the spectrum in the
range 3000È2800 and from 1800È1000 cm~1, respectively. The thin
curve illustrates the spectral change when the sample is kept in the
dark for 4 h. Thick curves illustrate the change caused by photoirra-
diation for 7 and 23 h.
dioxide and the surface species. Fig. 3 shows the TPD proÐles
hydrogen over ZrO . The development of surface formate is
of (a) adsorbed CO and (b) CO formed by the decomposition
2
2
also observed on the Rh/TiO system in CO ] H reaction.1
of surface species on ZrO used for photoreduction of carbon
2
2
2
2
Consequently, the decomposition of surface formate results in
the formation of CO by heat treatment at 673 K after photo-
reaction of CO and H .
Two reactions can be presumed as routes to formate
decomposition; decomposition to carbon monoxide and
water, or decomposition to carbon dioxide and hydrogen. The
contribution of the latter route was negligible, since very small
dioxide with hydrogen. This indicates that almost all of
adsorbed carbon dioxide will be desorbed by evacuation at
573 K, whereas the surface species will not be decomposed to
2
2
CO completely until the evacuation temperature reaches 673
K. Accordingly, if the sample is evacuated at 573 K, only the
surface species will remain on the surface of ZrO . Therefore
2
the di†erence spectrum between the two samples evacuated at
amounts of H were observed by heat treatment after photo-
573 and at 673 K will give the absorption bands of the surface
2
reaction.19 Hence, the amount of CO evolution by heat treat-
species. Fig. 4 illustrates such a di†erence spectrum. Three
peaks positioned at 1566, 1390 and 1372 cm~1 were clearly
observed. The species giving these peaks is not the same as
that formed by heat treatment at 573 or 673 K, since these
peaks are also observed at almost the same position in Fig.
ment should be directly equal to the amount of surface
formate formed.
The absorption bands assigned to surface formate also
appeared upon photoirradiation even when only CO2
introduced and no hydrogen was contacted onto the ZrO
2
was
2
6
(b). These bands disappeared by continuous evacuation at
73 K. This is consistent with the result of the TPD experi-
surface. This result indicates that surface formate can be pro-
duced to some extent by a stoichiometric reaction between
ment, in which the surface species was completely decomposed
carbon dioxide and the ZrO surface. The reductant on the
2
Fig. 3 TPD proÐles of (a) adsorbed CO and (b) CO formed by the
2
decomposition of surface species on ZrO used for photoreduction of
2
Fig. 5 IR spectrum of formic acid adsorbed on ZrO . The adsorp-
2
carbon dioxide with hydrogen. The signal intensity is normalized.
tion was carried out at 573 K followed by evacuation.
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1877

View Article Online
surface may be a coordinatively unsaturated Zr3` site, the
existence of which has been suggested from an EPR experi-
ment.31 However, we have already conÐrmed that no CO is
evolved by photoirradiation without hydrogen.19 We further
conÐrmed in this study that the amount of CO evolution by
heat treatment was very small in the absence of hydrogen
the photoreaction, since it is a reductant of carbon dioxide.
Therefore we postulate that surface formate is oxidized to
surface carbonates (carbonate or bicarbonate) or gaseous
carbon dioxide. The carbonates may be re-reduced to formate
when hydrogen coexists in the reactor.
In conclusion, surface formate was found to be a reaction
intermediate. Photoreduction of carbon dioxide with hydro-
(
0.07 lmol) compared to that with hydrogen (3.3 lmol). As the
amount of CO evolution after heat treatment after photoreac-
tion reÑects the amount of surface formate, the amount of
surface formate formed without hydrogen is very small. From
this, we conclude that the surface formate formed without
hydrogen had little inÑuence upon photoreduction of carbon
gen over ZrO consisted of the following two steps: (i) forma-
tion of surface formate from carbon dioxide and hydrogen; (ii)
reduction of carbon dioxide to carbon monoxide by surface
formate. Both these steps are photoreactions.
2
dioxide over ZrO .
Mechanism of formate formation
2
We now examine detailed mechanisms of photoreduction of
carbon dioxide with hydrogen over ZrO . It has been report-
Reactions with formic acid. It has been veriÐed that surface
formate is formed during photoreaction of CO and H over
2
2
2
ed that CO and H are activated on ZrO under photoirra-
ZrO . However, it, as yet, cannot be concluded that the
2
2
2
2
diation to yield surface formate. In this study, we reveal
primarily the role of carbon dioxide in the photoreaction of
CO and H to yield surface formate.
surface formate is a reaction intermediate. Rather, it might be
a “spectatorÏ intermediate32 which does not react to give any
product. Therefore we examined whether the surface formate
reacted to give carbon monoxide. For this purpose, we carried
out reactions under a variety of conditions using formic acid
as a substrate. As mentioned above, formic acid is adsorbed
2
2
Dependence on the amount of introduced CO . We tried to
obtain information about the di†erence between the e†ects of
2
adsorbed and gaseous carbon dioxide upon reaction. For this
purpose, we should Ðrst measure the saturated amount of
adsorption of CO on ZrO . To investigate this, the following
on ZrO as surface formate. Therefore, if the surface formate
2
is a “trueÏ intermediate, the evolution of carbon monoxide will
be observed in the reaction with formic acid under appropri-
ate reaction conditions.
2
2
method was adopted. Known amounts of CO were intro-
2
duced on ZrO , and after sufficient time for adsorption at
2
The results of such reactions are summarized in Table 1.
When formic acid only was introduced on zirconium oxide,
either no reaction proceeded or no CO was evolved even
under photoirradiation. This was also the case when formic
acid and hydrogen were introduced and photoirradiation per-
formed. These results show that the carbon monoxide evolved
by photoreaction was not formed either by decomposition of
surface formate or by reduction of surface formate with
hydrogen. On the other hand, when carbon dioxide was intro-
duced with formic acid and photoirradiation was performed
room temperature, gaseous carbon dioxide was collected in a
liquid nitrogen trap. The di†erence between the introduced
amount and the collected amount gives the saturated amount
of adsorption. By this method, we determined that the amount
of adsorption saturation of CO
2 was 45 lmol for 0.3 g of
ZrO .
2
We then investigated the dependence of the reaction on the
amount of introduced CO . Fig. 6 shows the dependence of
2
the amount of evolved CO after heat treatment at 673 K after
photoreaction upon the amount of introduced CO . The
amount of introduced hydrogen was 50 lmol in each run. As
mentioned previously, the amount of CO evolved by heat
treatment indicates the amount of surface formate.
The amount of CO evolved by heat treatment revealed a
on ZrO , the reaction proceeded and a signiÐcant amount of
2
2
carbon monoxide (0.7 lmol) was evolved. However, when
photoirradiation was not provided, no CO evolution was
observed even when carbon dioxide and formic acid were
introduced. When 13CO was used for the reaction of formate
2
strong dependence on the amount of CO introduced when
and CO , evolution of 13CO was observed. From these
2
2
the amount of CO was far smaller than the adsorption satu-
results, we conclude that formate reacted with carbon dioxide
2
ration. This increase just stopped when the amount of CO
to convert CO to CO, and that this reaction is also a photo-
2
2
reached the adsorption saturation (ca. 50 lmol). This indicates
reaction, as is the formation of surface formate. The surface
that the formation of surface formate is closely connected with
the amount of adsorbed carbon dioxide. It has already been
established by IR spectroscopy that surface formate is produc-
ed from carbon dioxide adsorbed on ZrO . The result
obtained here clearly further supports this conclusion.
formate acts as a reductant of carbon dioxide to give carbon
monoxide. This is also supported by the result that the IR
bands assigned to surface formate decreased during photo-
reaction between surface formate and carbon dioxide. In
summary, the surface formate can be regarded as a “trueÏ reac-
tion intermediate.
2
It is not clear as to what the surface formate is converted to
upon photoreaction with carbon dioxide. However, the
surface formate must be oxidized by carbon dioxide during
Table 1 Results of reactions using formic acid as one of substrates
CO/lmol
substrate(s)
reaction condition
amount of evolved
HCOOHa
under photoirradiation
under photoirradiation
under photoirradiation
in the dark at 313 K
0.0
0.0
0.7
0.0
HCOOH ] H b
2
HCOOH ] CO c
2
HCOOH ] CO d
2
Reaction time \ 6 h; reaction temperature, room temperature unless
otherwise noted, ZrO sample, 0.3 g. Introduced amount of
2
substrate(s): a HCOOH (10 lmol), b HCOOH (5 lmol), H (50 lmol),
Fig. 6 Plots of the amount of evolved CO by heat treatment after
2
c HCOOH (5 lmol), CO (150 lmol), d HCOOH (5 lmol), CO
lmol).
2 (150
photoreaction vs. the amount of introduced CO . The amount of
2
2
introduced hydrogen was 50 lmol, and irradiation time was 6 h.
1878
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

View Article Online
Dependence on the wavelength of illumination light. The
dependence of CO evolution upon the wavelength of illumi-
because of its long lifetime, and the photoluminescence tech-
nique detects the emission from the triplet state selectively.
nation light was investigated. ZrO is a semiconductor with a
wide band gap (5.0 eV), and does not absorb light with wave-
Phosphorescent excitation spectra of ZrO are shown in
2
2
Fig. 8. Curve (a) shows the spectrum of bare ZrO after
2
pretreatment, and curve (b) the spectrum of ZrO adsorbing a
2
length [250 nm as conÐrmed by the measurement of the
UVÈVIS spectrum of ZrO used in this study. Therefore, if
saturated amount of carbon dioxide. The emission light was
2
ZrO is the photoactive species in this reaction, it is expected
monitored at 440 nm, because the maximum emission inten-
sity was obtained at around that wavelength. We normalized
all the spectra by calculating so as to let the maximum emis-
sion intensity be unity. Looking at the spectrum (a), we can
see that the maximum emission intensity was obtained at 270
nm, i.e. [250 nm. This phenomenon has already been report-
ed in the literature,33 and the emitting species has been assign-
2
that the reaction should scarcely proceed upon irradiation by
light with j [ 250 nm.
Photoreactions were carried out under illumination of light
with various wavelengths. Short wavelengths of the irradiation
light were cut o† by a series of glass Ðlters which permitted
light with wavelengths j [ 290, 310, 330 and 370 nm, and
results are shown in Fig. 7. The results were obviously not in
ed to the coordinatively unsaturated site of ZrO .
2
accord with the optical properties of ZrO . Indeed, while the
Comparing spectra of Fig. 8(b) with Fig. 8(a), we can see an
2
amount of CO evolution decreased obviously as the cut wave-
increase in emission intensity at above 300 nm by the intro-
length became longer, light with wavelength [290 nm could
promote the reaction and surface formate was produced even
by light with j [ 330 nm (recall that the amount of CO
formed by heat treatment reÑects the amount of surface
duction of carbon dioxide over ZrO . Fig. 9 shows the di†er-
ence spectra between before and after adsorption of given
2
amounts of carbon dioxide. Subtraction was carried out after
the intensity of the original spectra was normalized to give the
same value at the maximum point. This Ðgure more clearly
illustrates the increase in intensity above 300 nm caused by
the introduction of carbon dioxide. The increase in emission
formate). These results suggest that excitation of bulk ZrO is
2
not required for this reaction: that is, the photoactive species
is not simply ZrO , and other photoactive species are
2
expected to exist. It is also suggested that photoactive species
intensity increased as the amount of CO was increased.
2
of formate formation from CO ] H and CO formation from
However, the introduction of an oversaturating amount of
2
2
formate ] CO are di†erent, because the dependence on the
CO caused a slight decrease in emission intensity. In other
2
2
wavelength was di†erent between these two steps. The latter
words, gaseous carbon dioxide did not contribute to an
step is thought to require shorter wavelength light than the
former step.
increase in emission. This shows that the changes in the phos-
phorescence spectra by introduction of carbon dioxide are not
caused by gaseous CO2 but rather by CO2 adsorbed on the
Proposed photoactive species. Other photoactive species
than bulk ZrO are expected for both steps of the CO ] H
2
2
2
reaction. Of the two steps, the photoactive species of the Ðrst
step reaction (formate formation from CO and H ) was con-
2
2
sidered in this study. Since adsorbed carbon dioxide is deeply
related to the mechanism of formate formation, as described
above, it can be assumed that a new photoactive species is
formed by adsorbed carbon dioxide. If the assumption is
valid, some spectral changes are expected in the adsorption of
carbon dioxide in di†use reÑectance UV spectra. However, the
recorded spectra did not show any new peaks caused by the
formation of the new photoactive species, although special
attention was paid to the spectral changes caused by the
adsorption of carbon dioxide. This seems to be mainly
because of the low selectivity of UV spectra in detection of
photoactive species. Hence, photoluminescence spectra of
ZrO were recorded since a triplet excitation state is often
expected to be a photoactive species in photoreactions
Fig. 8 Phosphorescence excitation spectra of ZrO (a) after pretreat-
2
2
ment and (b) after adsorption of a saturated amount of CO . Emis-
2
sion at 440 nm was monitored. The intensity of the spectra was
normalized to give the same value at the maximum point.
Fig. 7 Relationship between the wavelength cut-o† of irradiation
light and the amount of evolved CO. The dark and light bars illus-
trate the amount of CO evolved during photoreaction and that of CO
formed by heat treatment after photoreaction, respectively.
Fig. 9 Di†erence phosphorescence excitation spectra of ZrO
2
between before and after adsorption of given amounts of CO . The
2
amount of CO is (a) 10 lmol, (b) 30 lmol, (c) 50 lmol and (d) 90
2
lmol, respectively.
J. Chem. Soc., Faraday T rans., 1998, V ol. 94
1879

View Article Online
9
0
1
B. J. Liu, T. Torimoto, H. Matsumoto and H. Yoneyama, J.
Photochem. Photobiol. A: Chem., 1997, 108, 187.
L. B. Khalil, N. S. Youse†, W. Rophael and M. M. Moawad, J.
Chem. T echnol. Biotechnol., 1992, 55, 391.
surface. Furthermore, considering the two results that the
formate formation reaction proceeds under irradiation of light
with wavelength j [ 300 nm, and that surface formate is gen-
1
1
erated by adsorbed CO , we can conclude that adsorbed
M. A. Malati, L. Attubato and K. Beaney, Sol. Energy Mater. Sol.
Cells, 1996, 40, 1.
2
carbon dioxide is the new photoactive species predicted in the
previous section.
12 M. Kanemoto, T. Shiragami, C. Pac and S. Yanagida, J. Phys.
Chem., 1992, 96, 3521.
3
Introduction of hydrogen onto ZrO pre-adsorbed with
2
1
S. Yamagata, M. Nishijo, N. Murao, S. Ohta and I. Mizoguchi,
Zeolites, 1995, 15, 490.
H. Fujiwara, H. Hosokawa, K. Murakoshi, Y. Wada, S. Yana-
gida, T. Okada and H. Kobayashi, J. Phys. Chem. B, 1997, 101,
8270.
carbon dioxide decreases the intensity of the new peak at
above 300 nm. This result indicates that hydrogen interacts
with adsorbed carbon dioxide.
1
4
We propose that adsorption of carbon dioxide results in the
formation of the new photoactive species that can be excited
by light with wavelength [300 nm. We predict that the pho-
toactive species is either carbonate or bicarbonate because
these species were observed by IR spectroscopy. However, as
yet, we have not identiÐed the actual species.
We have succeeded in identifying the “trueÏ reaction inter-
mediate as surface formate in the photoreduction of carbon
dioxide with hydrogen over zirconium oxide. Although the
role of hydrogen and the mechanism of its activation are now
under investigation, the reaction mechanism is suggested as
15 M. Kanemoto, H. Hosokawa, Y. Wada, K. Murakoshi, S. Yana-
gida, T. Sakata, H. Mori, M. Ishikawa and H. Kobayashi, J.
Chem. Soc., Faraday T rans., 1996, 92, 2401.
1
6
M. Anpo, H. Yamashita, Y. Ichihashi, Y. Fujii and M. Honda, J.
Phys. Chem. B, 1997, 101, 2632.
K. Sayama and H. Arakawa, J. Phys. Chem., 1993, 97, 531.
1
7
18 K. Sayama and H. Arakawa, J. Photochem. Photobiol. A: Chem.,
1996, 94, 67.
19
20
21
Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, Chem.
Commun., 1997, 841.
Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, Chem. L ett.,
1
997, 993.
follows: (i) carbon dioxide adsorbed on ZrO surface in the
S. Takenaka, T. Tanaka, T. Funabiki and S. Yoshida, J. Chem.
Soc., Faraday T rans., 1997, 93, 4151.
2
dark creates a new photoactive species, and the species is
excited by light even with j [ 300 nm; (ii) the excited pho-
toactive species reacts with hydrogen to produce surface
formate and (iii) the surface formate reduces carbon dioxide
under photoirradiation to yield carbon monoxide.
22 N. E. TretÏyakov, D. V. Pozdnyakov, O. M. Oranskaya and V. N.
Filimonov, Russ. J. Phys. Chem., 1970, 44, 596.
2
3
J. Kondo, H. Abe, Y. Sakata, K. Maruya, K. Domen and T.
Onishi, J. Chem. Soc., Faraday T rans. 1, 1988, 84, 511.
M.-Y. He and J. G. Ekerdt, J. Catal., 1984, 87, 381.
T. Onishi, H. Abe, K. Maruya and K. Domen, J. Chem. Soc.,
Chem. Commun., 1985, 617.
2
2
4
5
This work was partially supported by Grant-in-Aids from the
Ministry of Education, Science, Sports and Culture of Japan
26 J. Kondo, Y. Sakata, K. Domen, K. Maruya and T. Onishi, J.
Chem. Soc., Faraday T rans., 1990, 86, 397.
(
no. 09650902 and 08405052). We thank Dr. Sakae Takenaka
27
28
29
30
31
K.-H. Jacob, E. Kn oŽ zinger and S. Benfer, J. Chem. Soc., Faraday
at Tokyo Institute of Technology for discussion.
T rans., 1994, 90, 2969.
A. Trunschke, D. L. Hoang and H. Lieske, J. Chem. Soc.,
Faraday T rans., 1995, 91, 4441.
References
F. Ouyang, J. N. Kondo, K. C. Maruya and K. Domen, J. Chem.
Soc., Faraday T rans., 1996, 92, 4491.
1
2
F. Solymosi, A. Erdo
A. Erdohelyi, M. Pa
66.
Ž
helyi and T. Ba
 sztor and F. Solymosi, J. Catal., 1986, 98,
 nsa gi, J. Catal., 1981, 68, 371.
Ž
F. Ouyang, J. N. Kondo, K. Maruya and K. Domen, J. Chem.
Soc., Faraday T rans., 1997, 93, 169.
1
3
A. T. Ashcroft, A. K. Cheetham, M. L. H. Green and P. D. F.
Vernon, Nature (L ondon), 1991, 352, 225.
M. F. Mark and W. F. Maier, J. Catal., 1996, 164, 122.
O. Ishitani, C. Inoue, Y. Suzuki and T. Ibusuki, J. Photochem.
Photobiol. A: Chem., 1993, 72, 269.
C. Morterra, E. Giamello, L. Orio and M. Volante, J. Phys.
Chem., 1990, 94, 3111.
4
5
3
3
2
3
S. T. Oyama and W. Zhang, J. Am. Chem. Soc., 1996, 118, 7173.
S.-L. Moon, H. Yamashita and M. Anpo, in AcidÈBase Catalysis
II, ed. H. Hattori, M. Misono and Y. Ono, Kodansha, Tokyo,
6
7
K. Tennakone, Sol. Energy Mater., 1984, 10, 235.
F. Saladin, L. Forss and I. Kamber, J. Chem. Soc., Chem.
Commun., 1995, 1995, 533.
1
994, pp. 479.
8
K. R. Thampi, J. Kiwi and M. Gra
Ž tzel, Nature (L ondon), 1987,
327, 506.
Paper 8/01055B; Received 5th February, 1998
1880
J. Chem. Soc., Faraday T rans., 1998, V ol. 94