Mechanistic study of CO2 photoreduction in Ti silicalite molecular sieve by FT-IR spectroscopy

Article abstract of DOI:10.1021/jp001470i

Reduction of CO₂ by visible light through an endoergic reaction is one of the most attractive, yet-unrealized goals in solar energy-to-fuel conversion. Most approaches to CO₂ reduction using heterogeneous photochemistry involve the liquid-solid or gas-solid interface of semiconducting materials. A heterogeneous approach to CO₂ photoreduction uses microporous silicates containing Ti centers in the framework. In-situ FTIR monitoring of the Ti-O ligand-to-metal charge-transfer-induced photoreduction of gaseous CO₂ in Ti silicalite sieve utilizing methanol as donor showed formic acid as the primary 2-electron-transfer product. This established C-H bond formation during the initial steps of CO₂ photoreduction. Growth kinetics indicated that CO is formed by secondary pyrolysis of HCO₂H. The proposed mechanism involved photoinduced one-electron transfer to produce transient CHO₂ radical.

Full text of DOI:10.1021/jp001470i

7834
J. Phys. Chem. A 2000, 104, 7834-7839
Mechanistic Study of CO2 Photoreduction in Ti Silicalite Molecular Sieve by FT-IR
Spectroscopy
N. Ulagappan and H. Frei*
Physical Biosciences DiVision, Mailstop CalVin Laboratory, Lawrence Berkeley National Laboratory,
UniVersity of California, Berkeley, California 94720
ReceiVed: April 18, 2000; In Final Form: June 16, 2000
The initial products of the photoreduction of gaseous CO
2
in Ti silicalite molecular sieve using methanol as
electron donor have been monitored by in-situ FT-IR spectroscopy. Reaction was induced by 266 nm excitation
+
IV
-II
+III
-I
of the Ti -O f Ti -O ligand-to-metal charge-transfer transition of the framework center. HCO
CO, and HCO CH were the observed products. CO escaped instantaneously into the gas phase and was
recorded at high spectral resolution. The origin of the products was elucidated by infrared analysis of
2
H,
2
3
1
8
13
13
experiments with C O
of HCO H, while HCO
oxidation product of methanol. The key finding is that formic acid is the primary 2-electron reduction product
of CO at the LMCT-excited Ti centers. This implies that C-H bond formation occurs in the initial steps of
CO activation at the gas-micropore interface.
2
, CO
2
, and CH
3
OH. The results show that CO originates from secondary photolysis
2
2
CH emerges mainly from spontaneous Tishchenko reaction of CH
3
2
dO, the initial
2
2
Introduction
the mechanism of CO2 photoreduction in Ti framework sub-
stituted molecular sieves remains unknown as monitoring of
the reaction has thus far been conducted by gas-chromatographic
analysis of the desorbed final products (CH3OH, CH4). Knowl-
edge of elementary steps of this reaction at the gas-micropore
interface would furnish crucial insights for the selection of new
framework metals and porous structures that may afford
Reduction of CO2 by visible light through an endoergic
reaction is one of the most attractive, yet-unrealized goals in
solar energy-to-fuel conversion. Most approaches to CO2
reduction using heterogeneous photochemistry involve the
liquid-solid or gas-solid interface of semiconductor materials.
Examples are TiO2, SrTiO3, or SiC in single-crystal or powder
1
,2
activation of CO at longer photolysis wavelengths.
form. The bulk of the work thus far has been conducted on
aqueous suspensions of colloidal particles of TiO2, ZnS, ZnSe,
CdSe, etc., often surface-modified to enhance efficiency,
selectivity, or wavelength response.3 Reduction is typically
initiated by transfer of photogenerated conduction band electrons
to surface-adsorbed CO2 or carbonate. Excitation wavelengths
are limited to the UV region for those materials that are stable
under use (metal oxides). Visible-light-induced reduction of CO2
to CO or formic acid has been achieved with low band gap
semiconductor particles such as ZnSe, ZnS, or CdS, but these
materials require sacrificial reductants in order to suppress
irreversible oxidation of the material. Likewise, visible light-
activation of CO2 in the presence of transition-metal colloids
requires sacrificial organic donors.5
2
In this paper, we report an in-situ FT-IR spectroscopic study
of the photoreduction of CO in Ti silicalite (TS-1) molecular
2
,4
sieve by UV light using CH3OH as electron source. Methanol
was chosen because the 2-electron-transfer reaction of CO2 with
CH3OH involves endothermicities which, in principle, render
this transformation accessible to visible photons.
Experimental Section
TS-1 molecular sieve was prepared according to the promoter-
11,12
induced enhancement method reported by Kumar et al.
A
gel of molar composition SiO2:0.02 TiO2:0.4 TPAOH:0.067 H3-
A second, very recent heterogeneous approach to CO2
photoreduction employs microporous silicates containing Ti
centers in the framework (TS-1, TiMCM-41 and -48).
i
PO4:0.75 PrOH:25 H2O was prepared using tetraethyl ortho-
silicate (Aldrich, 98%), tetrabutyl orthotitanate (Fluka), and
tetrapropylammonium hydroxide (TPAOH, Fluka), phosphoric
6
,7
Reactions are conducted in aqueous suspensions or on molecular
sieve loaded with gaseous CO2 and H2O. Efficiencies of CO2
reduction to methanol or methane in these micro- or mesoporous
materials by UV light are found to be substantially enhanced
compared to semiconductor TiO2 particles. This effect, already
noted in the case of dispersed Ti oxide species anchored on
i
acid (87%, Fluka), and 2-propanol ( PrOH) as respective source
materials. The gel was autoclaved at 433 K for 8-12 h.
Calcination was conducted at 723 K for 10 h under flow of
oxygen. ICP-AES combined with chemical analysis indicated
a Ti/Si ratio of 2.6 ( 0.2%. Characterization by powder XRD
(Cu KR) was conducted in the range 5° < 2θ < 40° at 0.05°/s.
porous glass8 or occluded in zeolite cages, is attributed to
the presence of isolated, tetrahedrally coordinated Ti centers.
Lifetimes of the excited Ti-O charge transfer state of such
framework centers were found to be several orders of magnitude
,9
10
The resulting pattern matches exactly a simulated MFI pattern.
No additional peak was found, which confirms that the material
1
2,13
is TS-1 with no other phase present.
phase could be discerned from SEM pictures. The silicalite (S-
) sieve used in this study was prepared by the same procedure
as TS-1 without the titanium source.
Also, no additional
6
longer than in the case of dense phase TiO2 particles. However,
1
*
Author to whom correspondence should be addressed.
1
0.1021/jp001470i CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/29/2000

CO2 Photoreduction in TS-1 Sieve
J. Phys. Chem. A, Vol. 104, No. 33, 2000 7835
Figure 1. XANES spectra of TS-1 molecular sieve (a) and of anatase
Figure 2. Diffuse reflectance spectra of pressed wafers of TS-1
molecular sieve, neat anatase, and a mixture of anatase (2%) and TS-1
crystallites.
(b). Samples were prepared in the form of pressed wafers. Energy is
referenced to the inflection point of the K-edge.
Ti K-edge XANES spectra were recorded at beamline 9.3.1
of the Advanced Light Source at LBNL. The beamline is
equipped with a Si (111) double crystal monochromator, and
ring currents were in the 200-400 mA range (1.5 GeV).
Measurements were conducted in transmission mode using a
Si photodiode detector, Hamamatsu model S2744-08. For energy
calibration, a 5 µ Ti foil was used, and the beam intensity was
monitored by the current generated at an Al foil. Self-supporting
pellets of about 30 mg and 12 mm diameter were used. Samples
were evacuated at 2 × 10 Torr, but no heating for removal
of residual water was performed. Spectra were recorded in the
range 4900-5100 eV at 0.5 eV sampling intervals and 3 s
sampling time per data point. Figure 1 shows XANES spectra
of TS-1 (trace a) and anatase (trace b). The sharp preedge peak
of the TS-1 spectrum (fwhm ) 2.7 eV) is characteristic of Ti
in tetrahedral substitution environment of the silicalite frame-
Photochemical experiments were conducted using equipment
2
0
and procedures described previously. In brief, about 22 mg
of TS-1 powder was pressed into a self-supporting wafer (3
5
min under a pressure of 2 × 10 kPa). The TS-1 wafer was
held in a miniature infrared vacuum cell, equipped with CaF2
windows, which was mounted inside an Oxford cryostat model
Optistat. Dehydration of the sieve was accomplished by heating
to 473 K under high vacuum (temperature controller Oxford
model ITC-502). Reactants were loaded into the molecular sieve
from the gas phase through a vacuum manifold. Chemistry was
monitored in situ by infrared spectroscopy, using an IBM-Bruker
-
7
-
1
-1
spectrometer model IR-97 at 1 cm and 0.1 cm resolution.
Photolysis was typically conducted at room temperature using
the 266 nm emission of a pulsed Nd:YAG laser at 10 Hz (model
Quanta-Ray DCR-2A with GCR3 upgrade).
Methanol (EM Science, 99%), 13CH3OH (Aldrich, 95% 13C),
CD3OD (Aldrich, 99.5% D), formic acid (Aldrich, 95%), and
work (A1-T2 transition). Trace a agrees with the spectrum of
pure TS-1 samples reported in the literature.1
4,15
By contrast,
1
3
13
H CO2H (Aldrich, 99% C) were degassed by vacuum
distillation. Paraformaldehyde (Aldrich, 95%) was depolymer-
the XANES spectrum of anatase shows three preedge peaks
corresponding to Ti in octahedral environment.15 No such
21
ized and purified as described in the literature. CO2 (Matheson,
9.995%), CO2 (ISOTEC, Inc., 99.1% C, 99.97% O), C O2
Prochem, 99.4% O), CO (Matheson, 99.99%), CO (ISO-
TEC, Inc., 99% C), and C O (ISOTEC, Inc., 95% O) were
used as received.
features are observed in the TS-1 sample.
13
13
16
18
9
(
A UV-vis diffuse reflectance spectrum of calcined TS-1 in
ambient air is displayed in Figure 2. It was recorded on a
Shimadzu model UV-2100 spectrometer equipped with an
integrating sphere model ISR-260. For this measurement, the
sample was prepared in the form of a pressed wafer, and BaSO4
powder used as reference. The spectrum agrees with that of TS-1
1
8
13
1
3
18
18
Results
1
4,17
materials reported in the literature
and is assigned to the
1. Infrared Spectrum of CO2 Adsorbed on TS-1 Sieve.
Loading of CO2 gas into the IR cell containing TS-1 sieve gave
rise to infrared absorptions at 1381, 1663, 2276 (natural
+IV
-II
+III
-I
Ti -O f Ti -O ligand-to-metal charge-transfer transi-
tion of tetrahedrally coordinated Ti (perfect or distorted).
1
8,19
1
3
-1
There is no indication of extraframework Ti-containing species,
which are known to give rise to absorption at wavelengths longer
than 330 nm.14,17 Specifically, we do not observe any shoulder
at 370 nm, indicative of anatase impurity. Spectra of neat anatase
and of anatase (2%) mixed with TS-1 are shown in Figure 2
for comparison. On the basis of comparison of our XANES
abundance of CO2), 2346, 3597, 3629, and 3704 cm . For
1
8
C O2, corresponding bands were at 1334, 1633, 2310, 3515,
3534, 3625, and 3656 cm , and for CO2 at 1366, 1619, 2280,
3511, 3541, 3625, and 3646 cm . Comparison with literature
spectra shows that the intense asymmetric stretch at 2346 cm
( C O2) and the combination tones in the 3500-3700 cm
region originate from both gas-phase and adsorbed CO2 (the
volume of the infrared vacuum cell is 3.4 cm , with a path length
-1
13
-
1
22
-1
-1
1
2 16
1
4,17
and UV-vis spectra with data reported in the literature,
we
3
conclude that the TS-1 sample contains no extraframework
species.
of 3 cm). The same observation was made for the corresponding

7836 J. Phys. Chem. A, Vol. 104, No. 33, 2000
Ulagappan and Frei
Figure 4. Infrared difference spectra upon 4 h irradiation at 266 nm
Figure 3. Infrared difference spectra upon loading of 20 Torr of CO
2
-
2
1
8
(30 mW cm ) of a mixture of CH
3 2
OH (5 Torr) and CO (40 Torr)
(
a) or C O
2
(b) into TS-1 sieve at room temperature. Bands marked
adsorbed into TS-1 or silicalite (S-1) sieve.
2
by asterisk are unique to adsorbed CO .
bands of 13CO2 and C O2. By contrast, the broad absorption
18
-
1
-1 12 16
around 1670 cm (main peaks at 1663 cm ( C O2), 1633
-
1
18
-1 13
cm (C O2) and 1619 cm ( CO2)) and the sharp feature at
1
-1
18
-1 13
1
381 cm- (1334 cm (C O2) and 1366 cm ( CO2)) are
only observed for CO2 adsorbed on TS-1. Corresponding spectra
18
-1
for CO2 and C O2 are shown in Figure 3. The 1381 cm band
is assigned to the ν1 symmetric stretch as the frequency and
isotope shifts agree with reported Raman spectra.2 Since this
is an infrared forbidden mode for the unperturbed molecule, it
is induced by interaction with the micropore environment.
Framework Ti centers seem principally responsible for activation
of this band because its intensity is greatly diminished in neat
2
-
1
18
silicalite sieve. The 1663 cm absorption exhibits an O shift
-
1
13
-1
of 30 cm and a C shift of 44 cm , typical for ν(CdO) of
23
a OdC-O moiety. The absorption can readily be removed
by evacuation of the sieve, which suggests that it originates
from weakly adsorbed CO2 molecules. Since loading of the gas
into silicalite sieve also results in an absorption around 1670
cm , the band is attributed to interaction of CO2 with silicous
regions of the micropores.
18
Figure 5. Infrared difference spectra following 4 h photolysis of C O
2
OH (a), 13CO
+ CH
OH (b), and CO
+
13
+
CH
3
2
3
2
3
CH OH (trace c)
-1
loaded into TS-1 sieve. Spectra in the 2200-2000 cm region were
recorded at 0.1 cm resolution; those in the fingerprint region at 1
-
1
-1
-
1
cm resolution.
2
. Photoreaction of CO2 and CH3OH in TS-1. Loading of
Torr of methanol gas into dehydrated TS-1 sieve at room
temperature resulted in infrared absorptions at 1350, 1450, 1463,
1
5
the product peak at 1380 cm- originates from formic acid. The
CdO stretch of HCO2H overlaps with the HCO2CH3 absorption
at 1717 cm . HCO2H loaded into TS-1 exhibits, in addition, a
broad band at 1594 cm that coincides with a similarly shaped
band in the CO2 + CH3OH photoreaction spectrum. Since this
feature is only present in silicalite sieves containing framework
Ti, and is at a frequency typical for the asymmetric CO2 stretch
of formate, it is attributed to νas(CO2) of formate interacting
with Ti centers, HCO2- - -Ti. This assignment is confirmed by
-
1
-1
2
847, 2952, 2988, 3280, and 3623 cm . The bands at 3250
and 3623 cm signal the presence of CH3OH molecules with
free OH groups (3623 cm ) as well as molecules H-bonded to
the pore wall (3250 cm , very broad). Coadsorption of 40 Torr
CO2 into the sieve did not result in any reaction. When
irradiating the loaded matrix with 266 nm light, depletion of
CO2 and methanol was observed under concurrent growth at
-
1
-1
-
1
-1
2
5
-
1
13
18
1
380, 1434, 1461, 1594, 1717, 2856, and 2963 cm . An FT-
the expected C, O, and D frequency shifts. The only other
IR difference spectrum following the 4 h irradiation at 266 nm
product absorptions of CO2 + CH3OH photolysis in TS-1 were
-2
-1
(30 mWcm ) is shown in Figure 4. When conducting the same
those due to gas-phase CO in the 2100-2200 cm region.
irradiation experiment on a neat silicalite matrix (S-1) loaded
with CO2 and CH3OH, no product growth was observed (Figure
Recording of these bands required a spectral resolution of 0.1
-1
cm . The fact that carbon monoxide produced in TS-1 desorbs
quantitatively from the sieve is not surprising given the weak
adsorption tendency of CO in room-temperature silicalite.
To determine the origin of these products, photolysis experi-
4
). Clearly, photochemical reaction requires the presence of Ti
-
1
centers. The bands at 1717 and 1434 cm agree with those of
an authentic sample of HCO2CH3 in TS-1, as do the ν(C-H)
-
1
18
13
13
absorptions at 2856 and 2963 cm . Methyl formate has an
ments were conducted with C O2, CO2, and CH3OH. Figure
5 shows the infrared difference spectra following 266 nm
-
1
additional peak at 1456 cm , which overlaps in the CO2 +
1
8
CH3OH photolysis spectrum with the decreasing methanol
irradiation of TS-1 sieve loaded with C O2 + CH3OH (trace
-
1
-1
13
3
absorption at 1450 cm and, hence, is shifted to 1461 cm .
These methyl formate bands agree well with the infrared
spectrum of matrix isolated HCO2CH3 reported in the litera-
a), CO2 + CH3OH (trace b), and CO2 + CH3OH (trace c).
-
1
Spectra shown in the 1800-1600 cm region were recorded
-1
-1
at 1 cm resolution; those in the 2200-2000 cm region were
2
4
-1
ture. Spectra of HCO2H gas loaded into TS-1 indicate that
taken at 0.1 cm resolution. Comparison of product bands in

CO2 Photoreduction in TS-1 Sieve
J. Phys. Chem. A, Vol. 104, No. 33, 2000 7837
Figure 6. Kinetics of infrared absorbance growth of carbon monoxide
1
8
13
upon photolysis of C O
2
+ CH
3
OH, CO
2
+ CH
3 2
OH, and CO +
13
CH
3
OH in TS-1 sieve. Curves represent the average of the normalized
1
8
OH, 13CO
+
3 2
Figure 8. Absorbance growth behavior of C O
2
+ CH
peak absorbance growth of all ro-vibrational bands of a given isotopic
species, shown in Figure 5. For each reaction, the curves of the two
isotopic CO products reflect the relative amounts produced.
13
CH
3
OH, and CO
2
+
CH OH upon 266 nm photolysis. (a) Infrared
difference spectra after 1 h irradiation. Intensities are amplified by a
factor of 2 for clarity. (b) Difference between 4 and 1 h photolysis.
3
upon 1 h photolysis (trace a) and of the time window between
1
and 4 h irradiation (trace b) for all three isotopic systems.
Clearly, relative intensities of spectra (a) differ significantly from
13
those of spectra (b). In the case of the CO2 + CH3OH reaction,
-
1
absorbance growth at 1682 cm lags behind that of bands at
-
1
1
736, 1717, and 1695 cm at late photolysis times (trace b)
-
1
compared to start of photolysis (trace a). The 1682 cm peak
is due to H CO2H according to the H CO2H spectrum in TS-1
Figure 7) and literature data. The 1736, 1717, and 1695 cm
13
13
27
-1
(
bands are attributed to different sites of HCO2CH3 (ν(CdO))
based on authentic spectra of this molecule in TS-1. The leveling
1
3
off of H CO2H growth is consistent with secondary photolysis
1
3
of the acid, signaled by the appearance of CO (Figure 5). The
1
3
formation of H CO2H implies that carbon dioxide is photore-
duced to formic acid. Close inspection of the band intensities
-
1
-1
at 1736, 1717, and 1695 cm reveals that the 1717 cm peak
grows faster than the other two upon prolonged photolysis. Since
1
3
Figure 7. Photolysis of H CO
(
2
H loaded into TS-1 sieve at 266 nm
-
2
30 mW cm ). (a) Infrared difference spectrum following 15 min
irradiation, (b) 30 min irradiation. Spectra in the 2200-2000 cm
region were recorded at 0.1 cm resolution, those below 2000 cm
-1
HCO2H also absorbs at 1717 cm , this suggests increased
-
1
HCO2H growth at late photolysis times. Formation of HCO2H
is confirmed by the appearance of CO, its secondary photolysis
product (Figure 5).
-
1
-1
-
1
at 1 cm resolution.
-
1
the 2200-2000 cm1 region with gas-phase spectra of authentic
These assignments are fully supported by the analysis of the
8
13
13
samples of CO, C O, and CO affords complete assignment
of all ro-vibrational peaks.26 According to Figure 5, C O is
CO2 + CH3OH system, the isotopic counterpart to the
18
reactions just presented. Here the absorbance growth at 1717
18
-1
-1
the main carbon monoxide species formed upon C O2 + CH3-
OH photolysis, while 13CO dominates in the CO2 + CH3OH
reaction, and parent CO is the main species produced upon CO2
cm falls behind that of bands at 1696, 1679, and 1647 cm
at late photolysis times, as can clearly be seen in Figure 8. The
13
-
1
1717 cm absorption is assigned to HCO2H, and the early
leveling off of its growth is consistent with the appearance of
the parent isotope CO as the dominant carbon monoxide product
of the reaction (Figure 5). This confirms that CO2 is photore-
duced to HCO2H. The 1696, 1679, and 1647 cm triplet is
attributed to H CO213CH3. Analogous to the CO2 + CH3-
OH case, the 1679 cm band outpaces the growth at 1696 and
1
3
+
CH3OH photolysis. The growth kinetics of the CO product
displayed in Figure 6 for each of the three isotopic experiments
clearly exhibits an induction period indicative of secondary
photolysis. Irradiation of TS-1 matrices loaded with authentic
samples of formic acid indicate that carbon monoxide is
generated by efficient photolysis of the acid. This is illustrated
in Figure 7 by infrared difference spectra recorded upon 266
-
1
1
3
13
-
1
-
1
1647 cm at long photolysis times. This indicates formation
13
13
nm irradiation of H CO2H in TS-1 for 15 min (trace a) and 30
of H CO2H upon prolonged reaction, and is confirmed by the
-
2
13
min (trace b) at 30 mW cm . We conclude that the carbon
monoxide product of the carbon dioxide + methanol photore-
action originates from secondary photolysis of formic acid. By
contrast, only traces of methyl formate photodissociation were
detected under the same irradiation conditions.
concurrent production of CO (Figure 5). We conclude from
1
3
13
the isotope experiments with CO2 and CH3OH that carbon
dioxide is directly reduced to formic acid while the carbon of
the methanol appears in methyl formate and, upon prolonged
photolysis, also in formic acid.
In the CdO stretching region, the simple product spectrum
of the parent reaction CO2 + CH3OH (Figure 4) is replaced by
Observation of methyl formate and formic acid as methanol
oxidation products, rather than the expected 2-electron oxidation
product formaldehyde, can be understood by the results of CH2d
O loading into TS-1, shown in Figure 9. Spectra recorded
immediately after loading of formaldehyde into the sieve exhibit
1
8
13
a complex pattern of bands when using C O2, CO2, or
1
3
CH3OH as reactants (Figure 5). Analysis of these spectra
reveals the origin of the products. Most useful is a comparison
of spectra at early photolysis times with those recorded upon
prolonged irradiation. Figure 8 shows infrared difference spectra
-
1
the characteristic CH2dO peaks at 1503 and 1730 cm . The
-
1
2813, 2909, and 2975 cm absorptions in the C-H stretching

7838 J. Phys. Chem. A, Vol. 104, No. 33, 2000
Ulagappan and Frei
SCHEME 1
Figure 9. Spontaneous dark reaction of formaldehyde loaded into TS-1
sieve. (a) Infrared spectrum recorded immediately after loading of 5
2
Torr of CH dO. (b) Spectrum after 1 h in the dark at room temperature.
region are also observed (not shown).28,29 After 1 h in the dark
SCHEME 2
(trace b), the CH2dO spectrum has practically disappeared and
is replaced by absorptions of HCO2CH3 (1717, 1456, 1436
-
1
-1
cm ). A weak band at 1383 cm indicates the formation of
-
1
small amounts of HCO2H (which also absorbs at 1717 cm ),
-
1
and of HCO2- - -Ti absorbing at 1596 cm . We attribute the
principal product, HCO2CH3, to the well-known Tishchenko
reaction,30 CH2dO + CH2dO f HCO2CH3. Small amounts
of HCO2H (HCO2- - - Ti) indicate that Cannizzaro reaction
CH2dO + CH2dO + H2Of HCO2H + CH3OH involving
residual H2O still present after dehydration of the sieve also
plays a role. The latter was the dominant process upon loading
of formaldehyde into FeAlPO4-5 sieve.20 We conclude from
CH2dO spectroscopy in TS-1 that the observed methyl formate
and a small fraction of the formic acid produced upon CO2 +
CH3OH photoreaction are secondary thermal products of initially
generated formaldehyde.
Upon continued photolysis, formaldehyde is consumed by the
Cannizzaro reaction opened up by the formation of H2O through
secondary photolysis of formic acid. Scheme 1 summarizes the
photochemistry derived from the spectral and kinetic analysis.
18
Analysis of the C O2 + CH3OH spectra and comparison with
Discussion
1
3
the C results furnishes a detailed insight into the role of
Tishchenko and Cannizzaro reactions of CH2dO, the proposed
initial oxidation product of methanol. According to Figure 5,
The main result of this study is the detection of formic acid
as the primary 2-electron-transfer product of CO2 photoreduction
at the gas-micropore interface of Ti silicate sieve. While the
elucidation of the secondary photochemical and dark processes
proved essential for establishing the nature of the initial
2-electron redox process, it is the latter that is of foremost
interest; secondary reactions, in particular the photodissociation
of formic acid, may not occur once materials are found that
allow initiation of the photochemistry at substantially longer
wavelengths.
1
8
18
C O is the dominant carbon monoxide product of C O2 +
CH3OH photolysis. Its growth exhibits an induction period,
which confirms that C O2 is reduced to HC O2H in the initial
step. The latter absorbs at 1682 cm and is most probably the
main contributor to the band at that frequency in the early
photolysis spectrum (trace a of Figure 8). In addition, spectrum
18
18
-
1
1
8
(
1
a) of the C O2 + CH3OH reaction features peaks at 1736 and
717 cm , which are assigned to HCO2CH3, most probably
-
1
formed by Tishchenko reaction of CH2dO, the primary oxida-
Our proposed mechanism for Ti-O LMCT-induced reaction
of CO2 and CH3OH consists of two one-electron-transfer steps.
tion product of CH3OH. Interestingly, additional peaks appear
-1
+III
in the region below 1700 cm upon prolonged photolysis (trace
b of Figure 8), indicating incorporation of 18O into the secondary
thermal products of CH2dO. This can readily be explained by
Upon excitation of the metal center, transient Ti reduces CO2
-
to CO2 under concurrent oxidation of methanol to the radical
cation by the hole on the oxygen ligand, as shown in Scheme
1
8
18
-
the formation of H2 O upon secondary photolysis of HC O2H.
It opens up the Cannizzaro reaction with the continuously
2. Proton transfer quenching of the highly unstable CO2 by
+
31
the strongly acidic CH3OH
would result in CHO2 radical
18
18
generated CH2dO to form HC OOH and CH3OH. HC OOH
and methanol form HC OOCH3 by subsequent ester condensa-
tion. HC OOCH3 isotopomers with a Cd O group give rise
to bands at 1700, 1685, and 1659 cm . HC(d O)OH absorbs
at 1685 cm as well. We conclude from the analysis of the
isotopic data that formaldehyde is the initial oxidation product
of methanol. In the early stages of photolysis, the aldehyde is
converted to methyl formate by the Tishchenko mechanism.
and CH3O radical. The latter most probably isomerizes to the
18
29,32
more stable CH2OH radical.
Subsequent H atom transfer to
1
8
18
yield HCO2H and CH2dO affords overall 2-electron reduction
of CO2.
-1 18
18
-
1
27
The energy profile of the proposed mechanism depends to a
substantial degree on the structure of the proposed CHO2
intermediate. Gas-phase CHO2 radical is known to exist in two
33
isomeric forms, namely, HCO2 (formate) and HOCO (hydroxy

CO2 Photoreduction in TS-1 Sieve
J. Phys. Chem. A, Vol. 104, No. 33, 2000 7839
carbonyl).34,35 On the basis of enthalpies of formation
Sciences Division of the U.S. Department of Energy under
Contract No. DE-AC03-76SF00098.
-
1 36
(
∆Hf°(CO2) ) -94.05 kcal mol
; ∆Hf°(CH3OH) ) -48.0
-1 36
-1 37
kcal mol
)
; ∆Hf°(CH2OH) ) -2 kcal mol
; ∆Hf°(HCO2)
-31 kcal mol- ; ∆Hf°(HOCO) ) -51 kcal mol
1 33
-1 33
), we
References and Notes
-
1
estimate an endothermicity of 109 kcal mol for the CO2 +
(
1) Hemminger, J. C.; Carr, R.; Somorjai, G. A. Chem. Phys. Lett.
1978, 57, 100.
2) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979, 277,
-
1
CH3OH f HCO2 + CH2OH path, and 87 kcal mol for the
HOCO + CH2OH route. Hence, the formation of HOCO
intermediate is thermodynamically preferred, and is consistent
with the fact that the reaction can be initiated by 266 nm quanta
(
37.
6
(3) Halmann, M. M. Chemical Fixation of Carbon Dioxide; CRC
Press: Boca Raton, 1993; p 131.
-
1
(
107 kcal mol ). By contrast, the photolysis photon energy is
(4) Linsebigler, A. L.; Lu, G.; Yates, J. T., Jr. Chem. ReV. 1995, 95,
735.
-1
a few kcal mol short of generating formate radical. However,
the thermochemical data refer to the free species in the gas
phase. Interaction of the transient radicals with the Ti center
may lower these energies substantially. While we cannot be
more specific on the CHO2 intermediate, this work establishes
conclusively that single photon-induced 2-electron reduction of
CO2 at Ti centers in the silicate framework results in the
formation of a C-H bond, and not in abstraction of oxygen to
yield CO.
(
5) Mandler, D.; Willner, I. J. Am. Chem. Soc. 1987, 109, 7884.
(6) Anpo, M.; Yamashita, H.; Ikene, K.; Fujii, Y.; Zhang, S. G.;
Ichihashi, Y.; Park, D. R.; Susuki, Y.; Koyano, K.; Tatsumi, T. Catal. Today
998, 44, 327, and references therein.
7) Zhang, S. G.; Fujii, Y.; Yamashita, H.; Koyano, K.; Tatsumi, T.;
1
(
Anpo, M. Chem. Lett. 1997, 659.
(8) Chiba, K.; Anpo, M. J. Mol. Catal. 1992, 74, 207.
(
9) Anpo, M.; Yamashita, H. In Surface Photochemistry; Anpo, M.,
Ed.; Wiley: Chichester, 1996; p 117.
10) Anpo, M.; Yamashita, H.; Ichihashi, Y.; Fujii, Y.; Honda, M. J.
Phys. Chem. B 1997, 101, 2632.
11) Kumar, R.; Bhaumik, A.; Ahedi, R. K.; Ganapathy, S. Nature 1996,
81, 298.
(
(
Anpo and co-workers have proposed a mechanism for CO2
photoreduction by H2O in Ti silicate sieves,6,7 based on
spectroscopic studies using highly dispersed Ti oxide anchored
3
(
12) Bhaumik, A.; Mukherjee, P.; Kumar, R. J. Catal. 1998, 178, 101.
(13) Treacy, M. M. J.; Higgins, J. B.; van Ballmoos, R. Zeolites 1996,
3
8,39
16, 524.
on porous glass, or by using powdered TiO2 catalyst.
ESR
(14) Trong On, D.; Kaliaguine, S.; Bonneviot, L. J. Catal. 1995, 157,
measurements at 77 K of CO2 photoreduction at these surfaces
indicate the formation of Ti3 ions, H atoms and OH radicals,
and C radicals. When conducting the photoreaction at 323 K,
CH3OH, CH4, and CO were detected as products. From these
results, it was proposed that CO2 reduction and H2O decomposi-
tion proceed competitively at the LMCT- excited Ti-O centers;
CO2 is reduced to CO, and subsequently to C radicals while
H2O photodecomposes to H and OH radicals. Reaction of H
and OH with the carbon species is thought to yield CH3OH
2
35.
+
(15) Vayssilov, G. N. Catal. ReV. 1997, 39, 209, and references therein.
(16) Geobaldo, F.; Bordiga, S.; Zecchina, A.; Giamello, E.; Leofanti,
G.; Petrini, G. Catal. Lett. 1992, 16, 109.
17) Van der Pol, A. J. H. P.; Verduyn, A. J.; van Hoof, J. H. C. Appl.
Catal. A 1992, 92, 113.
(18) Duprey, E.; Beaunier, P.; Springuel, M. A.; Bozon, F.; Fraissard,
J.; Manoli, J. M.; Bregeault, J. M. J. Catal. 1997, 165, 22.
(
(19) Lamberti, C.; Bordiga, S.; Arduino, D.; Zecchina, A.; Geobaldo,
F.; Spano, G.; Genoni, F.; Petrini, G.; Carati, A.; Villain, F.; Vlaic, G. J.
Phys. Chem. B 1998, 102, 6382.
6,38,39
(20) Ulagappan, N.; Frei, H. J. Phys. Chem. A 2000, 104, 490.
and CH4.
It is not known how many photons are consumed
(21) Spence, R.; Wild, W. J. Chem. Soc. (London) 1935, 338.
by these transformations. While the mechanism proposed by
Anpo cannot be compared directly with the CO2 photoreduction
reported here because of the use of different electron donors, it
is important to note that, according to our results, formic acid
photodissociates efficiently to CO and H2O under UV irradiation
of the Ti-O LMCT transition. Hence, formation of HCO2H
might escape detection under continuous UV photolysis condi-
tions.
(
22) Herzberg, G. Infrared and Raman Spectra; Van Nostrand: New
York, 1945; p 274.
(23) Jacox, M. E.; Milligan, D. E. J. Mol. Spectrosc. 1974, 52, 363.
(
24) M u¨ ller, R. P.; Hollenstein, H.; Huber, J. R. J. Mol. Spectrosc. 1983,
00, 95.
25) Colthup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared
and Raman Spectroscopy, 3rd ed.; Academic Press: New York, 1990; p
17.
26) Plyler, E. K.; Danti, A.; Blaine, L. R.; Tidwell, E. D. J. Res. Natl.
Bur. Stand. 1960, 64, 29.
1
(
3
(
(
27) Redington, R. L. J. Mol. Spectrosc. 1977, 65, 171.
(28) Herzberg, G. Infrared and Raman Spectra; Van Nostrand: New
Conclusions
York, 1945; p 300.
(29) Jacox, M. E. Chem. Phys. 1981, 59, 213.
(
30) March, J. AdVanced Organic Chemistry, 4th ed.; Wiley: New York,
In-situ FT-IR monitoring of the Ti-O LMCT-induced
photoreduction of gaseous CO2 in Ti silicalite sieve using
methanol as donor has revealed formic acid as the primary
1
5
992; p 1233.
(31) Hammerich, O.; Parker, V. D. AdV. Phys. Org. Chem. 1984, 20,
5.
(
32) (a) Radford, H. E. Chem. Phys. Lett. 1980, 71, 195. (b) Radford,
2-electron-transfer product. This establishes C-H bond forma-
H. E.; Evenson, K. M.; Jennings, D. A. Chem. Phys. Lett. 1981, 78, 589.
(33) Kim, E. H.; Bradforth, S. E.; Arnold, D. W.; Metz, R. B.; Newmark,
D. M. J. Chem. Phys. 1995, 103, 7801.
tion during the initial steps of CO2 photoreduction. While carbon
monoxide product is also observed, the growth kinetics indicate
that CO is formed by secondary photolysis of HCO2H. This
study constitutes the first insight into the initial steps of CO2
photoreduction in a framework Ti molecular sieve. The proposed
mechanism involves photoinduced one-electron transfer to
produce transient CHO2 radical. Time-resolved infrared spec-
troscopic studies on the nanosecond time-scale will be needed
(
(
(
34) Jacox, M. E. J. Chem. Phys. 1988, 88, 4598.
35) Milligan, D. E.; Jacox, M. E. J. Chem. Phys. 1971, 54, 927.
36) Benson, S. W. Thermochemical Kinetics; Wiley: New York, 1968.
(37) Tsang, W. In Energetics of Organic Free Radicals; Martinho-
Simoes, J. A., Greenberg, A., Liebman, J. F., Eds.; Blackie Academic and
Professional: London, 1996; p 22.
(38) Anpo, M.; Chiba, K. J. Mol. Catal. 1992, 74, 207.
(39) Anpo, M.; Yamashita, H.; Ichihashi, Y.; Ehara, S. J. Electroanal.
Chem. 1995, 396, 21.
40-42
to detect this crucial intermediate and elucidate its structure.
(40) Vasenkov, S.; Frei, H. J. Am. Chem. Soc. 1998, 120, 4031.
(41) Vasenkov, S.; Frei, H. J. Phys. Chem. A 2000, 104, 4327.
(42) Vasenkov, S.; Frei, H. In Molecular and Supramolecular Photo-
Acknowledgment. This work was supported by the Director,
Office of Science, Office of Basic Energy Sciences, Chemical
chemistry; Ramamurthy, V., Schanze, K. S., Eds.; Marcel Dekker: New
York, 2000; Vol. 5, p 299.