Photoreduction of carbon dioxide by hydrogen over magnesium oxide

0 downloads 0 Views 233KB Size Report
Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, ... Magnesium oxide was found to show activity for the reduction of carbon dioxide to carbon monoxide under ... produced from the surface formate in the presence of ... to be active for the methane coupling reaction without oxygen.
Photoreduction of carbon dioxide by hydrogen over magnesium oxide Yoshiumi Kohno, Haruka Ishikawa, Tsunehiro Tanaka, Takuzo Funabiki and Satohiro Yoshida Department of Molecular Engineering, Graduate School of Engineering, Kyoto University, Kyoto 606-8501, Japan Received 6th November 2000, Accepted 17th January 2001 First published as an Advance Article on the web 22nd February 2001

Magnesium oxide was found to show activity for the reduction of carbon dioxide to carbon monoxide under photoirradiation using hydrogen as a reductant. Fourier transform infrared spectroscopy was applied to a study of the reaction mechanism by detection and identiÐcation of the surface species arising during the photoreaction. The formation of surface formate ion was observed during the photoreaction. Since CO was produced from the surface formate in the presence of CO under irradiation, the surface formate was a 2 reaction intermediate which acted as a reductant and converted another CO molecule to CO. The correlation 2 of the reaction activity with the amount of introduced CO indicated that adsorbed carbonate was reduced by 2 H to the surface formate, and that the surface formate also reduced the adsorbed carbonate to CO. The IR 2 spectra showed the di†erence in the adsorption form of CO between the adsorbed carbonate reduced by H 2 2 to the surface formate and that reduced by the surface formate to CO.

Introduction Photocatalytic reduction of carbon dioxide has attracted chemists because this photoreaction can activate the chemically inert carbon dioxide by the e†ect of photoirradiation. At the same time, this reaction is interesting from the viewpoint of environmental protection, since it is a removal of a greenhouse e†ect gas under mild conditions. Many kinds of heterogeneous catalysts have been tested for the photoreduction of carbon dioxide. Of these, semiconductors are considered as hopeful materials for photocatalysts, mainly because it is believed that the excitation of electrons from the valence band to the conduction band is essential for the photocatalytic reactions.1 Metal loadings are often applied to semiconductor photocatalysts in order to enhance the charge separation across the band gap.2 Among the semiconductors, zirconium oxide was found to be active for the photoreduction of aqueous carbonate to carbon monoxide.3,4 Zirconium oxide is a unique catalyst in that it does not require any modiÐcation such as metal loading to exhibit sufÐcient photocatalytic activity. Previously, we applied zirconium oxide to the photoreduction of carbon dioxide at the gas/solid interface, and investigated the reaction mechanism.5h9 Zirconium oxide was active for the photoreduction of gaseous CO to CO using hydrogen 2 as a reductant.5 During the study we found that band-gap excitation of zirconium oxide is unnecessary for the photoreduction of CO .7 This suggests that the reaction can be cata2 lyzed by materials which are not semiconductors. There are several reports that photoreactions proceed over materials which do not have semiconductivity in the normal state. For example, a dispersed metal oxide should not be a semiconductor since it cannot exhibit band structure. Atomically dispersed vanadium oxide shows photocatalytic activity for partial oxidation of hydrocarbons and alcohols10,11 or selective photoreduction of NO to N with NH in the pres2 3 ence of O .12 Highly dispersed titanium oxide supported on 2 silica does not show semiconductivity, but still acts as a photocatalyst for propane oxidation.13 Silica itself has also been reported to show activity for the photometathesis reaction of propene,14,15 epoxidation of propene16 and photoÈ 1108

oxidation of CO.17 More recently, silicaÈalumina was found to be active for the methane coupling reaction without oxygen under irradiation.18 Emeline and co-workers are currently investigating photoprocesses taking place on the surface of several metal oxides which have a wide band gap and do not show semiconductivity.19 Magnesium oxide is an insulator and has been reported to show activity for several photoreactions such as the HÈD exchange reaction,20 isomerization of butenes,21 CÈH bond dissociation of methane22 and CO ~ anion radical formation 2 by adsorption of CO on irradiated MgO.23 Among these 2 reactions, CO ~ formation is noteworthy since the formation 2 of the CO ~ radical has also been observed on ZrO under 2 2 irradiation.8 This radical species is a stabilized form of photoexcited CO , and considered as an important reactive species 2 in the photoreduction of CO over ZrO . Therefore, we can 2 2 readily expect that the photoreduction of CO will proceed 2 over MgO if the CO ~ radical is stabilized on MgO. In addi2 tion, MgO has a rock-salt structure with a simple surface constitution compared with ZrO . This makes the investigation 2 of the reaction mechanism easier because the expected structure of the active sites is less varied than for ZrO . 2 Taking these properties into consideration, MgO was employed for the photoreduction of carbon dioxide in this study. It was conÐrmed that the photoreduction of CO to 2 CO takes place with H over MgO ; IR spectroscopy was used 2 for the identiÐcation of the surface species, which was later conÐrmed to be a reaction intermediate. The relationship between the adsorbed CO species and the reaction is also 2 discussed.

Experimental The magnesium oxide used in this study was supplied by Merck. The MgO was hydrated in distilled water for 6 h at 353 K, followed by calcination at 873 K for 3 h. The BET speciÐc surface area of the hydrated sample was 110 m2 g~1. Reactants were puriÐed prior to use for reactions in the following manner. Hydrogen was puriÐed by passing it through a liqueÐed nitrogen trap. Carbon dioxide was puriÐed by

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113 This journal is ( The Owner Societies 2001

DOI : 10.1039/b008887k

vacuum distillation at the temperature of liquid nitrogen. Formaldehyde was obtained by heating paraformaldehyde in vacuum. The photoreaction was carried out in a closed static system connected to a vacuum line. A 0.3 g amount of magnesium oxide was spread on the Ñat bottom of a quartz reactor. For conditioning, the sample was heated at 673 K for 30 min in air and evacuated for 30 min at the same temperature, followed by treatment with 8 kPa O for 75 min and evacuation for 30 2 min at 673 K. A mixture of CO (150 lmol) and H (50 lmol) 2 2 was introduced into the reactor, and the total pressure in the reactor was ca. 25 kPa. A 500 W ultrahigh-pressure mercury lamp (Ushio Denki USH-500D) was used as the light source, and the reactor was irradiated from the bottom. The area subjected to illumination was 12 cm2. The photoirradiation was carried out typically for 6 h. After each reaction, the gaseous products were analyzed, and after 5 min evacuation at room temperature the sample was heated at 673 K for 20 min and the desorbed gases were also analyzed. The analysis of the products was performed with an on-line TCD gas chromatograph (Shimadzu GC-8A) equipped with a column packed with Molecular Sieve 5A using Ar as the carrier gas. When formaldehyde was used as a reaction substrate, 5 lmol of formaldehyde were introduced onto 0.1 g of MgO with 150 lmol of CO or 50 lmol of H in the reactor. 2 2 Infrared spectra were recorded with a Perkin-Elmer PARAGON 1000 PC Fourier transform IR spectrometer. A magnesium oxide sample (ca. 40 mg) was pressed into a disc (diameter \ 10 mm) at a pressure of 2 MPa and suspended with platinum wire in a conventional in situ cell with NaCl windows. The cell allowed us to perform heating, O treat2 ment, introduction of substrates, UV irradiation and measurements of spectra without exposure of the sample to air. Before a measurement, the sample disc was conditioned by evacuation at 673 K for 60 min, treatment with 8 kPa O for 75 2 min and evacuation for 30 min at the same temperature. After cooling, CO (2.7 kPa) was admitted to the sample at room 2 temperature followed by evacuation, and H (4.0 kPa) was 2 then introduced. A 250 W ultrahigh-pressure mercury lamp (Ushio Denki USH-250D) was used as the light source for irradiation of the disc. For each spectrum, data from 100 scans were accumulated at a resolution of 4 cm~1.

Results and discussion 1. Photoreduction of CO with H over MgO 2 2 Attempts were made to reduce carbon dioxide by hydrogen over magnesium oxide under photoirradiation. After photoirradiation of MgO in the presence of CO and H for 6 h, 2.9 2 2 lmol of CO were detected in the gas phase. The amount of CO increased with the irradiation time. The open squares of Fig. 1 illustrate the time course of CO production in the photoreaction. After irradiation for 40 h, the amount of CO produced reached 8.3 lmol. When CO was introduced solely 2 into the reactor, the amount of CO produced was negligible (0.1 lmol) after irradiation for 6 h, indicating that H was 2 required as a reductant of CO . On the other hand, when H 2 2 only was introduced onto MgO, CO was not produced after irradiation for 6 h. This result shows that the CO came from introduced CO through the reduction by H . 2 2 Without photoirradiation, no CO was detected even when both CO and H were introduced onto MgO. This obviously 2 2 indicates that this reaction is a photoreaction. Irradiation of MgO with light with a wavelength longer than 290 nm brought about only a small amount of CO (0.4 lmol) from CO and H . We can thus conclude that the reaction requires 2 2 UV light with, at least, a wavelength shorter than 290 nm. The photoactive species of this reaction is not clear. However, the intrinsic band gap excitation of MgO is not essential in this

Fig. 1 Time course of CO production during photoreaction between CO and H over MgO. Open squares indicate the amount of CO 2 2 evolved to the gas phase during the photoreaction, and closed circles indicate the amount of CO produced by the heat treatment after the photoreaction.

photoreaction because the irradiation light source does not supply light with sufficient energy to excite the intrinsic band gap of MgO. The existence of a reaction intermediate remaining on the surface was suspected. To decompose the intermediate, if any, the MgO sample was heated at 673 K in vacuum after the photoreaction. The heat treatment caused additional evolution of CO. The closed circles of Fig. 1 illustrate the amount of CO produced by the heat treatment after the photoreaction. This result suggests the formation of a surface species which is decomposed to CO by heat treatment, although one cannot exclude the possibility that CO, once produced, is adsorbed on MgO.24,25 On the other hand, after reaction of CO and H over MgO for 6 h in the dark, the amount of CO 2 2 desorbed by heat treatment was less than 0.1 lmol. It is evident that the formation of this surface species occurred predominantly under irradiation. As this surface species was supposed to be a reduced form of CO , we tentatively assumed 2 this surface species to be a reaction intermediate. 2. Observation of surface species by IR spectroscopy To observe the surface species arising during the photoreaction of CO and H on MgO, IR spectra of the MgO sample 2 2 were recorded before and after its use for the photoreaction. Fig. 2 represents the IR spectra of MgO (a) in vacuum after pre-treatment, (b) after introduction of CO and H into (a), 2 2 and (c) after irradiation of (b) for 15 h with CO and H . In 2 2 Fig. 2 (a), a peak assigned to surface hydroxyl groups is observed at around 3760 cm~1. This indicates that dehydration of MgO was not complete by evacuation at 673 K. Peaks were also found at 1422 and 846 cm~1, which are both

Fig. 2 IR spectra of MgO (a) in vacuo after pre-treatment, (b) in the presence of CO and H and (c) after irradiation for 15 h in the presence of CO and2 H . 2 2 2

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113

1109

assigned to magnesium carbonate, MgCO .26 Even evac3 uation at 673 K cannot remove MgCO completely, since the 3 decomposition temperature of MgCO is higher than 673 K. 3 However, as CO was not produced under photoirradiation without the introduction of CO (see the above section), CO 2 2 remaining after pre-treatment in the form of MgCO is not 3 involved in the reaction. Introduction of CO and H caused the formation of 2 2 several surface carbonate species. Their characteristic bands appeared in the range from 1800 to 1000 cm~1, as shown in Fig. 2(b). Bands at 1660, 1628, 1310, 1014 and 974 cm~1 are assigned to bidentate surface carbonates.27 Bands at 1684, 1380 and 1216 cm~1 are assigned to a surface bicarbonate species.26,28 Since surface hydroxyl groups remained on MgO to some extent, the formation of the bicarbonate ion was not illogical. Fig. 2(c) represents the IR spectrum of MgO after photoirradiation for 15 h in the presence of CO and H . Changes 2 2 were observed in the IR spectra in the range from 2900 to 2700 cm~1, and from 1800 to 1000 cm~1. To make the spectral change caused by irradiation clearer, we calculated a difference spectrum between the spectra of the samples before and after photoirradiation. Fig. 3 depicts the subtraction of the IR spectrum before photoirradiation from that after photoirradiation. The inset shows the CÈH stretching band region, and we can see the growth of two bands at 2822 and 2728 cm~1 in this region during photoirradiation. Since surface carbonates have no vibration mode of CÈH stretching, the appearance of these bands indicates the formation of a surface species containing a CÈH bond, which we expected to be a reaction intermediate. In the range from 1800 to 1000 cm~1, bands at 1598 and 1358 cm~1 grew under irradiation, whereas bands at 1684, 1666, 1630 and 1216 cm~1 decreased. The decreasing bands are all assigned to bidentate surface carbonate and bicarbonate. Therefore, the negative bands indicate the consumption of the adsorbed carbonates by the reaction under irradiation. The positive bands at 1598 and 1358 cm~1 were not observed when irradiation was not applied. The bands in the CÈH stretching region at 2822 and 2728 cm~1 also did not appear without irradiation. From this, we concluded that the bands at 1598 and 1358 cm~1 together with those at 2822 and 2728 cm~1 were attributable to a surface species which is expected to be an intermediate, and that the surface species was formed only under irradiation. This conclusion is also supported by the fact that the additional evolution of CO by the heat treatment after the photoreaction is not observed without irradiation. Other positive bands were seen at around 1500 cm~1 in a CO and H atmosphere under irradiation of MgO, as shown 2 2 in Fig. 3. These broad bands appeared even when MgO was treated in the dark with CO and H for 17 h, and can be 2 2 assigned to monodentate carbonate species.28 Taking into

Fig. 3 Di†erence IR spectra of MgO in the presence of CO and H 2 before and after 15 h of photoirradiation. The inset illustrates the2 CÈH stretching band region.

1110

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113

consideration that the bands assigned to bidentate carbonates decreased to some extent even in the dark, it is suggested that bidentate carbonates were converted to monodentate carbonates over a period of time in the dark.29 It is known that hydrogen is adsorbed on MgO in a dissociated form yielding OÈH and MgÈH species even at room temperature.30 However, we failed to detect absorption bands assigned to the adsorbed hydrogen species in the IR spectra of MgO with CO and H . This is because the dissociative 2 2 adsorption of hydrogen requires evacuation of MgO at a temperature higher than 1073 K.31 In this study, MgO was evacuated at 673 K, so we could not observe the adsorbed hydrogen species. 3. IdentiÐcation of the surface species Attempts were made to identify the surface species arising on MgO during photoirradiation in CO and H by IR spectra. 2 2 The adsorption of several molecules was tested, and it was found that a similar spectrum to that of the surface species was obtained when formaldehyde was adsorbed on the surface of MgO. The IR spectrum of the adsorbed formaldehyde is shown in Fig. 4, together with the spectrum of the species remaining on the surface of MgO after the photoreaction of CO and H . Although the species existing on the surface 2 2 after the photoreaction includes the adsorbed carbonates or bicarbonates, in Fig. 4[B] we can clearly see the absorption bands at 1598 and 1360 cm~1, attributed to the surface species arising on MgO during irradiation in CO and H . 2 2 The IR spectrum of the adsorbed formaldehyde showed bands at 2824, 2728, 1602 and 1362 cm~1. The frequencies of these bands are almost identical with those of the bands seen in the surface species remaining after the photoreaction. Therefore, we judged that the surface species formed by the photoreaction was the same as that obtained by the adsorption of formaldehyde on MgO. The species formed by the adsorption of formaldehyde was assigned to a bidentate surface formate ion.32 The nature of the adsorption is shown in Scheme 1. On adsorption onto the MgO surface, formaldehyde loses one hydrogen to involve one oxygen atom of MgO and forms a bidentate surface

Fig. 4 IR spectra of (a) formaldehyde adsorbed on MgO and (b) the surface species existing on MgO after irradiation for 15 h in CO and H atmosphere : [A] in the region between 2650 and 2900 cm~1,2 and 2 between 1200 and 1800 cm~1. [B]

formate ion. Therefore, we concluded that the surface species arising during the photoreaction between CO and H is a 2 2 surface formate ion. This conclusion seems reasonable considering that the surface formate is reported to be formed in the (thermal) water-gas shift reaction on MgO,33 which is the reverse reaction of the reduction of CO to CO by H . 2 2

Scheme 1

Two reactions can be proposed as a decomposition route of the surface formate on MgO : decomposition to H and CO , 2 2 or decomposition to CO and H O.33 In our case, only a small 2 amount of H was detected by the thermal decomposition of 2 the surface formate at 673 K after the photoreaction. Therefore, most of the formate ion is probably decomposed to CO and H O. This means that the amount of the surface formate 2 ion is equal to the amount of CO produced by the heat treatment after the photoreaction. 4. Reactivity of the surface formate It was conÐrmed that the surface formate is formed during the photoreaction between CO and H over MgO. The formate 2 2 ion in the same adsorption form can be created by the adsorption of formaldehyde on MgO. However, it is still not clear whether this formate ion is a reaction intermediate or not. If the surface formate is a true reaction intermediate, the Ðnal product, CO, should be obtained by using formaldehyde as a reaction substrate. To conÐrm this, some reactions were carried out using formaldehyde as one of the reactants. The results of these reactions are summarized in Table 1. When formaldehyde solely was introduced onto MgO, no CO was evolved even after photoirradiation for 6 h. CO was not obtained when MgO was irradiated for 6 h in the presence of formaldehyde and H . However, when CO was introduced 2 2 together with formaldehyde, 2.0 lmol of CO were evolved after 6 h irradiation of MgO. This result indicates that the surface formate ion gives gaseous CO in the presence of CO . 2 Therefore, it is suggested that the surface formate acts as a reductant and converts another CO molecule to CO. On the 2 other hand, without irradiation no CO was detected even if CO and formaldehyde were in contact with MgO. From this, 2 the evolution of CO from the surface formate ion with CO 2 on MgO was a photoreaction, as is the formation of the surface formate from CO and H . 2 2 To conÐrm further that the surface formate is able to reduce another CO molecule, the carbon atom of the CO molecule 2 2 was labelled with 13C and the reaction of H12CHO and 13CO on MgO was carried out under irradiation. 13CO was 2 detected as a product by a quadrupole mass spectrometer. Although it has been reported that the decomposition of surface formate ion is suppressed by the existence of CO on 2 MgO,34 in our case CO was certainly reduced to CO by the 2

surface formate under irradiation. However, it is not clear whether the carbon atom of the surface formate was released into the gaseous phase as CO or not, because 12CO was hardly distinguished from N in the background air by the 2 mass spectrometer. It is possible that, on receiving one oxygen atom of CO , the surface formate was oxidized to a surface 2 carbonate or bicarbonate ion. From these results, we concluded that the surface formate is a reaction intermediate of the photoreaction between CO 2 and H over MgO. It does not decompose to yield CO 2 directly by itself, but acts as a reductant and converts another CO molecule to gaseous CO under irradiation. Such a pro2 perty of the surface formate ion has also been seen in the photoreduction of CO by H over zirconium oxide : 7 over 2 2 ZrO , the surface formate arises during the photoreaction of 2 CO and H , and it acts as a reductant and converts another 2 2 CO molecule to CO. This fact suggests that a common reac2 tion mechanism exists in the photoreaction of CO and H 2 2 over MgO or ZrO . 2 5. Dependency of the reaction on the amount of CO

2 It was found that the surface formate ion is formed during the photoreaction between CO and H on MgO, and that the 2 2 surface formate reduces another CO molecule to yield 2 gaseous CO. The surface formate was produced from the adsorbed CO , because the formation of the surface formate 2 was observed in the IR spectrum when gaseous CO was 2 absent. However, it was not determined whether gaseous CO 2 or adsorbed CO reacted with the surface formate to yield 2 CO. In order to clarify this, the dependency of the photoreaction on the amount of introduced CO was investigated. Prior 2 to this investigation, it was conÐrmed that 0.3 g of MgO captures about 40 lmol of CO under vacuum. This result means 2 that the introduction of less than 40 lmol of CO onto 0.3 g 2 of MgO brings about no gaseous CO . 2 Fig. 5 represents the relationship between the amount of introduced CO and the activity of the photoreaction where 2 50 lmol of H were introduced and irradiation was applied 2 for 6 h. Open squares indicate the amount of gaseous CO, and closed circles the amount of CO evolved by the heat treatment at 673 K after the photoreaction. As described above, the amount of CO evolved by the heat treatment after the photoreaction reÑects the amount of surface formate ion. Until the amount of introduced CO reached 20 lmol, the 2 amount of surface formate increased with an increase in the introduced CO . This indicates that the adsorbed CO is 2 2 reduced by the reaction with H to form the surface formate. 2 Since the adsorbed CO increases with an increase in the 2

Table 1 Reactions using formaldehyde as one of the reactantsa Substrates

Reaction conditions

Amount of CO/lmol

HCHO Under irradiation 0.0 HCHO ] H Under irradiation 0.0 2 HCHO ] CO Under irradiation 2.0 HCHO ] CO 2 b In the dark 0.0 2 a Initial amount of HCHO, 5 lmol ; H , 50 lmol ; CO , 150 lmol. 2 Reaction time, 6 h ; reaction temperature, room 2temperature. b Conditions as in a except that the reaction temperature was 313 K.

Fig. 5 Dependencies of the amount of CO evolved by the photoreaction (open squares) and by the heat treatment after photoreaction (closed circles) on the initial amount of introduced CO . The amount 2 of the surface of CO evolved by the heat treatment reÑects the amount formate, see text.

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113

1111

amount of introduced CO , the formation of the surface 2 formate ion also increases. When the CO introduction was increased further, the 2 surface formate decreased whereas the production of gaseous CO became signiÐcant. In this region, the reduction of CO to 2 CO by the surface formate occurs. Since the surface formate is consumed by this reaction, the amount of surface formate remaining after the photoreaction decreases. The CO reduced 2 by the formate is adsorbed CO , because gaseous CO does 2 2 not exist in this region. The introduction of more than 40 lmol of CO did not 2 have a marked inÑuence on the amount of both surface formate and gaseous CO. Therefore, the presence or absence of gaseous CO has little inÑuence on the reaction. This result 2 further supports the suggestion that the reaction of CO with 2 the formate occurs between the adsorbed molecules. It is interesting that the amount of introduced CO con2 trolled the reaction. The adsorption form of CO is suspected 2 to be di†erent depending on the amount of introduced CO : 2 when a small amount of CO is introduced, the adsorbed CO 2 2 is favorable for the reaction with H to form the surface 2 formate, whereas when the amount is large, the adsorbed CO 2 is favorable for the reaction with the surface formate to yield gaseous CO. 6. Changes in the adsorption form of carbonate species caused by the amount of CO 2 IR spectra of the adsorbed carbonate species were recorded under various degrees of CO coverage, in an attempt to 2 detect the di†erence in the adsorption forms of CO . Fig. 6 2 shows the di†erence IR spectrum between bare MgO and MgO with various amounts of adsorbed CO up to 4.2 lmol 2 for 40 mg of MgO. In Fig. 6, we can see the increase in the absorbance of several carbonate species along with the introduction of CO . The mainly growing bands at the earlier 2 stages of CO adsorption were di†erent from those at the later 2 stages. This phenomenon is more clearly shown in Fig. 7, where the spectral changes caused by the additional introduction of CO are illustrated by the subtraction of spectra 2 before and after CO addition. The increase in the absorbance 2 of the bands at 1660 and 1310 cm~1 stopped after the amount of introduced CO exceeded 3.0 lmol. The growth of the 2 bands at 1682 and 1624 cm~1 replaced that at 1660 cm~1. Simultaneously, the bands at 1216 and 1384 cm~1 increased when a sufficient amount of CO had been introduced. In 2 addition, a band at 3622 cm~1 appeared when a large amount of CO was introduced whereas the band at 3768 cm~1 2

Fig. 6 IR spectra of carbonate species adsorbed on MgO surface at room temperature. The amount of introduced CO was varied from 2 0.6 to 4.2 lmol.

1112

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113

Fig. 7 IR spectral changes caused by the additional adsorption of CO onto the MgO surface with previously adsorbed CO . Each 0.6 2 of CO was introduced onto MgO with pre-adsorbed 2 CO up lmol 2 2 to 4.2 lmol. Amount of pre-adsorbed CO : (a) 0, (b) 0.6, (c) 1.2, (d) 2 1.8, (e) 2.4, (f ) 3.0 and (g) 3.6 lmol, respectively.

decreased (data not shown). These results indicate that an adsorbed CO species with bands at 1660 and 1310 cm~1 2 appears at lower coverages of CO , and at higher coverages, 2 adsorbed CO with bands at 3622, 1682, 1624, 1384 and 1216 2 cm~1 is predominantly formed. As described in the previous section, the bands at 1660 and 1310 cm~1 are assigned to a surface bidentate carbonate ion. This may be the species formed at the earlier stages of CO 2 adsorption. The bands at 1682 and 1384 cm~1 are considered to be due to a surface bicarbonate ion on MgO together with the bands at 1216 and 3622 cm~1.26 On the other hand, the band appearing at around 3760 cm~1 is assigned to a surface hydroxyl species of MgO.35 Therefore, the formation of the bicarbonate species occurs by the reaction between CO and 2 the surface hydroxyl group of MgO in proportion to an increase in the surface coverage of CO . The band at 1624 2 cm~1, which appears on the introduction of a large amount of CO , may be due to another form of adsorbed carbonate 2 species. The band can be assigned to a monodentate-like incomplete bidentate carbonate.27 It is suggested that at lower coverages of CO the surface carbonate may be advantageous 2 to the reaction with H yielding the surface formate, and at 2 higher coverages the carbonate may take a form that is unfavorable for the reaction with H . 2 The introduction of 3 lmol of CO to 40 mg of MgO corre2 sponds to that of ca. 20 lmol of CO onto 300 mg of MgO. 2 We have already conÐrmed that the introduction of more than 20 lmol of CO onto 300 mg of MgO causes the forma2 tion of gaseous CO from the adsorbed formate and CO . At 2 the same time, it is shown here that the introduction of this amount of CO is a critical point between the formation of 2 bidentate carbonate and bicarbonate or another form of bidentate carbonate. From this, we conclude that the bicarbonate or the other bidentate carbonate is reduced by the surface formate to yield gaseous CO, i.e., adsorbed CO 2 species as precursors are di†erent between the formate and the gaseous CO. The surface formate ion originates from the bidentate carbonate on MgO during the photoreduction of CO by H . 2 2 The most stable form of the bidentate carbonate is reported to be the side-on adsorption-type form, where the carbon atom of CO interacts with the oxygen atom of MgO and the 2 oxygen of CO with the magnesium of MgO.36,37 At low 2 coverages, the adsorbed CO may take the most stable form. 2 This type of adsorption is similar to that of the surface

formate ion observed in this study, as described in Section 3. Therefore, one can speculate that this type of bidentate carbonate is converted to the surface formate in the reaction with hydrogen. At present, we do not have enough information about the reaction of adsorbed CO and H on MgO. However, the 2 2 interaction of adsorbed CO and H on ZrO under irradia2 2 2 tion has already been clariÐed ;8 the adsorbed CO is excited 2 and stabilized as a CO ~ anion radical by irradiation, and H 2 2 interacts with the CO ~ radical in the dark to form the 2 surface formate. The formation of the CO ~ radical has also 2 been reported on MgO, although the formation process is different.23 These Ðndings suggest that the adsorbed CO is 2 excited by irradiation to CO ~ also on MgO, followed by the 2 reaction with H to yield the surface formate. The source of 2 electrons for CO ~ may be the oxygen 2p orbital of MgO. 2 However, it is still not clear how the adsorbed CO is 2 photoexcited. An MgÈO pair with low coordination number has been reported to be photoexcited by UV irradiation and to act as an active site for some photoreactions occurring on the MgO surface.30 Therefore, the low coordinated MgÈO site can be regarded as an active center of the photoreaction. On the other hand, the idea that photoactive species are formed by adsorbed molecules has also been proposed.7,38,39 The clariÐcation of the photoexcitation process in the photoreduction of CO by H over MgO requires more detailed 2 2 investigation. A study of the activation mechanism of the adsorbed CO by irradiation is currently in progress. 2

Conclusion Carbon dioxide is reduced to CO selectively on MgO under photoirradiation, using hydrogen as a reductant. This reaction proceeds via a surface formate ion as an intermediate. The surface formate does not decompose to yield CO but acts as a reductant and converts another CO molecule to CO under 2 irradiation. Thus, this reaction consists of two independent photoreactions. The surface formate and gaseous CO are both produced from adsorbed CO . The adsorbed carbonate, as a 2 precursor for the surface formate, is suggested to be di†erent from that for the gaseous CO.

7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

References 1 H. Yoneyama, Catal. T oday, 1997, 39, 169. 2 A. Mills and S. LeHunte, J. Photochem. Photobiol. A, 1997, 108, 1. 3 K. Sayama and H. Arakawa, J. Phys. Chem., 1993, 97, 531. 4 K. Sayama and H. Arakawa, J. Photochem. Photobiol. A, 1996, 94, 67. 5 Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, Chem. Commun., 1997, 841. 6 Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, Chem. L ett., 1997, 993.

33 34 35 36 37 38 39

Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, J. Chem. Soc., Faraday T rans., 1998, 94, 1875. Y. Kohno, T. Tanaka, T. Funabiki and S. Yoshida, Phys. Chem. Chem. Phys., 2000, 2, 2635. T. Tanaka, Y. Kohno and S. Yoshida, Res. Chem. Intermed., 2000, 26, 93. T. Tanaka, Y. Nishimura, S. Kawasaki, M. Ooe, T. Funabiki and S. Yoshida, J. Catal., 1989, 118, 327. T. Tanaka, M. Ooe, T. Funabiki and S. Yoshida, J. Chem. Soc., Faraday T rans. 1, 1986, 82, 35. T. Tanaka, K. Teramura and T. Funabiki, Phys. Chem. Chem. Phys., 2000, 2, 2681. S. Yoshida, S. Takenaka, T. Takana, H. Hirano and H. Hayashi, Stud. Surf. Sci. Catal., 1996, 101, 871. H. Yoshida, T. Tanaka, S. Matsuo, T. Funabiki and S. Yoshida, J. Chem. Soc., Chem. Commun., 1995, 761. T. Tanaka, S. Matsuo, T. Maeda, H. Yoshida, T. Funabiki and S. Yoshida, Appl. Surf. Sci., 1997, 121, 296. H. Yoshida, T. Tanaka, M. Yamamoto, T. Yoshida, T. Funabiki and S. Yoshida, J. Catal., 1997, 171, 351. A. Ogata, A. Kazusaka and M. Enyo, J. Phys. Chem., 1986, 90, 5201. Y. Kato, H. Yoshida and T. Hattori, Chem. Commun., 1998, 2389. A. V. Emeline, G. N. Kuzmin, D. Purevdorj, V. K. Ryabchuk and N. Serpone, J. Phys. Chem. B, 2000, 104, 2989. C. G. Harkins, W. W. Shang and T. W. Leland, J. Phys. Chem., 1969, 73, 130. M. Anpo, Y. Yamada, S. Coluccia, A. Zecchina and M. Che, J. Chem. Soc., Faraday T rans. 1, 1989, 85, 609. T. Ito, T. Watanabe, T. Tashiro, M. Kawasaki, K. Toi and H. Kobayashi, in AcidÈBase Catalysis, ed. K. Tanabe, H. Hattori, T. Yamaguchi and T. Tanaka, Kodansha, Sapporo, 1988, p. 483. J. H. Lunsford and J. P. Jayne, J. Phys. Chem., 1965, 69, 2182. R. M. Morris and K. J. Klabunde, J. Am. Chem. Soc., 1983, 105, 2633. R. M. Morris, R. A. Kaba, T. G. Groshens, K. J. Klabunde, R. J. Baltisberger, N. F. Woolsey and V. I. Stenberg, J. Am. Chem. Soc., 1980, 102, 3419. J. V. Evans and T. L. Whateley, T rans. Faraday Soc., 1967, 63, 2769. Y. Yanagisawa, K. Takaoka, S. Yamabe and T. Ito, J. Phys. Chem., 1995, 99, 3704. A. A. Davydov, N. A. Rubene and A. A. Budneva, Kinet. Catal., 1979, 19, 776. R. S. C. Smart, T. L. Slager, L. H. Little and R. G. Greenler, J. Phys. Chem., 1973, 77, 1019. S. Coluccia and A. J. Tench, Proc. Int. Congr. Catal., 7th, 1981, B, 1154. M. C. Paganini, M. Chiesa, E. Giamello, S. Coluccia, G. Martra, D. M. Murphy and G. Pacchioni, Surf. Sci., 1999, 421, 246. G.-W. Wang and H. Hattori, J. Chem. Soc., Faraday T rans. 1, 1983, 80, 1039. T. Shido, K. Asakura and Y. Iwasawa, J. Catal., 1990, 122, 55. T. Shido and Y. Iwasawa, J. Catal., 1993, 140, 575. A. A. Tsyganenko and V. N. Filimonov, J. Mol. Struct., 1973, 19, 579. H. Tsuji, T. Shishido, A. Okamura, Y. Gao, H. Hattori and H. Kita, J. Chem. Soc., Faraday T rans., 1994, 90, 803. H. Takahashi, K. Okuda and T. Nitta, Bull. Chem. Soc. Jpn., 2000, 73, 315. V. N. Parmon and V. S. Zakharenko, Z. Phys. Chem., 1999, 213, 9. A. M. Volodin, Catal. T oday, 2000, 58, 103.

Phys. Chem. Chem. Phys., 2001, 3, 1108È1113

1113