2572
J. Phys. Chem. 1984,88, 2512-2515
on the basis of the concept that the interaction of alkenes with the trapped e- (Ti3+)and h+ (OH) pairs involves electron transfer from alkenes to the OH moiety, Le., the sites trapping positive holes.3 As described above, water molecules and not surface OHgroups are responsible for the photohydrogenation of alkenes (or alkynes) in the present system. A similar conclusion has been reached by Domen et al.,I3 who investigated the photodecomposition of water on Ni0-SrTiO3 catalyst. According to their views, the oxidation of water proceeds via the attack of the radicals formed from the OH- groups where the equilibrium H20ad, H+ OH- is set up. The OH- groups responsible for the reaction should be distinguished from the stable surfce OH- groups remaining after evacuation at a high temperature. In the present system it appears that the proton exists not as such but in combination with the 012- ions on the surface, Le., the acidic OHgroups as proposed by Munuera et ale7 By losing H+ these OH-
+
(13) K. Domen, S.Naito, T. Onishi, K. Tamaru, and M. Soma, J . Phys. Chem., 86, 3657 (1982). (14) M. Anpo, N. Aikawa, and Y. Kubokawa, J . Phys. Chem., in press.
groups would revert to the 012- ions. By repetition of the cycle, the supply of H+ to the active sites would proceed. Although the nature of the OH- groups associated with the oxidation reactions is unclear, it appears almost certain that the OH- groups remaining after the removal of H+ from HzO is responsible for the oxidation reactions. The significance of the supply of water molecules in the progress of the reaction is obvious from the results shown in Figure 3, where the product yields level off after a certain reaction time where the H 2 0 pressure is very low. Furthermore, it should be noted that the initial rate of the photoreaction is higher the lower the H 2 0 pressure. This suggests that the existence of coordination vacancies, Le., of vacant sites for the adsorption of alkenes (or alkynes), is necessary for the occurrence of the photoreaction, in agreement with the reaction scheme described above. Registry No. HzO, 7732-18-5; TiOz, 13463-67-7; C H z C H , 74-86-2; CH,C=CH, 74-99-7; C2HSCSCH, 107-00-6; CH2=C=CH2, 463-490; CHz=CHz, 74-85-1; c - C ~ H ~75-19-4; , CH4, 74-82-8; n-C4H8, 25 167-67-3; C3H8, 74-98-6; CzHs, 74-84-0; CHjCH=C=CH2, 59019-2; CdH10, 106-97-8; CHz=CHCH=CH2, 106-99-0; ( C H , ) Z h C H z , 115-11-7; CH$H=CHz, 115-07-1; DzO,7789-20-0.
Quantum Chemical and ‘*O, Tracer Studies of the Activation of Oxygen in Photocatalytic Oxidation Reactions Masakazu Anpo,* Yutaka Kubokawa, Department of Applied Chemistry, College of Engineering, University of Osaka Prefecture, Sakai, Osaka 591, Japan
Tsuneo Fujii,* and Satoshi Suzuki Department of Engineering Chemistry, Faculty of Engineering, Shinshu University, Wakasato, Nagano, Nagano 380, Japan (Received: September 9, 1983)
Mechanisms for the photocatalytic activation of oxygen have been investigated by EPR techniques and lSO2tracer experiments as well as by the quantum chemical approach for the structure and reactivity of intermediate species. UV irradiation of porous Vycor glass in the presence of a mixture of I8O2and CO leads to the formation of C160180 alone. Addition of CO to the 0,- species leads to the disappearance of their EPR signals and to the formation of COz. These results indicate that the oxygen in the gas phase, and not lattice oxygen, is incorporated into COzmolecules. The C N D 0 / 2 and INDO calculations of the 0,- anion radical indicate that the 0-0 bond of the oxygen is weakened through the interaction of it with the 0hole center and also the total energy of the 03-species shows two minima at the LOO0 bond angle (e) of 60 and 120’. From these results it is concluded that oxygen molecules undergoes photoactivation, Le., weakening of the 0-0 bond, which results in the formation of active oxygen species responsible for photooxidation reactions on surfaces.
Introduction Photocatalytic oxidation is one of the most important fields in photocatalysis on metal oxides.] The nature of the active oxygen species responsible for photocatalytic oxidation has been discussed by a number of workers.’” Pichat et al. have suggested that 0,(1) (a) K. Tanaka and G. Blyholder, J . Phys. Chem., 76, 1807, 3184 (1972); (b) M. Formenti and S.J. Teichner, “Catalysis”, Vol. 2, C. Kemball, Ed., The Chemical Society, London, 1978, p 87; (c) Y. Kubokawa and M. Anpo, Shokubai, 22, 189 (1981); (d) R. I. Bickley, “Catalysis”, Vol. 5 , G . C. Bond and G. Webb, Ed., The Royal Society of Chemistry, London, 1982, p 308. (2) K. M. Sancier and S. R. Morrison, Surf. Sci., 83, 29 (1979). (3) J. Cunningham, B. Doyle, and E. M. Leahy, J . Chem. SOC.,Faraday Trans. 1 , 75, 2000 (1979). (4) S.Yoshida, Y. Matsumura, S.Noda, and T. Funabiki, J . Chem. Soc., Faraday Trans. 1,17, 2237 (1981). (5) H. Courbon, M. Formenti, and P. Pichat, J. Phys. Chem., 81, 550 (1977). (6) Y. Kubokawa, M. Anpo, and C. Yun, Proc. In?. Congr. Catal. 7th, 1170 (1981).
species are intermediates for the hydrocarbon photooxidation as well as the photoinduced 1602-1802 isotopic exchange reaction over Ti02.s Recently, we have investigated the photooxidation of alkenes over porous Vycor glass (PVG) by EPR measurements of the photoformed oxygen species as well as by analysis of the reaction products. We showed that the 0- hole center, formed at low-coordination sites by a charge-transfer process, i.e.
reacts easily with oxygen to form an unstable 03-species, which is closely associated with the photooxidation of alkenes as well isotopic exchange reaction over as the photoinduced 1602-1802 PVG.6 It seems important to clarify the mechanism by which the 03species brings about the oxidation of hydrocarbons. For this purpose, the present work was undertaken to investigate the photooxidation of a simple molecule such as CO using l8OZtracer. Furthermore, quantum chemical studies of the structure and
0022-3654/84/2088-2512$01.50/00 1984 American Chemical Society
Oxygen Activation in the Photocatalytic Oxidation Reactions
+
R
*
3
1
The Journal of Physical Chemistry, Vol. 88, No. 12, 1984 2573
- o ~ / - oxygen,0.12 torr "0"
C O , 0031 torr
I
off
03;
1
v
- -.-.- .
.
w
llghtoff
.
on
I
- 0 0 .c
0
LT
reactivity of the 03species have been carried out in order to obtain a theoretical foundation for the proposed reaction mechanism. The problem has been dealt with by Kazansky et al. who investigated the structure and reactivity of Z 0 2 peroxy-type radicals (Z = F, H, 0-, et^.).^
Experimental Section Photoreactions. Details of the experiments were described previously.6 The gases were 99.5% pure (Takachiho Kogyo Co.) (90 at. %, E. and were used without further purification. 1802 Merck Co.) was used. UV irradiation was carried out at 77 or a t 300 K with a 500-W high-pressure mercury lamp through a water filter. The isotopic content of oxygen was determined with a Shimazu quadrupole mass spectrometer (Maspeq-070) a t 300 K. No oxygen isotopic exchange reaction occurred on PVG at room temperature under dark conditions. The BET surface area of PVG was about 160 m2/g. EPR measurements were carried out a t 77 K with a JES-ME-1 (X-band). Mn2+ ions in MgO powder were used for g value and sweep calibrations. The reaction products were separated by low-temperature fractional distillation and analyzed with a gas chromatograph or a quadrupole mass spectrometer. Quantum Chemical Calculations. The calculations for O,O-, and 03were carried out according to the C N D 0 / 2 and INDO programs on a HITAC M-240 computer.8 T a t calculations gave essentially the same results as was reported by Pople et al? Figure 1 shows the model of the 03-anion radical for which the calculations were carried out. Results and Discussion Photooxidation of CO Molecules with I8O2. It has been reported previously that the photoinduced oxygen isotopic exchange reaction, which proceeds via 03-intermediates, is efficiently inhibited by the addition of alkenes, and the reaction of 03-species with alkenes under dark conditions gives products similar to those observed for the photooxidation of alkenes, leading to the conclusion that photooxidation of alkenes proceeds via 03-intermediates.6 The effects of the addition of C O molecules instead of alkenes upon the photoinduced 1602-1802 isotopic exchange reaction have been investigated. As seen in Figures 2 and 3, the 1602-1802 isotopic exchange reaction at 300 K is inhibited by the presence of a small amount of CO molecules. Simultaneously, formation of C02 is observed. The lack of adsorption of C O on PVG even under UV irradiation excluded the possibility that the inhibition is caused by competitive adsorption of C O and oxygen. Thus, it is concluded that the inhibition by added C O is attributable to the removal of the 03species by C O by a mechanism similar to that observed for the photooxidation of alkenes. Such a high reactivity of the 03-species toward CO molecules is confirmed by the following 1 8 0 2 tracer experiments: UV irwas carried out at 160 radiation of PVG in the presence of 1802 (7) N. D. Chuvylkin, G. M. Zhidomirov, and V. B. Kazansky, Kinet. Katal., 19, 561 (1978). (8) J. A. Pople and G. A. Segal, J. Chem. Phys., 44, 3289 (1966); J. A. Pople, D. L. Beveridge, and P. A. Dobosh, ibid., 47, 2026 (1967). (9) J. A. Pople and D. L. Beveridge, 'Approximate Molecular Orbital Theory", McGraw-Hill, New York, 1970.
400
200
Figure 1. Model of 03anion radical.
Time,
600
800
1000
rnin
Figure 2. Effect of the addition of CO upon the photoinduced 1602-1802 isotopic exchange reaction over PVG at 300 K. (The symbol C means peak height of the molecule with a given mass.)
-
y2.0 0
oxygen masure ~0.1Torr
P
c 0
a 100 n'100
50 50
Reaction
200 250 ' 300 400 200' '250 ' 300 "400 '
.
'
Time , min
Figure 3. Photoinduced 1602-1802 isotopic exchange reaction over PVG at 300 K. (The symbol C means peak height of the molecule with a given mass.) 2.0109
I
Esr spectrum of oxygen anion rujtcds on PVG ai 77 K Initio1 oxygen wesoure - 0 9 Torr UV inndintion time -30
Min
2.0053
0; anion radicals
T
2 003,
Figure 4. EPR spectra of oxygen anion radicals on PVG at 77 K and the effect of addition of CO upon them (CO pressure, ca. 1.0 torr).
K, and then the EPR spectrum was measured at 77 K. As shown in Figure 4, the formation of 03-species as well as 02-species is detected. After introduction of C O to the system at 77 K, the temperature was raised to room temperature. The 03-EPR signals disappeared instantaneously, while the 02-signals remained unchanged in a manner similar to that observed with alkenes.1° Simultaneously, C 0 2 formation occurred. The C 0 2 formed was found to consist of C 1 6 0 1 8 0alone. As shown in Figure 5 , UV and a small irradiation of PVG in the presence of a mixture of 1802 amount of CO leads to the formation of Cl60l8Oalone, with no dilution of l 8 0 content in the oxygen taking place. Figure 5 also (10) This disappearance is not due to warming but to the reaction with CO. species has relative long lifetimes in the As described previously,6 the 03presence of oxygen at 300 K. Its concentration decreases to half its initial value after 7 rnin at 0.24 torr and after 31 min at 0.96 torr of oxygen.
2514
The Journal of Physical Chemistry, Vol. 88, No. 12, 1984
Anpo et al.
Q 3
5
.
0
P 2 14.0
UV Irradiation time
-55.20
? 21
P
E
-55.30
0 c
0
k-
-55,40
,min
Figure 5. Photoformation of C160180 over PVG in the presence of a mixture of CO and l8OZat 300 K ('*OZ pressure, 0.18 torr; CO pressure, 0.06 torr).
I I
I
1-55.379
40
60
80 Angle
100
120
140
160
180
( e ) , degree
Figure 6. Total energy of the 03anion radical obtained by CNDO/2
calculations. shows that no oxygen isotopic exchange reaction occurs in the presence of sufficient C O molecules. As expected from such behavior no change in the l60content of the gas-phase oxygen occurred during the photoinduced oxygen exchange reaction. These results clearly indicate that oxygen in the gas phase, and not lattice oxygen, is incorporated into C 0 2 molecules. Thus, it is concluded that the oxygen molecule undergoes photoactivation, Le., weakening of the 0-0 bond through its interaction with the 0- hole center, resulting in the formation of active oxygen species. Teichner et al." and Pichat et aL5J2 have proposed that the photooxidation of alkenes over T i 0 2 is caused by the photoformed atomic oxygen species, although the proposed processes for its formation are different. It is worth noting that Vedrine and Nacchache have suggested the possibility of the activation of oxygen by its interaction with the 0- hole centers formed on a zeolite catalyst by y4rradiati0n.I~ 1.1 1.2 1.3 1.4 1.5 The features observed with the 0- hole centers on PVG are quite Bond distance (R12) , A different from those with the 0- species formed by the adsorption of oxygen on ZnO, where its removal by C O is very e a ~ y . ~ ~ , ~ ~ , ~ ~ Such difficulty in the removal of the 0- hole centers on PVG is The contour lines of total energy of 0; supported by the lack of formation of C 0 2 under UV irradiation in the presence of C O alone. It is well-known that the reactivity Figure 7. The contour lines of total energy of the 0,- anion radical obtained by CNDO/2 calculations. of the 8-species is quite different on various oxides, being strongly dependent on the contribution of covalency to the bond between 60 and 120°, the minimum value at 0 = 60' being much lower. the 0- radicals and the surfaces.I6-l8 It is worth noting that the This suggests that there are two configurations, A (8 = 60') and photoformed 0- hole centers on supported transition-metal oxides B (8 = 120°), for the 03-species, the former being more stable. having a surface metal-oxygen double bond character such as It should be noted that configuraion A has no spin population in VzOs and Moo3 are reactive toward C O molecules, resulting in T-AO,in contrast to the appreciable population with configuration COS formation in the presence of C O a10ne.l~ B. With both configurations a significant weakening of the 0-0 Quantum Chemical Studies of 03Intermediates. The 03bond is caused by the formation of the O< species, its extent being species may be considered to have the geometry shown in Figure larger for configuration A. The extent of weakening is similar 1, where O1refers to the 0- hole center, while O2and O3represent to that reported by Kazansky et al.7 the oxygen m ~ l e c u l e .The ~ total energy as well as the electron The total energy of the 03-species has been calculated as a and spin populations in the T - A 0 have been calculated as a function of R12and R23 values with a fixed value for the angle function of the angle 0, taking fixed values for R12and R23 from 0. The results are represented as the contour lines (Figure 7), C N D 0 / 2 and INDO methods. The calculation has been carried which indicate points of equal energy for the 0; species. It should out at intervals of 10". The principal results of the C N D 0 / 2 and be emphasized that as the oxygen molecule approaches the 0INDO calculations are available as supplementary material. (See hole center, Le., with decreasing bond distance, R12,the nuclear paragraph at end of text regarding supplementary material.) distance of the 0-0 bond of oxygen, R23, becomes larger than These results are in good agreement each other. As seen in Figure that of a free oxygen molecule (ca. 1.20 A). The energy minimum 6, the total energy of the 0,- species shows two minima at B = is observed at R I 2= R2, = 1.280 A. This distance is larger than the internuclear distance of oxygen, in agreement with the conclusion that the 0-0 bond in the 03-species is weakened as (1 1) P. C. Gravelle, F. Jullet, P. Meriaudeau, and S . J. Teichner, Proc. Intn. Congr. Catal., 5th, 2, 1011 (1972). compared to that in oxygen molecules. (12) J. M. Herrmann, J. Disdier, M. N. Mozzaneja, and P. Pichat, J. As described previously, the g values for the 03-species on PVG Catal., 60, 369 (1979). are g, = g2 = 2.003 and g3 = 2.008K6 Che et aL20 and Kazansky (13) J. C. Vedrine and C . Naccache, J. Phys. Chem., 77, 1606 (1973). et aLzl have found that the 03-species on V205/Si02 and (14) K. Tanaka and K. Miyahara, J . Phys. Chem., 78, 2303 (1974). (15) K. M. Sancier, J . Catal., 9, 331 (1967). TiO2/§iO2 exhibits g values similar to those with PVG and (16) J. H. Lunsford, Catal. Rev., 8, 135 (1973). (17) M. Che and A. J. Tench, Adu. Catal., 31, 77 (1982). (18) M. Iwamoto, J . Synth. Org.Chem., Jpn., 40, 694 (1982). (20) B. N. Shelimov, C. Naccache, and M. Che, J . Catal., 37,279 (1975). (19) M. Anpo, I. Tanahashi, and Y. Kubokawa, J . Phys. Chem., 84, 3440 (1980); M. Anpo, I. Tanahashi, and Y. Kubokawa, ibid., 86, 1 (1982).
(21) W. W. Nikisa, S . A. Surin, B. N. Shelimov, and V. B. Kazansky, React. Kine?. Catal. Lett., 1, 141 (1974).
J. Phys. Chem. 1984, 88, 2575-2581 proposed that they have a T-shaped configuration with two equivalent oxygen atoms. These 03-species on V205/Si02, TiOz/Si02, and PVG are all very reactive toward the oxygen isotopic exchange reaction. It is worth noting that the g values for the 0; species on MgO are quite different from those values, having a classical ozonide ion structure, and that the 03-species on MgO is very stable and less reactive.I6 From the stability of configuration A described above, it appears that the shape of such a T-shaped configuration proposed for the 03-species may be a regular triangle. Thus, quantum chemical calculations of the 03intermediates support the weakening of the 0-0 bond of the oxygen molecule through its interaction with 0- hole centers, a result which has been suggested from the results of lSO2tracer experiments. It should be noted that there is a source of electrons in the neighborhood of the photoformed 0- hole center, since it constitutes the charge-transfer excited complex, [Me("-')+-O-] *, described previously? The population of such electron is expected to enhance back-donation of electron to 7r* orbitals of the 03species with the consequent much easier rupture of the 0-0bond, leading to the formation of active oxygen species. In other words, the energy produced when photoformed electrons and holes re-
2575
combine is used for 0-0 bond fission. Such a situation has been suggested for the photoinduced metathesis reaction of propene over supported M o o 3 catalyst, in which C=C bond cleavage of propene by the excited complex [Mo5+-0-]* is the important step in the reaction.22 To settle the problem, further quantum chemical studies using a more advanced model are in progress in this laboratory.
Acknowledgment. The authors are much indebted to Professor M. Che of Universite de Paris VI for his very helpful advice about the structure of the 0,- species. Registry No. CO, 630-08-0; 02, 7782-44-7; '*02, 14797-71-8; C'60'80,18983-82-9; O