Comparative Study of Mo2O x S y (cys) 22-Complexes as Catalysts for

Mar 19, 1997 - Comparative Study of Mo2OxSy(cys)22- Complexes as Catalysts for Electron ... At high light intensity, efficiencies for electron transfe...
0 downloads 0 Views 204KB Size
Langmuir 1997, 13, 1571-1576

1571

Comparative Study of Mo2OxSy(cys)22- Complexes as Catalysts for Electron Transfer from Irradiated Colloidal TiO2 to Acetylene Robert R. Kuntz Department of Chemistry, University of Missouri, Columbia, Missouri 65211 Received August 30, 1996. In Final Form: January 2, 1997X A comparative study of the complexes Mo2OxSy(cys)22- (x ) 4, 3, 2; y ) 4 - x) as catalysts for electron transfer from irradiated TiO2 to acetylene is reported. These catalytic species all show similar behavior with respect to variations in intensity, loading, and pH. Their ability to facilitate electron transfer from TiO2 to substrate increases in the order Mo2O2S2(cys)22- > Mo2O3S(cys)22- > Mo2O4(cys)22- . At high light intensity, efficiencies for electron transfer from the colloidal TiO2 surface are comparable to those obtained with Pt under similar conditions. Efficiencies increase with decreased light intensity, giving product yields which account for 7-9% of the incident light at the lowest intensities studied. All three catalysts produce H2 in N2- or C2H4-saturated solutions, but only Mo2O2S2(cys)22- produces H2 in the presence of C2H2. The effects of pH, temperature, intensity, and loading on the catalytic process are consistent with a proposed mechanism in which the catalytic site associated with the TiO2 surface promotes the transfer of electrons to the bound substrate. For each complex, a single catalytic site appears to be responsible for all reduction processes. A comparison of the photocatalytic properties of species containing the Mo(V)2 core is included.

Introduction Molybdenum-sulfur compounds participate in a variety of catalytic activities ranging from hydrodesulfurization (HDS) in the petroleum industry1 to nitrogen fixation by the nitrogenase enzyme.2 The mechanism and structure of the high-temperature HDS catalysts are a subject of much study,1,3-11 and are generally thought to involve MoS3 and MoS2 microcrystallites which utilize bridging sulfur atoms in a Mo cluster. MoS2 deposited on semiconductors also shows electrocatalytic and photocatalytic activity for hydrogen evolution in aqueous media.12 A salient feature of nitrogenase is its ability to reduce many substrates other than nitrogen under mild conditions. The active site of the enzyme contains Mo-Fe-S clusters about which considerable experimental data has been gathered.13 Recent structural information on the FeMo-cofactor14 shows the Mo has an octahedral geometry with three sulfurs bridging to the iron center. Whether direct binding to Mo is involved in the catalytic process is still unclear. In attempts to understand the nitrogenase mechanism, Schrauzer et al.15-20 constructed an artificial active site utilizing molybdenum complexes with sulfur-containing X Abstract published in Advance ACS Abstracts, February 15, 1997.

(1) Gates, B. C.; Katzer, J. R.; Schuit, G. C. A. In Chemistry of Catalytic Processes; McGraw Hill: New York, 1979; Chapter 5, pp 390-445. (2) Molybdenum Enzymes; Spiro, T. G., Ed.; John Wiley: New York, 1985. (3) Prins, R.; DeBeer, V. H. J.; Somorjai, G. A. Catal. Rev.sSci. Eng. 1989, 31,122. (4) Stiefel, E. I.; Halbert, T. R.; Coyle, C. L.; Wei, L.; Pan, W.-H.; Ho, T. C.; Chianelli, R. R.; Daage, M. Polyhedron 1989, 8, 1625. (5) Drew, M. G. B.; Mitchell, P. C. H. Polyhedron 1989, 8, 1814. (6) Komatsu, T.; Hall, W. K. J. Phys. Chem. 1991, 95, 9966. (7) Komatsu, T.; Hall, W. K. J. Phys. Chem. 1992, 96, 8131. (8) Muller, B.; van Langeveld, A. D.; Moulijn, J. A.; Knozinger, H. J. Phys. Chem. 1993, 97, 9028. (9) Startsev, A. N. Catal. Rev.sSci. Eng. 1995, 37, 353. (10) Weber, Th.; Muijsers, J. C.; Niemantsverdriet, J. W. J. Phys. Chem. 1995, 99, 9194. (11) Hong, Z.; Regalbuto, J. R. J. Phys. Chem. 1995, 99, 9452. (12) Sobczynski, A. J. Catal. 1991, 131, 156. (13) For a review, see: Burgess, B. K. Chem. Rev. 1996, 96, 2983 and references therein. (14) Kim, J.; Rees, D. C. Science 1992, 257, 1677. (15) Schrauzer, G. N.; Doemeny, P. A. J. Am. Chem. Soc. 1971, 93, 1608.

S0743-7463(96)00850-5 CCC: $14.00

ligands and a reducing agent (NaBH4 or Na2S2O4) to investigate reduction of N2, C2H2, nitriles, and many other substrates. The success of these model active site studies led us to investigate the possibility of using conduction band electrons from band-gap irradiation of TiO2 as a source of the reducing equivalents for electron transfer to substrates through the Mo center. Band-gap irradiation of TiO2 produces electron-hole pairs which are trapped at the surface of small particles. Recent reviews describing electron-hole trapping and transfer dynamics and energetics have appeared.21-23 The holes are initially trapped either as subsurface oxygen anion radicals24 or on surface OH groups,25 and have the potential to oxidize most organic species. Trapped electrons may be transferred directly to substrates or to catalytic surface sites for further participation in reduction processes. The mechanism of photocatalytic reactions at semiconductor surfaces has been the subject of several reviews.22,23,26 Previous studies in this laboratory have utilized C2H2 reduction to evaluate catalytic effectiveness with a variety of Mo-based electron transfer catalysts.27-32 Generally, the systems containing Mo-S linkages are much more (16) Schrauzer, G. N.; Doemeny, P. A.; Kiefer, G. w.; Frazier, R. H. J. Am. Chem. Soc. 1972, 94, 3604. (17) Schrauzer, G. N.; Doemeny, P. A.; Frazier, R. H., Jr.; R. H.; Kiefer, G. W. J. Am. Chem. Soc. 1972, 94, 7378. (18) Schrauzer, G. N.; Kiefer, G. W.; Doemeny, P. A.; Kisch, H. J. Am. Chem. Soc. 1973, 95, 5582. (19) Schrauzer, G. N.; Kiefer, G. W.; Tano, K.; Doemeny, P. A. J. Am. Chem. Soc. 1974, 96, 641. (20) Robinson, P. R.; Moorehead, E. L.; Weathers, B. J.; Ufkes, E. A.; Vickrey, T. M.; Schrauzer, G. N. J. Am. Chem. Soc. 1977, 99, 3657. (21) Hagfeldt, A.; Gra¨tzel, M. Chem. Rev. 1995, 95, 49. (22) Linsebigler, A. L.; Lu, G.; Yates, J. T. Jr. J. T. Chem. Rev. 1995, 95, 735. (23) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. Rev. 1995, 95,69. (24) Micic, O.; Zhang, Y.; Cromack, K. R.; Trifunac, A. D.; Thurnauer, M. C. J. Phys. Chem. 1993, 97, 7277. (25) Martin, S. T.; Herrmann, H.; Choi, W.; Hoffmann, M. R. J. Chem. Soc., Faraday Trans. 1994, 90, 3315. (26) Fox, M. A.; Dulay, M. T. Chem. Rev. 1993, 93, 341. (27) Cai, Z.-S.; Kuntz, R. R. Langmuir 1988, 4, 830. (28) Al-Thabaiti, S.; Kuntz, R. R. Langmuir 1990, 6, 782. (29) Lin, L.; Kuntz, R. R. Langmuir 1992, 8, 870. (30) Lin, L.; Al-Thabaiti, S.; Kuntz, R. R. J. Photochem. Photobiol., A: Chem. 1992, 64, 93.

© 1997 American Chemical Society

1572 Langmuir, Vol. 13, No. 6, 1997

Kuntz

efficient than similar compounds with Mo-O bonding. MoS42- and Mo2S4(S2C2H4)22- are nearly as effective as Pt for the conversion of C2H2 to C2H4 and C2H6.29 The current study was designed to determine the effect of µ-sulfido bonding on the photocatalytic efficiency of Mo(V)2 complexes. A homologous series of compounds of the formula Mo2OxSy(cys)22- (x ) 2, 3, 4; y ) 4 - x) were evaluated in their ability to transfer electrons from irradiated colloidal TiO2 to C2H2. Experimental Section Sodium di-µ-oxobis(L-cysteinato)oxomolybdate(V) pentahydrate, Na2Mo2O4(cys)2‚5 H2O (4,0), was prepared according to the literature33 by direct reaction between Mo2O42- and cysteine. Sodium µ-oxo-µ-sulfidobis(l-cysteinato)oxomolybdate(V) tetrahydrate, Na2Mo2O3S(cys)2‚4H2O (3,1), was prepared by a similar literature method34 using H2S in 3 M HCl to replace one of the µ-oxo groups with sulfur. The final compound, sodium di-µsulfidobis(l-cysteinato)oxomolybdate(V) tetrahydrate, Na2Mo2O2S2(cys)2‚4H2O (2,2), was prepared by reduction of sodium molybdate with H2S, as described previously.34 Elemental analysis was consistent with the formulations. The absorption characteristics of these compounds are similar to those reported.34 Preparation of transparent sols of TiO2 and their characterization has been described previously.27,28 To 8 mL of a stock solution containing 5-8 nm diameter TiO2 particles at pH 1.6 was added 3 mL of 2% poly(vinyl alcohol) (PVA), 3 mL of H2O, 4 mL of MeOH, and 2 mL of an aqueous solution of Na2 Mo2OxSy(cys)2 at the appropriate concentration. pH was adjusted with NaOH or HCl. The final solution has a volume of 20 mL and contained the appropriate amount of catalyst, 1.0 g/dm3 of TiO2 and 3.0 g/dm3 of PVA in a 4/16 (v/v) MeOH/H2O solution. PVA served doubly as a colloidal stabilizer and as a sacrificial electron donor. Solutions were deaerated and saturated with C2H2, C2H4, or N2 by bubbling for 30 min. Irradiations were carried out with a 200 W high-pressure Hg lamp filtered with water and a 320 nm cutoff filter. The lamp intensity was monitored periodically with ferrioxalate actinometry. The intensity available to the sample in the 320-360 nm range, where the TiO2 sol absorbs most of the radiation, was either 0.49 × 10-3 or 0.79 × 10-3 Einsteins/h, with the change accompanying lamp replacement and refocusing. The 50 mL reaction cell contained 20 mL of solution and was placed in a thermostated jacket. A few experiments, performed for quantum efficiency determination, were carried out with a 313 nm narrow band filter. Neutral density filters were inserted in the lamp beam for intensity dependence measurements. C2H2 was prepurified by bubbling through three water traps prior to introduction into the reaction vessel. All other reagents were the highest purity commercially available and were used without further purification. Product formation was monitored by analysis of the head space gases in the photolyzed vessel. The vessel was equipped with an evacuable calibrated sampler into which aliquots of the head space gases could be expanded without contaminating the photolysis mixture with O2. Aliquots were withdrawn from the sampler by gas syringe and analyzed chromatographically. A 6 ft × 1/4 in. Porapak N column was used for hydrocarbon products (He carrier, TCD), and a 5 ft × 1/4 in. Molecular Sieve 5A column for H2 analysis (N2 carrier, TCD). Because of the change in intensity necessitated by lamp replacement during the course of this investigation, product yields were normalized to the total incident intensity in the 320-360 nm range. Duplicate experiments normalized in this manner gave a typical reproducibility of 10-15%.

Results The absorption spectra of the three catalysts (Figure 1) illustrate minimal light absorption beyond the 360 nm (31) Lin, L.; Kuntz, R. R. J. Photochem. Photobiol., A: Chem. 1992, 66, 245. (32) Kuntz, R. R. J. Photochem. Photobiol., A: Chem. 1994, 84, 75. (33) Melby, L. R. Inorg. Chem. 1969, 8, 349. (34) Ott, V. R.; Swieter, D. S.; Schultz, F. A. Inorg. Chem. 1977, 16, 2538.

Figure 1. Absorption spectra for Mo2O4(cys)22- (4,0), Mo2O3S(cys)22- (3,1), and Mo2O2S2(cys)22- (2,2) in CH3OH. Table 1. Effect of pH on Normalized Electron Transfer Yieldsa catalyst 4,0

3,1

2,2

pH

H2b

totalc

H2b

totalc

H2b

totalc

3 4 5 6 7 8 10

0.022 0.052 0.052 0.046 0.050 0.036 0.020

0.44 0.61 0.76 1.17 1.24 1.30 1.22

0.074 0.098

1.89 2.07

0.924 0.980

3.10 3.29

0.096

2.33

0.058 0.032

2.13 1.93

0.504 0.276 0.176 0.094

3.38 3.35 2.50 2.26

a All solutions have a loading of 26 catalyst molecules/TiO 2 particle. Yields are reported as the percent of electrons transferred b c per photon in the 320-360 nm range. %(H2) × 2. Electron yields are calculated as (%(H2) × 2 + %(C2H4) × 2 + %(C2H6) × 4).

band-gap threshold for this preparation of TiO2. Blank experiments with catalysts, but without TiO2 , show no photoreduction of C2H2 even in the absence of the screening effect of TiO2. Ethylene is the only acetylene reduction product formed by photolysis of the standard reactant mixture without catalyst. In the presence of catalyst, C2H6 and H2 are also produced, and the C2H4 yields are altered. As in previous studies, product formation is characterized by a catalyst “preparation period”, after which the yields of all products increased linearly with time for a substantial period. To minimize analysis errors, a linear least squares fit of the production data after the preparation period was used as the effective photocatalytic rate. pH Effect. The formation rates of reduction products as a function of pH are shown for the three catalysts in Figure 2. The Mo2O2S22- species is the more effective catalyst over the entire range. Several distinctions among the catalysts appear: (1) only 2,2 is effective at H2 production; (2) 2,2 becomes effective at lower pH than the other members of the group; and (3) above pH 6, all members of the comparison group give approximately equal efficiency for the four-electron reduction process to form C2H6. The normalized yield of electrons transferred for each of the three catalysts is listed in Table 1. Loading Effect. Increasing the concentration of catalyst generally increases the yield of reduction products for all three catalysts. Figure 3 shows these data presented as a ratio of the number of catalyst molecules/ TiO2 particles where the TiO2 particle concentrations were calculated from an average diameter of 6 nm,27,32 a density of 3.8 g/cc, and a concentration of 1.0 g/L. Even low catalyst loading led to the evolution of H2 and C2H6 and greatly increased the rate of production of C2H4. A notable

Mo2OxS6(cys)22-

Figure 2. Effect of pH on the normalized yields of H2 (top), C2H4 (middle), and C2H6 (bottom). Loading was fixed at 26 catalyst molecules/TiO2 particle. The normalized yield is the production rate (%) per incident photon in the 320-360 nm range at full lamp intensity: (9) 4,0 catalyst; (b) 3,1 catalyst; (2) 2,2 catalyst.

exception is 4,0, for which the C2H4 yield appears to be relatively independent of catalyst loading and very similar to its value in the absence of catalyst. H2 yields are significant only for 2,2, for which H2 production increases with loading. Ethane production follows a similar profile for all three catalysts, again increasing with loading. C2H4, the major reduction product, shows a differential behavior with loading. 4,0 is not effective in C2H4 production, while 3,1 exhibits a maximum near a loading of 20-30 catalyst molecules/TiO2 particle, and 2,2, while giving considerable scatter in yields, appears to plateau above a loading of about 10 molecules of catalyst/TiO2 particle. For comparison purposes, subsequent studies were conducted at 26 molecules of catalyst/TiO2 particle. Intensity Effect. The effect of incident intensity in the 320-360 nm region on the normalized product yield is shown in Table 2 for the three catalysts. The general trend is to increase quantum efficiencies for C2H2 reduction as intensity is lowered, consistent with a competition between electron transfer to the catalyst and electronhole recombination. At higher intensities, this dependence is relatively weak for 3,1 and 2,2. A noticeable difference is the independence of the C2H4 yield on incident intensity for 3,1. Hydrogen production, while not of great significance except for 2,2, shows the opposite dependence on intensity, increasing as intensity increases but also showing a reversal of trend for 2,2. Temperature Effect. A typical Arrhenius plot is shown in Figure 4 for 4,0 in the 25-45 °C range. Error

Langmuir, Vol. 13, No. 6, 1997 1573

Figure 3. Loading effect on the normalized yields for H2 (top), C2H4 (middle), and C2H6 (bottom). pH was fixed at 6. Yields were normalized to incident intensity in the 320-360 nm range: (9) 4,0 catalyst; (b) 3,1 catalyst; (2) 2,2 catalyst. Table 2. Effect of Intensity on the Normalized Photoreduction Yielda intensity (320-360 nm) (µEinsteins/h)

H2

yield C2H6

C2H4

“electrons”b

2-

50.0 159 493

Mo2O4(cys)2 0.010 0.648 0.014 0.486 0.023 0.286

0.687 0.303 0.136

4.07 2.21 1.17

85.8 248 780

Mo2O3S(cys)220.001 0.713 0.364 0.041 0.793 0.162 0.048 0.794 0.161

2.87 2.32 2.33

50.0 159 493

Mo2O2S2(cys)220.110 1.58 0.288 0.364 1.12 0.192 0.252 1.14 0.148

4.56 3.74 3.38

a All solutions have a loading of 26 catalyst molecules/TiO 2 particle. Yields are reported as molecules produced per 100 photons in the 320-360 nm range. b Electron yields are calculated as (%(H2) × 2 + %(C2H4) × 2 + %(C2H6) × 4).

bars on this plot reflect a (15% deviation in the initial rate data, which is within the typical reproducibility range. The data for all three catalysts are summarized in Table 3. Within the error of measurement, the two µ-sulfido catalysts show very modest increases in efficiency with rising temperature. 4,0, however, increases efficiency by about 60% over this temperature range. As indicated by the error bars in Figure 4, the data do not distinguish differential activation energies for products from a given catalyst.

1574 Langmuir, Vol. 13, No. 6, 1997

Kuntz Table 5. Turnover Studiesa catalyst

irradiation time (h)

H2

Na2Mo2O4(cys)2 Na2Mo2O3S(cys)2 Na2Mo2O2S2(cys)2

33 h 47 h 41 h

1.2 2.9 15.7

turnover C2H4 C2H6 “electrons”b 14.1 63.3 32.7

6.1 9.9 5.7

55 (65%) 172 (40%) 119(35%)

a Turnover numbers are reported as moles of products/mole of catalyst. b Number in parentheses is the average percent of original activity over the irradiation time indicated.

Table 6. Comparison of Substrate Efficienciesa catalyst

Figure 4. Arrhenius plots for the 4,0 catalyst. Error bars reflect (15% variation in individual yield determinations. Table 3. Arrhenius Activation Energies for the Na2Mo2OxSy(cys)2 Catalysts activation energy (kJ/mol) catalyst

H2

C2H4

C2H6

“electrons”

Na2Mo2O4(cys)2 Na2Mo2O3S(cys)2 Na2Mo2O2S2(cys)2

20.0 1.7 6.8

25.1 13.2 8.6

12.7 (-1.3) 15.8

19.1 9.2 7.5

Table 4. Photonic Efficiency for Photocatalytic Reduction of C2H2a photonic efficiency catalyst

H2

C2H4

C2H6

“electrons”

Na2Mo2O4(cys)2b Na2Mo2O3S(cys)2c Na2Mo2O2S2(cys)2b

0.026 0.027 0.310

1.01 1.95 2.85

1.40 0.649 0.709

7.67 6.56 9.21

a Solutions were at pH 6 with a loading of 26 with irradiation through a 313 narrow band-pass filter. b Intensity is 21.5 µEinsteins/h. c Intensity is 38.0 µEinsteins/h.

Photonic Efficiency. The photonic efficiency,35 describing the effectiveness of light conversion to products in heterogeneous, light-scattering systems, may be considered a lower limit for the more commonly used quantum yield in homogeneous photochemical systems. The photonic efficiency of the present system was determined using a 313 nm narrow band-pass filter to insure all incident light was absorbed (or scattered) by the system. Intensities incident on the system were 21.5 or 38 µEinsteins/h, as determined by ferrioxalate actinometry. The results of these measurements appear in Table 4. The last column in Table 4 demonstrate that all of these catalysts are reasonably efficient at a low intensity of 313 nm light. Under this condition, the preferred C2H2 reduction product for 3,1 and 2,2 is the two-electron product, C2H4, while 4,0 gives better yields of the four-electron product, C2H6. Only 2,2 shows significant activity for H2 production. If that contribution to the “electron” column is removed, all three catalysts have similar overall efficiencies. Turnover Studies. Long term irradiations were performed with all three catalysts to insure that the process was catalytic rather than stoichiometric and to determine the catalyst stability. These data, shown in Table 5, suggest that all three systems are catalytic with respect to all products. Also, all of the catalysts decompose during the process, with 4,0 having the greatest stability. Reactions Involving Other Substrates. Solutions which were saturated with C2H4 or N2 instead of C2H2 also exhibited photocatalytic activity with H2 as the principal reduction product (Table 6). These data clearly indicate that C2H2 is not required for H2 production and (35) Serpone, N.; Terzian, R.; Lawless, D.; Kennepohl, P.; Sauve´, G. J. Photochem. Photobiol., A: Chem. 1993, 73, 11.

Na2Mo2O4(cys)2 Na2Mo2O4(cys)2 Na2Mo2O4(cys)2 Na2Mo2O3S(cys)2 Na2Mo2O3S(cys)2 Na2Mo2O3S(cys)2 Na2Mo2O2S2(cys)2 Na2Mo2O2S2(cys)2 Na2Mo2O2S2(cys)2

saturation gas N2 C2H4 C2H2 N2 C2H4 C2H2 N2 C2H4 C2H2

normalized yield of products H2 C2H4 C2H6 “electrons” 0.13 0.103 0.023 0.394 0.409 0.048 1.37 1.33 0.252

n.d.b

n.d. 0039 0.286 0.136 n.d. n.d. 0.026 0.794 0.161 0.002 n.d. 0.011 1.14 0.148

0.26 0.28 1.17 0.79 0.87 2.33 2.74 2.68 3.38

a Studies were conducted at pH 6 and a loading of 26. Yields are reported as moles/Einstein (320-360 nm) at full lamp intensity. b n.d. ) none detected.

apparently competes for the same catalytic site(s), since the H2 yield decreases in the presence of C2H2. Ethylene is a poor substrate for the catalyst, since it is not reduced to C2H6 and does not compete effectively with H2 production. In the absence of C2H2, the ability of the catalyst to promote H2 production increases significantly with S content in the bridging positions. Discussion Control experiments in the absence of TiO2 demonstrate that direct photolysis of the catalyst does not contribute to the observed products. While cooperative effects between the excited catalyst and TiO2 cannot be eliminated, the similarity of results at 313 nm, where the catalysts do not contribute substantially to absorbed light, and low-intensity irradiation above 320 nm suggest that this effect is minimal. Consequently, photoreduction is initiated by band-gap irradiation of TiO2. Electron transfer from the colloidal surface to the catalytic site and subsequently to the substrate is a complex process from which the efficiency-limiting step cannot be extracted from product analysis alone. The initial events following electron-hole formation, (h+, e-), are a competition between charge carrier recombination and trapping. This is followed by a slower competition between trapped carrier (h+tr, e-tr) recombination and interfacial charge transfer with the recombination process dominating at higher intensities.23 This rationale is consistent with intensity data for the catalysts shown here, which show a greater efficiency for interfacial charge transfer at lower intensity due to a decrease in the (h+tr, e-tr) recombination rate. In the current system, some electron transfer occurs directly to C2H2 from the e-tr, as evidenced by C2H4 formation in the absence of catalyst. For 3,1 and 2,2, C2H4 production increases dramatically when catalyst is added, suggesting that the two-electron reduction process occurs dominantly through the catalytic site. In the case of 4,0 the evidence is not so clear, but we believe from studies of other Mo2O42+ complexes36 that ethylene is produced through the catalytic site. The relatively high photonic efficiencies found for all three systems are (36) Kuntz, R. Unpublished data.

Mo2OxS6(cys)22-

Langmuir, Vol. 13, No. 6, 1997 1575

consistent with intimate contact between the catalyst and the colloidal surface. Mechanistically, the “preparation period” is seen as the initial reduction of the catalyst on the TiO2 surface shown in generic form in reactions 1 and 2,

TiO2 + Mo2OxSy(L)22- H TiO2/Mo2OxSy(L)22- (1) TiO2/Mo2OxSy(L)22- + nhν (+ PVA) f TiO2/Mo2(L)2 (+ PVAox) (2) where L ) cysteine, and PVAox is the oxidized form of PVA following scavenging of h+, and TiO2/Mo2(L)2 refers to the active catalytic form of TiO2/Mo2OxSy(L)2 with undetermined oxidation state. The reversibility of reaction 1 is indicated by the general increase in product formation with loading and the lack of a sharp falloff seen in similar systems.28,30,36 C2H2 is a better substrate for the active site than either C2H4 or the H2 precursor, as shown in Table 6. The decrease in H2 yields in the presence of C2H2 suggests that the same catalytic site is involved in H2 formation and C2H2 reduction. The only plausible source of H2 in these systems is the solvent system (H2O/MeOH). The catalytic site is believed to be involved in the H2 production, since alternative mechanisms such as the oxidative elimination of hydrogen from methanol (reaction 3) should not be altered by the

TiO2(2h+) + CH3OH f CH2O + H2 + TiO2

(3)

presence of C2H2 nor the identity of the catalyst. In fact, only 2,2 gives good yields of H2 in the presence of C2H2, and these yields are maximized at low pH. This suggests that H+ rather than H2O is the H2 precursor, as described in reaction 4. For each of the catalyst systems, H2 yields

TiO2/Mo2(L)2 + nhν + 2H+ + PVA f TiO2/Mo2(L)2 + H2 + PVAox (4) are similar for either N2- or C2H4-saturation, indicating that these gases do not compete with solvent for the catalytic site. In the absence of C2H2, however, H2 yields increase with S content. In the presence of C2H2, the occupancy of the catalytic site(s) is dominated by C2H2 and its partial reduction

TiO2/Mo2(L)2 + C2H2 H TiO2/Mo2(L)2(C2H2) (5) products. Because C2H4 is not a very good substrate for these catalytic sites, it appears that bound C2H2 is retained at the site until either two-electron or four-electron reduction occurs.

TiO2/Mo2(L)2(C2H2) + 2e-tr + 2H+ + PVA f TiO2/Mo2(L)2(C2H4) + PVAox (6) TiO2/Mo2(L)2(C2H4) f TiO2/Mo2(L)2 + C2H4 (7) TiO2/Mo2(L)2(C2H4) + 2e-tr + 2H+ + PVA f TiO2/Mo2(L)2 + C2H6 + PVAox (8) The effects of pH, loading, and temperature and intensity data for C2H2 reduction products can be explained by this general mechanism where the relative product mix is indicative of competition between reactions 7 and 8. There is no information on the nature of substrate binding to the catalyst. However, it seems unlikely that

C2H4 would be bound in its molecular form, since it is a poor substrate for the catalytic site. The intensity dependence for C2H2 reduction is consistent with competition between (h+tr, e-tr) recombination and interfacial electron transport to the catalytic site. Unlike these products, however, H2 photonic yields tend to decrease at lower intensities for all three catalysts. This behavior suggests a higher order dependence on light intensity for H2 formation. One possibility is that the catalytic process for H2 formation involves hydride intermediates, TiO2/Mo2L2(H)x, with subsequent lightinduced formation of H2 either by H atom formation or direct H2 elimination (reaction 9).

TiO2/Mo2(L)2(H)2 + hν f TiO2/Mo2(L)2 + H2 (9) These possibilities have not been studied in detail. Complexes 3,1 and 2,2 show very low activation energies, suggesting there are no significant electron transfer barriers in reactions 1-9 for these species, while complex 4,0 exhibits a higher barrier. In examining the cause of this difference, attention is drawn to the difference in pH dependence of C2H6 formation for the three catalysts. The TiO2 surface is positively charged in acid solution but becomes neutral around pH 4.6,37 and negative in basic solutions. However, the redox potential of the trapped electrons increases with pH.38 To a first approximation, the three catalytic species with similar structure and charge should have approximately equal affinity for the colloidal surface under all pH conditions. At acid pH, 4,0 is inefficient at the four-electron transfer process while 3,1 and 2,2 have similar higher efficiencies. As pH rises, all three catalysts attain nearly equal yields. Therefore, on the basis of the increasing reducing potential as pH increases, a good candidate for the dominant activation barrier step is the interfacial electron transfer process. This circumstantial evidence is consistent with the study of redox potentials of complexes containing the Mo2OxSy core,34,39 which shows that the ease of reduction increases in the sequence Mo2O4 < Mo2O3S < Mo2O2S2 < Mo2OS3 < Mo2S4. The electrochemical study of Mo2OxSy(cys)22- complexes in aqueous buffer34 characterizes the reduction process as a single four-electron reduction from Mo(V)2 to Mo(III)2 and indicates that the reduced product becomes increasingly unstable with S substitution for O. However, in nonaqueous media (DMSO), reduction occurs by successive one-electron steps from Mo(V)2 to Mo(IV)2 and the stability increases in the order Mo2O4 < Mo2O3S < Mo2O2S2. Neither of these studies correspond exactly to the photocatalytic conditions. Whereas the medium is aqueous, the electrons probably are transferred sequentially. Also, the combined effects of surface interaction and of bound C2H2 are unknown. It is true that the catalyst long term stability matches the order in the aqueous electrochemical study, and this may be the primary cause of catalyst degradation. The current evidence is not adequate to identify the nature of the reduced Mo2 species responsible for the catalysis. Electrochemical studies would suggest that Mo(III)2 is the lowest feasible oxidation state for the catalytic species, but electron transfer could also occur from mixed valence species. Under optimal conditions of low light intensity, the photonic efficiency for all of these catalysts is quite high, (37) Brown, G. T.; Darwent, J. R.; Fletcher, P. D. I. J. Am. Chem. Soc. 1985, 107, 6446. (38) Duonghong, D.; Ramsden, J.; Gra¨tzel, M. J. Am. Chem. Soc. 1982, 104, 2977. (39) Schultz, F. A.; Ott, V. R.; Rolison, D. S.; Bravard, D. C.; McDonald, J. W.; Newton, W. E. Inorg. Chem. 1978, 17, 1758.

1576 Langmuir, Vol. 13, No. 6, 1997

Kuntz

Table 7. Comparison of the Photonic Efficiencies for Several Mo(V)2 Speciesa photonic efficiency catalys 2+

Mo2O4 Mo2O4(dedtc)2b Mo2O4(cys)22Mo2O3S(cys)22Mo2O2S2(cys)22Mo2S4(S2C2H4)22PtCl62-

C2H4

C2H6

“electrons”

ref

0.29 0.41 0.15 0.41 0.58 0.75 0.26

0.14 0.09 0.07 0.08 0.08 0.10 0.18

1.14 1.18 0.60 1.19 1.73 1.90 1.24

29 36 this datac this datac this datac 29 31

a Data are reported for optimum loading and pH 6 conditions at full light intensity, with yields reported as percent of incident light giving products. b dedtc ) diethyl dithiocarbamate. c In order to compare yields, it was necessary to renormalize the current data to account for light intensities in the 320-385 nm range used for earlier determinations (decreased by a factor of (intensity at 320385 nm)/(intensity at 320-360 nm) ) 1.95).

utilizing 7-9% of the light to transfer electrons to substrate. A comparison of photonic efficiencies observed for C2H2 reduction with Mo(V)2 catalysts on TiO2 under similar conditions appears in Table 7. Since the overall efficiency is related as much to hole-scavenging efficiency by PVA as to electron transport to the catalyst, these yields are better appreciated by comparison to those for Pt. Photoreduction of PtCl62- onto the colloidal surface results in formation of Pt islands,40-42 which serve as excellent electron traps and reducing sites. We see that all the complexes in this study are comparable to Pt in efficiency of C2H2 reduction. In the absence of C2H2, however, Pt is approximately twice as efficient as Mo2S4(S2C2H4)22- in the production of H2.30 Using this ratio and the highest yield for H2 production in the presence of N2 from Table (40) Kiwi, J.; Gra¨tzel, M. J. Phys. Chem. 1984, 88, 1302. (41) Furlong, D. N.; Wells, D.; Sasse, W. H. F. J. Phys. Chem. 1985, 89, 626. (42) Furlong, D. N.; Wells, D.; Sasse, W. H. F. J. Phys. Chem. 1985, 89, 1922.

6, we would project an electron transfer efficiency of 2(1.37 × 2) ) 5.48% for the Pt system. Thus, using the Table 6 data for the 4,0, 3,1, and 2,2 catalysts suggests they are 20%, 40%, and 60%, respectively, as efficient as Pt for trapping electrons from the TiO2 surface. The very similar distribution of products and yields for 2,2 and Mo2S4(S2C2H4)22- (Table 7) would suggest that the increase in efficiency with substitution of S for O is primarily due to the µ-sulfido groups rather than the ModS groups. Conclusion The Mo2OxSy(cys)22- species are all effective as catalysts to transfer electrons from TiO2 to C2H2. The efficiency of electron transfer is greater for the µ-sulfido complexes than the µ-oxo counterparts, and efficiency increases with increasing S substitution. The difference between catalyst efficiency diminishes at low light intensity, suggesting that interfacial electron transport is the rate-limiting factor. Mo2O4(cys)22- is similar in efficiency to the Mo2O42aquo ion, suggesting that the Mo(V)2 core rather than the ligand system is primarily responsible for the catalytic activity. These catalysts are comparable to Pt in their ability to accept electrons from the colloidal surface and transfer them to a substrate. The increase in electron transfer efficiency afforded by increasing sulfur content is somewhat negated by the higher degradation rate of the S-containing catalysts in long term studies. The high efficiency of electron transfer gives hope, however, that complexes with high efficiency and which exhibit greater stability and selectivity for specific reduction processes can be designed. Acknowledgment. This work was supported in part by NSF Grant CTS-9224318, which is gratefully acknowledged. LA960850L