J. Phys. Chem. 1989, 93, 1542-1548
1542
TABLE IV: Most Relevant Data of the Two Lowest Energy Minima Found in the Analysis of the Interactions between NaCDC Micelle and A and B Enantiomers of BR" enantiomer rLI &, & tr tv t7 energy A
B
21 -34
-28 -43
20 -31
14.9 14.9
-2.8 -2.7
0.3 0.7
-24.0 -22.6
"The rotation angles, the translations, and the energy are given in deg, A, and kcal/mol, respectively. For the interatomic interactions a cutoff distance of 7 8,was assumed, together with angular and translational increments for BR progressively decreasing from 20° to 1O and from 0.5 to 0.1 A, respectively. All the six-dimensional parametric space was explored without resorting to minimization procedures, which often suffer from the drawback of not being able of locating all the minima, in such a way as to identify the deepest ones. The van der Waals energy map presents a limited number of acceptable minima. The deepest ones for the A and B enantiomers of BR are defined in Table IV. The situation corresponding to the two minima is shown in Figures 9 and 10. The interaction with the A enantiomer is favored by about 1.4 kcal/mol. Although the semiempirical potentials do not allow the attainment of quantitative agreement, reliable results can be achieved for a qualitative c o m p a r i ~ o n .The ~ ~ energy difference between the two minima is not so high that an exclusive selection of the A enantiomer from the chiral helices can be supposed, and hence, a preferential selection very likely occurs. By inspection of the shortest intermolecular distances it appears that the best interactions arise from hydrophobic contacts involving mainly NaGDC and BR methyl groups, the helical methyl group most protruding outside (Ck9)being particularly active. NaDC behaves likewise when it interacts with a cholestane spin label' and with hydrocarbons,] so that there is more evidence of the similarity between
the micellar structures of NaDC and NaGDC. In both minima BR subsides with its concave side on the cylindrical surface of the helix, in front of the A, B, and C rings of the steroid skeleton of three adjoining NaGDC anions and the side chains of other three anions shifted along OZ of 11.7 8,. However, there are no good contacts between the atoms of BR and those of the side chains. The two wings, which characterize the butterfly-shaped BR molecule, have opposite screw sense in the A and B enantiomers. The A screw sense coincides with that of the NaDC, RbDC, and NaGDC helices, thus allowing that this conformer fits these helices better than B. As far as NaTDC is concerned, it is difficult to infer an interaction model with BR for the reasons previously mentioned. Since it is very probable that NaTDC micelles are highly hydrated,27in accordance with the models of Figure 5 which display large empty regions,8v26no definite models can be assumed if the bound water is taken into account. On the other hand, the CD experimental data seem to point out that BR is capable also of interacting by means of polar interactions with the hydrophilic surface of the NaTDC helix, if one of our models, at least, is correct. Acknowledgment. We thank Prof. L. Mazzarella of the University of Napoli and Prof. V. Albano of the University of Bologna for data collection. This work was sponsored by the Italian Consiglio Nazionale delle Ricerche-Progetto Finalizzato Chimica Fine e Secondaria, and by the Italian Minister0 della Pubblica Istruzione. Registry No. NaGDC, 16409-34-0; NaTDC, 1180-95-6; NaGDC. 3/2H20,117800-82-5; Na, 7440-23-5. Supplementary Material Available: Listing of observed and calculated structure factors for NaGDC-1S H 2 0 (6 pages). Ordering information is given on any current masthead page.
Semiconductor Effect on the Selective Photocatalytic Reaction of a-Hydroxycarboxylic Acids H. Harada,* T. Ueda, Department of Chemistry, Faculty of Science and Engineering, Meisei University, Hino, Tokyo 191, Japan
and T. Sakata* Institute for Molecular Science, Myodaiji, Okazaki 444, Japan (Received: June 7, 1988)
Photocatalyticand photoelectrochemical reactions of a-hydroxycarboxylic acids were compared for various types of semiconductor electrode (TiO,, CdS, SrTi03, and ZnO) and suspension of particulate semiconductor (TiO,, CdS, MoS,, and ZnS). These reactions were found to depend strongly on the type of semiconductor studied. In the cases of Pt/CdS and ZnS photocatalysts, the hydroxy group of the acids was oxidized selectively into the corresponding keto acids, whereas in the cases of Pt/Ti02, decarboxylation took place in addition to dehydrogenation. The same dependence was observed in the photoelectrochemical reactions with semiconductor single-crystalelectrodes. For the TiOz electrode, the reaction depends strongly on pH, whereas it does not for CdS. The results of pH effects, electrochemical reaction with various metal electrodes, and Fenton reaction in a homogeneous solution suggest the importance of adsorption of the reactants on the semiconductor and metal surfaces for the selective reaction.
Introduction Photocatalytic reactions of particulate semiconductors have been studied not only from the viewpoint of solar energy conversion but also from that of organic synthesis.' From the latter viewpoint, the specificity of a photocatalyst is important to control the reaction path. For instance, the Pt/CdS photocatalyst can decompose ethanol into hydrogen and acetaldehyde but cannot (1) Fox, M. A . N o w . J . Chim.1987, 11, 129.
0022-365418912093-1542$01.50/0
decompose acetic acid, because the oxidation potential of acetic acid is located more positive than the valence band edge of CdS. On the other hand, Pt/TiO, photocatalyst can decompose both of these compounds.2 We have already briefly reported the fact that the photocatalytic reaction of lactic acid depends clearly on the kind of semiconductors: In the case of Pt/CdS photocatalyst, pyruvic acid is produced in the liquid phase, while in the case of (2) Sakata, T ; Kawai, T. J . Synth. Org. Chem. Jpn. 1981, 39, 589.
0 1989 American Chemical Society
The Journal of Physical Chemistry, Vol. 93, No. 4, 1989 1543
Photocatalytic Reaction of a-Hydroxycarboxylic Acids
TABLE I: Photocatalytic Reaction Products from Lactic Acid-Water Solution'sb product/mmol cat.c
HZ
co2
CH3CHO
CZHSOH
CH$OOH
CH3COCOOH
Pt/TiO,
1.21 1.20
1.43 0.015
1.08 d
0.047 d
0.151 d
0.02 0.8
Pt/CdS
"Reference 3. bIrradiation with 1-kW Xe lamp (under 500-W operation) for 4 h. c300 mg of catalyst was used. dBelow the detection limit.
Pt/Ti02 photocatalyst acetaldehyde is p r o d ~ c e d . This ~ reaction is one of the clear examples in which reaction selectivity is controlled by the kind of semiconductors. Relating with this reaction, Feiner et al. reported photoelectrochemical oxidation of formates and some hydroxycarboxylates at CdS photo anode^.^ And Cuendet and Gratzel also succeeded in an efficient reduction of pyruvate into lactate under illumination of aqueous suspensions of Ti02 p ~ w d e r . ~ Here we report a detailed experimental study on the semiconductor dependence of photocatalytic reactions of lactic acid and other a-hydroxycarboxylic acids by using various particulate semiconductors such as Ti02, CdS, ZnS, and MoS,. Besides photocatalytic reactions with particulate semiconductors, photoelectrochemical studies have been done by using several photoanodes such as TiO,, CdS, SrTi03, and ZnO. Relating with the electrocatalysis on the surfaces, electrochemical studies with various electrodes such as Pt, Au, and glassy carbon have been carried out at various pH values. On the basis of the results of the semiconductor and pH dependence, we will discuss the reaction mechanism and several factors that control the above photocatalytic reactions. Experimental Section Materials. Various kinds of powdered semiconductors were used for the experiments: CdS, Katayama Kagaku, cubic and average size 0.6 pm; Ti02, Furuuchi Kagaku, rutile and average size 0.5 Mm; MoS2, Kanto Kagaku or Soekawa Rikagaku; ZnO, Katayama Kagaku; ZnS, Nakarai Kagaku; and CdSe, Furuuchi Kagaku. As semiconductor electrodes, single crystals of CdS, Ti02, SrTiO,, MoS2, ZnO, and CdSe were used. As conductive electrodes, glassy carbon (Tokai Carbon) was used in addition to the noble metals of Pt and Au. As a-hydroxycarboxylic acids, glycolic acid (Wako), lactic acid acid (Aldrich) were used. (Wako), and 2-hydroxy-3-methylbutyric For loading Pt on semiconductors, K2PtC1, (Soekawa Rikagaku) and ethanol (Wako) were ~ s e d . ~Colloidal ,~ ZnS was prepared from ZnS04 (Wako) and Na2S (Wako).8 The pH of the solution was varied with NaOH. All commercial materials were of the highest grade and were used without further purification. Photocatalytic Reactions. The photocatalyst of Pt/Ti02 was prepared by depositing Pt photochemically on the surface of Ti02 particle^.^,^ Other photocatalysts were prepared by mixing semiconductor with 5% Pt black (Nippon Engelhard) in an agate m ~ r t a r . As ~ reactant mixtures, 50 mL of glycolic acid-water (1:lO in the ratio of volume content; hereafter the same abbreviation is used), lactic acid-water (l:lO), and 2-hydroxy-3methylbutyric acid-water (1:20) was used. After evacuation, the Pyrex glass bulb containing each photocatalyst (300 mg) suspended in reactant mixture was irradiated from the bottom with a 1-kW Xe lamp (Ushio Denki, under 500-W operation) or with a 500-W Xe lamp (Ushio Denki). In the case of Pt/MoS2 photocatalyst, Ag ion (AgN03 or Ag2S04) was added into the solution as an electron acceptor. (3) Harada, H.; Sakata, T.; Ueda, T. J. Am. Chem. SOC.1985, 107, 1773. (4) Feiner, A. S.; McEvoy, A. .I.; Vlachopoulos, N.; Gratzel, M. Ber. Bunsen-Ges. Phys. Chem. 1987, 91, 399. (5) Cuendet, P.; Gratzel, M. J . Phys. Chem. 1987, 91, 654. (6) Kraeutler, B.; Bard, A. J. J . Am. Chem. SOC.1978, 100, 2239. (7) Sakata, T.; Kawai, T. Chem. Phys. Lett. 1981, 80, 341. (8) Yanagida, S.;Azuma, T.; Sakurai, H. Chem. Lett. 1982, 1069. Yanagida, S.; Azuma, T.; Kawakami, H.; Kizumoto, H.; Sakurai, T. J . Chem. SOC.,Chem. Commun. 1984, 21. (9) Kawai. T.; Sakata, T. J . Chem. SOC.,Chem. Commun. 1980, 647. In the case of Pt/TiOl, the mixing method showed lower activity than the photochemical method. In both methods, the ratio of reaction products was not changed.
For the determination of quantum yield, a photocatalyst and lactic acid-water (1:l) stirred regularly in a Pyrex glass cylinder were irradiated from one side with monochromatic radiation from a 150-W Xe lamp/monochromator (Jobin Yvon HL) combination. Quantum yields were evaluated on the basis of the incident photon flux, which was measured by a thermopile (Eppley).*O Electrochemical Reactions. An In-Ga alloy was used to maintain Ohmic contact with each semiconductor electrode. The entire assembly, except for the electrode surface, was covered with Araldite (Ciba-Geigy) and Torr Seal (Varian). Current-potential curves were obtained by using a function generator (Hokuto Denko, HB-107), a potentiostat (Hokuto Denko, HA-104), and a X-Y recorder (Riken Denshi, D72-BP). Potentiostatic reactions were carried out with the potentiostat referred to SCE with a coulomb meter (Hokuto Denko, HF-201). As a supporting electrolyte, K2S04 or Na2S04was used and the solution was constantly stirred by a magnetic stirrer. The irradiation source for the photoreaction was a 500-W Xe lamp. The measurement conditions for dark reactions were the same as those in the illuminating case. Analyses. The gaseous reaction products were trapped at -196 "C (using liquid N2) or at about -50 OC (using a mixture of liquid N2 and ethanol) and analyzed by a quadrupole mass spectrometer (Anelva, AGA-360), as described previo~sly.~J'The amounts of gaseous products were determined from the volume content multiplied by the pressure, which was measured by a highly sensitive pressure gauge (Datametrics, barocel pressure sensor). In order to separate carbon dioxide and ethane, a NaOH trap was used. The liquid reaction products were filtered (by a MILLEX FILTER, 0.45 pm, Millipore) and analyzed by a steam carrier gas chromatograph (Ohkura Denki, Model 103, G-0 or 1-0 column) and by a liquid chromatograph (Shimadzu, LC4A, Shodex (2-81 1 column, measuring wavelength 210 nm). The amount of cadmium in the solution was determined by atomic absorption spectroscopy (with a Hitachi 208 atomic absorption spectrometer). Results Photocatalytic Reaction Products of a-Hydroxycarboxylic Acids. ( a ) Lactic Acid. As shown in Table I, clear differences in the reaction products were observed depending on the kind of semiconductor used. The gaseous products were H2and C 0 2 for Pt/Ti02 and only H2 for Pt/CdS. On the other hand, the main reaction product in aqueous medium was acetaldehyde for Pt/ Ti02, whereas it was pyruvic acid for Pt/CdS. These results are summarized as
for Pt/Ti02 CH,CH(OH)COOH
-
H2
+ C02 + CH3CHO
(1)
for Pt/CdS CH,CH(OH)COOH
4
H2 + CH3COCOOH
(2)
The production of ethanol and acetic acid can be explained by the decarboxylation of lactic acid and the oxidation of acetaldehyde in water, respectively.' CH,CH(OH)COOH C 0 2 + CH3CH20H (3) +
CH3CHO
+ H2O
4
H2
+ CH3COOH
(4)
The results in Table I should be considered as in an initial stage of the total reaction since a large excess of lactic acid, about 30 (10) Sakata, T.; Hashimoto, K. N o w . J. Chem. 1985, 9, 699. ( 1 1) Kawai, T.;Sakata, T. Nature 1979, 282, 283.
1544 The Journal of Physical Chemistry, Vol. 93, No. 4, 1989 TABLE II: Photocatalytic Reaction Products after Long Irradiation from Lactic Acid-Water Solution
Pt/Ti02 irradiat time/h product/mmol H2
co2
CH3CHO CH3CH2OH CH3COCOOH
16
21
27
16
21
27
4.76 4.70 2.86 0.087 0.17
6.74 6.32 4.02 0.102 0.21
9.26 8.48 6.04 0.12 0.35
4.46 0.160
5.91 0.179
7.77 0.303
a
a
u
a 4.73
a a
2.06
cat. H, Pt/Ti02 210 Pt/CdS 480
wavelenath/nm
Quantumvield
420 400 380 360 520 500 480 460 440
0.03 0.43 0.64 0.7 1 0.08 0.12 0.21 0.26 0.38
2(CH3)+2HCHOH
'Irradiated with 1-kW Xe lamp (under 500-W operation) for 5 h. b300 mg of catalyst was used. CBelowthe detection limit. mmol, was used as the reactant. We tested the effects of irradiation for long duration, as shown in Table 11. The amount of pyruvic acid increased slowly with irradiation time for Pt/Ti02. No acetaldehyde, however, was detected even after long irradiation for Pt/CdS. Therefore, it is concluded that Pt/CdS photocatalyst can oxidize only the hydroxyl group of lactic acid and never oxidize the carboxyl group. The pH of the lactic acid solution used for the above experiment was about 2. In the basic solution, the rates of the photocatalytic reactions were very slow, so a small amount of acetaldehyde and pyruvic acid were detected. The ratio of pyruvic acid to acetaldehyde increased with increasing pH. The pH effect will be discussed on later section. The quantum yields of hydrogen production were quite high as shown in Table 111. The wavelength dependence of the quantum yield indicates that the band-gap excitation of the semiconductor is required for the photocatalytic reaction. ( b ) Glycolic Acid. Table IV shows the photocatalytic reaction products of glycolic acid. No formaldehyde was detected for Pt/CdS. On the other hand, decarboxylation as well as dehydrogenation occurred for Pt/TiO,. These results suggest that the following reactions take place: for Pt/Ti02