J. Phys. Chem. lQ83, 87, 801-805
80 1
Hydrogen Generatlon by Photocatalytic Oxidation of Glucose by Platinized n-TiO, Powder Mlchael R. SI. John," Alan J. Furgala, and Anthony F. Sammells Institute of Gas Technology, Chicago, Illlnois 60616 (Received:April 28, 1982; In FIn8l Form: September 24, 1982)
The photocatalytic reaction of glucose in aqueous solution containing suspended platinized n-Ti02 (anatase) powder was found to occur in mildly acidic media, pH 4.5. The gaseous products were determined to be H2 and C 0 2when conducted under an inert atmosphere and C 0 2alone under an O2atmosphere. For a 2.8 m glucose solution and a temperature of 60 "C, the initial H2 to COz ratio was near 10 and decreased to a value near 3 for times greater than 100 h. Initial rates of gas evolution were near 1.0 mL/h (volume at STP) and declined with time. Mott-Schottky plots on flamed Ti02 electrodes determined in solutions containing glucose were found to yield flat-band potential values 210-mV more negative than measurements made in its absence; such a shift would enhance the capability for H2 evolution. Isotopic experiments using D20 were found to support the occurrence of conventional H2 evolution without glucose involvement.
The photoelectrochemical oxidation of organic compounds Introduction by platinized n-type titanium dioxide (hereafter, Pt/nThe conversion of solar energy to fuels and chemicals Ti02) powder has been shown to occur quite readily,'6-22 from illuminated liquid junction semiconductor devices is with the photo-Kolbe reaction of carboxylic acids being a promising method of solar energy utili~ation.'-~ The the most widely d i s c ~ s s e d . ' In ~ ~addition, ~ ~ ~ ~ ~a ~me~~ photoelectrochemical decomposition of water for hydrogen chanically prepared mixture of Pt/Ti02/Ru02, exhibiting generation has been of particular interest (reviews 5-7 and the ability to oxidize sucrose under illumination with the references therein). Of recent interest is the simplification generation of H2 and COz, was reported.20 Consequently, of the conventional photoelectrochemical cell through the the existing evidence suggested that the oxidation of gluuse of solution suspensions of semiconductor powders to cose and concurrent H2 evolution observed to occur eleccarry out photocatalytic or photoelectrosynthetic reactrochemically could be performed photoelectrochemically tions.'" The suspended particles have consisted of a single with an illuminated, metallized semiconductor powder. semiconductor alone,&12metallized semiconductor^,^^-^^ The results of the photoelectrochemical reaction of and p-n semiconductor the latter two cases aqueous glucose solution on illuminate Pt/n-Ti02 are recontaining the necessary ohmic contacts where required. ported in this paper. Evidence on the origin of the hyThe term photochemical diodes has been applied to medrogen evolved is also presented. tallized semiconductors and p-n semiconductor pairs on the basis of their fabrication.14J5 The photochemical Experimental Section diodes have been envisioned as short-circuited photoeMaterials. Titanium dioxide powder (reagent grade, lectrochemical cells. Baker Chemical Co.) was obtained in the anatase crystal Possible electrochemical routes for nonbiological constructure as determined by powder X-ray diffraction. The version of materials derivable from biomass to fuels and/or as received Ti02 was capable of passing through 20-25-~m chemicals are being explored, and glucose, the most comfilter paper which provides an upper limit to the initial mon biomass building block, has exhibited potential as an anode depolarizer for electrolytic hydrogen g e n e r a t i ~ n . ~ ~ particle size. The anatase was doped by reduction in a hydrogen atmosphere a t 650 OC for 8 h. X-ray powder diffraction after the reduction showed that the anatase (1) Bard, A. J. Science 1980, 207, 139. structure was maintained. (2) Bard, A. J. J. Photochem. 1979, 10, 59. Chloroplatinic acid hexahydrate (reagent grade, Baker (3) Bard, A. J. J.Phys. Chem. 1982, 86, 172. (4) Nozik, A. J. Annu. Rev. Phys. Chem. 1978, 29, 189. Chemical Co.), trisodium citrate dihydrate (reagent grade, (5) Harris, L. A.;Wilson, R. H. Annu. Rev. Mater. Sci. 1978, 8, 99. Baker Chemical Co.), ruthenium dioxide (Alfa Products), (6) Rajeshwar, K.; Singh, P.; DuBow, J. Electrochim. Acta, 1978,23, and Amberlite MB-3 ion-exchange resin (Sigma Chemical 1117. (7) Maruska, H. p.; Ghosh, K. Solar Energy 1978,20, 443. Co.) were used as received in the preparation of the Pt/ (8) Freund, T.;Gomes, W. P. Catal. Rev. 1969, 3, 1. n-Ti02 particles. (9) Frank, S. N.;Bard, A. J. J. Am. Chem. SOC.1977, 99, 303. The components of the reaction solutions, anhydrous (10) Frank, S. N.; Bard, A. J. J. Phys. Chem. 1977,81, 1484. (11) Watanabe, T.; Takizawa, T.; Honda, K. J. Phys. Chem. 1977,81, D-glucose (reagent grade, Baker Chemical Co.), sodium 1845. dihydrogen phosphate monohydrate (reagent grade, Baker (12) Miyake, M.; Yoneyama, H.; Tamura, H. Bull. Chem. SOC.Jpn. Chemical Co.), and deuterium oxide (either Alfa Products 1977,50, 1492. (13) Wrighton, M. S.; Wolczanski, P. T.; Ellis, A. B. J. Solid State or Aldrich Chemical Co.), 99.8 mol 5% deuterium) were also Chem. 1977,22, 17. used as supplied. Solutions were prepared with reagent (14) Nozik, A. J. U S . Patent 4094751, June 13, 1978. grade water provided by a Continental Milli Q/Millipore (15) Nozik, A. J. Appl. Phys. Lett. 1977, 30, 567. water purification system yielding water with a specific (16) Kraeutler, B.; Bard, A. J. J. Am. Chem. SOC.1978, 100, 2239. (17) Kraeutler, B.; Bard, A. J. J. Am. Chem. SOC.1978, 100, 5985. resistance > 1 MQ. (18) Izumi, I.; Dunn, W. W.; Wilbourn, K. 0.;Fan, F. F.; Bard, A. J. Preparation of Catalysts. Pt/n-Ti02 catalyst was preJ. Phys. Chem. 1980,84, 3207. pared in two steps. In the first a platinum colloid was (19) Izumi, I.; Fan, F. F.; Bard, A. J. J. Phys. Chem. 1981, 85, 218. (20) Kawai, T.; Sakata, T. Nature (London) 1980,286, 474. produced by a method similar to that described by (21) Kawai, T.; Sakata, T. J.Chem. Soc., Chem. Commun. 1980,694.
(22) Kraeutler, B.; Bard, A. J. Nouv. J. Chim. 1979, 3, 31.
(23) Metlee, H.;Otvos, J. W.; Calvin, M. Solar Energy Mat. 1981,4, 443. 0022-3654/03/2087-080 1$01.50/0
(24) St. John, M. R.; Furgala, A. J.; Sammells, A. F. J. Electrochem.
SOC.1981, 128, 1174.
0 1983 American Chemical Society
002
The Journal of Physical Chemistry, Vol. 87, No. 5, 1983
Q
k
GAS COLLECTION
St. John et al.
TABLE I: Effect of Catalyst on the Photoelectrochemical Results for the Reaction of GlucoseGb
irradn time, h
catalyst
I
Pt /n-Ti0 , Pt/n-TiO, Pt /n-Ti0 , n-TiO, suspended Pt suspendedPt t n-TiO, Pt/n-TiO,/RuO, Pt/n-TiO,/RuO,
I
Figure 1. Schematic of photoelectrolysis assembly.
G r a t ~ e l . ~In~the , ~ procedure ~ adopted, 250 mL of 9.76 X M H2PtC1, was mixed with 35 mL of 4.06 X M trisodium citrate which was refluxed a t 90 OC until the absorbance a t 450 nm was 0.50 (1 cm path length). A period of 12 h was required to reach this absorbance value. The solution was filtered through the Amberlite MB-3 ion-exchange resin to remove the ionic components leaving only the colloidal Pt. The colloidal solution gave a negative test for C1- ion when tested with AgN03. In the second step, the Pt was deposited onto the Ti02. The deposition was accomplished by subjecting a slurry of 100 mg of TiOz in 100 mL of Pt colloid to an ultrasonic cleaner for 30 min followed by evaporation of the solvent in an oven a t 85 "C. On the basis of the amounts of starting Pt and TiOz, Pt loading could be no greater than 12 wt % . In addition, several catalysts were prepared with the incorporation of R u 0 2 by mixing 17 wt % R u 0 2 into the Ti02prior to deposition of the Pt. Apparatus. The simple volumetric apparatus shown in Figure 1was used to measure the rate of gas evolution and for gas collection. The round-bottom Pyrex reaction flask was maintained at 60 f 5 OC in a water bath. The apparatus was allowed to thermally equilibrate for 0.5 h before illumination. Suspension of the particulate catalysts was provided by continuous stirring by a Teflon magnetic stirring bar. Illumination of the sample was provided by a Hg lamp (Ultra-Violet Products Blak-Ray Longwave) with appropriate filtering capable of yielding a rated intensity of 7 mW/cm2 at a wavelength of 366 nm. The aqueous reaction mixtures were prepared from 20 mL of 0.50 M NaH2P04(pH 4.5 and p250c = 1.0652 g/mL), 10.8 g of D-glucose, and 100 mg of catalyst. In the isotopic experiments performed, D20 was used in place of H20. In the majority of experiments, solutions were purged with either high-purity N2 or Ar prior to illumination. When experiments were conducted over an extended time, the system was periodically repurged in order to obtain more representative gas compositions for each period. Following illumination, the reaction assembly was allowed to equilibrate to room temperature and atmospheric pressure before measuring the volume of gas evolved. A sample of the gas was then collected and analyzed both quantitatively and qualitatively by mass spectrometry with a CEC 21-104 mass spectrometer. Analyses were performed according to ASTM procedure^.^^ Mott-Schottky Measurements. Capacitance measurements as a function of electrode potential were made with (25) Kiwi, J.; Borgarello, E.; Pelizzetti, E.; Visca, M.; Gratzel, M. Agnew. Chem., Int. Ed. Engl. 1980, 19,646. (26) Borgarello, E.; Kiwi, E.; Pellizzetti, E.; Visca, M.; Gratzel, M. Nature (London) 1981,289, 158. (27) ASTM D1137-53, 'Analysis of Natural Gases and Related Twes of Gaseous Mixtures by the Mass Spectrometer".
av rate of gas evolution,e mL/h
15.00 17.00 11.75 19.25 22.00 15.00
0.60 1.12 1.14 0.00
17.50
18.00
mol % mol % H,f CO,f 92.4 90.6 90.3
7.6 9.4 9.7
0.00 0.00
NAd NA NA
NA NA NA
0.34 0.62
c
C
93.2
6.8
a Temperature = 60 * 5 "C. Electrolyte: 2.8 m gluNot measured. NA = not cose in 0.50 M NaH,PO,. Percentages deapplicable. e Milliliter of gas at STP. termined from mass spectrometry.
17 -
- 15 E
3
i
0
>
/I--
7 -
, P 0
1
2
3
4
5
6
7
8
TIME, h
Flgure 2. Raw gas volume as a function of time. Data for the third entry in Table I. The line corresponds to a least-squares fit of the data with a correlation coefficient = 0.996.
an H P 4265B universal bridge operating at 1 kHz. Polycrystalline Ti02electrodes prepared by heating Ti foil (Alfa Products) in a natural gas/air flame were investigated in 0.50 M NaHzP04solutions with and without 2.8 m glucose. Powder X-ray diffraction of the flamed foil revealed only the rutile crystal structure.
Results Photocatalytic Activity in Glucose Solution. The results of illuminating the weakly acidic glucose solutions containing potential photocatalytic powders and their component parts are given in Table I. Presented as the first three entries in Table I are the results for the illumination of three different preparations of Pt/n-Ti02. As shown, in the presence of Pt/n-TiO, macroscopic gas evolution was observed to occur with rates near 1.0 mL/h (volume at STP) being typical. In all cases where gas evolution occurred, the raw gas volume increased linearly with time over the periods specified as shown for example in Figure 2. The composition of the initial evolved gas consisted of Hz and COz in a ratio of about 10 to 1. Contamination of the gas samples by air made it difficult to rule out the possibility of O2 evolution, but the measured O2 to Ar ratio or N2 to O2 ratio depending on the purge gas used was always typical of air within experimental error suggesting that no O2 evolution occurred. Some COPwas undoubtedly trapped in the aqueous phase, but this amount was calculated to be at most 10 mol % of the total C 0 2collected.2s The
Photocatalytic Oxidation of Glucose by Platinized n-TiO,
TABLE 11: Extended Illumination Experiments, Time Dependence of Evolved Gas Composition, and Average Gas Evolution Rate for Illuminated Pt/n-TiO, in 2.8 m Glucose in 0.50 M NaH,PO,
illumination time interval, h
av rate of gas evolution: mL/h
mol% H, in evolved gasb
mol% CO, in evolved gasb
0-11.8 11.8-34.4 34.4-78.8 78.8-100.4
Experiment 1 1.14 90.3 83.7 0.49 0.54 79.9 0.53 76.4
9.7 16.3 20.1 23.6
0.0-24.5 24.5-48.0' 48.0-113.5 113.5-153.9 153.9-176.8
Experiment 2 0.66 93.6 0.54 73.8 0.29 78.1 0.30 77.6 0.17 68.7
6.4 26.2 21.9 22.4 31.3
0.0-2 3.5 23.5-47.3 47.3-134.3 134.3-157.0
Experiment 3 0.68 90.5 0.65 88.7 0.32 79.0 0.28 74.2
9.5 11.3 21.0 25.8
Mol % based o n mass spectroma Gas volume at STP. etry results. Leakage of air into the gas sampling bulb used t o collect the gas contained in the reaction cell occurred while the sample for this interval was waiting for mass spectrometry analyses. Therefore, the gas evolution rate is unaffected but the H, and CO, mol ?hare unreliable for this interval. As each time interval is an independent experiment, the others were unaffected by this leakage.
predominance of H2 suggests that oxidized fragments of glucose remain in solution; the composition of the liquid phase was not determined, however. Each component of the photocatalyst was tested independently for evidence of activity (Table I). Neither nTiOz or Pt alone showed any evidence of gas evolution under illumination. Mixtures of suspended Pt plus n-TiOz, i.e., not treated by ultrasonic bath followed by evaporation, also exhibited no activity. The oxygen evolution catalyst, Ru02, was thought to improve oxidation kinetics,20 but photocatalysts prepared with R u 0 2 in this work showed neither an enhanced rate of gas evolution or an increase in C 0 2evolution (Table I) over those photocatalysts prepared without it. That the presence of glucose is a necessary requirement for reaction was demonstrated by illuminating the Pt/nT i 0 2 suspension in 0.50 M NaH,P04 without glucose for 21.75 h with no detectable gas evolution. The activity of the Pt/n-TiO, particles used in this blank was ensured by testing a portion of the particles derived from the same batch on the glucose reaction with positive results. Extended Illumination. Three separate experiments were performed with illumination periods greater than 100 h. Table I1 lists the average rate of gas evolution and the composition of the evolved gas over selected time intervals (28) The estimated value of 10 mol % (molesof C 0 2 in solution/moles
of COz in gas phase X 100%) was based on the following reasoning. First,
the amount of COzdissolved in pure water at 20 "C was calculated with Henry's law and the partial pressure of COz above the solution determined by the measured amount of COBin the gas phase and found to be near 20 mol % for all samples. The effect of higher temperature (-25 "C) and particularly the high salt and glucose concentrations is to reduce this value. Based upon the COz solubilities given by Linke and Seidell (Linke, W. F.; Seidell, A. 'Solubilities of Inorganic and Metal Organic Compounds"; Van Nostrand: New York, 1958; 4th ed, pp 459-86), the COzsolubility should be reduced by greater than a factor two at the salt and glucose concentration used. Therefore, a value of 10 mol % was conservativelychosen for the amount of C02remaining in the electrolyte.
The Journal of Physical Chemistry, Vol. 87, No. 5, 1983
803
TABLE 111: Reaction of Glucose o n Pt/n-TiO, Photochemical Diodes under an 0, Atmosphere with and without Illumination,a Gas Evolution Rates and Composition av gas evolution timeof rate,c mol% mol% exptl conditions expt, h mL/h 0, e CO, e
0 , atm;b no illumination 0, atm; with illuminat ionf
17
-0.38d
92
8
25
-0.34
64
36
a Common experimental conditions: catalyst, 100 mg; electrolyte, 20 m L of 2.8 m glucose in 0.50 M NaH,PO,; temperature, 65 "C. Solution was black at end of experiment. Volume at STP. Negative sign indicates volume decrease. e Gas analyzed by mass spectrometry. Illumination of Blak-Ray; 7 mW/cm* rating.
for these three experiments. In between time intervals, the reaction vessel was repurged with either N2 (experiment 1) or Ar (experiments 2 and 3). Therefore the gas composition is representative of only a single interval and does not represent a cumulative composition. Although the general trend in the average rate of gas evolution for an interval is a decline as a function of time, the progression toward lower rates varies erratically among the three experiments conducted. The decline in the amount of total gas evolved with time is found to be associated with a decline in hydrogen production since the C 0 2 evolution is roughly constant. Averaging the CO, evolution rates over the intervals contained in experiments 1 , 2 , and 3 (Table 11) yields values of 0.11 (0.02), 0.06 (0.01), and 0.07 (0.01) mL/h, respectively [volumes at STP, standard deviations in parentheses, and the second interval in experiment 2 was not included in the average because of its inaccuracy due to leaking]. The relative amount of C02increases with time resulting in a change of the H2 to C02 ratio from approximately 1 0 1 in the initial interval to a relatively stable value of 3:l in later intervals. This increase in the relative amount of C 0 2 evolved would be expected on the basis of two facts. First, the electrochemical oxidation of glucose is known to proceed with the formation of carboxylic acids, e.g., gluconic or g l ~ c a r i c . ~Second, ~ ? ~ ~ as discussed earlier, the photoKolbe decarboxylation reaction is known to occur quite readily.16J7r22Therefore, as the photoelectrochemical oxidation of glucose proceeds, more carboxylic acid groups are expected to form allowing the photo-Kolbe reaction to occur and produce more C02. Because of the high glucose concentration in the solution, such an explanation is plausible only if the decarboxylation reaction occurs at a rate faster than the time required for the newly formed carboxylic acid to diffuse into the bulk solution. This would suggest that the glucose molecule and its oxidized derivatives are rather tightly bound. Effect of O2 on Glucose Reaction. The presence of O2 is expected to exhibit some effect on the reaction based on prior literature. The catalytic oxidation of carbohydrates to organic acids by 0, in alkaline solution using a P t catalyst is w e l l - k n ~ w n ,and, ~ ~ *since ~ ~ P t is present in these experiments, oxidation could possibly take place even though the solution is mildly acidic. Electrochemically, the reduction of O2 to H 2 0 zshould occur before the reduction of H+ to H2. Under such circumstances, the 129) Fedoronko, M. A d u . Carbohydr. Chem. Biochem. 1974,29,107. (30) Heyns, K.; Paulsen, H. A d u . Carbohydr. Chem. 1962, 17, 169. (31) Dirkx, J. M. H.; van der Baan, H. S. J . Catal. 1981, 67, 1.
St. John et al.
TABLE IV: Comparison of Calculated and Measured H/D Ratios in Hydrogen Evolved during Electrolysis of Glucose in D 2 0 / 0 . 5 0 M N a H 2 P 0 4 . H , 0Solutiona
0 WITH GLUCOSE A WITHOUT GLUCOSE 12.-
5
'Ol
no.
08
1
=c X
2
A
1Q
0 I
-08
I I -06 - 0 4
I
1
I
02 0 E v s SCE, V
I
02
I
0 4
I
3
06
Flgure 3. Mott-Schottky curves for a flamed titanium foil electrode 2.8 m glucose. in 0.5 M NaH,PO, and 0.5 M NaH,PO,
+
carbohydrates would be oxidized both directly by semiconductor holes and indirectly by H202formed in the cathodic reaction of 02. The results obtained with an O2 atmosphere with and without near-UV illumination are given in Table 111. A decrease in gas volume was observed indicating consumption of 02,as expected. The similar rates of volume decrease initially suggested that the presence of the photochemical diodes had little effect. However, analyses of the remaining gas showed that substantially more reaction, both in terms of O2 consumption and glucose oxidation, occurred when the diodes were present. No H, was evolved, substantiating the fact that the O2 is reduced instead. On the other hand, complete oxidation of the carbohydrates to C 0 2 was enhanced. Hydrogen Origin. The substantial amounts of H2 evolved in these experiments elicits questions as to its origin because the experimental evidence from conventional photoelectrochemical cells using n-Ti02 has indicated the inability to evolve H2 a t an appreciable rate without the application of an external bias of some sort (ref 6 and references therein). The normal explanation given for this inability is that the flat-band potential or flat-band potential plus hydrogen overpotential for n-Ti02 is not sufficiently negative enough to allow H, evoluition to occur via reduction of aqueous H+. Because significant H2 was evolved during the experiments presented here, one of the following two circumstances would seem to be required to explain this result. Either the flat-band potential is shifted to more negative values in the electrolyte used or the H2 d 'es not originate from reduction of aqueous H+ but via some other reaction. Flat-band potential shifts caused by electrolyte species have been r e p ~ r t e d , ~and ,-~~ therefore, the possibility el sts that glucose could have such an effect. However, results reported on the photoKolbe reaction utilizing Pt/n-Ti02 particles have indicated that anodic and cathodic processes may not take place completely independent of one another. In the case of the photo-Kolbe reaction of acetate on Pt/n-Ti02, CH4 and C 0 2 rather than the expected mixture of C2H6, CO,, and H2 were identified.17i22This fact was explained by the possibility of the oxidation intermediates, namely, methyl radicals, participating in the cathodic reaction made possible by the spatial proximity of the anode and cathode (32) Ohnishi, T.; Nakato, Y.; Tsubomura, H. Ber. Bunsenges. Phys. Chem. 1975, 79, 523. (33) Watanabe, T.; Fujishima, A.; Tatsuoki, 0.; Honda, K. Bull. Chem. SOC.Jpn. 1976, 49, 8. (34) Wilson, J. R.; Park, S. J . Electrochem. SOC.1981, 128, 2369.
4 5
6
7
method of determination or preparation Calculated based upon the statistical ratio of H and D of t h e water in solution when the exchange of glucose hydroxyl hydrogen is taken into account based o n the electrolysis o n Pt of t h e water described in (1)with the kinetic isotope effect considered at 25 O C , separation factor equal 7.47 from ref 35 based upon statistical ratio of H and D o n the glucose in solution when the exchange of the glucose hydroxyl hydrogen taken in account Conventional Electrolysis ratio measured in gas evolved from electrolysis of the D,O/glucose solution o n Pt at 60 OC, expt 1 ratio measured in gas evolved from electrolysis of the D,O/glucose solution o n Pt at 60 O C , expt 2 Photoelectrolysis* rate measured in gas evolved for t h e photoelectrolysis of t h e D,O/glucose solution Pt/n-TiO, during t h e time interval 0.0-21.0 h rate measured in gas evolved for the photoelectrolysis of t h e D,O/glucose solution Pt/n-TiO, during the time interval 21.0-45.5 h
H/D ratio
0.18
1.38
1.84
0.92 0.91
0.75
0.90
a Solution composition: 19.9385 g of D 2 0 ;1 . 4 3 1 g of NaH,P04.H,0; 11.0409 g of D-glucose (C,H,,O,). Photoelectrolysis experiments o n Pt/n-TiO, particles equivalent to those previously discussed except H,O replaced by D,O.
sites in the these metallized semiconductor particles.16J7 In light of these results, the glucose molecules could be the origin of the H, evolved. In an attempt to clarify this situation mott-Schotky plots were obtained for n-Ti02 in electrolytes differing only by the presence of glucose, and isotopic reactions were performed with H 2 0 replaced by DZO. The Mott-Schottky plots of C2vs. electrode potential for a polycrystalline Ti02 electrode in an electrolyte of 0.50 M NaH2P04with and without glucose are given in Figure 3. Although flat-band measurements using MottSchottky data collected at a single frequency should be viewed only tentatively, the measurements made here, Figure 3, indicate a 210-mV negative shift when glucose is present. This flat-band shift in the presence of glucose is being investigated further. Such a negative shift would enhance the capability for H2 evolution. In the isotopic experiments conducted, photoelectrolysis and conventional electrolysis were performed under identical conditions using D 2 0instead of HzO. If the H/D ratio of the evolved gas for the photoelectrolysis experiment is the same as the gas evolved through simple electrolysis on platinum in the same solution, the inference is that the cathodic reaction on the diodes is simple reduction of hydrogen ion on the platinum portion of the diode. Table IV lists the expected H/D ratios of the evolved gas calculated under various assumptions and those determined experimentally, both from photoelectrolysis and conventional electrolysis. The two different experimental methods, photoelectrolysis and direct electrolysis, agree
The Journal of Physical Chemistry, Vol. 87, No. 5, 1983
Photocatalytic Oxidation of Glucose by Platinized n-TIO,
reasonably well and match best to the calculated value based on the YHz”originating from water, which has equilibrated with the glucose hydroxyl hydrogens, by direct electrolysis in which the kinetic isotope effect is accounted for. Agreement between calculated and experimental values would be even better if the differences in temperature were taken into account. Raising the temperature to 65 “C, the temperature of the experiments, would tend to decrease the 1.38 H / D ratio calculated because the kinetic isotope effect would be less.
Discussion A conclusion which can be drawn from this work and that of Kawai and SakatamP2lis that the hdyroxy functional group on organic compounds and probably the aldehydes as well are capable of photoelectrochemical oxidation on n-Ti02 with simultaneous production of H2 on the platinized area of the photochemical diode. The data presented are felt to be most consistent with initial oxidation of the organic alcohol or aldehyde in a conventional in manner as given in the general eq 1 and 2. R-CHO R’-CH,-OH
+ H20 + 2p+ + H 2 0 + 4p+
hu
R-COOH
+ 2H+
(1)
&R’-COOH + 4H+
(2) If the compounds in eq 1 and 2 are glucose, then the open chain structures are as follows: I
805
The carboxylic acids generated by reactions 1 and 2 are then available for decarboxylation via a photo-Kolbe reaction resulting in eventual C02 evolution (4) If R is a carbohydrate, the end carbon group resulting from (4) is an alcohol which can continue oxidation by eq 1 and 2. The end gaseous products are then only H, and CO,. Alcohols and aldehydes are usually easily oxidized electrochemically but have greater difficulty in oxidizing completely to C 0 2 because of the difficulty in decarboxylation which requires carbon-carbon bond c l e a ~ a g e . ~ ~ , ~ ~ With the Pt/n-Ti02 photocatalysts, this problem is overcome by the ease of the photo-Kolbe decarboxylation. Hydroxycarboxylic acids resulting from partial oxidation or carbohydrates are common waste products of biomass processing industries, e.g., pulp and paper,38 or sugar.39 These products are, in general, harmful to the environment, virtually impossible to separate from the aqueous phase, and have such low heat content that they are poor boiler fuels. Their photoelectrochemical gasification to H, and C 0 2 using photocatalysts has potential for cleaning up waste streams, an application previously n ~ t e d , ~ J O J ~ while simultaneously providing a source of H2without the generation of an explosive H 2 / 0 2mixture.
Acknowledgment. We thank Dr. B. Solka and Mr. J. Neuzil for assistance in the mass spectrometry measurements and Dr. P. F. Richardson for performing the Mott-Schottky measurements. Registry No. Glucose, 50-99-7; hydrogen, 1333-74-0; carbon dioxide, 124-38-9;platinum, 7440-06-4;titanium dioxide, 1346367-7.
R
R’
Hydrogen ions then consume the photogenerated electrons in a conventional reduction 2H+ 2eH2 (3)
+
-
(35) Rowland, P. R. Nature (London) 1968, 218,945. (36) Gileadi, E.;Piersma, B. Mod. Aspects Electrochem. 1966,4, 47. (37)Parker, V. D. In “Organic Electrochemistry”; Baizer, M. M.; Marcel Dekker: New York, 1973; Chapter XV. (38) Sjmtrom, E. “Wood Chemistry, Fundamentals and Applications”; Academic Press: London, 1981. (39) Paturau, J. M.“By-Products of the Cane Sugar Industry”; Elsevier: New York, 1969.
