Investigation of the Mechanism for the Controlled Periodic Illumination

These investigations indicated that the periodic illumination effects reported for this reaction, and for other photocatalytic oxidations, resulted fr...
0 downloads 0 Views 62KB Size
Ind. Eng. Chem. Res. 2001, 40, 1097-1102

1097

Investigation of the Mechanism for the Controlled Periodic Illumination Effect in TiO2 Photocatalysis Karen J. Buechler,† Theresa M. Zawistowski,*,† Richard D. Noble,† and Carl A. Koval‡ Departments of Chemical Engineering and of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309

The photocatalytic oxidation of formate ion at illuminated TiO2 particles was investigated in an annular slurry reactor and in a rotating disk reactor that utilized TiO2 particulate films. Rates of CO2 formation were determined during continuous and periodic illumination over a range of oxygen and formate ion concentrations. These investigations indicated that the periodic illumination effects reported for this reaction, and for other photocatalytic oxidations, resulted from intraparticle diffusion in flocculated particles, mass transport of oxygen to the catalyst surface, slow or weak adsorption of formate ion, or a combination of these processes. Introduction One solution that has been proposed to address the low energy efficiency of photocatalytic processes is the use of controlled periodic illumination (CPI) instead of continuous illumination. Sczechowski et al.1 initially demonstrated the usefulness of this technique on the oxidation of formate ion in aqueous slurries of TiO2. Estimations of the flux of reagents, both oxygen and formate, did not support the CPI effect, being due to a diffusional mass-transfer limitation to the outside of the TiO2 flocculated particles. However, enough information to support any specific mechanism for the enhancement seen in their work was not provided. Attempts to replicate this result for the aqueous oxidation of formate ion on films of TiO2 showed that the enhancement was only observed when the reaction was limited by the bulk diffusion of O2 to the surface of the catalyst.2 This requirement for a bulk mass-transfer limitation to observe a CPI enhancement did not appear to be consistent with the early report. This work is a continued investigation into the mechanism for CPI to enhance the apparent quantum yields of photocatalyic oxidation of aqueous formate. Experiments designed to identify the differences in the oxidation of formate ion on films of catalyst versus slurries of catalyst were performed. Background Previous CPI Investigations. The initial experiments to investigate the CPI hypothesis were performed in aqueous slurries of TiO2.1 A novel design for the photocatalytic reactor was implemented: a trough reactor. In this system, the slurry flow was maintained in the laminar flow regime. This ensured that, as the fluid moved into and out of the illuminated regimes, the same packet of fluid saw the pulsed illumination. Sczechowski et al. were able to show that at short light times (40 ms) and long dark times (1.5 s) the apparent quantum yield (also referred to as the photoefficiency) was increased by a factor of 5. These experiments † ‡

Department of Chemical Engineering. Department of Chemistry and Biochemistry.

showed that CPI could enhance the apparent quantum yield for a variety of photocatalyic processes. However, the investigation did not result in a clear explanation of why the apparent quantum yield enhancement occurred. One possible mechanism for the enhancement in photocatalytic efficiency was bulk mass-transfer limitations to the particles. To eliminate this possibility, Sczechowski et al. performed calculations on the bulk mass-transfer limitations in aqueous slurries of TiO2. Mass transfer in slurries has been thoroughly investigated by theoretical and experimental techniques. Satterfield3 presented a relationship which related the Peclet number to the Sherwood number for slurries. Brian and Hales4 developed this relationship, which allowed an overall mass-transfer coefficient to be calculated. In the flow regimes and O2 concentrations used by Sczechowski et al., the O2 bulk diffusion-limited rate was at least a factor of 7 higher than the measured reaction rate. This supported the hypothesis that the CPI effect was not caused by an O2 mass-transfer limitation to the outer surface of the flocculated particles. Sczechowski et al. also performed calculations related to the maximum use of O2 by the reactor. These calculations showed that at no time in the experiments did the O2 concentration drop to less than 95% of its initial value. These calculations assumed that all of the reaction takes place at the surface of the 25 µm flocculated particles. Upadhya and Ollis5 performed a detailed modeling of the experimental work by Sczechowski et al. They assumed that the surface coverage of the formate ion (or hydroxyl radicals) is constant during the reaction or that the rate of formate adsorption at least equals the rate of reaction. Adsorption of O2 on the surface and/ or the transfer of conduction band electrons to adsorbed O2 was assumed to be the rate-limiting step. Upadhya and Ollis’ modeling provided a reasonable explanation for the CPI results obtained by Sczechowski et al. Stewart and Fox suggested that a dark recovery time improves the apparent quantum yield for photooxidations.6 Stewart used a nonaqueous media in order to eliminate water-derived oxidation products. Both a photocatalytic one-step oxidation (1-octanol to 1-octanal) and a one-step reduction (p-nitroacetophenone to p-

10.1021/ie0004592 CCC: $20.00 © 2001 American Chemical Society Published on Web 01/26/2001

1098

Ind. Eng. Chem. Res., Vol. 40, No. 4, 2001

aminoacetophenone) were investigated. A pulsed laser light source was used to illuminate the reaction chamber. The apparent quantum yield of the oxidation increased by approximately 1.8 times at the lower pulse frequency (longer dark time). The apparent quantum yield of the reduction showed no measurable improvement due to the longer dark time. Stewart suggested that the reduction of oxygen could be the important secondary reaction that causes the CPI effect. The importance of the kinetics of oxygen reduction to the understanding of the CPI effect follows from the earlier calculations reported by Heller and Gerischer, which concluded that O2 reduction would limit most photocatalytic processes.7 Recent studies from our group on the photocatalytic oxidation of aqueous formate on thin films of TiO2 have not yielded the same enhancement due to CPI2 that was reported by Sczechowski et al. These experiments were performed in a reactor designed to provide controllable bulk mass transfer to the surface of the catalyst. The reactor used a thin film of TiO2 affixed to a stainless steel disk. Rotation of the disk in a sealed cylinder provides uniform access to the surface of the catalyst by aqueous reagents through Levich flow. At the light (0.6 s) and dark (2.0 s) times investigated in this reactor, the only conditions under which an enhancement due to CPI is observed is when the reaction is limited by the bulk mass transfer of O2 to the surface of the catalyst. Mass-Transfer Limitations in Photocatalytic Reactions. Most of the results from prior studies suggested that the CPI effect is due to one of the chemical or mass transport steps involved in the overall photocatalytic reactions being slow rather than photon absorption or the migration of photogenerated charge carriers in the semiconductor. In addition to the reaction steps, reagents (organic and O2) are proceeding through a series of mass-transfer steps. Mass-transfer processes are very complicated in gas/liquid/solid systems; however, the basic principles behind these processes are understood.8 A gas-phase reagent must undergo four separate mass-transfer steps before being able to participate in a photocatalytic reaction. First, it must be transported to the gas/liquid interface. This is typically an extremely fast step, especially when the gas phase is present in the system in great excess. Next, the gas must be absorbed into the liquid. The rate of this step is highly dependent upon the solubility of the gas into the liquid and the contact area between the gas and the liquid. The reagent must then be transported through the liquid to the surface of the catalyst. This requires the reactant to pass through the bulk of the liquid as well as the boundary layer adjacent to catalyst particles or films. Transport through the bulk of the fluid can be rapid if the primary mode of transport is convection rather than diffusion. Fickian diffusion through the liquid boundary layer adjacent to the catalyst particle can be slow and overall rate-determining, especially if the boundary layer is thick. Finally, it may be necessary for the reagent to adsorb to the catalyst surface prior to the reaction. Investigation of Mass Transport Steps Potentially Responsible for the CPI Effect. Through this study, we investigate the importance of several of the mass transport steps outlined above either directly or indirectly. The different mass transport geometries discussed in the paper are shown in Figure 1. The gas to liquid transport steps have been omitted because they

Figure 1. Diagrams showing the various geometries for liquidsolid mass transport in the photocatalytic system discussed: (A) mass transport through the bulk fluid; (B) Fickian diffusion through the outer boundary layer; (C) diffusion through the flocculated particle.

are the same as those for all geometries discussed. Mass transfer to an individual particle of TiO2 is shown in Figure 1(I). The reagents transfer through the bulk fluid to the boundary layer. Then they slowly cross the boundary layer to the surface of the catalyst by diffusion. When particles agglomerate together, as shown in Figure 1(II), the total diffusional path length is much larger for particles toward the center of the agglomerate. Thus, the time required for reagents to reach these particles is much longer than that with the nonagglomerated particles. In the rotating disk reactor (RDR), the TiO2 particles are affixed as a film to a rotating solid support [Figure 1(III)]. Convective transport rapidly brings the reagents up to the thin boundary layer between the catalyst film and the bulk solution. Diffusion controls the mass transfer of the reagents through the boundary layer. Experimental Section RDR. The oxidation of aqueous formate was studied as a function of the oxygen concentration, formate concentration, light intensity, and light and dark times in the RDR. The details of this reactor are given in Buechler et al.2 TiO2 (Degussa P-25) was affixed to a stainless steel disk that rotates in the reaction mixture. An array of UV-A lamps illuminated the entire surface of the catalyst. A mechanical shutter was positioned

Ind. Eng. Chem. Res., Vol. 40, No. 4, 2001 1099

Figure 2. Slurry reactor experimental setup. The mixing vessel is used to purge the reactants with oxygen, air, or nitrogen. The pump circulates the slurry through the system at a volumetric flow rate of 0.08 L/s. The volume of reactant slurry used in all experiments is 0.7 L. Periodic illumination is created by masking portions of the bulb with aluminum foil.

between the lamps and provided variable light and dark times. The primary advantage of this design was the uniform and controllable mass transfer to the surface of the catalyst. The formate ion reaction rate was measured by monitoring CO2 evolution as a function of time. The reactor was charged with 250 mL of sodium formate in water. The reactor was then sealed to gases and liquids. Samples were removed with a 5 mL syringe for analysis. The CO2 concentration was measured using a Fisher/ Accumet gas-sensing electrode specific for CO2 with a Corning 220 pH meter. The rate of CO2 evolution was found by monitoring the CO2 concentration as a function of time. Sczechowski et al. verified this method with total organic carbon (TOC) measurements. It was confirmed that a mass balance was closed. Reactions were primarily performed with a pure O2 atmosphere in the RDR gas phase. This was achieved by purging the reagents with O2 for at least 30 min to ensure saturation of the liquid phase and purging of the air from the gas phase. The headspace was also continuously replenished with humidified O2 during the course of the reactions. Annular Slurry Reactor (ASR). The annular reactor used by Sczechowski,9 Shoffner,10 and Foster-Mills11 was used to perform slurry experiments (Figure 2). The annulus was 0.400 m in length with a 0.003 m annular gap and a volume of 0.145 L. Illumination was through a 9 W black light blue fluorescent lamp positioned in the center of the annulus. The illuminated surface area was 450 cm2. The intensity of the lamp incident upon the surface of the liquid was measured as a function of the length of the bulb, with the bulb placed inside and outside the annulus. The intensity on the reaction fluid was estimated by averaging these two values to take into account the absorption of UV light by the inner Pyrex annulus walls. A peristaltic pump circulated reagents and the catalyst. The pump operated at a volumetric flow rate of 0.11 L/s. This created a residence time of 6.33 s. The fluid was recirculated through a loop that included a reservoir to allow for mixing, saturation with gases, and removal of samples. Experiments without UV light showed no detectable CO2 formation in this system over 3 h. Several experiments were designed to investigate the effect of flocculated particle size. A portion of the Tygon tubing that was external to the reactor itself was coiled in a sonic mixer. The sonicator dispersed the flocculated particles, which should have increased the overall mass-

transfer rate to the catalyst surface. The temperature in the sonicator was maintained at approximately 25 °C by the addition of ice throughout the experiment. A control experiment demonstrated that no reaction occurred with formate, O2, and sonication without UV light. Apparent Quantum Yield. It is important to point out that the apparent quantum yield values (moles of reaction per mole of photons entering the reactor) reported in this paper were low estimates of the true quantum yield (moles of reaction per mole of absorbed photons). The number of photons absorbed by the catalyst was lower than the number of photons entering the reactor because of scattering and reflective losses. Some rudimentary attempts to quantify these losses have been made by Sczechowski. In an annular reactor he varied the concentration (C) of TiO2 in a 4.2 × 10-3 mol/L sodium formate solution. By mounting a 1 cm2 sensor to the outside of the annular reactor, he was able to measure the transmitted (and presumably scattered) light. He related his transmitted light intensity (I)

log(I/I0) ) -Cl to the incident light intensity (I0) and calculated a linear function of C in C:

C ) 0.47C + 0.39 This relationship was developed using experimental values of C up to 1.6 g/L. Results and Discussion Mass Transport to Particles in Slurry Reactors. In the trough reactor used by Sczechowski et al. and in other slurry reactors, the individual 25 nm TiO2 particles flocculated together into larger 25 µm particles. When Sczechowski et al. calculated the effects of reagent mass transport in their reactor, they assumed that the reagents (oxygen and formate) were only consumed at the external surface of the flocculated particles. Therefore, as shown in Figure 1, mass transfer occurred as shown by arrow B through the solution adjacent to the flocculated particles. However, the flocculated particles also had an internal surface area where the reaction could have occurred. Access of reagents to this internal surface area through solution contained in the flocculated particles is depicted in Figure 1 by arrow C. If mass transport within the flocculated particles was ratelimiting in the slurry reactors, it could explain the difference between the observed CPI effects observed in slurry reactors and those at nonporous thin films. Binary diffusion coefficients (D) for formate ion and oxygen in water are approximately 1.5 × 10-9 m2/s. According to Fick’s laws of diffusion, the root-meansquare displacement for the reagents, which equals (2Dt)1/2, for the reagent molecules would be 50 µm in the 2 s CPI dark time observed by Sczechowski et al. Therefore, formate and oxygen would have access to at least a portion of the internal surface area of a flocculated particle in the 2 s dark time. Comparison of CPI Effects for Nonporous Films with Flocculated Particles. The possible importance of mass transport within flocculated TiO2 particles was explored experimentally using nonporous TiO2 films in the RDR. As shown in Figure 1(III), mass transport of reagents to the external surface of the TiO2 films in the

1100

Ind. Eng. Chem. Res., Vol. 40, No. 4, 2001

Table 1. Light and Dark Conditions for the ASRa illumination condition

tl (s)

td (s)

tl/td

unmasked three slits two slits

1.81 0.16 0.03

4.52 6.17 6.30

0.400 0.025 0.004

Table 2. Summary of Experiments Performed at Various Illumination Conditions: Sonication vs Nonsonication and Air vs O2 Saturationa sonication of the slurry

oxygen or air

illumination condition

rate × 107 (mol/s)

adjusted rate × 107 (mol/s)

yes yes yes no no no no yes

air air air air air air O2 O2

unmasked three slits two slits unmasked three slits two slits unmasked unmasked

5(1 0.8 ( 0.1 0.3 ( 0.04 1.6 ( 0.4 0.8 ( 0.2 0.3 ( 0.08 4.3 ( 0.4 6(1

13 ( 3 32 ( 4 75 ( 10 4(1 32 ( 8 75 ( 20 11 ( 1 15 ( 3

a The light and dark times reported are for the total time that the slurry is illuminated and the total time that the slurry is in the dark during the 6.33 s circulation time.

RDR (arrow B) is controlled by Levich flow. By the appropriate choice of bulk reagent concentrations and disk rotation rate in the RDR, photocatalytic experiments were performed that duplicated the reagent fluxes to the external surfaces of flocculated particles that were presumed in Sczechowski’s experimental work. A formate concentration of 8.9 × 10-3 mol/L and an oxygen concentration of 1.9 × 10-3 mol/L were used. At a rotation rate of 80 rpm, the maximum flux to the external catalyst surface is 1.6 × 10-6 mol/cm2‚s formate and 3.4 × 10-7 mol/cm2‚s oxygen. In the RDR, the reaction progressed at a rate of 7.3 × 10-10 mol/cm2‚s under conditions of continuous illumination. During periodic illumination, with a light time of 0.1 s and a dark time of 2.0 s, the observed rate (moles of CO2 evolved per total time per cross-sectional area) was 3.7 × 10-11 mol/cm2‚s. If this rate is adjusted for the fraction of time that the catalyst is illuminated, then this adjusted rate (moles of CO2 evolved per total light time per cross-sectional area) was 7.6 × 10-10 mol/cm2‚s, e.g., the same as the continuous illumination rate. The apparent quantum yield (moles of CO2 evolved per moles of photons entering the reactor) was 20% for both experiments. The lack of a CPI effect for nonporous TiO2 films at similar fluxes of reagents and similar light and dark times as those used by Sczechowski in slurries, where an increase in the apparent quantum yield from 12% to 20% was observed, is consistent with the hypothesis above. It suggested that the CPI effect observed by Sczechowski was associated with masstransfer limitations within the flocculated particles rather than mass transfer to the external surface of the flocculated particles. Effects of CPI, Sonication, and Reagent Concentrations Using an ASR. Another technique used to investigate the effect of diffusion into the flocculated particles was to disperse the individual TiO2 particles using sonication. The ASR was used with Tygon tubing looped through the sonicator. Periodic illumination of the slurry in the ASR was accomplished by blocking a portion of the light with foil (see Figure 2) so that light and dark times similar to those used by Sczechowski et al were achieved. Three different illumination conditions were investigated as shown in Table 1. Even when the annular reactor was unmasked, a dark time was associated with the time required for the slurry to pass through the mixing vessel, peristaltic pump, and sonicator. By analogy to the RDR, the observed CO2 evolution rates observed for different illumination conditions in the ASR were adjusted to account for the fraction of time that the slurry spent in the dark using the equation

rateadj )

CO2 evolution rate tl/(tl + td)

Tables 2 and 3 show the observed and adjusted CO2 evolution rates for the ASR with formate concentrations

a

The initial concentration of sodium formate is 4.2 × 10-3 mol/

L. Table 3. Summary of Experiments Performed at Various Illumination Conditions: Sonication vs Nonsonication and Air vs O2 Saturationa sonication of the slurry

oxygen or air

illumination condition

rate × 107 (mol/s)

adjusted rate × 107 (mol/s)

yes no yes yes no no

air air O2 O2 O2 O2

three slits three slits unmasked three slits unmasked three slits

1.6 ( 0.3 1.5 ( 0.2 21 ( 1 1.3 ( 0.2 18 ( 0.4 1.3 ( 0.1

64 ( 12 60 ( 8 53 ( 3 52 ( 8 45 ( 1 52 ( 4

a

The initial concentration of sodium formate is 4.2 × 10-2 mol/

L.

of 4.2 × 10-3 and 4.2 × 10-2 mol/L, with air and oxygen saturation of the slurry, and with and without sonication. There were several interesting trends associated with these data. For the lower formate and oxygen concentrations (air source) in Table 2, there appeared to be a dramatic CPI effect because the adjusted CO2 evolution rates increased substantially as the light time was decreased from 1.82 to 0.03 s. However, sonication only had an effect when the reactor was unmasked and the observed CO2 evolution rates, and thereby the reagent consumption rates, were highest. In this case, the adjusted rate increased from 4 × 10-7 to 13 × 10-7 mol/s as a result of sonication. When the reactor was unmasked and the slurry was not sonicated, an increased oxygen concentration produced a significant increase in the CO2 evolution rate; however, sonication produced a slightly higher rate that was virtually independent of whether air or pure oxygen was used. Although most previous studies assumed that reduction of oxygen by photogenerated electrons would be slower than oxidation of formate ion by holes, these results suggested that both processes might be contributing to the CO2 evolution rates. Additional experiments performed at 10 times greater formate ion concentrations confirmed this possibility. The most striking features of the data obtained at higher formate ion concentrations (Table 3) are that all of the CO2 evolution rates are considerably greater than those in Table 2 and that neither periodic illumination, sonication, nor oxygen concentration have a noticeable effect on the adjusted rates. These results suggested that while mass transport limitations produced the CPI effects in Table 2, the limitation may have been produced by consumption of formate ion rather than oxygen. Effect of Formate Ion Concentrations in the RDR. The possibility that low formate ion concentrations contributed to CPI effects observed in slurry

Ind. Eng. Chem. Res., Vol. 40, No. 4, 2001 1101

Figure 3. Adjusted CO2 rate of evolution as a function of the initial formate concentration. The lamp intensity was 1.1 mW/ cm2 for all experiments. The dashed line represents the maximum rate of bulk diffusion of formate to the catalyst surface. The solid line represents the maximum rate of bulk diffusion of oxygen to the catalyst surface. The experiments at low formate concentration would not be limited by the bulk rate of diffusion of either oxygen or formate. The experiments at higher formate concentration would be limited by bulk diffusion of oxygen to the catalyst surface.

reactors under certain conditions was explored at nonporous TiO2 disks in the RDR, where the mass transport of reagents to the catalyst surface is uniform and controlled. These experiments were performed at 1 mW/cm2 incident light intensity and with oxygen saturation. The initial formate concentration was varied from 1 × 10-3 to 0.5 mol/L. This is a range that includes the initial formate concentration used in the slurries in this study and the previous work. Observed and adjusted CO2 evolution rates over 3 orders of magnitude in formate concentration are displayed in Figure 3 for both continuous and periodic illumination. Also shown in Figure 3 are lines depicting the diffusion-limited rates for formate ion and oxygen that were calculated from the Levich equation. As shown in Figure 3, the adjusted CO2 evolution rates for continuous and periodic illumination are similar, especially at high formate ion concentrations; however, the observed rates were approximately 100 times smaller for the periodic illumination experiments because the disk was only illuminated about 1% of the time. Clearly, the CO2 evolution rates for continuous illumination were dependent upon the formate ion concentration over the entire range, most noticeably at the lower formate concentrations. This was surprising because the photocatalytic reaction rates were several orders of magnitude lower than the maximum rates at which formate was continuously transported to the catalyst surface. The dependence of CO2 evolution rate on the formate ion concentration was less pronounced for periodic illumination, although a slight effect is seen at low concentrations. This is expected because the nonadjusted rate of formate consumption is much lower when the disk is periodically illuminated. At the highest formate ion concentrations, the adjusted CO2 evolution rates approached the oxygen diffusion-limited rate and there was little or no CPI effect. For periodic illumination, the observed CO2 evolution rates were below the oxygen diffusion-limited rate. This suggested that, during the 0.1 s light time, the ability of oxygen mass transport to limit the photocatalytic reaction was the same as that when the catalyst was continuously illuminated. At lower formate ion concentrations, the adjusted CO2

evolution rates for continuous and periodic illumination dropped below the oxygen diffusion-limited rate and an apparent CPI effect was observed. The observed CO2 evolution rates for periodic illumination were below the rates for continuous illumination. (For continuous illumination, the observed and adjusted rates are the same.) However, unlike the situation at high formate ion concentration, the adjusted rates for periodic illumination remained below the diffusion-limited rates for oxygen and formate. Under these conditions, periodic illumination of the catalyst improved the adjusted CO2 evolution rate. This observation, coupled with the fact that the adjusted CO2 evolution rates are orders of magnitude lower than the formate ion diffusion-imited rates calculated from mass transport equations, suggested that the limitation imposed by formate is not simply mass transport. Apparently, there was some other processes involving the formate ion that proceeded during the 10 s dark time that allowed the photocatalytic reaction to be more effective during the light time. One explanation for the effect of the formate ion concentration on the photocatalytic rates and for the CPI effect is that the formate ion needed to be adsorbed on the TiO2 catalyst in order to react efficiently with holes. At low formate ion concentrations, the surface concentration of the formate ion was too small to allow the efficient formation of CO2 under continuous illumination. Under periodic illumination, adsorbed formate ion was depleted during the light time and replenished during the dark time. At higher formate ion concentrations, the surface concentration of the formate ion became adequate to react with holes at a sufficiently high rate so that oxygen reduction became rate-limiting. An alternative explanation would result if adsorption of the formate ion were extremely weak and the amount of adsorbed formate ion were small over the entire range of the bulk concentration investigated. In this case, the rate of adsorption of the formate ion, which would presumably have had a first-order dependence on the bulk concentration of the formate ion, would have limited the overall photocatalytic rate at low concentration and become sufficiently rapid at high concentrations. Distinguishing between these two possibilities would have required a determination of the adsorption isotherm for the formate ion on the TiO2 catalyst. This determination was not carried out in the present study and would be difficult if the adsorption were weak. Conclusions and Comparison with Other CPI Studies. Sczechowski et al. demonstrated the CPI effect using two types of slurry reactors in which the photocatalytic oxidation of the formate ion was studied.1,9,12,13 Based on the results presented in this study and in two previous papers by Buechler et al., apparent improvements in the efficiency of photocatalytic processes can be attributed either to mass-transfer limitations or to slow adsorption/reaction steps. For the gas-phase photocatalytic oxidation of trichloroethylene, CPI was found to have no effect on the apparent quantum yield or on the selectivity of product formation except in flow regimes where the reaction was mass transport limited.14 In this study and in an earlier description of the RDR,2 apparent CPI effects were attributed to mass transport limitations of either oxygen and/or formate ion to the catalyst surface. Calculations and experiments indicated that in slurry reactors diffusion of reagents through solution trapped inside flocculated TiO2 particles imposed additional mass transport limi-

1102

Ind. Eng. Chem. Res., Vol. 40, No. 4, 2001

tations when compared to experiments involving highly dispersed particles or particles immobilized on surfaces. Under conditions where the reduction of oxygen by electrons limits the photocatalytic rate, apparent CPI effects were observed at relatively high light intensities and when the light time to dark time ratio was high.2 When the light time to dark time ratio was low, as in the present RDR studies, the adjusted photocatalytic rate was controlled by the convection/diffusion process that transported oxygen to the catalyst surface, not by adsorption or reaction steps, and there was no CPI effect. The key result from the present study, however, was the observation that the formate ion concentration contributed to limiting the photocatalytic rate and that this limitation could not have been due to transport to the catalyst surface. At low formate ion concentrations, the largest apparent CPI effects were observed and were attributed to slow or weak adsorption. The CPI effects reported by Foster et al. for a TiO2 catalyst film could also have been associated with slow or weak adsorption of the organic reagent involved, 2-methyl-1,4-hydroquinone at millimolar concentration.15 The large initial photocurrents that were observed and the necessity for having a large dark time to achieve them could have resulted from slow or weak adsorption of the organic prior to photoinduced reaction. The results of Dulay and Fox demonstrated an enhancement due to CPI in a TiO2 catalyst slurry. CPI was observed for the oxidation of 1-octanol but not for the reduction of nitroacetophenone. This could have occurred if 1-octanol had to adsorb prior to reaction but nitroacetophenone did not. Their reactions in a slurry could have shown the effects of flocculation and intraparticle diffusion limitations. The model proposed by Upadhya and Ollis is accurate for systems, which are limited by the adsorption or reaction of O2. However, this formulation of the model failed to account for the enhancement caused by CPI when the photocatalytic rate was limited by adsorption of the formate ion. Acknowledgment We acknowledge the support of a National Science Foundation Minority Graduate Fellowship, the University of Colorado UROP program, and NSF Grant CTS9903511.

Literature Cited (1) Sczechowski, J. G.; Koval, C. A.; Noble, R. D. Evidence of Critical Illumination and Dark Recovery Times for Increasing the Photoefficiency of Aqueous Heterogeneous Photocatalysis. J. Photochem. Photobiol., A 1993, 74, 273. (2) Buechler, K. J.; Nam, C. H.; Zawistowski, T. M.; Noble, R. D.; Koval, C. A. Design and Evaluation of a Novel Controlled Periodic Illumination Reactor to Study Photocatalysis. Ind. Eng. Chem. Res. 1999, 38, 4. (3) Satterfield, C. N. Mass Transfer in Heterogeneous Catalysis; MIT Press: Cambridge, MA, 1970. (4) Brian, P. L. T.; Hales, H. B. AIChE J. 1969, 15, 419. (5) Upadhya, S.; Ollis, D. F. Simple Photocatalysis Model for Photoefficiency Enhancement via Controlled Periodic Illumination. J. Phys. Chem. B 1997, 101 (14), 2625. (6) Stewart, G.,; Fox, M. A. The Effect of Dark Recovery Time on the Photoefficiency of Heterogeneous Photocatalysis by TiO2 Suspended in Non-Aqueous Media. Res. Chem. Intermed. 1995, 21, 8/9. (7) Gerischer, H.; Heller, A. Photocatalytic Oxidation of Organic Molecules at TiO2 Particles by Sunlight in Aerated Water. J. Electrochem. Soc. 1992, 139. (8) Bideau, M.; Claudel, B.; Faure, L.; Kazouan, H. Diffusional Limitations in Liquid-Phase Photocatalysis. Prog. React. Kinet. 1994, 19. (9) Sczechowski, J. G. Increasing the Photoefficiency in Heterogeneous Photocatalysis through Controlled Periodic Illumination. Ph.D. Thesis, University of Colorado, Boulder, CO, 1994. (10) Shoffner, G. Improving the Photoefficiency of the Oxidation of Proponic Acid Through Periodic Illumination of a Semiconductor Photocatalyst. M.S. Thesis, University of Colorado, Boulder, CO, 1994. (11) Foster-Mills, N. S. Remediation of Aqueous Waste Streams Containing Metal Ions and Organic Compounds Using Semiconductor Photocatalysis. Ph.D. Thesis, University of Colorado, Boulder, CO, 1994. (12) Sczechowski, J. G.; Koval, C. A.; Noble, R. D. A Taylor Vortex Reactor for Heterogeneous Photocatalysis. Chem. Eng. Sci. 1995, 50. (13) Sczechowski, J. G.; Koval, C. A.; Noble, R. D. In Photocatalytic Purification and Treatment of Water and Air; Ollis, D. F., Al-Ekabi, H., Eds.; Elsevier: Amsterdam, The Netherlands, 1993. (14) Buechler, K. J.; Noble, R. D.; Koval, C. A.; Jacoby, W. A. Investigation of the Effects of Controlled Periodic Illumination on the Oxidation of Gaseous TCE Using a Thin Film of TiO2. Ind. Eng. Chem. Res. 1999, 38, 3. (15) Foster, N. S.; Koval, C. A.; Sczechowski, J. G.; Noble, R. D. Investigation of Controlled Periodic Illumination Effects on Photo-Oxidation Processes at Titanium Dioxide Films Using Rotating Ring Disk Photoelectrochemistry. J. Electroanal. Chem. 1996, 406.

Received for review May 5, 2000 Revised manuscript received November 9, 2000 Accepted November 28, 2000 IE0004592