481
Langmuir 1992,8, 481-486
Role of Adsorption in Photocatalyzed Reactions of Organic Molecules in Aqueous Ti02 Suspensions Claudio Minero, Flavio Catozzo, and Ezio Pelizzett? Dipartimento di Chimica Analitica, Universith di Torino, 10125 Torino, Italy Received June 12, 1991.I n Final Form: September 30, 1991 The photocatalyzed transformation of chemical compounds strongly adsorbed on a particle surface has been investigated in the presence of different photoactive and "inert" supports. For several compounds, such as dioctylquinol and crysene, the rate of degradationis only slightly affected by the initial adsorption onto nonphotocatalytic materials (Si02, Al203) when irradiated in a slurry with added micrometer size Ti02 particles. A rapid exchange of the substrate between the different inorganic supports was experimentally observed and explains the photocatalytic results. Decafluorobiphenyl (DFBP), which adsorbs tenaciously on A1203, degrades slowly when irradiated in the presence of Ti02 particles. Measurements confirm that DFPB is poorly exchanged from alumina to TiOz. Comparison with the results obtained using colloidal Ti02 or silica particles, and with the behavior of pentafluorophenol,under otherwise identical conditions, suggests that the photogenerated oxidizing species does not migrate far from the photogenerated active centers and that the degradation process occurs at the surface or within a few monolayers around the photocatalytic particles. Introduction The efficacy of heterogeneous photocatalysis for the total oxidation of organic contaminants in water has been demonstrated in numerous reports.lI2 This has led to considerable interest in examining the mechanistic details of the Ti02 photoreactivity in an effort to improve its photocatalytic a ~ t i v i t y . ~ , ~ Material modification and the practical immobilization of the photocatalyst require a detailed understanding of the mechanism by which Ti02 photodegrades pollutant species.kl0 An interesting aspect, both from mechanistic as well as engineering points of view, is represented by the location of the degradation reaction, that is whether the reactive site is a t the surface of the catalyst and/or in the bulk solution. Although the kinetics of degradation often obey the Langmuir-Hinshelwood model, thus suggesting that the reaction may be occurring at the surface of the catalyst, it was recently noted that wherever the reaction occurs, at the surface or in the bulk or via a Rideal mechanism, the rate equations have the same analytical form as the Langmuir-Hinshelwood expression.ll Several reports have dealt with the nature of the photogenerated reactive species (direct electron transfer to the holes vs reactions with oxidizing radicals, such as hydroxy radicals) and with its mobility into the solution. (1)Ollis, D. F.;Pelizzetti, E.; Serpone, N. Enuiron. Sci. Technol. 1991, 25, 1522.
(2)Matthews, R. W. In Photochemical Conversion and Storage of Solar Energy; Pelizzetti, E., Schiavello, M., Eds.; Kluwer: Dordrecht, 1991. (3)Photocatalysis-Fundamentals and Applications; Serpone, N., Pelizzetti, E., Eds.; Wiley: New York, 1989. (4)Bahnemann, D. W. In Photochemical Conuersion and Storage of Solar Energy; Pelizzetti, E., Schiavello, M., Eds.; Kluwer: Dordrecht, 1991. ( 5 ) Serpone, N.; Borgarello, E.; Harris, R.; Cahill, P.; Borgarello, M.; Pelizzetti, E. Sol. Energy Mater. 1986,14, 121. (6)Matthews, R. W.J. Phys. Chem. 1987,91, 3328. (7)Sabate, J.;Anderson, M. A.; Kikkawa, H.; Edwards, M.; Hill, C. G. J. Catal. 1991,127, 167. (8)Pichat, P. In Photochemical Conuersion and Storage of Solar Energy; Pelizzetti, E., Schiavello, M., Eds.; Kluwer: Dordrecht, 1991. (9)Pichat, P.; Hermann, J. M. In ref 3,Chapter 8. (IO)Cunningham, J.; Al-Sayyed, G. J. Chem. SOC.,Faraday Trans. 1990,86,3935. (11)Turchi, C. S.; Ollis, D. F. J. Catal. 1990,122, 178.
The conclusions are under active current debate.12-15It is worth noting that a recent report16 on the photodegradation of hydrophilic compounds has inferred that under proper conditions of pH, when the species are adsorbed because of their electric charge, the reaction takes place at the surface of the catalyst. In this paper, we examine the role of adsorption on the photocatalytic process.1° The results allow formulation of a hypothesis on the reaction location and indicate the photocatalytic process as a potential route for the detoxification of contaminated supports. Experimental Section Materials. Decafluorobiphenyl (DFBP, Fluorochem 99 5% ), pentafluorophenol (PFP, Aldrich >99 %), 2,5-bis(1,1,3,3-tetramethylbutyl)benzene-1,4-diol(later referred as dioctylquinol), and crysene (Aldrich)were used as received. Silica (Si02 gel 60, 70-230 mesh, Merck) and alumina (A1203 60 active, basic for column chromatography,70-230 mesh, Merck) were passed over a 160-meshsieve to collect the biggest fraction later used in the experiments (100-200rm). Ti02was used in particles of different sizes (see Table F).Ti02 P25 (Degussa,surface area 55 m2 g-l) was used as received. Ti02 beads were prepared by high-tem perature sintering of small particles prepared by wet hydrolysis of Tic&. The resulting material is a highly porous material (70 m2g-l) in 1-mm sphericalparti~1es.l~ Ti02 colloidswere prepared by hydrolysis in water at 0 "C of TiCL, followedby dialysisagainst bidistilled water. The obtained gel is collected, stored in refrigerator at 4 "C for 1month, and again dialyzed until the pH of the colloidal solution of Ti02 (5 g L-l) was around 2.7. Colloids were characterized by dynamic light scattering and the hydrodynamic radius was found to be about 60 nm.17 HzOz (Merck) was diluted to 0.2 M and standardized with permanganate. Other reagents were at least analytical grade. Water was passed over a mixed bed of ion exchange resin and double distilled. (12) Draper, R. B.; Fox,M. A. Langmuir 1990,6,1396. (13)Peterson, M. W.; Turner, J. A.; Nozik, A. J. J.Phys. Chem. 1991, 95, 221. (14)Lawless, D.; Serpone, N.; Meisel, D. J. Phys. Chem. 1991,95, 5166. (15)Serpone, N.;Lawless,D.; Terzian, R.; Meisel, D. To be submitted
for publication. (16)Kormann, C.; Bahnemann, D. W.; Hoffmann, M. R. Enuiron. Sci. Technol. 1991,25,494.
(17)Pelizzetti, E.; Minero, C.; Carlin, V.; Borgarello, E.; Miano, F.; Serpone, N. Paper presented at IPS-8 Conference,Palermo, Italy, 1990.
0743-7463/92/2408-0481$03.00/00 1992 American Chemical Society
482 Langmuir, Vol. 8, No. 2, 1992 Table I. Size of Metal Oxide Particles Used through the Present Work material size, pm 0.06 Ti02 colloids 0.5 Ti02 P25 Ti02 beads lo00 100-200 Si02 100-200 A1203 Procedures. Deposition of DFBP, dioctylquinol, and crysene on supports (TiOz, alumina, and silica) was performed by suspending a weighted amount of support (typically 2 g) in petroleum ether, later adding the suitable amount of a concentrated solution of organics (typically lo00 ppm) in petroleum ether and then evaporating the solvent under vacuum and vigorousstirring. Theprocedureledto0.5-0.6% (w/w) oforganics on supports and ensured a quite homogeneous dispersion and deposition of the organic molecules. Blank extraction of the organics from the support was performed by mixing 55 mg of supported oxide with 5 mL of acetonitrile, followed by intense stirring. The solution was then filtered and the organic content in solution was evaluated. Yields were always >90 % with respect to the content expected taking into account the amount of organics used prior to the deposition procedure. Deposition of PFP was performed as before but using dichloromethane as solvent. Irradiation experiments were carried out by exposing 5 mL of solution, containing inert supports (55 mg in 5 mL) and/or Ti02 (always a t 1g L-l as a final concentration for P25 and colloids, 10% (w/w) for beads), to a UV-vis light generated by a 1500-W xenon lamp (Solarbox, Cofomegra, Milan). The cell geometry and light flux are reported elsewhere.18 Average temperature in the lamp housing was 60 O C . Values of pH = 4.5 were measured prior to the irradiation and the initial concentration of organics was 40-70 ppm. Concentration values for each experiment are reported in Tables I1 and 111. After the proper time of exposure, the cell was removed and the slurry was passed through a 0.45Mm filter (Millex, HV, Millipore) or a sieve for P25 and Ti02 beads, respectively. When organics are strongly hydrophobic (DFBP, dioctylquinol, and crysene), the filtered solution was discarded (content of the initial organic compound C 0.1 ppm). Five milliliters of acetonitrile was used to clean the cell, then flushed through the filter collecting the oxide particles, and fiially analyzed. When PFP was used, the slurry was filtered and the aqueous solution was directly analyzed. Because Ti02 colloids block the filter, after irradiation (at pH = 2.5 to reduce colloid aggregation) the slurry was dried under vacuum (Rotavapor, Biichi) and 5 mL of acetonitrile was used to clean the irradiation cell and to recover from solid materials the residual initial compound. Irradiation experiments in the presence of H2Oz were performed without optical filter (wavelength 2 300 nm) on solutions prepared by mixing the supports with 5 mL of 0.05 M HzOz. During the experiment the ['OH] is probably not under steadystate conditions, but addition of H202 every 15 min does not change the degradation kinetics. The partition of organics between silica or alumina and Ti02 was evaluated by mixing 0.50 g of silica or alumina containing the organics with 50 mL of Ti02 P25 slurry (1g L-I) in a stirred vessel. The ratio of silica or alumina to Ti02 was always 1O:l. After the proper delay time, the stirring was stopped and two different procedures were followed: (a) The silica or alumina were allowed to settle on the bottom of the vessel (step duration about 20 8 ) . Because of the different particle sizes, in this time the Ti02 P25 particles did not settle and 1 mL of Ti02 slurry was taken out and collected on a 0.45-pm filter (Millex HV, Millipore). The filter and the Ti02 were then flushed with 2 mL of acetonitrile and the organics analyzed. (b) The whole slurry was passed on a sieve able to keep back the silica or alumina particles. The whole Ti02 slurry passed over the sieve, but without silica or alumina, was then collected on a filter, washed with acetonitrile, and analyzed as before. The two procedures give similar results. Procedure a was preferred, since it permitted several (18) Pelizzetti, E.; Maurino, V.; Minero, C.; Sclafani, A.; Serpone, N.; Hidaka, H. Enuiron. Sei. Technol. 1989,23, 1380.
Minero et al. small samples during the same experiment and ensured improved reproducibility. Analytical determinations of extracted organics were carried out on a Merck-Hitachi HPLC equipped with two pumps, a UVvis detector, and a reverse-phase CIScolumn (125mm long, 5 pm, Lichrwpher, Merck). DFBP was detected at 233 nm (detection limit < 0.1 ppm a t SIN > lo), with a retention time of 3.4 min using a eluent 2080 HzO-acetonitrile (v/v). PFP was detected a t 265 nm with a retention time of 1.4 min using as eluent 55:45 HzO-acetonitrile. Intermediates formed during the irradiation experiment had smaller retention times and did not interfere. Determinations of fluoride and COZare reported elsewhere.17 Kinetics of t h e Degradation Process. In a variety of previously investigated aromatic compounds it has been observed that the disappearance of the organic compound under irradiation in the presence of Ti02 follows generally a pseudo-first-order kinetics.l9 The change in time of the initial substrate concentration, reported in the f i i e s , are then fittedwith an exponential equation
[A,I/[A,l= exp(-kt)
(1)
where [A01 and [At] represent the initial concentration of the substrate and the concentration after time t of irradiation, respectively,and k represents the apparent kinetic rate constant [min-'1. Values appropriate for each experiment are reported in Tables I1 and 111. Reported errors refer to the experiment to which eq 1 was applied. The mean error on k calculated from different experiments is about 20 % .
Results and Discussion Primary Events. The absorption of UV light by Ti02 excites electronsfrom t h e valence band into the conduction band, leaving a hole i n t h e valence band. Conduction band electrons and valence band holes, which escape recombination, are further separated b y an electric field that favors migration of the holes to the irradiated surface. The reducing electrons migrate to the dark side of the particle and later are scavenged by the adsorbed oxygen or b y other electron acceptor^.^ The holes reaching the surface are trapped at t h e highly hydroxylated Ti02 surface. The fate and dynamics of the surface reactive species (02'-and 'OH radicals) are of considerable importance in the subsequent degradation processes. These primary events, as might occur o n colloidal TiOz, have been a matter of extensive ~ t u d i e s . ~
Degradation Process. Once the active oxidizing species are generated at the catalyst surface, the subsequent reaction sequence, which leads to the final degradation of the pollutant, can be envisaged t o occur through several steps. If t h e reaction site is at the surface of the photocatalyst, the following processes can occur: (i) diffusion of t h e reactants from the bulk solution (i,a) or from an "inert" surface (i,b) to the photocatalyst surface; (ii) diffusion of t h e reactants from surface (ii,a) or pore (ii,b) sites t o the active centers at the surface of the photocatalyst; (iii) reaction at the catalytic centers; (iv) diffusion of t h e reaction products from the surface to t h e bulk solution. If the reaction occurs exclusively (or also) i n the bulk solution, the steps are (v) diffusion of photogenerated reactive species from t h e surface to the bulk solution and (vi) reactions in solution or, if the compound t o b e degraded is adsorbed o n an "inert" surface, (vii) reaction of the photogenerated reactive species originating from the catalyst at the surface of t h e "inert" support after migration. An interesting aspect shown by the accumulation of kinetic data on related compounds is that their rates of (19)Terzian, R.; Serpone,N.; Minero, C.; Pelizzetti, E. J . Catal. 1991, 128, 352,and references cited therein.
Role of Adsorption in Photocatalyzed Reactions
disappearance are quite similar.20~21 This is the case for p-alkylphenols which, on going from the parent phenol (52% adsorption on TiOz) to pnonylphenol(>99% adsorbed on TiOz), show only a 3-fold increase in the degradation rate.18 For a highly adsorbing species, paths i,b-iii or v and vii are the two expected routes, whereas for a weakly (or non) adsorbed species, paths i-vi could be contributing. Degradation of Strongly Adsorbed Species. The photocatalyzed degradation of species which are highly adsorbed were investigated by running the following experiments: (a) The extent of adsorption on Ti02 in the dark, as well as on some “inert” supports, such as Si02 and A1203 was verified. Crysene, dioctylquinol, and DFBP are adsorbed 199%. (b) The kinetics and the extent of exchange between the “inert” support and Ti02 in the dark were determined.22 Compounds were adsorbed first on the “inert” support and then mixed with naked photocatalyst. The amount of compound that was transferred to the catalyst as a function of the time was then measured. (c) Finally, the photodegradation of these compounds was investigated. The experiments can be classed, according to the type of support on which the compound is adsorbed a t the beginning of the photodegradation experiment and to the possible presence of the other support, in the following groups: group 1, compounds are adsorbed on TiOz; group 2, compounds are adsorbed on the “inert” support; group 3, compounds are adsorbed on Ti02 and subsequently photodegraded in the presence of the “inert” support; group 4, compounds are adsorbed on the “inert” support and then photodegraded in the presence of TiOz. As far as dioctylquinol and crysene are concerned, the results show that the two compounds are rapidly partitioned between the two support materials and the equilibrium is attained in a few minutes. We noted that a close similarity of the degradation curves exists for the experiments of group 3 and group 4. Although the photodegradation rate obtained in the experiments of group 3 is slightly less than that obtained in group 1,probably due to scattering of the incident light, experiments of group 3 and group 4 suggest that steps ib and ii are fast compared to step iii. Unfortunately, this deduction precludes any ,conclusion as to the ability of the photogenerated active species to leave the catalyst surface. However, it is worth noting that these two hydrophobic compounds can be degraded if they are adsorbed on an “inert” support. To differentiate between steps i-iii and v and vii, it was necessary to find a compound that is strongly adsorbed and that diffuses slowly from surface or pore sites of the “inert” support to the active centers at the surface of the photocatalyst. After screening several compounds, we found that decafluorobiphenyl (DFBP) is strongly adsorbed on A1203and exchanged slowly. Figure 1 shows that DFBP adsorbed on Si02 exchange and equilibrates with Ti02 within 40 min. In the reported experimental conditions, it is partitioned ca. l/3 on Ti02 at equilibrium. When DFBPis initiallyadsorbed onA1203,1 5 %is present on Ti02 after 120 min (see Table 11,experiments Al-A3). (20) Ollis, D. F. Enuiron. Sci. Technol. 1985, 19, 480. (21) Matthews, R. W. A u t . J. Chem. 1987,40,667. (22) (a) Photoadsorption of organic substrates has been shown to be negligibleby measuring the concentration of cyanuric acid (an unusually resistant compound to the photocatalytic degradationzzb) upon UV illumination of a previouslydark-equilibratedsystem (unpublished, C.M. and E.P.). Similar conclusionsemerged from experiments carried out to detect any enhanced capacity of preilluminated TiOdH20 suspensions for adsorbing 2-chlorophenolinjected into the slurry within seconds from the end of illumination (J. Cunningham, personal communication). (b) Pelizzetti, E.; Maurino, V.; Minero, C.; Carlin, V.; Pramauro, E.; Zerbinati, 0.; Tosato, M. L. Enuiron. Sci. Technol. 1990, 24, 1559.
Langmuir, Vol. 8, No. 2, 1992 483 0.4
90,
0
30
60
90
120
time (min.)
Figure 1. Fractional amount of decafluorobiphenyl found on Ti02 P25 particles as function of the time of exchange. DFBP was initially adsorbed on silica or alumina particles (1W200pm
size). Conditions are reported in the Experimental Section. Bars indicate the estimated experimental error (&3%).
It is known that the sorption approach to equilibrium is dependent on the extent of the relative adsorption coefficients and on the particle surface area.23~~~ Then the system DFBP/Ah03/Ti02 seems suitable for obtaining some information on the role of adsorption on photocatalytic degradation processes. In the following section we report the degradation behavior of DFBP in conditions of group 1-group 4, using as inert support alumina or silica. The degradation kinetics of the related but poorly adsorbed compound, the pentafluorophenol (PFP),are reported for reference. Photocatalyzed Degradation of DFBP and PFP. The extent of exchange of the adsorbed molecule between two different surfaces and the degradation experiments are summarized in the Tables I1 and 111. Details of the degradation mechanism of the two compounds will be published e l s e ~ h e r e .From ~ ~ the point of view of this paper, the relevant features of these degradation processes are that a stoichiometric conversion to C02 and fluoride has been assessed at long irradiation times and in the presence of excess 02.25,26 It is worth noting that, in order for the initial compound to disappear, a displacement of one of the fluorine atoms has to occur. Verification of fluoride formation is necessary to test the effective phototransformation and to avoid the problem of mistaking compound disappearance arising from any (unknown) cause, with its very photodegradation. The formation of fluoride is indeed observed in all the experiments reported. The initial formation of an adduct with a reactive species (e.g. *OH)27can give rise to an intermediate which is in equilibrium with the initial compound. The initial transients formed with PFP have also been identified.26 Several interesting features emerge from the results reported in Tables I1 and I11 and depicted in Figures 2-6. DFBP is strongly adsorbed on A1203 and has little, if any, ability to exchange with other materials in the time scale of the photodegradation experiments (experimentsA1 and A3bZ8 Also, the exchange from pores to the surface sites is strongly reduced, as shown by the degradation exper(23) Wu, S.; Gschwend, P. M. Enuiron. Sci. Technol. 1986, 20,717. (24) Karickhoff, S. W.; Morris, K. R. Enuiron. Toxicol. Chem. 1988, 4. 469. ’ (26 Pelizzetti, E.; et al. in preparation. (26) Minero, C.; Aliberti, C.; Pelizzetti, E.; Terzian, R.; Serpone, N. Langmuir 1991, 7, 928. (27) Terzian, R.; Serpone, N.; Draper, R. B.; Fox,M. A.; Pelizzetti, E. Langmuir 1991, 7, oo00. (28)The labelsquoted in the paper and usedto identify the experiments refer to Tables I1 and 111.
Minero et al.
484 Langmuir, Vol. 8, No. 2, 1992
Table 11. Summary of Exwriments on DecafluorobiDhenYP experiment ppm k,mi+ comments 299% adsorbed on Ti02 A1 extent of adsorption P25, Al203, and Si02 equilibrium reached within ca. 40 min A2 adsorbed on SiOz, exchange with Ti02 P25 1 5 % exchanged A3 adsorbed on Al2O3, exchange with Ti02 P25 adsorbed on Ti02 P25 (7.3 f 0.3) x rapidly degraded 56 + hv (2300 nm) A4 39 (3.5 f 0.3) x + hv (2340 nm) A5 degraded 50 (6.8 f 0.6) X A6 adsorbed on Si02 + hu (2300 nm) adsorbed on A1203 (6.4 f 0.1) x 10-3 degraded 47 + hv (2300 nm) A7 49 (6.0 f 0.5) X + hv (2340 nm) A8 67 (1.2 0.2) x 10-3 initially degraded (ca. 20%), then stops A9 adsorbed on Ti02 beads + hv rapidly degraded, just little less than A5 (1.4 f 0.2) X 36 A10 adsorbed on Ti02 P25 + A 1 2 0 3 + hv (4.6 f 0.3) x rapidly degraded, just little less than A4 57 A l l adsorbed on Si02 + Ti02 P25 + hv (2300 nm) 48 (7.4 f 0.1) x 10-3 degraded (similar to AS) A12 adsorbed on A 1 2 0 3 + Ti02 P25 + hv 55 (1.5 f 0.2) X initially rapidly degraded, then rate reduced A13 adsorbed on 4 2 0 3 + Ti02 colloids + hv A14 adsorbed on Si02 + Ti02 beads, not stirred, + hv (2300 nm) 66 (2.8 f 0.5) X l W 3 slowly degraded degraded, faster than A14 70 (8.7 f 0.5) X A15 adsorbed on Si02 + Ti02 beads, stirred, + hv (2300 nm) 50 (1.2 f 0.2) x 10-3 slowly degraded (similar to A9) A16 adsorbed on A 1 2 0 3 + Ti02 beads + hv (1.3 f 0.2) x degraded (see A18) 65 A17 adsorbed on Si02 + HzOz + hv (2300 nm) 55 (1.3 f 0.2) X degraded, no inhibition of A1203 A18 adsorbed on A 1 2 0 3 + HZOZ+ hv (2300nm) surface on 'OH reactivity a Whereas not indicated, light wavelength is 2340 nm; p p m is the initial concentration of DFBP in mg L-I; k is calculated from eq 1; concentrations of TiOz, silica, and alumina are reported in the Experimental Section. ~
no.
Table 111. Summary of Experiments on PentafluorophenoP exDeriment PPmo k,min-1 comments 1 5 % adsorbed on TiOz, SiOz, and 4 2 0 3 extent of adsorption exchanges with solution adsorbed on exchange with Ti02 50 (1.0 f 0.1) x 10-3 very slowly degraded aqueous solution + hv 49 (2.5 f 0.4) X slowly degraded adsorbed on A1203 + hv 50 (8.6 f 0.3) x rapidly degraded (50 mg L-I TiOz) adsorbed on Ti02 P25 + hv 50 (1.3 f 0.1) X degraded adsorbed on Ti02 beads + hv 49 (2.6 f 0.4) X degraded adsorbed on A1203 + Ti02 P25 + hv
no. B1 B2 B3 B4 B5 B6 B7
Light wavelength is 2340 nm; other conditions as in legend of Table 11.
-
p--* --...-
l a d t o r b e d on TiO,/
-
0.8
m
m
-
-
0.6
I
L
\
\
n
E
0.8
no
no
-- ---a
0.6
n
0.4
0.4
n
---
P
I
I
0.2
0.2
-----A
0 0
30
60
90
120
150
180
irradiation time (rnin.) Figure 2. Photodegradation of DFBP adsorbed on Ti02 beads (A9), on Ti02 P25 (A4, A5), and on Ti02 P25 in the presence of alumina particles (A10). Conditions are reported in the Experimental Section and in Table 11.
iment A9; on large and highly porous spheres of TiO2, only a small fraction of DFBP at the surface is degraded prior to the slowing down of the reaction rate. However, under the same conditions, PFP is efficiently degraded (experiment B7 compared with A12 and B6 compared to A9). DFBP is easily degraded when adsorbed on Ti02 (experiments A4 and A5) but very poorly degraded by light when adsorbed on A1203 (experiments A7 and A8) or when a mixture of Ti02 and of compound adsorbed on A1203 is exposed to the light (experiment A12, to be compared with experiments A10). Even more evident is the absence of degradation when DFBP is adsorbed on A1203 and then mixed with Ti02 beads (experiment A16).
0
60
120
180
240
irradiation time (min.) Figure 3. Photodegradation of DFBP adsorbed on Si02 particles (100-200pmsize) in absence of Ti02 (A6), with addedTiO2 beads (A14 = not stirred, A15 = stirred), and with added Ti02 P25 (All). Conditions are reported in the Experimental Section and in Table 11. The stirring of the slurry in the presence of Ti02 beads causes the subdivision of beads and increases the rate of degradation toward the value obtained in the presence of Ti02 P25.
By contrast, when DFBP is adsorbed on Si02, it is easily degraded in the presence of Ti02 P25 (experiment A l l ) or, although slightly less, in the presence of beads (experiments A14 and A15). Degradation in the presence of silica can be usefully compared with the blank run (experiment A6) and with the case of adsorption on Ti02 (experiment A4). On comparison with experiments AlA3, all these experiments suggest that the degradation rate of DFBP is dominated by a low exchange rate (steps ib and iib very slow) and the inability of photogenerated oxidizing species to reach the location of the substrate
Langmuir, Vol. 8, No. 2, 1992 485
Role of Adsorption in Photocatalyzed Reactions
0.2
-j ladsorbed on A120,
o , ,
I
I
0
I
,
,
1
I
~
60
30
,
,
I
90
120
irradiation time (min.1 Figure 4. Photodegradation of DFBP adsorbed on A1203 particles (100-200 pm size) in absence of Ti02 (A@, with added Ti02 beads (A16 = not stirred),with added Ti02 P25 (A12),and in the presence of Ti02 colloids (A13). Conditions are reported in the Experimental Section and in Table 11. Ti02 beads scatter and absorb light, thus preventing the direct photodegradation by light.
0.4! 0
I
I
,
I
10
20
30
40
7
I
I
7
50
'
60
irradiation time lmin.1 Figure 5. Photodegradation of DFBP under direct light exposure (A6, A7) and in the presence of H202 (A17, AB). The rate of degradation is almost indifferent of the supporting material. Conditions are reported in the Experimental Sectionand in Table 11. 1
k -
0.8
ti"
0.6
I
\
ti
k
0.4
I
0.2 0 , 0
r
I
m.,
30
'. - - _ _ _ _ _ _ _ _
, "
I
,
60
>
-
90
"
'
,
'
'
120
I
150
,
'
180
irradiation time (min.1 Figure 6. Photodegradation of pentafluorophenol adsorbed on alumina, without TiOz (B4) and in the presence of Ti02 P25 (B7),and when PFP is adsorbed on Ti02 P25 (B5) or Ti02 beads (B6L Homogeneous photodegradation is reported in B3. Conditions are reported in the Experimental Section and in Table 111.
when it is adsorbed on a inert support or it is located in pores far from the surface of the photocatalytic particle. Finally, the easily exchangeable PFP, like dioctylquinol and crysene, is degraded rapidly, whether it is initially
adsorbed on Al2O3, TiO2, or Ti02 beads (experiments B5, B6, and B7; blank experiment B4). An alternative explanation to the slow degradation rate in the presence of alumina is the possible inhibition of the kinetic rate of degradation due to the surface local environment or to the possible chemisorption of DFBP on alumina. In order to check that the surface of A1203 does not inhibit the degradation of DFBP in case vii, i.e. the reaction of the photogenerated reactive species at the surface of the "inert" support, we run experiments A17A18 and A13. In experiments A17 and A18, DFBP was adsorbed on silica and respectively, and then was illuminated with UV light and in the presence of H202 (see Experimental Section). The compound is degraded when adsorbed on A203 and on Si02 with a the same apparent kinetic constant. This fact let us conclude that, whatever the steady state concentration of the 'OH radical is, the two surfaces do not inhibit the intrinsic degradation rate of the compound. Conversely, in experiment A13, DFBP was adsorbed on A1203 to which Ti02 colloids were later added. During the experiment, the Ti02 colloids tend to aggregate, which increases in the irradiation time and in the sample temperature. Moreover, Ti02 colloids can stick on alumina and can bring reactive photocatalytic centers at the surface or in the pores of alumina, near the locations where DFBP is adsorbed. Experiment A13, when compared with experiments A10 or A12, well support the last hypotheses. In conclusion,both experiments A18 and A13 show that degradation of adsorbed DFBP takes place when photochemically generated 'OH reach the surface of alumina (inexperiment A18) or oxidizingreactive species are locally photogenerated from Ti02 (experiment A13). Thus, from the experiments with DFBP, it can be suggested that steps ib and iib are slow compared to step iii, while steps ia and vi are negligible, because of the extremely low concentration of the compound in solution, and steps v and vii do not take place appreciably. Then, the experiments decidedly provide support for the hypothesis that very little, if any, of the active species generated at the Ti02 surface can reach the surface of the other support. This is in agreement with the suggestion of Turchi and Ollis.ll They calculated the average distance that a hydroxyl radical, the most probable reactive species, can diffuse into solution before reacting. For a slurry, where the particle-to-particle distances are of the order of micrometers, the diffusion of 'OH is limited to a very narrow width near the photocatalyst surface. The experiments with PFP, on the other hand, suggest only that steps ia and ii for this compound are fast within respect to step iii and that no information can be deduced on paths v, vi, and vii. Whereas the present results do not allow clarification of the nature of the reactive species, either trapped holes or hydroxyl radicals, which are responsible of the degradation p r o ~ e s s , they ~ ~ J seem ~ to strongly support that the location site of the reaction is at the surface of the photocatalytic particle, as implied by the results already rep~rted.~~.~~J~ Conclusions
Photocatalyzed degradation processes involvingstrongly adsorbed species indicate that, for many compounds, the degradation rate is not strongly affected by their actual location, whether in solution, on the photocatalyst, or on another "inert" support. A large rate of exchange of the compound between the two supports can quickly supply
486 Langmuir, Vol. 8, No. 2, 1992 the reactive surface with the compound, leading to a facile degradation. Yet, decafluorobiphenylis stronglyadsorbed and exchange with other supports is slow. The absence of appreciable photocatalyzed degradation when it is adsorbed on alumina shows that in some cases the exchange from pores or from the "inert" surface to the active centers of the photocatalyst may be rate determining, thus suggesting that the migration of photogenerated active species through the solution may have very little relevance in the overall degradation process. The experiments reported here suggest that the degradation reaction occurs within a few nanometers of the photocatalytic surface, to which the substrates migrate and ultimately degrade. This conclusion has relevance in the design of supported catalysts and reactor kinetics. On
Minero et al. the other hand, for most of the hydrophobic and strongly adsorbed compounds, the degradation process also takes place when they are initially located on an "inert" surface, indicating a possible photocatalyzed detoxification of contaminated s ~ p p o r t s . ~ ~ * ~ ~
Acknowledgment. This research received generous support from EEC (Contract EV4V-0068-C(CD)),Regione Piemonte and Eniricerche. Registry No. DFBP, 434-90-2; PFP, 771-61-9; SiOz, 763186-9;Azo3,1344-28-1;TiOz, 13463-67-7; HzOz, 7722-84-1; HzO, 7732-18-5; dioctylquinol, 903-19-5; crysene, 218-01-9. (29) Pelizzetti, E.; Carlin, V.; Maurino, V.; Minero, C.; Dolci, M.; Marchesini, A. Soil Sci. 1990, 150, 5230.