Photocatalytic Degradation of Aromatic Pollutants: A Pivotal Role of

Apr 12, 2012 - Photocatalytic Degradation of Aromatic Pollutants: A Pivotal Role of Conduction Band Electron in Distribution of Hydroxylated Intermedi...
0 downloads 0 Views 345KB Size
Article pubs.acs.org/est

Photocatalytic Degradation of Aromatic Pollutants: A Pivotal Role of Conduction Band Electron in Distribution of Hydroxylated Intermediates Yue Li, Bo Wen, Wanhong Ma, Chuncheng Chen,* and Jincai Zhao Beijing National Laboratory for Molecular Sciences, Key Laboratory of Photochemistry, Institute of Chemistry, The Chinese Academy of Sciences, Beijing 100190, China S Supporting Information *

ABSTRACT: The modulation of the yield distribution of intermediates formed in the photocatalytic degradation of organic pollutants is of extreme importance for the application of photocatalysis in environmental cleanup, as different intermediates usually exhibit distinct biological toxicity and secondary reactivity. In this paper, we report that the distribution of monohydroxylated intermediates (m-, p- and o-) formed during the photocatalytic oxidation of aromatic compounds changes with the variation of reaction conditions, such as O2 partial pressure and substrate concentration. By detailed product analysis, theoretical calculation, and oxygen isotope labeling experiments, we show that these changes are due to the selective reduction of HOadduct radicals (the precursors of hydroxylated intermediates) by conduction band electrons (ecb−) back to the original substrate, that is, p- and o-HO-adduct radicals are more susceptible to ecb− than the m- one. Our experiments give an example that, even under oxidative conditions, the yield distribution of isomeric intermediates can be modulated by ecb−-initiated reduction. This study also illustrates that the unique redox characteristics of photocatalysis, that is, both oxidation and reduction reactions take place on or near the surface of a single nanoparticle, can provide opportunities for the reaction control.



INTRODUCTION Semiconductor photocatalysis has been extensively studied in the past 30 years as a promising method of environmental cleanup and sterilization.1 During photocatalysis, the absorption of light with energy larger than the band gap of a semiconductor, such as TiO2, generates oxidative valence band holes (hvb+) and reductive conduction band electrons (ecb−), which can initiate various redox reactions. Because of the strong oxidation ability of hvb+ and reactive oxygen species (·OH, ·OOH, and H2O2), which are formed from the hvb+ oxidation of H2O and ecb− reduction of O2, most organic compounds can be oxidized, even mineralized to CO2 and H2O in the photocatalytic systems. However, it is frequently observed that, prior to the complete mineralization, various intermediates are formed and accumulate in the reaction systems. They may be much more toxic than their parent pollutants.2,3 Therefore, a better insight and then modulation of the formation of these intermediates are of fundamental importance for the practical application of photocatalysis. Aromatic ring is the basic constituent of many kinds of organic pollutants, such as dyes, explosives, pesticides, and pharmaceuticals. Aromatic compounds, such as benzoic acid, are the most frequently used model substrates to investigate the photocatalytic mechanism4,5 and to test the activity of the photocatalysts.6,7 In the photocatalytic degradation of aromatic pollutants, hydroxylated products are always among the main © 2012 American Chemical Society

intermediates, especially at the beginning stage of the reaction.8,9 Hydroxylation was also regarded as the ratedetermining step of the whole photocatalytic mineralization of aromatic pollutants.10 Because the aromatic ring has different positions that the hydroxyl group may add onto, different regioisomeric hydroxylated products might be formed. For example, the hydroxylation of a monosubstituted benzene derivative can give rise to three isomers, i.e., meta- (m-), para- (p-), and ortho- (o-) hydroxylated products with regard to the original substituent. The yield distribution of these isomers is crucial to the photocatalytic detoxification of aromatic pollutants, because the position of the hydroxyl group usually affects the metabolism and toxicity of the substituted phenol (for example, osubstituted chlorophenols are generally less toxic than the mand p- ones),6,11,12 as well as its reactivity in the secondary oxidation.13,14 However, in the literature, the reported yield distribution of hydroxylated intermediates varied drastically in the regular TiO2 photocatalytic systems even for the same substrate, as shown in Table S1 in the Supporting Information by taking benzoic acid as an example.4,15−18 It is indicated that Received: Revised: Accepted: Published: 5093

February April 11, April 12, April 12,

16, 2012 2012 2012 2012

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

In this study, the effect of reaction conditions on the yield distribution of hydroxylated intermediates formed in the photocatalytic degradation of aromatics is investigated. The primary model substrate used in this study is benzoic acid (BA), which is one of the main decomposition products of alkyl-substituted benzenes.25 Its hydroxylated intermediates, mono- and poly hydroxylbenzoic acids (e.g., syringic and protocatechuic acids), widely exist in most plant tissues as secondary metabolites,26 and in agroindustrial effluents,27 which are usually characterized by low biodegradability and high ecotoxicity. We find that the yield distribution of hydroxylated intermediates can be changed by reaction conditions, such as O2 partial pressure (PO2) and substrate concentration. By detailed product analysis, theoretical calculation and oxygen isotope labeling experiments, we show that the selective reduction of the formed HO-adduct radicals by ecb− (process III in Scheme 1) remarkably affects the yield distribution of hydroxylated intermediates. These findings would be greatly helpful to our understanding about the photocatalytic hydroxylation of aromatics, and provide the guideline for the modulation of the hydroxylation regioselectivity.

the distribution of hydroxylated intermediates is highly sensitive to reaction conditions. How reaction conditions influence this isomers distribution is unclear now. Evidently, an in-depth study on the photocatalytic hydroxylation mechanism, particularly on the key step that determines the position of the hydroxyl group, is desirable for our understanding about this issue and the control of isomeric intermediates distribution. Due to the complexity of heterogeneous photocatalytic systems, little specific mechanism information on the regioselectivity of hydroxylation has been gathered so far. For the photocatalytic hydroxylation of aromatic compounds, the products yield distribution is believed to be primarily determined by the electronic property of the original substituent. For example, it was reported that electron donor group conduces to the production of p- and o-hydroxyl derivatives, while all three monohydroxyl derivatives are obtained if the original substitute is an electron withdrawing one.17,19 As shown in earlier studies, the hydroxylation of aromatic compounds involves a key intermediate, HO-adduct radical (Scheme 1), which is formed via the addition of ·OH Scheme 1. Hydroxylation of Aromatic Compounds under Photocatalytic Conditions



EXPERIMENTAL SECTION Materials. TiO2 (P25, ca. 80% anatase, 20% rutile; surface area, ca. 50 m2/g) was kindly supplied by the Degussa Company. Benzoic acid (BA), nitrobenzene (NB), and their hydroxylated products (m-, p-, o-HO-BAs and m-, p-, o-HONBs) were all of analytical grade, and offered by the Beijing Chemical Company. H218O (18O: 98%) was purchased from Jiangsu Changshu Chemical Limited. Chlorotrimethylsilane (TMSCl) and 1,1,1,3,3,3-hexamethyldisilazane (HMDS) were purchased from Acros Organics. All reagents were used as received without further purification. Photocatalytic Reactions. The light source used in this study was a 100 W high pressure Hg Lamp (ToshibaIn SHL100UVQ). In a typical photocatalytic reaction, 0.1 g of TiO2 was dispersed in 100 mL of aqueous solution of BA. Prior to irradiation, the suspension was magnetically stirred in the dark for about 30 min to ensure the establishment of an adsorption/ desorption equilibrium. The photocatalytic oxidation was carried out in aerial or mixed O2/Ar atmosphere (O2%: 3, 10, 20, 40, 60, 80 or 100, modulated by flow-rate controllers) with total pressure of 1 atm. To obtain higher PO2, pure O2 atmosphere with 2 atm of pressure was also used. In pHcontrolled experiments, the pH of the reaction system was adjusted by HClO4 or NaOH. During the first 90 min of the reaction, the pH change of the suspension was found to be less than 0.1 in these experiments. At a given time, ca. 5-mL aliquot was collected, centrifuged, and then filtered to remove the photocatalyst. The filtrate was subject to HPLC analysis. During HPLC analysis, each species was quantified at its absorption maximum (255 nm for p-HO-BA, 236 nm for mand o-HO-BAs, 230 nm for BA) and compared with authentic sample. The repeatability error of HPLC quantification is less than 1%. The photocatalytic degradation of NB was carried out under the same conditions as those of BA, and its hydroxylated intermediates were also quantified by HPLC (at 208 nm). Control experiments showed that both TiO2 and UV irradiation are necessary for the degradation of BA and NB. Isotope Labeled Reactions. In oxygen isotope labeling experiments, BA was photocatalytically oxidized in 1 mL of TiO2/H218O suspension. Because the volume of the suspension

Scheme 2. Formation of HO-Adduct Radicals4,15,20

(path a in Scheme 2)4,15 or via direct hvb+ oxidation followed by hydrolysis (path b in Scheme 2)20 under photocatalytic conditions. The further oxidation of this radical would lead to the formation of the hydroxylated product (process II). Accordingly, both the generation (process I) and further reactions (including process II) of HO-adduct radicals are important for the yield distribution of hydroxylated products. Previous reports on the distribution of hydroxylated products, in both photocatalysis and homogeneous oxidation systems, focused mainly on the formation of HO-adduct radicals (process I).4,21 Through either path a or b, the HO-addition position of process I is influenced only by the electronic structure of the aromatic ring. In process II,20,22−24 HO-adduct radicals undergo simply one-electron oxidation and deprotonation, and transform to the corresponding hydroxylated products, which means that the yield distribution of isomeric HO-adduct radicals with different HO positions would be retained to the steady-state hydroxylated products. Obviously, only processes I and II cannot explain the sensitivity of isomeric products distribution to photocatalytic conditions. 5094

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

Figure 1. Change of the yield distribution of HO-BAs formed in the photocatalytic oxidation of BA along with (A) O2 partial pressure (PO2) at cBA0 = 20 mM and (B) initial concentration of BA (cBA0) under open air conditions (PO2 ∼0.21 atm).

performed with a generalized gradient approximation (GGA) exchange-correlation functional of B-P86 and a basis set of defSV(P), as incorporated to the TURBOMOLE (V6.0) package. The reduction potentials of the HO-adduct radicals of BA and NB were computed at B3LYP/6-311++G(2df,2p)//B3LYP/631+G(d) level, using polarized continuum model (PCM) to describe the solvent.28,29 More details on the calculation are given in the Supporting Information.

is small, to ensure the photocatalyst concentrations would be identical, higher TiO2 concentration (2 g/L) was applied in these experiments. The oxidant O2 was in its natural isotope abundance (from atmosphere or O2 cylinder, 16O2 for short). After irradiation, the oxygen isotope abundances of m-, p- and o-HO-BAs were evaluated by HPLC-ESI method (Agilent LC 1200/Ion Trap 6310) with C-18 column (250 mm × 2.1 mm), and the results were corrected with the oxygen isotope abundance of solvent H2O and the natural isotope abundance of the hydroxylated intermediate to determine the proportion of the origination of the hydroxyl O-atom: H 2O% =

Ap − An Aw − An

× 100

O2 % = 100 − H 2O%



RESULTS AND DISCUSSION The photocatalytic degradation of BA was carried out in aqueous suspension of TiO2 with UV irradiation. An analysis of the consumption of BA and the total yield of monohydroxylated intermediates (m-, p-, and o-HO-BAs) showed that HOBAs are the main primary photocatalytic oxidation intermediates of BA (the overall formation rate of three HO-BAs accounting for ca. 50% of the consumption rate of BA, see Figure S1A for more details). In this study, we focus mainly on the formation proportions of these three HO-BAs. The yield distribution of HO-BAs is expected to be determined by their formation and secondary consumption. To minimize the disturbance caused by the secondary reactions of these hydroxylated intermediates, only the initial stage of the photocatalytic oxidation is analyzed. In this period, the concentrations of HO-BAs are low, and their secondary reactions are not remarkable, so the accumulation of HO-BAs is determined mainly by their formation. As shown in Table S2, in all the tested systems, the difference in the relative yield of HO-BA between 45 and 90 min reactions was normally less than 1%, which means that the yield distribution of hydroxylated intermediates is stable against irradiation time in the first 90 min of the reaction, although this yield distribution may change slightly with further prolongation of the reaction time (insets of Figures S1B and S1C). Therefore, in the following study, the average values of the relative yields of 45 and 90 min reactions are used to characterize the yield distribution of HO-BAs. The yield distribution of HO-BAs was first examined under different PO2. As shown in Figure 1A, in the open air conditions (PO2 ∼0.21 atm), the relative yields of m-, p-, and o-HO-BAs were 58.3%, 19.3%, and 22.4%, respectively. At lower PO2, the relative yield of m-HO-BA increased, while the proportions of p- and o-HO-BAs decreased accordingly. For example, in the Ar atmosphere with 3% O2 (PO2 = 0.03 atm), the relative yields of these three hydroxylated intermediates became 76.3%, 14.0%, and 9.7%, respectively, that is, the yield ratios of m- to p- and o-

(1) (2)

Ap, An, and Aw are the 18O percentages of the measured isotope abundance of the intermediate, natural isotope abundance of the intermediate, and measured isotope abundance of solvent H2O, respectively. In the measurement of the isotope abundances of the hydroxyl O-atoms in the hydroxylated intermediates (m-, pand o-HO-NBs) formed in the photocatalytic oxidation of NB, the reacted suspension was extracted with 5 mL of ethyl ether. The organic phase was dried with anhydrous Na2SO4 and then evaporated to about 0.1 mL by Ar purging. The concentrated sample was treated with 100 μL of HMDS and 50 μL of TMSCl, and then centrifuged. The TMS derivations of HONBs were analyzed by GC-MS (Thermo-Finingan; Trace 2000/Trace DSQ) with DB-5MS column (30 m × 0.25 mm).8 The MS peaks corresponding to [M − CH3]+ (m/z = 196, 198) were used to calculate the oxygen isotope abundance. The possibility of the O-atom scramble from solvent H2O into BA, NB, or their hydroxylated intermediates was excluded by blank exchanging experiments. Quantum Calculations. The TiO2 cluster with 108 atoms was initially constructed on an anatase lattice. The cluster adsorbed HO-adduct radical was optimized to its lowest energy configuration by spin-unrestricted DFT. No constraint was imposed on the geometry. After that, the single-point energy calculation on the corresponding anion was carried out to model the reduction of TiO2/HO-adduct radical complex by extra ecb−. The electron density difference between neutral TiO2/HO-adduct radical complex and the corresponding anion would reflect the distribution of this extra ecb−. The geometry optimizations and single-point energy calculations were 5095

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

HO-BAs increased from 3.0 to 5.5 and from 2.6 to 7.9, respectively. On the other hand, with the increase of PO2, the formation of m-HO-BA is depressed relative to the p- and oones. The relative yields were 47.1%, 26.0%, and 26.9%, respectively, when PO2 was 2 atm. Similar effect of PO2 on the hydroxylated intermediates distribution was also observed when nitrobenzene (NB), another pollutant, was used as substrate (Table 1). In the

To shed light on the role of this recombination process in the photocatalytic hydroxylation, we calculated the electron density change after another electron is given to TiO2 (modeled by a TiO2 cluster with 108 atoms) with an adsorbed HO-adduct radical of BA by DFT method. As shown in Figure 2, in the p-

Table 1. Yield Distribution of HO-NBs Formed in the Photocatalytic Oxidation of NB under Different PO2a

a

PO2 (atm)

m-HO-NB (%)

p-HO-NB (%)

o-HO-NB (%)

0.1 0.2 1.0

63.1 47.1 42.9

25.7 30.8 33.9

11.2 22.1 23.2

1 g/L TiO2 (P25), cNB0 = 5 mM, in mixed O2 and Ar atmosphere.

Figure 2. Electron density difference of TiO2-adsorbed (A) m-, (B) p-, and (C) o-HO-BA radicals in the absence and presence of one extra electron.

photocatalytic hydroxylation of NB, as PO2 increased from 0.1 to 1 atm, the relative yield of m-HO-NB decreased from 63.1% to 43.0%, while the proportions of the p- and o-hydroxylated ones increased from 25.8% to 33.9% and from 11.2% to 23.2%, respectively. We also examined the photocatalytic degradation of phenol and aniline, but no m-hydroxylated intermediate was detected in these systems (discussed below). It was also observed that the yield distribution of hydroxylated intermediates was strongly dependent on the concentration of the substrate (Figure 1B). At 1 mM of BA, the relative yields of m- and p-HO-BAs were 53.2% and 24.7%, respectively. They changed gradually to 58.8% and 18.0%, respectively, as the initial BA concentration increased to 25 mM. This means that high substrate concentration favors the formation of m-HO-BA, while it disfavors the generation of pHO-BA. Because the adsorption ability and consumption rates of m- and p-HO-BAs are nearly identical (Figures S2 and S3), it is unreasonable to ascribe the opposite trends in their relative yields to their secondary consumption. Therefore, the change in the yield distribution would be induced during the formation of hydroxylated intermediates. As discussed above (Scheme 1), if the photocatalytic hydroxylation only contains processes I and II, neither PO2 nor substrate concentration is able to influence the products yield distribution. Therefore, we consider the reverse of process I, that is, the conversion of HO-adduct radical back to the original substrate molecule (process III). This process is in competition with process II, and determines the efficiency of the conversion from HO-adduct radical to the corresponding hydroxylated product. Because the addition of ·OH onto aromatic ring is highly exothermic (normally 10−20 kcal/mol, the corresponding equilibrium constant is 107−1014),30,31 the direct dissociation of HO-adduct radical to the substrate molecule and ·OH is not likely to play a significant role in the reaction. However, if the HO-adduct radical accepts an electron, it would dissociate facilely to the original aromatic molecule and OH− (eq 3). In photocatalytic oxidation systems, the dominant reductive species is ecb−. Accordingly, it is entirely possible that ecb− initiates the reductive dissociation of HOadduct radical. This process can be regarded as one of the recombination pathways between hvb+−ecb− pair or back reactions, as proposed by Minero, et al.7,32,33

and o- cases, the added electron distributes mainly on the HOBA radical. In contrast, this electron spreads predominantly over the d-orbits of Ti-atoms, which construct the conduction band of TiO2, when m-HO-BA radical is adsorbed on the TiO2 cluster. These results imply that p- and o-HO-BA radicals can be easily reduced by ecb−, while the reduction of m-HO-BA radical is hindered. Confirmatively, we also calculated the standard reduction potentials (E0 vs NHE) of these HO-adduct radicals (Table S3) by DFT method in combination with polarized continuum solvation model (PCM).28 The calculated E0s of p- and o-HO-BA radicals are −0.22 and −0.27 V, respectively, while m-HO-BA radical has a E0 of −0.66 V. It is indicated that p- and o-HO-BA radicals are much easier to be reduced than the m- one. On the other hand, the E0 of the ecb− of TiO2 is estimated to be −0.29 V in the present photocatalytic system,34 which is slightly lower than the E0s of p- and o-HO-BA radicals, but much higher than that of mHO-BA radical. These results mean that ecb−, if available, is able to reduce p- and o-HO-BA radicals, whereas the recombination between m-HO-BA radical and ecb− is thermodynamically unfavorable, which is in good agreement with the calculation shown in Figure 2. Similar results were also obtained in the calculations about the E0s of the HO-adduct radicals of NB (Table S3). On the basis of these calculation results, we attribute the yield distribution change of hydroxylated intermediates with reaction conditions to the selective reductive dissociation of HO-adduct radicals initiated by ecb− (Scheme 3), that is, the difference in the reduction susceptibility among HO-BA radicals makes the modulation of the relative yields of hydroxylated intermediates possible. It was reported that the reduction of O2 on TiO2 does not occur at high enough rate to match the oxidation of BA,35 so the accumulation of ecb− is inevitable during the photocatalytic reaction. Because O2 is the main electron acceptor in the photocatalytic systems, the decrease of PO2 would aggravate the accumulation of ecb− and favor the reduction of HO-adduct radicals. In addition, O2 is able to oxidize HO-BA radicals to the final hydroxylated products. Therefore, in the cases of p- and o-HO-BA radicals, 5096

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

Scheme 3. Proposed Mechanism for the Effect of ecb− on the Yield Distribution of Hydroxylated Intermediates

radicals, and thus lower the relative yields of p- and ohydroxylated intermediates. It should be pointed out that the increase of BA concentration is also able to facilitate charge separation by capturing hvb+ and/or inhibit the scavenging of ecb− by the formed H2O2.24 Both effects would lead to the enhancement of the accumulation of ecb−. It should also be pointed out that, although this study highlights process III, the yield distribution of hydroxylated intermediates under certain reaction conditions is determined by all the processes shown in Scheme 1. As for phenol and aniline, rare m-hydroxylated intermediates were detected in their photocatalytic degradation. The lack of m-hydroxylated products was also observed in earlier studies on the radiationinduced homogeneous hydroxylation of these substrates,41,42 in which process III or similar back reactions is unviable. These results suggest that few m-HO-adduct radicals are formed during their hydroxylation, probably due to the electrondonating property of the hydroxyl and amido groups. In other words, the products distribution (lack of m- products) during the photocatalytic hydroxylation of phenol and aniline is mainly determined in process I. In contrast, all the three HO-adduct radicals are formed in the hydroxylation of BA and NB, accordingly, the effect of process III on the products distribution would be relatively remarkable. Our previous work24 indicated that the hydroxylation process on the TiO2 surface incorporates more of the O-atom from H2O into the hydroxyl group of the product than that occurring in the bulk solution, which forms more of the hydroxylated product containing O2-derived O-atom. According to our mechanism, because the proposed recombination process can only take place on the surface of TiO2, the products distribution of surface and bulk solution hydroxylation should be different. When the photocatalytic reaction is carried out in oxygen isotope labeled systems (18O2/H216O or 16O2/H218O), it is anticipated that the formed isomeric hydroxylated intermediates should have different hydroxyl oxygen isotope abundances. Therefore, we carried out the photocatalytic oxidation of BA in 18O-enriched water (H218O) by using 16O2 as oxidant. As shown in Figure 3, besides the overall change of the hydroxyl oxygen isotope abundances of HO-BAs,24 a remarkable isotope abundance difference among three hydroxylated isomers was observed in all the experiments. This difference can reach as high as 11% (between m- and oHO-BAs, cBA0 = 25 mM, in open air), so it cannot be explained

lower PO2 would facilitate the proposed recombination process and depress their further oxidation, and consequently lower the yields of corresponding hydroxylated intermediates. On the other hand, the accumulation of ecb− would have little effect on the formation of m-HO-BA, since the reduction of m-HO-BA radical by ecb− is thermodynamically unfavorable. As a result, the relative yield of m-HO-BA increases at low PO2. It has been proposed that high substrate concentration favors the back reactions in the photocatalysis systems.7,32,33 The effect of substrate concentration on the regioselectivity of hydroxylation may also be explained by the competitive adsorption between BA and O2. Both BA and O2 are adsorbed on surface Tiatoms.36,37 The measured adsorption constant of BA on TiO2 is 2.78 mM−1 (Figure S4), which is comparable to that of O2 (8.8 mM−1).38 At the typical BA concentration applied in this study (20 mM), up to 98% of the maximum adsorption of BA is achieved. Hence, the adsorption of substrate would impede the O2 adsorption on TiO2 surface.39,40 The increase of substrate concentration would have a similar effect to the decrease of PO2, i.e., higher BA concentration would also exaggerate the accumulation of ecb−, favor the reduction of p- and o-HO-BA

Figure 3. Abundances of 18O in the hydroxyl groups of HO-BAs formed in H218O isotope labeled photocatalytic systems with (A) cBA0 = 20 mM, under various PO2 (16O2) and (B) different cBA0, under aerated conditions. Two mg of TiO2 (P25), 1 mL of H218O, irradiation time = 2 h. The insets illustrate the change of the oxygen isotope abundance difference with reaction conditions. 5097

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

the isotope abundance difference observed in photocatalytic reactions. Moreover, the dehydration−hydration process should be highly pH-dependent.20,23 However, such dependence was not experimentally observed, that is, the yield distribution of HO-BAs was not remarkably affected by the pH of the photocatalytic system (the yield ratio of m-HO-BA to pHO-BA was 2.25 when pH was 2.8, 2.22 when pH was 3.6, 2.11 when pH was 5.8, cBA0 = 2 mM, at irradiation of 90 min). Therefore, we conclude that this equilibrium, even though it occurs, can only play a negligible role in the photocatalytic hydroxylation of BA.

by the error of MS measurement (ca. 0.6%, identified by the standard deviation of repetitive measurements). The 18O abundance of m-HO-BA was generally higher than those of the p- and o-hydroxylated ones. Similar isotope abundance difference was also observed in the photocatalytic hydroxylation of NB (Figure S5). In the bulk solution, few reductive species are present, so all the formed m-, p-, and o-HO-BA radicals would remain and produce the corresponding hydroxylated intermediates. However, on the surface of TiO2, due to the proposed recombination process which consumes part of pand o-HO-BA radicals, the relative yields of p- and o-HO-BAs are reduced and the proportion of m-HO-BA is enhanced correspondingly. To verify this argument, we displaced surfaceadsorbed BA by F−, thus making the hydroxylation reaction occur mainly in the bulk solution7 and inhibited the surface reduction of HO-adduct radicals. The measured relative yield of m-HO-BA was 39.4%, which was much lower than that of naked TiO 2 photocatalytic system (58.3%), while the proportions of p- and o-HO-BAs were higher (21.8% vs 19.3% and 38.8% vs 22.4%, respectively). It is indicated that this surface recombination process really enhances the relative yield of m-HO-BA and lowers the proportions of p- and ohydroxylated intermediates. Because the oxygen incorporation from H218O into the product is dominant for the surface hydroxylation process, which gives a higher m-HO-BA proportion, the overall 18O abundance in m-HO-BA should be higher than those in p- and o-HO-BAs. In addition, it was also observed that the difference in the hydroxyl oxygen isotope abundance increased stably with the decrease of PO2 and the increase of substrate concentration (insets of Figure 3), which is a trend similar to that of the yield distribution, suggesting that this isotope abundance difference is really caused by the regioselectivity difference between surface and bulk solution hydroxylation. The difference in the oxygen isotope abundance also confirms that the susceptibility of the yield distribution to reaction conditions originates from the change in the generation distribution of hydroxylated intermediates, but not from their different secondary consumption rates, otherwise, no difference in the isotope abundance should be observed. It has been proposed that some HO-adduct radical of aromatic compound can exchange its HO with solvent H2O through the dehydration−hydration process as shown in Scheme 4,20,23 which can introduce the 18O-atom of H218O

hv

H 2O2 → 2·OH

(4)

In summary, based on our experiments and theoretical calculations, we show that the products distribution of photocatalytic hydroxylation can be modulated by the selective recombination between surface adsorbed HO-adduct radical and ecb−. The present study would be helpful to our understanding of photocatalytic hydroxylation and the interpretation of the diversity in the reported yield distribution of isomeric hydroxylated intermediates. Our results suggest that, by tuning the Fermi level of the photocatalyst or by controlling the accumulation of ecb−, the hydroxylated intermediates distribution in the photocatalytic degradation of aromatic pollutants can be modulated. This finding is significant to the practical application of photocatalysis, because the formation of the intermediate with low toxicity and high degradability is desirable for environmental remediation. Our study also implicates that, even under oxidative conditions, the yield distribution of intermediates can be modulated by the reduction of the formed radical species. Considering that radical species are ever-present and diverse in photocatalysis, our concept would be applicable to the photocatalytic degradation of other pollutants.



ASSOCIATED CONTENT

S Supporting Information *

Tables S1−S4 and Figures S1−S5. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; phone/fax: (+86)-10-8261-6495.

Scheme 4. Dehydration−hydration Mechanism20,23

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Financial support from 973 project (2010CB933503), NSFC (20920102034, 21137004, and 20907056), and CAS is gratefully acknowledged.

into the hydroxylated intermediate. The observed isotope abundance difference might also be explained if these isomeric HO-adduct radicals of BA have different exchanging rates. To rule out this possibility, we used H216O2 photolysis as the source of ·16OH (eq 4) to induce the hydroxylation of BA (no TiO2), and the reaction was carried out in H218O (Table S4). Only a trace amount of 18O incorporation was observed (for example, 2.3%, 3.1%, and 3.4% for m-, p-, and o-HO-BAs, respectively, 2 h of irradiation), and the isotope abundance difference among isomeric products was less than 1.1%. In addition, unlike photocatalytic systems, the 18O abundance in m-HO-BA was lower than those in the p- and o-hydroxylated ones. Thus, this exchanging mechanism is insufficient to explain



REFERENCES

(1) Mills, A.; Davies, R. H.; Worsley, D. Water-purification by semiconductor photocatalysis. Chem. Soc. Rev. 1993, 22, 417−425. (2) Svenson, A.; Kaj, L. Photochemical conversion of chlorinated phenolic substances in aquatic media as studied by AOX and microtox tests. Sci. Total Environ. 1989, 78, 89−98. (3) Herrmann, J. M.; Guillard, C.; Arguello, M.; Agüera, A.; Tejedor, A.; Piedra, L.; Fernández-Alba, A. Photocatalytic degradation of pesticide pirimiphos-methyl: Determination of the reaction pathway and identification of intermediate products by various analytical methods. Catal. Today 1999, 54, 353−367.

5098

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099

Environmental Science & Technology

Article

(4) Brezova, V.; Ceppan, M.; Brandsteterova, E.; Breza, M.; Lapcik, L. Photocatalytic hydroxylation of benzoic-acid in aqueous titaniumdioxide suspension. J. Photochem. Photobiol., A 1991, 59, 385−391. (5) Mrowetz, M.; Selli, E. H2O2 evolution during the photocatalytic degradation of organic molecules on fluorinated TiO2. New J. Chem. 2006, 30, 108−114. (6) Pera-Titus, M.; Garcia-Molina, V.; Banos, M. A.; Gimenez, J.; Esplugas, S. Degradation of chlorophenols by means of advanced oxidation processes: A general review. Appl. Catal., B 2004, 47, 219− 256. (7) Vione, D.; Minero, C.; Maurino, V.; Carlotti, A. E.; Picatonotto, T.; Pelizzetti, E. Degradation of phenol and benzoic acid in the presence of a TiO2-based heterogeneous photocatalyst. Appl. Catal., B 2005, 58, 79−88. (8) Li, X.; Cubbage, J. W.; Tetzlaff, T. A.; Jenks, W. S. Photocatalytic degradation of 4-chlorophenol. 1. The hydroquinone pathway. J. Org. Chem. 1999, 64, 8509−8524. (9) Canle, M.; Santaballa, J. A.; Vulliet, E. On the mechanism of TiO2-photocatalyzed degradation of aniline derivatives. J. Photochem. Photobiol., A 2005, 175, 192−200. (10) Thuan, D. B.; Kimura, A.; Ikeda, S.; Matsumura, M. Determination of oxygen sources for oxidation of benzene on TiO2 photocatalysts in aqueous solutions containing molecular oxygen. J. Am. Chem. Soc. 2010, 132, 8453−8458. (11) Demarini, D. M.; Brooks, H. G.; Parkes, D. G. Induction of prophage-lambda by chlorophenols. Environ. Mol. Mutagen. 1990, 15, 1−9. (12) Oconnor, O. A.; Young, L. Y. Effects of six different functional groups and their position on the bacterial metabolism of monosubstituted phenols under anaerobic conditions. Environ. Sci. Technol. 1996, 30, 1419−1428. (13) Parra, S.; Olivero, J.; Pacheco, L.; Pulgarin, C. Structural properties and photoreactivity relationships of substituted phenols in TiO2 suspensions. Appl. Catal., B 2003, 43, 293−301. (14) Karunakaran, C.; Dhanalakshmi, R. Substituent effect on nano TiO2- and ZnO-catalyzed phenol photodegradation rates. Int. J. Chem. Kinet. 2009, 41, 275−283. (15) Matthews, R. W. Hydroxylation reactions induced by nearultraviolet photolysis of aqueous titanium-dioxide suspensions. J. Chem. Soc., Faraday Trans. I 1984, 80, 457−471. (16) Chan, A. H. C.; Chan, C. K.; Barford, J. P.; Porter, J. F. Solar photocatalytic thin film cascade reactor for treatment of benzoic acid containing wastewater. Water Res. 2003, 37, 1125−1135. (17) Palmisano, G.; Addamo, M.; Augugliaro, V.; Caronna, T.; Di Paola, A.; Lopez, E. G.; Loddo, V.; Marci, G.; Palmisano, L.; Schiavello, M. Selectivity of hydroxyl radical in the partial oxidation of aromatic compounds in heterogeneous photocatalysis. Catal. Today 2007, 122, 118−127. (18) Velegraki, T.; Mantzavinos, D. Conversion of benzoic acid during TiO2-mediated photocatalytic degradation in water. Chem. Eng. J. 2008, 140, 15−21. (19) Palmisano, G.; Addamo, M.; Augugliaro, V.; Caronna, T.; Garcia-Lopez, E.; Loddo, V.; Palmisano, L. Influence of the substituent on selective photocatalytic oxidation of aromatic compounds in aqueous TiO2 suspensions. Chem. Commun. 2006, 1012−1014. (20) Goldstein, S.; Czapski, G.; Rabani, J. Oxidation of phenol by radiolytically generated ·OH and chemically generated SO4•‑. A distinction between ·OH transfer and hole oxidation in the photolysis of TiO2 colloid solution. J. Phys. Chem. 1994, 98, 6586−6591. (21) Eberhardt, M. K. Radiation-induced homolytic aromatic substitution. 6. The effect of metal ions on the hydroxylation of benzonitrile, anisole, and fluorobenzene. J. Phys. Chem. 1977, 81, 1051−1057. (22) Eberhardt, M. K. Radiation-induced homolytic aromatic substitution. 4. Effect of metal ions on the hydroxylation of nitrobenzene. J. Phys. Chem. 1975, 79, 1913−1916. (23) Walling, C.; Camaioni, D. M.; Kim, S. S. Aromatic hydroxylation by peroxydisulfate. J. Am. Chem. Soc. 1978, 100, 4814−4818.

(24) Li, Y.; Wen, B.; Yu, C.; Chen, C.; Ji, H.; Ma, W.; Zhao, J. Pathway of oxygen incorporation from O2 in TiO2 photocatalytic hydroxylation of aromatics: Oxygen isotope labeling studies. Chem. Eur. J. 2012, 18, 2030−2039. (25) Hansen, H. P. Photochemical degradation of petroleum hydrocarbon surface films on seawater. Mar. Chem. 1975, 3, 183−195. (26) Khadem, S.; Marles, R. J. Monocyclic phenolic acids; hydroxyand polyhydroxybenzoic acids: Occurrence and recent bioactivity studies. Molecules 2010, 15, 7985−8005. (27) Martins, R. C.; Quinta-Ferreira, R. M. Manganese-based catalysts for the catalytic remediation of phenolic acids by ozone. Ozone Sci. Eng. 2009, 31, 402−411. (28) Fu, Y.; Liu, L.; Yu, H.-Z.; Wang, Y.-M.; Guo, Q.-X. Quantumchemical predictions of absolute standard redox potentials of diverse organic molecules and free radicals in acetonitrile. J. Am. Chem. Soc. 2005, 127, 7227−7234. (29) Namazian, M.; Halvani, S. Calculations of pK(a) values of carboxylic acids in aqueous solution using density functional theory. J. Chem. Thermodyn. 2006, 38, 1495−1502. (30) Lundqvist, M. J.; Eriksson, L. A. Hydroxyl radical reactions with phenol as a model for generation of biologically reactive tyrosyl radicals. J. Phys. Chem. B 2000, 104, 848−855. (31) DeMatteo, M. P.; Poole, J. S.; Shi, X.; Sachdeva, R.; Hatcher, P. G.; Hadad, C. M.; Platz, M. S. On the electrophilicity of hydroxyl radical: A laser flash photolysis and computational study. J. Am. Chem. Soc. 2005, 127, 7094−7109. (32) Minero, C. Kinetic analysis of photoinduced reactions at the water semiconductor interface. Catal. Today 1999, 54, 205−216. (33) Minero, C.; Vione, D. A quantitative evalution of the photocatalytic performance of TiO2 slurries. Appl. Catal., B 2006, 67, 257−269. (34) Dung, D.; Ramsden, J.; Graetzel, M. Dynamics of interfacial electron-transfer processes in colloidal semiconductor systems. J. Am. Chem. Soc. 1982, 104, 2977−2985. (35) Izumi, I.; Fan, F. R. F.; Bard, A. J. Heterogeneous photocatalytic decomposition of benzoic acid and adipic acid on platinized TiO2 powder. The photo-Kolbe decarboxylative route to the breakdown of the benzene ring and to the production of butane. J. Phys. Chem. 1981, 85, 218−223. (36) Tunesi, S.; Anderson, M. A. Surface effects in photochemistry: An in situ cylindrical internal reflection−Fourier transform infrared investigation of the effect of ring substituents on chemisorption onto TiO2 ceramic membranes. Langmuir 1992, 8, 487−495. (37) Schaub, R.; Wahlstrom, E.; Ronnau, A.; Laegsgaard, E.; Stensgaard, I.; Besenbacher, F. Oxygen-mediated diffusion of oxygen vacancies on the TiO2(110) surface. Science 2003, 299, 377−379. (38) Okamoto, K.; Yamamoto, Y.; Tanaka, H.; Itaya, A. Kinetics of heterogeneous photocatalytic decomposition of phenol over anatase TiO2 powder. Bull. Chem. Soc. Jpn. 1985, 58, 2023−2028. (39) Zhang, H.; Chen, D.; Lv, X.; Wang, Y.; Chang, H.; Li, J. Energyefficient photodegradation of azo dyes with TiO2 nanoparticles based on photoisomerization and alternate UV−Visible light. Environ. Sci. Technol. 2009, 44, 1107−1111. (40) Rideh, L.; Wehrer, A.; Ronze, D.; Zoulalian, A. Photocatalytic degradation of 2-chlorophenol in TiO2 aqueous suspension: Modeling of reaction rate. Ind. Eng. Chem. Res. 1997, 36, 4712−4718. (41) Raghavan, N. V.; Steenken, S. Electrophilic reaction of the OH radical with phenol. Determination of the distribution of isomeric dihydroxycyclohexadienyl radicals. J. Am. Chem. Soc. 1980, 102, 3495− 3499. (42) Solar, S.; Solar, W.; Getoff, N. Resolved multisite OH-attack on aqueous aniline studied by pulse radiolysis. Radiat. Phys. Chem. 1986, 28, 229−234.

5099

dx.doi.org/10.1021/es300655r | Environ. Sci. Technol. 2012, 46, 5093−5099