Langmuir 2004, 20, 5911-5917
5911
Narrow-Band Irradiation of a Homologous Series of Chlorophenols on TiO2: Charge-Transfer Complex Formation and Reactivity Alexander G. Agrios,† Kimberly A. Gray,*,† and Eric Weitz‡ Institute for Environmental Catalysis, Department of Civil and Environmental Engineering, and Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208 Received November 17, 2003. In Final Form: April 5, 2004 The goal of this research was to investigate the formation and reactivity of charge-transfer complexes (CTCs) among a homologous series of chlorophenols on TiO2. We previously showed that 2,4,5-trichlorophenol (245TCP) forms a CTC with Degussa P25, a commercial preparation of TiO2. Here, we probe how light energy influences reactivity and product formation. Slurries of P25 containing 245TCP were irradiated at 360, 400, 430, 480, and 550 nm. At each wavelength, the amount of transformation of 245TCP correlates to the diffuse-reflectance absorbance of a 245TCP/P25 system, supporting the CTC as the cause of reaction. In addition, polymeric products are formed only under wavelengths that excite the CTC, indicating a different reaction mechanism for the CTC than for bandgap excitation of TiO2. We also found a higher quantum efficiency for CTC reactivity than for bandgap activation of the catalyst, suggesting that the photocatalytic efficiency and selectivity can be improved for certain compounds by designing catalytic materials that form CTCs with them. Furthermore, to determine how chlorine substitution patterns affected adsorption and sub-bandgap reactivity, P25 slurries containing phenol, 4-chlorophenol, 2,4-dichlorophenol, or 2,4,6-trichlorophenol were probed following dark contact or irradiation at 360, 430, or 550 nm. With respect to the extent of adsorption, complexation, reaction, and polymerization on P25, the behavior of 245TCP far exceeded that of the other chlorophenols. Among these chlorophenols, only 2,4-dichlorophenol produced a polymeric product. 245TCP is unique among this family of chlorophenols, which we attribute to a chlorine arrangement that leads to a favorable orbital overlap with TiO2 and sterically permits coupling reactions. Our results demonstrate the critical role that charge-transfer complexation can play in determining the rates and products of photocatalytic reactions.
Introduction Environmental applications of TiO2 photocatalysis have been intensively studied.1-5 Ultraviolet illumination of slurries of TiO2 leads to charge separation, reaction with solvent and/or solute molecules, and ultimately to the destruction of a very wide array of organic molecules.1 The two crystalline phases of TiO2 most important in photocatalysis are anatase and rutile. Degussa P25, a widely studied commercial preparation of TiO2, consists of about 80% anatase and 20% rutile intimately associated with each other within particles. Although the reasons are not completely understood, Degussa P25 exhibits significantly higher photocatalytic activity than either pure phase of TiO2 alone. This is due to a combination of effects including enhanced surface affinity, reduced rates of recombination, and an extended range of photoactivity.6 The bandgaps of anatase and rutile, 3.2 and 3.0 V versus NHE, respectively, correspond to wavelengths of 385 and * To whom correspondence should be addressed. Phone: 847/ 467-4252. Fax: 847/491-4011. E-mail:
[email protected]. † Institute for Environmental Catalysis and Department of Civil and Environmental Engineering. ‡ Institute for Environmental Catalysis and Department of Chemistry. (1) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. Rev. 1995, 95, 69-96. (2) Stafford, U.; Gray, K. A.; Kamat, P. V. Heterogen. Chem. Rev. 1996, 3, 77-104. (3) Mills, A.; Le Hunte, S. J. Photochem. Photobiol., A 1997, 108, 1-35. (4) Fox, M. A.; Dulay, M. T. Chem. Rev. 1993, 93, 341-357. (5) Linsebigler, A. L.; Lu, G.; Yates, J. T. J. Chem. Rev. 1995, 95, 735-758. (6) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C. J. Phys. Chem. B 2003, 107, 4545-4549.
410 nm, necessitating ultraviolet light for the excitation of an electron from the valence band of TiO2 to the conduction band. However, mechanisms are known for photocatalytic reactions promoted by visible light. Defect sites within TiO2 and the resulting surface states7-9 give rise to limited intrinsic sub-bandgap activity.10 Various dopants11-19 have been reported to extend the photoactivation threshold into the visible wavelength range. TiO2 can also be sensitized by dyes, whereby the dye absorbs a photon of visible light and enters an excited state, from which it injects an electron into the conduction band of TiO2. This can result in visible-light-activated destruction of the dye20-29 and potentially of other solutes.30 (7) Preusser, S.; Stimming, U.; Tokunaga, S. J. Electrochem. Soc. 1995, 142, 102-111. (8) Newmark, A. R.; Stimming, U. J. Electroanal. Chem. 1986, 204, 197-209. (9) Stimming, U. Langmuir 1987, 3, 423-428. (10) Cao, F.; Oskam, G.; Searson, P. C.; Stipkala, J. M.; Heimer, T. A.; Farzad, F.; Meyer, G. J. J. Phys. Chem. 1995, 99, 11974-11980. (11) Anpo, M.; Ichihashi, Y.; Takeuchi, M.; Yamashita, H. Res. Chem. Intermed. 1998, 24, 143-149. (12) Yamashita, H.; Ichihashi, Y.; Takeuchi, M.; Kishiguchi, S.; Anpo, M. J. Synchrotron Radiat. 1999, 6, 451-452. (13) Kesselman, J. M.; Weres, O.; Lewis, N. S.; Hoffmann, M. R. J. Phys. Chem. B 1997, 101, 2637-2643. (14) Zang, L.; Lange, C.; Abraham, I.; Storck, S.; Maier, W. F.; Kisch, H. J. Phys. Chem. B 1998, 102, 10765-10771. (15) Ohno, T.; Tanigawa, F.; Fujihara, K.; Izumi, S.; Matsumura, M. J. Photochem. Photobiol., A 1999, 127, 107-110. (16) Kutty, T. R. N.; Avudaithai, M. Chem. Phys. Lett. 1989, 163, 93-97. (17) Morris, D.; Dixon, R.; Jones, F. H.; Dou, Y.; Edgell, R. G.; Downes, S. W.; Beamson, G. Phys. Rev. B 1997, 55, 16083-16087. (18) Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B. J. Science 2002, 297, 2243-2245. (19) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269-271.
10.1021/la036165d CCC: $27.50 © 2004 American Chemical Society Published on Web 06/15/2004
5912
Langmuir, Vol. 20, No. 14, 2004
We recently reported31 on a charge-transfer complex (CTC) formed between 2,4,5-trichlorophenol (245TCP) and P25. Light of wavelengths as long as 520 nm can promote an electron from the highest occupied molecular orbital (HOMO) of 245TCP directly into the conduction band of P25. This differs from the sensitization mechanism in that the electron is not first promoted to the lowest unoccupied molecular orbital (LUMO) of the organic molecule. The CTC mechanism can therefore allow visible-light activation of a system in which neither the catalyst nor the adsorbate absorbs visible light by itself. For example, 245TCP is colorless in solution (no absorbance above 340 nm) but takes on a grayish color when adsorbed to P25. The products of sub-bandgap activation of 245TCP/P25 systems are different from those of UV activation. Under UV illumination, 245TCP is transformed into aliphatic products and ultimately CO2, but sub-bandgap light produces 245TCP coupling products ranging in size from dimeric species such as dioxins and dibenzofurans to polymers with atomic masses exceeding 1200 Da. Because they remain chlorinated, these coupling products are undesirable. However, these phenomena suggest that a different parent compound might produce useful polymers under conditions of CTC excitation. For example, unchlorinated polyphenols are manufactured for industrial use and their synthesis by other means is an area of active research.32-35 Our preliminary results with 245TCP raised questions concerning the characteristics of the CTCs formed by other chlorophenols on Degussa P25. For example, would the strength or peak absorbance of the CTC correlate with the number and pattern of chlorine substitutions? Particularly intriguing is the possibility that visible illumination of less-chlorinated chlorophenols on TiO2 might yield dechlorinated polymers and thus be a useful process for recovering polyphenols from pollutants. To characterize these types of trends and differences, experiments were conducted with a homologous series of five chlorophenols: phenol (P), 4-chlorophenol (4CP), 2,4-dichlorophenol (24DCP), 2,4,5-trichlorophenol (245TCP), and 2,4,6trichlorophenol (246TCP). (20) Zhao, J.; Wu, T.; Wu, K.; Oikawa, K.; Hidaka, H.; Serpone, N. Environ. Sci. Technol. 1998, 32, 2394-2400. (21) Wu, T.; Liu, G.; Zhao, J.; Hidaka, H.; Serpone, N. J. Phys. Chem. B 1998, 102, 5845-5851. (22) Liu, G. M.; Wu, T. X.; Zhao, J. C.; Hidaka, H.; Serpone, N. Environ. Sci. Technol. 1999, 33, 2081-2087. (23) Kamat, P. V.; Gevaert, M.; Vinodgopal, K. J. Phys. Chem. B 1997, 101, 4422-4427. (24) Tanaka, K.; Padermpole, K.; Hisanaga, T. Water Res. 2000, 34, 327-333. (25) Kiriakidou, F.; Kondarides, D. I.; Verykios, X. E. Catal. Today 1999, 54, 119-130. (26) Bandara, J.; Mielczarski, J. A.; Kiwi, J. Langmuir 1999, 15, 7680-7687. (27) Dieckmann, M. S.; Gray, K. A. Chemosphere 1994, 28, 10211034. (28) Vinodgopal, K.; Kamat, P. V. J. Photochem. Photobiol., A 1994, 83, 141-146. (29) Vinodgopal, K.; Bedja, I.; Hotchandani, S.; Kamat, P. V. Langmuir 1994, 10, 1767-1771. (30) Thurnauer, M. C.; Rajh, T.; Tiede, D. M. Acta Chem. Scand. 1997, 51, 610-618. (31) Agrios, A. G.; Gray, K. A.; Weitz, E. Langmuir 2003, 19, 14021409. (32) Tonami, H.; Uyama, H.; Kobayashi, S.; Higashimura, H.; Oguchi, T. J. Macromol. Sci., Pure Appl. Chem. 1999, A36, 719-730. (33) Oguchi, T.; Tawaki, S.; Uyama, H.; Kobayashi, S. Macromol. Rapid Commun. 1999, 20, 401-403. (34) Baesjou, P. J.; Driessen, W. L.; Challa, G.; Reedijk, J. Macromolecules 1999, 32, 270-276. (35) Higashimura, H.; Kubota, M.; Shiga, A.; Fujisawa, K.; Morooka, Y.; Uyama, H.; Kobayashi, S. Macromolecules 2000, 33, 19861995.
Agrios et al.
Figure 1. Illustration of the wavelengths used in this work and the excitations that they trigger.
In addition, we were interested in exploring how light wavelength could be used to control the mechanism and/ or product distribution of photocatalytic reactions. The effects of the different wavelengths are best understood with reference to Figure 1. Light of 360 nm can promote electrons across both the anatase and rutile bandgaps. At 400 nm, bandgap excitation occurs on rutile but not on anatase. At 430 nm, bandgap excitation is impossible, but the charge-transfer complex is activated. Based on the diffuse-reflectance spectra of 245TCP and P25, the activity of the charge-transfer complex is lower at 480 nm and nearly zero at 550 nm. We have used a xenon arc lamp apparatus to deliver narrow-band illumination (20 nm wide at 1% transmittance) at these wavelengths, allowing (1) the quantitative determination of the amount of 245TCP transformed at each wavelength and (2) a qualitative evaluation of the products formed at each wavelength. Correlation between the amount of transformation and the diffuse-reflectance absorbance of 245TCP on P25 over the five wavelengths tested provides further insight into the mechanism by which the chargetransfer complex causes chemical transformation under visible light. Differences in byproducts over the different wavelengths may indicate if the reaction can be steered toward useful products. A similar approach was employed with the other chlorophenols except that they were illuminated at 360, 430, and 550 nm. For all chlorophenols, charge-transfer complexes were characterized using diffuse-reflectance spectroscopy. In illumination experiments, the amount of reaction induced by narrow-band irradiation was quantified by highperformance liquid chromatography (HPLC). The byproducts of these reactions were extracted and analyzed by gas chromatography-mass spectrometry (GC-MS). Experimental Section Materials. Degussa P25 was provided courtesy of the DegussaHu¨ls Corp. Anatase was Aldrich, 99.9+%. Phenol and anhydrous sodium sulfate were Fisher, ACS grade. All chlorinated phenols were Aldrich, 98% or 99%. Acetone was Fischer, GC Resolv grade, and hexane was EM, Omnisolv grade. Methanol was EM, HPLC grade. Acetic acid was Fisher, HPLC grade. All chemicals were used as delivered without further purification. All water was treated first by reverse osmosis and ultrafiltration and then by one further treatment depending on the application. For general use (preparation of chlorophenol solutions, washing of glassware, etc.), water was distilled. Water was treated to a resistivity greater than 18 MΩ cm by a Barnstead system for HPLC. Adsorption/Reaction Experiments. Slurries containing TiO2 and a chlorophenol (CP) were prepared and exposed to light or kept in darkness. The loss of the CP from aqueous solution was measured and was taken as the sum of adsorption and reaction. Based on our previous results,31 the loss from solution in darkness was attributed to adsorption only. Sample preparation procedures were as previously described,31 with the following differences. All centrifuge tubes were prerinsed with 2 mL of acetone/hexane (1:1 v/v). All samples were prepared with an initial concentration of 2.5 mM, with two exceptions. Because 246TCP is not soluble in water at 2.5 mM, it was used
Narrow-Band Irradiation of Chlorophenols at an initial concentration of 1.0 mM. In addition to the 2.5 mM experiments, some experiments were conducted using 30 mM 4CP, to obtain an amount of surface adsorption similar to that of 2.5 mM 245TCP. Samples were irradiated for 42 h by a Xe arc lamp (Oriel) (see the Supporting Information for a diagram). A 150-W ozone-free Xe bulb (Oriel 6255) in a housing (Oriel 66902) equipped with a 3.8-cm f/1.0 lens was powered by an Oriel 68907 power supply. Affixed to the condenser lens was a water filter to remove heat (IR) from the light beam, a light-tight filter holder, a mirror to direct the beam downward, and a focusing lens (f ) 50 mm) to cause the collimated beam to converge and subsequently diverge before reaching the samples. The water filter was cooled with a circulating chiller (Neslab Digital One). Samples were positioned 35 cm below the focusing lens, where illumination covered a circle about 25 cm in diameter. The irradiance was measured as 0.6 W/m2 at 550 nm. Based on the projected surface area of the centrifuge tubes, each sample received a total of 184 J, or 5.1 × 1020 photons at 550 nm, over the 42-h experiment. Wavelength ranges were selected using five 5-cm narrow-band interference filters (Oriel). The filters had peak transmittance wavelengths of 360, 400, 430, 480, or 550 nm. The bandwidth was about 14 nm at 10% transmittance for each filter and about 20 nm at 1% transmittance. The filters were selected with the 1% transmittance in mind, so that the transmittance of light wavelengths differing from the center wavelength by more than 10 nm could be considered insignificant. Thus, the 400-nm filter would preclude excitation of anatase by 385-nm light, and the 430-nm filter would prevent excitation of rutile by 410-nm light. The interference filters had different peak transmittances, which tended to increase with wavelength. Therefore, a set of neutral density filters (Oriel) with optical densities of 0.1, 0.15, and 0.2 was used as appropriate to attenuate the light transmitted by higher-wavelength filters, with the result of nearly uniform irradiance of the samples over all wavelengths. Following illumination, tubes were centrifuged for 60 min at 5000 rpm in a Sorvall RC 5B Plus with an SA-600 rotor. Sampling of the supernate and calculations of loss from aqueous solution proceeded as described previously.31 All tubes were decanted following sampling, leaving only the TiO2 paste for extraction. Extraction. Polar (methanol) extractions of TiO2 were performed as previously described.31 Duplicate samples of 245TCP and P25 exposed to 360-nm light, 400-nm light, or darkness were triply extracted. The duplicate results were averaged to ascertain the amounts of adsorption and reaction of 245TCP under each light condition. All chlorophenols under all light conditions were subjected to extractions using a 1:1 (v/v) mixture of acetone/hexane to recover adsorbed nonpolar byproducts. Anhydrous Na2SO4 (2 g) was added as a desiccant to the TiO2 paste in each tube, followed by 20 mL of acetone/hexane. Each tube was vigorously shaken, wrapped in aluminum foil, and allowed to desiccate overnight. Subsequently, the TiO2 paste was dispersed and mixed using a poly(tetrafluoroethylene) (PTFE)-coated spatula cleaned with acetone/hexane. The tube was recapped and allowed to extract for another 24 h. The tube was then centrifuged for 15 min at 5000 rpm, and the supernate was withdrawn using a 5-mL glass syringe (B-D) and filtered through a 0.2-µm syringe filter (Gelman, 25 mm diameter) into a 24-mL borosilicate vial. Both the syringe and the vial were prerinsed with acetone/hexane. The vial was stored at 4 °C until analysis. Immediately prior to analysis, the extract was evaporated under a nitrogen stream to a final volume slightly greater than 1 mL, filtered (Gelman, 0.2µm pores, 13-mm diameter) into a 2-mL autosampler vial (Kimble) prerinsed with acetone/hexane, and evaporated further to 1 mL. To partially compensate for the reduced aqueous concentration of 246TCP, acetone/hexane extracts from duplicate 246TCP/TiO2 samples were pooled and concentrated to 1 mL, effectively doubling the product concentrations. Analysis. HPLC and GC-MS analyses, as well as diffuse reflectance (DR) UV-vis spectroscopy, were described previously.31 In HPLC, the elution times (tR) and analyzed absorbance wavelengths (λ) of the different chlorophenols were as follows: P (3.93 min, 273 nm); 4CP (5.35 min, 284 nm); 24DCP (8.00 min, 284 min); 245TCP (12.81 min, 296 nm); 246TCP (12.89 min, 296 nm). DR spectra were taken of the five phenolic compounds on
Langmuir, Vol. 20, No. 14, 2004 5913
Figure 2. Results of narrow-band irradiation of slurries containing 245TCP and P25. P25 and anatase. Each result is presented as a difference between the compound adsorbed on TiO2 and a control sample without the phenolic compound. These difference spectra were shifted vertically for ease of comparison. With the exception of 246TCP, all DR samples were prepared by mixing 0.5 g of the catalyst with 25 mL of a 2.5 mM aqueous solution of the compound. Because 246TCP is not soluble in water at 2.5 mM, it was dissolved at 2.5 mM in a solvent of 10% methanol/90% water (v/v). Its spectrum was compared to a control of TiO2 in the 10% methanol solvent. To examine how much the presence of methanol alters the DR spectra, 245TCP was prepared using 10% methanol, and its spectrum was compared to that without methanol.
Results and Discussion Wavelength-Dependent Reaction of 245TCP. Slurries containing P25 and 2.5 mM 245TCP were illuminated at light wavelengths of 360, 400, 430, 480, or 550 nm. The 245TCP concentration was measured before and after its exposure to TiO2 and light, and the difference is plotted for each wavelength in Figure 2. Within experimental error, no loss of 245TCP was observed in catalyst-free control samples. Therefore, all 245TCP loss in Figure 2 involves TiO2. Each bar in this graph represents a mean of two experiments. It is useful to compare these results with the spectrum of the CTC formed by 245TCP and P25. A spectrum of the complex itself, that is, the difference between the spectra of a P25/H2O paste and of a P25/ 245TCP/H2O paste, is shown in Figure 3. The shape of Figure 2 is consistent with that of the charge-transfer complex spectrum in Figure 3. Our previous study31 indicated that any increased loss of 245TCP under illumination versus in the dark is due to transformation; that is, adsorption is not increased by illumination. A small amount of chemical transformation occurs at 550 nm, which is likely due to an overlap of tails of the charge-transfer complex and the lamp illumination. Much more transformation is observed at 480 nm, which is consistent with the significant amount of charge transfer seen by diffuse reflectance at that wavelength. The highest amount of reaction occurs in the 400-430 nm range, which is consistent with the 415-nm peak absorbance of the charge-transfer complex. At 400 nm, activation of the rutile
5914
Langmuir, Vol. 20, No. 14, 2004
Agrios et al.
ζ)
Figure 3. A difference DR spectrum showing the 245TCPP25 charge-transfer complex.
Figure 4. Depletion from aqueous solution and recovery by methanol extraction for duplicate 245TCP/P25 samples exposed to light of 360 or 400 nm or darkness.
phase of P25 is possible. The fact that the degree of reaction at 400 nm is only slightly higher than that at 430 nm indicates that CTC excitation is a far more important process than rutile excitation at 400 nm. It was surprising that the extent of reaction was greater at 400 nm than at 360 nm. The onset of anatase excitation was expected to result in far more rapid transformation of 245TCP than excitation of the CTC. This expectation is supported by our previously reported DR spectrum of P25 with adsorbed 245TCP,31 which showed light absorption rapidly increasing with decreasing wavelength below about 410 nm. To confirm that there was more reaction at 400 nm than at 360 nm, methanol extractions were performed from duplicate 245TCP/P25 samples following illumination at each wavelength. As the results show (Figure 4), the amount of reaction, taken as the difference between the amount of 245TCP lost from solution and the amount of 245TCP recovered by extraction, was 3.7 ( 0.7 µmol 245TCP/g TiO2 at 360 nm and 6.2 ( 1.4 µmol/g at 400 nm. These results indicate that there is approximately 70% more transformation of 245TCP at 400 nm, where anatase does not absorb at all but the CTC absorbance is high, than at 360 nm, where anatase absorbs but the CTC absorbance is low. These findings indicate that the CTC mechanism has a higher photonic efficiency than bandgap excitation. Photonic efficiency, ζ, is defined by Serpone36 as
Nmolecules Nphotons
where Nmolecules is the number of molecules transformed, in this case of 245TCP, and Nphotons is the number of photons incident upon the sample. Note that photonic efficiency differs from quantum efficiency, Φ, where the denominator is the number of photons absorbed by the photoactive system.36-38 This quantity is difficult to determine in heterogeneous systems due to scattering and is obviated by the use of ζ instead of Φ. Given that the irradiance is nearly equal across all wavelengths studied and taking into account that the photon irradiance varies with wavelength, we calculate a ζ of 0.0033 at 360 nm and 0.0050 at 400 nm. The higher photonic efficiency at 400 nm is explained by the fact that every photon absorbed by the CTC results in (1) the transfer of an electron from the adsorbed organic molecule to the semiconductor and (2) the immediate generation of an organic radical. In contrast, 360-nm photons produce electron-hole pairs in anatase; these charges recombine rapidly. Recombination at 400 nm requires that the electron cross back from the inorganic system to the adsorbed organic layer. The higher ζ at 400 nm indicates that either (1) the total amount of charge separation through bandgap and CTC excitation is higher at 400 nm than at 360 nm (that is, more photons are absorbed at 400 nm) or (2) recombination at 400 nm is slower than electron-hole recombination at 360 nm (that is, not only the photonic efficiency but also the quantum efficiency is higher at 400 nm). Since the total absorbance of P25 with adsorbed 245TCP is much higher at 360 nm than at 400 nm (as shown by DR spectroscopy31), the first possibility above can be discounted. Therefore, the CTC recombination process must be slower than anatase electron-hole recombination; that is, CTC excitation has a higher quantum efficiency (Φ) than anatase bandgap excitation. Diffuse-Reflectance Spectra of a Series of Chlorophenols. The electronic interactions between TiO2 and P, 4CP, 24DCP, 245TCP, and 246TCP were probed by DR spectroscopy. The difference spectra observed on P25 are shown in Figure 5. All samples shown in Figure 5 have the same initial aqueous concentration of chlorophenol, namely, 2.5 mM, with the exception of 246TCP, which is not soluble in water at 2.5 mM. In this case, a solvent of 10% methanol in water was used. (Please see the Supporting Information for an explanation of the validity of comparing the 246TCP spectrum in 10% MeOH with the other spectra in pure water.) The absorbance of the CTC increases with the number of chlorines. This method cannot resolve whether higher-chlorinated phenols exhibit greater adsorption to P25 and, therefore, a greater total number of photons absorbed or a greater cross-section for photon absorbance by each adsorbed molecule. As a control, DR difference spectra were also obtained for most chlorophenols on anatase (Figure 6). The difference spectra reveal a broad but discernible absorption for 245TCP on anatase, and some absorbance is observed for all chlorophenols. However, the CTC absorbances are significantly lower on anatase than on P25. The difference is especially great for 245TCP, for which the magnitude of the CTC absorption band on P25 is nearly an order of magnitude greater than that on anatase. (36) Serpone, N. J. Photochem. Photobiol., A 1997, 104, 1-12. (37) Serpone, N.; Terzian, R.; Lawless, D.; Kennepohl, P.; Sauve´, G. J. Photochem. Photobiol., A 1993, 73, 11-16. (38) Serpone, N.; Salinaro, A. Pure Appl. Chem. 1999, 71, 303-320.
Narrow-Band Irradiation of Chlorophenols
Figure 5. DR difference spectra of chlorophenols on Degussa P25. (a) 245TCP; (b) 246TCP (in 10% MeOH); (c) 24DCP; (d) 4CP; (e) P.
Figure 6. DR difference spectra of chlorophenols on anatase. (a) 245TCP; (b) 24DCP; (c) 4CP; (d) P.
Figure 7. Results of exposure of chlorophenols to selected light wavelengths. All data are in reference to the left y-axis except those of 245TCP.
Wavelength-Dependent Reaction of a Series of Chlorophenols. The reaction of 245TCP/P25 slurries under different light wavelengths was compared to that of four other phenols. Results describing the loss from aqueous solution for P, 4CP, 24DCP, and 246TCP are shown in Figure 7. All samples had an initial chlorophenol concentration of 2.5 mM, with the exception of the lesssoluble 246TCP, which was tested at 1.0 mM. The data shown are the averages of duplicate experiments. In a
Langmuir, Vol. 20, No. 14, 2004 5915
preliminary set of data for the four chlorophenols under all five wavelengths, the data for 400 and 480 nm did not significantly contribute to the wavelength trends. Therefore, experiments were duplicated, and are shown below, only for the wavelengths 360, 430, and 550 nm and darkness. For comparison, the data for 245TCP are included. To maintain a readable scale for the other chlorophenols, data for 245TCP in the figure are referenced to the right y-axis, which has double the scale of the left y-axis. No loss of any chlorophenol was observed in catalyst-free controls. Two main trends are discernible. First, the extent of adsorption in the dark increases with the number of chlorines. The adsorption of 246TCP is higher than that of all lesser-chlorinated phenols, even though 246TCP was measured with a starting concentration of only 1.0 mM, compared to 2.5 mM for all other chlorophenols. However, the dark adsorption of 246TCP (3.6 ( 0.1 µmol/g TiO2) is much less than that of 245TCP, even when comparing identical concentration conditions of 1.0 mM (11.3 ( 0.4 µmol 245TCP/g TiO2, not shown in Figure 7). Second, except for 245TCP, the depletion of aqueous-phase chlorophenols increases with decreasing light wavelength. The results obtained at 550 nm are very similar to those obtained in darkness. This was expected based on the absence of measurable charge transfer in the diffusereflectance spectra (Figure 5). Slight increases in loss from aqueous solution are observed for both trichlorophenols at 550 nm versus darkness, presumably due to an overlap of the tails of CTC absorption and lamp emission. At 430 nm, however, significant increases in depletion from bulk solution are seen for all chlorophenols. As in the case of 245TCP, this is the result of CTC activation as observed by DR spectroscopy (Figure 5). Reaction under bandgapexciting conditions (λ ) 360 nm) is observed, as is expected for almost any organic compound. However, we cannot determine from these data whether Φ is higher at 430 nm than at 360 nm for the chlorophenols other than 245TCP. The aqueous-phase depletion of all chlorophenols on anatase under 430-nm light and in the dark was investigated. Within experimental error, none of the chlorophenols adsorbed in the dark, and none were depleted from the bulk water under illumination at 430 nm. The peak absorbance wavelengths of the chlorophenol-anatase CTCs vary considerably (see Figure 6). For some compounds (phenol and 4-chlorophenol), the complexes formed with anatase do not absorb at all at 430 nm. A slight amount of absorption is observed for 2,4-dichlorophenol, and the 2,4,5-trichlorophenol-anatase CTC shows a broad peak near 430 nm. Nevertheless, no reaction is observed for the anatase CTCs. This is consistent with our previous results:31 in contrast to the high amount of adsorption and reaction of 245TCP on P25 compared to the other chlorophenols, wide-spectrum illumination (λ > 415 nm) produced no effect on 245TCP in contact with anatase. Product Analyses. Nonpolar (acetone/hexane) extractions were performed for every chlorophenol under every light condition. The extracts were concentrated 20-fold and analyzed by GC-MS for product identification. No byproducts were observed in any chlorophenol/anatase systems, confirming the observations described above that no reaction on anatase was detected by monitoring parent compound loss from solution. Four compounds are found only in 245TCP/P25 samples irradiated at 400, 430, or 480 nm: pentachlorodibenzo-p-dioxin (dioxin-Cl5), hexachlorodibenzo-p-dioxin (dioxin-Cl6), hexachlorodibenzofuran (DBF-Cl6), and an unidentified molecule with a peak mass of 448 amu (M448; further data and possible
5916
Langmuir, Vol. 20, No. 14, 2004
structures are given in the Supporting Information). All coupling products were detected in trace amounts. Traces of dioxin-Cl5 were once observed in a 550-nm sample. The existence of such products at 550 nm is consistent with the observation of a small amount of photoinduced reaction of 245TCP at that wavelength (see Figure 7). Only one other chlorophenol yields an observable byproduct. Some samples of irradiated 24DCP contain tetrachlorodibenzofuran (DBF-Cl4). Of course, a tetrachloro-species is consistent with the dimerization of two dichlorophenols. But it is interesting that no dioxins or dibenzofurans are extracted from illuminated phenol, 4-chlorophenol, or 2,4,6-trichlorophenol slurries. We have suggested31 an orientation by which 245TCP dimerizes to dioxin-Cl6. If hydrogen abstraction ortho to the phenolic OH group is necessary in this dimerization, then 246TCP is clearly prohibited from coupling by this mechanism, as is consistent with the results showing no dimers formed from 246TCP. However, neither P nor 4CP is similarly hindered. Perhaps the radical formed from 245TCP is electronically more favorable toward reaction. Or perhaps the lesser extent of adsorption, charge transfer, and reaction that occurs on the less-chlorinated chlorophenols simply leads to undetectable amounts of polymer products. The latter hypothesis was tested by irradiating slurries containing 30 mM 4CP, resulting in a high surface concentration of 4CP. Comparison of 245TCP with 4CP at High Concentration. 245TCP exhibits a far greater extent of adsorption, magnitude of light absorption by the CTC, amount of reaction under sub-bandgap light, and formation of polymeric products than all other chlorophenols tested. We hypothesized that one key difference, the high extent of adsorption of 245TCP compared to that of other chlorophenols, may explain all of the other observed differences. In other words, the extent of adsorption may be the only quantity on which 245TCP exceeds the other chlorophenols on a per-molecule basis, and all other measures of 245TCP reactivity (DR absorbance, loss from aqueous solution, and polymer formation) are higher simply because there are more 245TCP molecules on the surface to react. This hypothesis was tested via experiments using an elevated aqueous concentration of 4CP (soluble to approximately 200 mM), which would force a higher surface concentration. In screening experiments, it was estimated that the surface adsorption of 30 mM 4CP to P25 was similar to that of 2.5 mM 245TCP. However, the loss from solution in 30 mM 4CP/TiO2 systems was barely above the analytical error of the HPLC measurements: a loss of 0.3 mM (as is observed in the case of 2.5 mM 245TCP) represents only a 1% change in a 30 mM 4CP system. The CTC formed between 30 mM 4CP and P25 was characterized by DR spectroscopy. Also, systems of 30 mM 4CP and P25 were subjected to reaction under the five different light wavelengths, extracted with acetone/hexane, and analyzed for products. In Figure 8, the 30 mM 4CP spectrum (dashed) is compared to the diffuse-reflectance results from Figure 5. The 30 mM 4CP absorbs somewhat less than 245TCP but far more than all other chlorophenols at 2.5 mM. Next, aqueous suspensions of P25 containing 30 mM 4CP were irradiated at 360, 400, 430, 480, and 550 nm. The surface concentration of 4CP for these slurry experiments, while not precisely quantifiable, was similar to that of 2.5 mM 245TCP samples and much higher than that of the 2.5 mM 24DCP samples that yielded dibenzofurans. Correspondingly, the DR spectrum of 30 mM 4CP showed visible light absorption within a factor of 2 of that of
Agrios et al.
Figure 8. DR difference spectra of chlorophenols on Degussa P25. (a) 245TCP; (b) 30 mM 4CP (dashed); (c) 246TCP (in 10% MeOH); (d) 24DCP; (e) 4CP; (f) P.
245TCP and far more than that of 24DCP. Since coupling is generally expected to require a high surface concentration, these conditions would tend to promote coupling of 4CP under CTC excitation. Nevertheless, no coupling products or other byproducts were detected when illuminated and dark 4CP samples were extracted and analyzed by GC-MS. Thus, the failure of 4CP to dimerize cannot be attributed to the extent of adsorption or complexation. Clearly, the chlorophenols studied differ significantly in their extent of adsorption to and reactivity with TiO2. The high degree of adsorption of 245TCP compared to that of P, 4CP, 24DCP, or 246TCP cannot be explained by pKa values (see the Supporting Information), which are equal for the two trichlorophenols. Likewise, the reasons for the different reactivities are not yet clear, but the chlorophenols have inherent differences in lability toward dimerization that are not simply explained by differences in adsorptivity. The unique behavior of 245TCP must ultimately be related to a chlorine arrangement that causes shifts of electron density within the molecule, resulting in more effective orbital overlap with active sites on P25. Likewise, the chlorine conformation may lower activation energies for polymerization, relative to the other chlorophenols, to form large chlorinated products, but little more can be said without knowing the precise mechanism of coupling. Conclusions TiO2 slurries containing 2,4,5-trichlorophenol were illuminated by narrow ranges of light at 360, 400, 430, 480, and 550 nm. The illuminated slurries were analyzed by the quantification of 245TCP depletion from solution and the extraction and analysis of byproducts. The results are entirely consistent with a reaction mechanism originating with excitation of the CTC identified by diffuse reflectance spectroscopy and electron paramagnetic resonance. The amount of reaction at each wavelength is well correlated with the absorbance of the CTC, and coupling products are formed only when the CTC is excited. Furthermore, the CTC has a higher quantum efficiency than bandgap excitation for transformation of 245TCP. A homologous series of chlorophenols was studied for comparison to 245TCP. The results show that 245TCP is unique among the compounds tested. Not only does it adsorb much more strongly to P25 than the other chlorophenols, leading to greater charge transfer under visible light, but it shows a much greater propensity to
Narrow-Band Irradiation of Chlorophenols
form coupling products than 4CP even when both species are at a similar surface concentration. We attribute the large degree of adsorption and reaction of 245TCP on TiO2 to a chlorine configuration that renders the molecule electronically favorable toward both processes. Although it was hypothesized that some of the chlorophenols tested would react on P25 under visible light to form useful, nonchlorinated polyphenols, no such polymers were detected. At all wavelengths tested and with all chlorophenols examined, P25 showed consistently and significantly higher reactivity than pure-phase anatase. In fact, anatase was found to be totally inert toward the chlorophenols in the absence of bandgap-energy light. On anatase, no adsorption occurred in the dark, and no reaction occurred under visible light. This further demonstrates that synergistic interactions in P25 between the anatase and rutile phases lead to unique sites that can exhibit very different chemical interactions and higher reactivity than either pure phase alone. Our results demonstrate the critical role that chargetransfer complexation can play in determining the rates
Langmuir, Vol. 20, No. 14, 2004 5917
and products of photocatalytic reactions. These findings suggest that materials could be designed to form a CTC with a specific compound, which can then be irradiated at the peak of the CTC absorption, resulting in more efficient and selective photocatalysis under visible light for applications such as targeted pollutant destruction or chemical synthesis. Acknowledgment. This work was supported by the EMSI program of the National Science Foundation and the Department of Energy (CHE-9810378) at the Northwestern University Institute for Environmental Catalysis. The authors thank Dr. Deanna Hurum for assistance with laboratory work and data interpretation and Dr. Bruce Ankenman for assistance with error analysis. Supporting Information Available: Diagram of illumination apparatus; discussion of the DR spectrum of 246TCP in 10% methanol; information on the structure of product M448; pKa’s of studied chlorophenols. This material is available free of charge via the Internet at http://pubs.acs.org. LA036165D
