5166
J. Phys. Chem. 1991,95, 5166-5170
Role of O H Radlcals and Trapped Holes In Photocatalysis. A Pulse Radlolysls Study D.Lawless, N.Serpone,* Chemistry Department, Concordia University, Montrtal. Qutbec, Canada H3G I M8
and D.Meisel* Chemistry Division, Argonne National Laboratory, Argonne, Illinois 60439 (Received: November 27, 1990)
The pulse radiolysis technique has been used to study the reaction of OH' radicals with 13-nm Ti02 particles. Hydroxyl radicals react with these particles in a near-diffusion-controlled rate. The hydroxyl is trapped on the TiO2 and exhibits a broad absorption band centered at about 350 nm. Oxygen has no effect on this reaction and its product characteristics. The product decays via first-order kinetics which are assigned to the collapse of two species trapped on the same particle to yield peroxides. The oxidative abilities of this product were confirmed by oxidation of SCN- to give the anion radical (SCN)2'. The identity of the product of the OH' reaction with the particles is discussed and is identified as a trapped hole at the particle surface. Mechanistic implications to photocatalysis are emphasized; it is argued that the trapped hole and a surface-bound OH' radical are indistinguishable species.
Introduction
Titanium dioxide is a stable highly photoactive semiconductor material often usad in photocatalytic applications. Recent reports' have detailed the complete mineralization of a number of organic pollutants such as cresols? chlor~phenols,~ and surfactants4as well as the removal of several toxic metals such as lead,s chromium: and mercury' from wastewaters. This has led to a considerable amount of interest in examining the mechanistic details of the T i 0 2 photoreactivity in an effort to improve its photocatalytic activity. Before the potential of Ti0, in photocatalytic processes can be maximized, a better understanding of the chemical nature of the photoformed electrons and holes and the role these species play in heterogeneous reactions at the Ti02/electrolyte interface is required.8 Recent advances in reproducible synthesis of transparent colloidal semiconductor solutions9 have permitted the use of a number of state-of-the-art optical techniques to characterize the electronic properties of the conduction and valence bands, as well as the events taking place at the particle surface traps, preexisting or formed upon band gap excitation of Ti02. Several reports'w16 detailing the optical characteristicsof the photogenerated electron have appeared; it now appears to be fairly well accepted that within few picoseconds, following bandgap excitation, the electron is trapped by titanium(1V) ions near or at the surface, resulting in the formation of a Ti3+center.I6 Far less clear, however, are the (1) For r a n t accounts see: (a) Photocaralysis-Fundamentals and Applications; Scrpone, N., Pelizzetti, E., Eds.; Wiley: New York, 1989. (b) Photocatalysis and Environment- Trends and Applications; Schiavello, M., Ed.; NATO AS1 Series C237; Kluwer Academic: Dordrecht, The Netherlands, 1988. (2) Terzian, R.; Scrpone, N.; Minero, C.; Pelizzetti, E. J . Catal., in press. (3) AI-Ekabi, H.; Serpone, N.; Pelizzetti, E.; Minero, C.; Fox, M. A,; Draper, R. B. Lungmuir 1989, 2, 250. (4) Hidaka, H.; Ihara, K.; Fujita, Y.; Yamada, S.;Pelizzetti, E.; Scrpne, N . J . Photochem. Photobiol. A 1988, 375. (5) Lawless, D.; Res, A.; Harris, R.; Serpone. N.; Minero, C.; Pclizzctti, E.; Hidaka, H. Chem. Ind. (Milano) 1990, 72, 139. (6) Domenech, J.; Munoz, J. Electrochim. Acta 1987, 32, 1383. (7) Serpone, N.; Ah-You, Y. K.; Tran, T. P.; Harris, R.; Pelizzetti, E.; Hidaka, H. Sol. EnergV 1987, 39, 491. (8) Anpo, M. Research on Chemical Intermediates-11; Elsevier Science: Amsterdam, 1989; p 67. (9) MatijeviE. E. Longmuir 1986, 2, 12. (10) Henglein, A. Top. Curr. Chem. 1988. 143, 113. (1 1) Henglein, A. Eer. Bunsen-Ges Phys. Chem. 1982,86, 241. (12) Duonghong, D.; Ramsdcn, J.; Grltzel, M. J . Am. Chem. Soc. 1982, 104. 2977. (13) Bahnemann, D.; Henglein, A.; Lillie. J.; Spanhel, L. J . Phys. Chem.
-.--.
1984.88. 709 -.
(14) Bahnemann, D.; Henglein, A.; Spanhel, L. Faraday Discuss. Chem. Soc. 1984, 78, 15 1 . (15) Rothenbergcr, G.; Moser, J.; GrBtzel, M.; Serpone, N.; Sharma, D. K. 1.Am. Chem. Soc. 1985, 107, 8054. (16) Howe, R. F.; GrBtzel, M. J. fhys. Chem. 1985, 89, 4495.
observations concerning the optical characteristics of the photogenerated hole species. A recent report" has proposed that the trapped hole in 120-A T i 0 2 particles is characterized by an absorption maximum at about 630 nm, a feature more in line with surfacetrapped electrons {Ti'"- - -e-}.15*m This is intriguing since the "hole" in Ti02 is essentially a Ti*v-o'-'o radical and such a surface radical is not expected to have an absorption peak in the visible region. As well, this contrasts sharply with the observations of Henglein and c o - w o r k e r ~ who ~ ~ J ~assigned a band at -430 nm to the trapped hole in 250-A platinized TiO, particles. The present paper reports our results in which the pulse radiolysis technique was used to study the reaction of OH' radicals with small T i 0 2 (- 13 nm) particles in aqueous solution. We argue that the species produced by the reaction of OH' radicals with TiO, particles is essentially identical with a trapped hole at the particle surface. The advantage of generating radicals this way, over flash photolysis techniques, iies in the ability to explore the chemical and physical properties of a charge carrier (hole in this case) without the fast recombination of charge carriers and without interference from absorption by the other carrier.,' Hole injection into CdS particles22has been reported to lead to formation of surface-trapped holes identified as S'- radicals. We herein examine the properties of the resulting {TiOz + OH') product and discuss its implications to the ongoing controversy surrounding the role of OH' radicals in the semiconductor-mediated photooxidative degradation of organic pollutants. Experimental Section
Materials. Colloidal solutions of T i 0 2 were prepared by controlled hydrolysis of TiCI4 at low temperature.u In a typical preparation, 5.2 mL of fresh, doubly distilled TiC14 (Fisher) was slowly added dropwise to 200 mL of doubly distilled deionized water (0 f 0.2 "C) under vigorous stirring. The solution was subsequently dialyzed (Viskcase membrane, presoaked for 24 h in distilled water and then thoroughly rinsed prior to use) against approximately 4 L of distilled water (replaced several times) for about 8 h. Approximately 20% titanium is lost during the dialysis ~~
(17) Arbour, C.; Sharma, D. K.; Langford, C. H. J . fhys. Chem. 1990, 94, 331. (18) DimitrijeviE, N . M.; SaviE, D.; MiEiE, 0. I.; Nozik, A. J. J. fhys. Chem. 1984,88,4278. (19) Henglein, A. Pure Appl. Chem. 1984.56, 1215. (20) GrBtzel, M. In Photoinduced Electron Tramfec Fox, M. A., Chanon, M., Eds.; Elsevier: Amsterdam, 1988; Vol. D, p 422. (21) Henglein, A. Chem. Reo. 1989, 89, 1861. (22) (a) Baral, S.;Fojtic, A.; Wcllcr, H.; Henglein, A. J. Am. Chem. Soc. 1986,108,375. (b) Kumar, A.; Janata, E.; Henglein, A. J. fhys. Chem. 1988, 92, 2587. (23) Moscr, J. Ph.D. dissertation, Thesis No. 616, Ecole Polytechnique FUdrale de Lausanne, Lausanne, Switzerland, 1986.
0022-365419 112095-5166S02.50lO 0 1991 American Chemical Society I
,
The Journal of Physical Chemistry, Vol. 95, No. 13, 1991 5167
Role of OH' and Trapped Holes in Photocatalysis
proced~re;~' the concentration of TiOz is calculated to be 15 g/L and the resulting pH of the colloidal solution was between 2.5 and 3. Dilution of the stock solution was done with water at pH 3, adjusted with HClO,. All experiments were performed with unstabilized sols in aqueous acidic solutions that had been degassed with purified N 2 0 gas for a minimum of 10 min by using the syringe technique. Deaeration with NzO was done to quantitatively convert all solvated electrons to the hydroxyl radical in the radiolysis experiments. For pulse radiolysis experiments, aliquots of the colloidal solution were placed in 1-cm cylindrical Suprasil cells by using the cell tilling technique as reported previou~ly.~~ Solutions were changed after each electron pulse. Dynamic light scattering and transmission electron microscopy measurements (Philips EM 420) indicated that the mean particle size of the material was approximately 133 A. Characterization of the Particles. The concentration of the particles used in this study was deduced from eq 1, using the average particle diameter of 133 A. In eq 1, w is the total TiOz
0
1
2 Time (us)
3
4
z .03
CI
.r(
m
c
g
.02
r (
[TiO], = 3w/4uN6R3 = 3.5
X
lO-'w
m
(1)
2
concentration in g/L; N is Avogadro's number; 6 is the density of the material, =3.84 g/cm3 (anatase TiO,); and R is the average radius of the particle, =66.5 A. The average aggregation number of the particles is calculated to be 3.6 X 104 TiOZmolecules per particle. Powder X-ray diffraction patterns of the T i 0 2 were obtained with a Philips PW 1050-25 diffractometer using the Ni-filtered K a radiation of copper (A = 1.5417 A). The powder, obtained from the slow air drying of the colloid followed by grinding with a mortar and pestle, was contained in a flat holder made of Plexiglas. The X-ray diffraction studies showed the material to be highly amorphous with the only crystal phase observable being that of anatase. The presence of a single anatase crystal phase was further confirmed by electron diffraction. Transmission electron microscopy indicates the Ti02 sample to have a narrow size distribution of spherical particles. BET analysis (low-temperature N2 adsorption, Micrometrics Instrument Corp.) of the material indicated that the specific surface area of the T i 0 was 162.8 f 0.9 m2/g with an average pore diameter of 20.7 and a micropore area of 21.57 m2/g. Pulse Radiolysis. Electron pulses, 2-40 ns width, from the Argonne electron linear accelerator were used to produce 1.7 x l o d to 2.5 X 1Vs M OH' radicals per pulse. A pulsed xenon lamp was used as the analyzing light. A more detailed description of the pulse radiolysis apparatus is given elsewhere.2s A NzOsaturated aqueous solution of 10 mM KSCN (G(0H') = G(SCN)2'- = 6.0 molecules/lOO eV; €480 = 7600 M-I cm-I) was used as a dosimeter to calculate the absorbed radiation dose.
0"
U
.Ol
n .10 .15 .20 Time (msl Figure 1. Formation (a) and decay (b) of transient species formed from the reaction of OH' radicals with Ti02 particles followed at 370 nm; M. Inset: dependence of [Ti021p= 1.0 X 10" M, [OH']= 1.5 X
-0
.05
pseudo-first-order rate constant on [OH'].
1
Results Irradiation of NzO-saturatedaqueous solutions with high-energy electrons generates primarily OH' radicals via eqs 2 and 3. Some H20 N 2 0 + ea(
+ H20
-
-
OH', ea(, H', H2, H 2 0 2 OH'
(2)
+ N 2 + OH-
(k = 8.7 x 109 M-1 s-I126 (3)
H' atoms are formed either directly (reaction 2) or by reaction of H20+ with the eaq- (reaction 4). At the pH of the sols, H30+
+ e,-
-
H' + H 2 0
(k = 2.3 X
1Olo
M-I s - I ) ~ ~ (4)
approximately 12%of the solvated electrons are lost via reaction 4. (24) Hart, E. J.; Anbar, M . The Hydrated Electron; Wiley-Interscience: New York, 1970; Chapter IX. (25) (a) Schmidt, K. H.; Gordon, S.Reo. Sci. Instrum. 1979, 50, 1656. (b) Schmidt, K. H.; Gordon, S.;Thompson, M.;Sullivan, J. C.; Mulac, W. A. Radiat. Phys. Chem. 1983. 21. 321. (26) Buxton, G.;Greenstock, C. L.; Helman, W. P.;Ross,A. B. J . Phys. Chem. Ref. Data I ? 1988, 513.
400 450 500 W a v e l e n g t h , nm Figure 2. Absorption spectrum obtained following reaction of TiOzwith OH'. Experimental conditions as in Figure I . 350
The Reaction of OH' Radicals with Ti02 Figure la illustrates a typical trace obtained at 360 nm following the reaction of OH' radicals with the Ti02 particles. The observed growth rate was linearly dependent on [Ti02], yielding k = 6.0 X 10" M-I s-I (concentration in terms of particles), close to the diffusion-controlled limit for this reaction. The rate also increased with the concentration of OH' radicals, suggesting a competing radicalradical reaction. This situation is not unexpected as the concentration of OH' radicals was typically higher than that of Ti02 particles. Addition of 0.1 M tert-butyl alcohol as OH' scavenger completely eliminates this signal. Figure 1 b illustrates the decay of the absorption at 370 nm of the product species. This decay obeys a first-order rate law, indicating a recombination reaction at the surface of a single particle. The decay rate, however, increases linearly with [OH']; a plot of k l o versus [OH'] at constant [TiOz] = 1.05 X 10-6 M yields a rate constant of 1.52 (hO.15) X lo9 d1 s-I (inset Figure lb). This rate constant is basd on the total concentration of OH' radicals. However, since the reaction is a two-dimensional surface reaction, the actual rate constant within the confined space is different. Such a process is expected to follow a dose dependent first-order rate law. The rate of decay was nearly independent of Ti02 particle concentration (at [OH']= 1.50 X lW5 M). This is attributed to two opposing effects: Increasing [Ti02] leads on the one hand to more efficient scavenging of OH' radicals (in competition with bulk OH' recombination) while, on the other hand, to dilution of the product among a large number of TiOz
5168 The Journal of Physical Chemistry, Vol. 95, No. 13, 1991
Lawless et al.
02i i i isctj-i
1
3
2
01 0
4
20
Figure 3. pH dependence of the absorbance formed upon reaction of TiO, with OH'. Conditions as in Figure 1 I
particles. Both the growth and decay rates were wavelength independent thereby implicating only one species responsible for the spectral feature monitored at 370 nm. The absorption spectrum of the transient taken 4 ps after the pulse is illustrated in Figure 2. The spectrum shows an onset of absorption at -470 nm rising toward the UV and reaching a maximum at -350 nm. At lower wavelengths, the sample absorbed too much of the probe light leading to unacceptable signal to noise ratio. pH Dependence. The dependence of the optical density of the species discussed above on pH, in the limited accessible range (2.0 I pH I3 . 9 , is shown in Figure 3. This dependence suggests an acid-base equilibrium with pK, = 2.8. Indeed the pK, measured in Figure 3 is very similar to the pK, of TiO, -OH2+.,' In the pH range examined, the Ti0, colloidal dispersion was stable and no protective agents were needed. Properties of the Species Formed. Addition of oxygen (5 X lo4 M 0,;8 X M NzO) to the solution had no effect on the growth or decay of the transient absorbance. Since oxygen is a good electron acceptor, the species formed on the TiO, particle surface is not an efficient reductant. Oxidizing properties of the (TiO, + OH'] transient were examined by addition of KSCN to the solution. The thiocyanate ion can be oxidized to the radical SCN' which in the presence of excess SCN- ion yields the radical dimer (SCN),' (eqs 5 and 6 ) . This radical dimer has a maximum
-
+ Ox
SCN'
+ SCN-
SCN' -.+
60
80
MI
-1
100
1SCN-I. i i M
PH
SCN-
40
110-5
+ Ox-
(5)
(SCN),'-
(6)
+
absorption at =478 nm,27aa region in which the (TiO, OH') product absorbs negligibly. Similarly, (SCN),'- absorbs weakly at 370 nm where the absorption by the (TiO, + OH') species provides optimal signal to noise ratio. As can be seen in Figure 4, the rate of decay of the absorption at 370 nm increases on increasing [SCN-1. Concomitant with the decay at 370 nm, the absorbance a t 478 nm increases owing to the formation of (SCN),', implicating a reaction between (Ti0, + OH'] and SCN-. At the end of the reaction, the yields of (SCN),'- corresponds to ca. 50% of the initial concentration of OH' radicals (at [SCN-] = IO4 M)while the absorbance at 370 nm decayed to ca. half of its maximum height. This observation implies equilibration between the (TiO, + OH'] product and the SCN-/(SCN),'couple. From the known redox potential of the latter (1.33 V vs NHE211b)we estimate the redox level of the (TiO, + OH') product to position at ca. 1.5 V, a strongly oxidizing species, yet a rather deep hole trap, approximately 1.3 eV above the valence band.
Discussion It is generally accepted that band gap excitation of Ti02leads to the photogeneration of electrons and holes. Both species rapidly migrate to the surface and are responsible for chemical reactions occurring at the TiO,/electrolyte interfacee2* Chemically, a (27) (a) Hug, G. L. "Optical Spectra of Nonmetallic Inorganic Transient Species in Aqueous Solution", NSRDS-NBS 69, 1981. (b) Wardman, P. J . Phys. Chem. Ref. Data 1989, 18, 1637.
Figure 4. Dependence of the pseudo-first-order rate constant for the decay at 370 nm on [SCN-1. Other conditions as in Figure 1. Inset: Langmuir-Hinshelwood analysis of the same results.
trapped electron at or near the particle surface can be described as (Ti1"---e-) or a Ti3+center;16 the trapped hole can be viewed as the radical TiIV-O'- i ~ n . ' ~The ? ' ~optical absorption of 0'radicals maximizes a t 240 nm;27anevertheless, a substantial red shift could result from the interaction with TilV at the surface. Henglein and c o - w ~ r k e r s lreported ~ + ~ ~ that the photogenerated trapped hole in platinized TiO, particles displays a broad spectral feature centered at -430 nm. However, using colloidal platinum deposits on TiO, to scavenge photogenerated electrons presents some problems. These platinized samples are colored and could lead to an observed red shift in the absorption band maximum. Furthermore, picosecond laser flash photolysis of colloidal platinum alone displays spectral features similar to those observed for Ti02/0.5 wt 3'% Pt under identical condition^.^^ A recent examination of the optical properties of the hole by depositing an electron-scavenging dye (CuPcTSe) onto Ti0217has deduced that the trapped hole absorbs at -630 nm. A difficulty in this assertion lies in the close resemblance of the reported difference spectrum for the hole to that of reduced TiOp'8-20 Furthermore, it is dinicult to rationalize a 400-nm shift in the absorption band of the hole (O*-) trapped on Ti02 particles. This intrigue could be best resolved if injecting a hole into TiOz via reaction with the OH' radical proves feasible, since the drawbacks of adding substances foreign to the particle surface could then be avoided. Furthermore, if the OH' radical adsorbs at the particle surface, it will be easily assimilated by the highly hydroxylated Ti0, in aqueous solution and will be indistinguishable from a surface-trapped hole. A major issue that needs to be addressed is, therefore, the nature of the product(s) formed from the reaction of Ti0, with the hydroxyl radical. Of major concern in the analysis of the results described above is the possibility that C1- ions, left over following the controlled hydrolysis of TiC14, might remain either free in solution or adsorbed on the surface of TiO,. The hydroxyl radical can then oxidize C1- according to eqs 7 and 8.
-
+ CI- C1' + OHC1' + c1- c12*-
OH'
(7) (8)
Several observations preclude this potentiality. First, TiO, sols prepared from titanium isopropoxide" instead of TiC14(dialyzed 24 h, pH 3 with HC104 to remove excess i-PrOH) displayed similar transient absorbance features, albeit with lower intensity. The smaller absorbance intensity is likely due to the smaller concentration ( 5 3 . 5 X lW7 M) of the larger Ti02particles (radius -200 A) achievable in this preparation in the absence of stab i l i z e r ~ . Second, ~~ the absorption spectrum in Figure 2 is redshifted relative to that for the Cl,' radical (A = 340 nm). Finally, blank experiments under identical conditions but without TiO, indicate that at least IO mM C1- will be required to generate the (28) Howe, R. F.; GrBtzel, M. J . Phys. Chem. 1987, 91, 3906. (29) Serpone, N.; Moser, J . Unpublished results, 1985. (30) Using titanium isopropoxide, the largest concentration which be obtained was ca. 1.2 g/L.
could
The Journal of Physical Chemistry, Vol. 95, No. 13, 1991 5169
Role of OH' and Trapped Holes in Photocatalysis same absorbance as in the TiO,-containing solution. Such high concentration of chloride ions could not remain in our solutions after the extensive dialysis utilized and they were not detected by the silver nitrate analytical procedure. Moreover, addition of 1 mM chloride to the Ti02 solution had no effect on the transient. Taken together, all these observations convince us that the transient observed is not due to the reaction of OH' with chloride. We have already noted that the product of the OH' reaction with TiO, is a strong oxidant. The dependence of the decay rate of this product on [SCN-] (Figure 4) and the concomitant formation of (SCN);- indicate that the production of (SCN)2'- is not due to a direct reaction between OH' and SCN- in the bulk of the solution. OH'
+ SCN-
SCN'
-
+ SCN-
SCN'
+
+ OH-
(9)
(SCN)2'-
(6)
The dependence of the decay rate on SCN- concentration is reminiscent of a Langmuir-type adsorption/desorption quasiequilibrium for SCN-. The rate of decay of the (Ti02 OH') product in the presence of SCN-, assuming that a fraction of the initial [SCN-] is adsorbed, may be expressed as
+
rate = kd,,[{TiO,
+ OH')] = klK [SCN-]
.
(k?+ 1 + K[SCN-]
[{Ti02+ OH')] (loa)
+
where kI0is the intrinsic decay of the (Ti02 OH'} species (5.4 X 10, s-') and kl is the SCN-dependent first-order decay of (Ti02 OH'], identified in eq 12. The mechanism for the oxidation of SCN- is given by the sequence of reactions 11-13 and the adsorption-desorption coefficient, K, is defined by eq 1 1. The
+
(TiO,)
+ SCN- & (SCN-),*
--
kb
(TiO,
+ OH'} + (SCN-),& {TiO, + OH-)-SCN'
kl
+
(1 2)
(TiO,} + SCN2'-
(13)
{Ti02 OH-I-SCN'
SCN-
(11)
electron-transfer reaction (12) is understood to occur between the two species, SCN- ion and OH' radical, both adsorbed at the surface of the same particle. While we assume that this reaction is the rate-determining step, it is quite possible that some other reaction, e.g., the dimerization or dimer desorption reaction, is rate limiting. The inset in Figure 4 presents the linear dependence expected from this mechanism (eq lob). The intercept and slope 1 1 = -1+ (1Ob) kdmy - klo kl klK[SCN-]
It is commonly accepted that a free hole will react with a water molecule upon arrival at the Ti02 surface to create an adsorbed OH' radical, a process which will occur on a small particle within a few picoseconds. We therefore will not be able to distinguish between a trapped hole and an adsorbed OH' radical. Thus, we identify the product of the OH' reaction as the trapped hole (adsorbed hydroxyl radical):
+
--
(Ti(IV)-02--Ti(IV)) OH' {{Ti(IV)-02--Ti( IV)}-OH'
{Ti(IV)-O'--Ti(
1V)WH-J (AI)
Given that the surface of Ti02 is covered with between 5 and 15 OH- g r ~ u p s / n m ~the ? ~additional hydroxyl group is unlikely to assert any influence on the properties of the surface. The pH dependence of the absorbance of Figure 3 is attributed to the acid-base equilibrium (Ti(IV)-02--Ti(IV))-OH'
+ H+ + (Ti(IV)-O'--Ti(IV))-H20 (A2)
The higher extinction coefficient of the protonated form, relative to the base form, is in line with the similar observation for free OH' radicals. The -9 pH units shift to acidic solutions in the pK, (2.8) relative to the pK, of OH' (1 1.8) is to be expected from the strong shift of charge density to the lattice oxygen as shown in the resonance structure of eq Al. On the other hand, the pK, of this species is essentially identical with the first pK,, of Ti0234 (where pzc = (pK,, + pK2)/2 = 4.823),indicating that the additional OH' group is indistinguishable from the original preexisting ones. Implications on Photooxidation Reactions on T i 4 . Recently, a great deal of discussion has centered on mechanistic details of the photooxidation of organic substrates mediated by irradiated Ti02 and on the role and importance of degradation by free versus surface-bound OH' radicals on the one hand, and versus direct hole oxidation on the other. Several researchers3538have proposed that the hydroxyl radical, produced by oxidation of water or OHby holes at the surface, will diffuse away from the surface to oxidize the organic compound in solution. This contention is based mainly on the fact that H202 and hydroxylated degradation products are formed during this r e a c t i ~ n .Further ~~ support for this mechanism comes from EPR ~ t u d i e s ~which ~ ~ ~show ' that the OH' radical is the most abundant radical produced on photoexcitation of TiO,; a kinetic deuterium isotope experiment48 shows that in the photooxidation of 2-propanol the rate-limiting step is formation of active oxygen species through a reaction involving the solvent water. In addition, a number of researche r have~ concluded ~ that ~ hydroxylation ~ ~ of the T i 0 2 surface
of thecurve yield kl = 3.7 X lo5 s-' and K = 1.5 X 10" M-I. The value of K is consistent with previous observations that SCN(34) Boehm, H. P. Discuss. Faraday Soc. 1971, 52, 264. strongly adsorbs to the surface of Ti02.12-14Further, the rate of (35) Ceresa, E. M.; Burlamacchi, L.; Visca, M. J . Mater. Sci. 1983, 18, formation of (SCN)2' is in good agreement with the rate recently 289. (36) Gonzalez-Elipe, A. R.; Munuera, G.;Soria, J. J . Chem. Soc.,Faraday reported by Draper and Fox3' using the diffuse reflectance flash Trans. I 1979, 75, 748. photolysis technique. The intrinsic decay, represented by kI0,is (37) VBIz, H. G.; Kimpf, G.;Klaeren. A. Farbe + h c k 1976, 82, 805. likely a recombination reaction of two adsorbed hydroxyl radicals (38) VBlz, H. G.; Kimpf, G.; Fitsky, H. G.; Klaeren, A. ACSSymp. Ser. (trapped holes) at the surface of a single particle yielding H 2 0 2 1981, 151, 163. (39) Augugliaro, V.; Palmisano, L.; Sclafani, A.; Minero, C.; Pelizzetti, (see Results section). E. Toxicol. Enuiron. Chem. 1988, 16, 89. The Identity of the ( T i 4 OH'}Product. The identity of the (40) AI-Ekabi, H.; Serpone, N.; Pelizzetti, E.; Minero, C.; Fox, M. A,; species produced by the reaction of OH' radicals with Ti02 can Draper, R. B. Lungmuir 1989, 5, 250. now be addressed. The hydroxyl radical is a strong oxidizing agent (41) Ollis. D. F.; Hisao, C.; Budiman, L.; Lee, C. J . Catal. 1984,88, 89. and previous studies have shown that it can be used to inject a (42) Barbeni, M.; Morello, M.; Pramauro, E.; Pelizzetti, E.; Vincenti, M.; Borgarello. E.; Serpone, N. Chemosphere 1987, 16, 1165. hole into other semiconductor colloid materials such as CdS,22 (43) Turchi, C . S.;Ollis. D. F. J . Catal. 1989, 119, 480. ZnS,m Cd3P2:h and In2Se3.32 The hole in the case of CdS was (44) Metelitsa, D. I. Russ. Chem. Reu., Engl. Trans. 1989, 119, 480. identified as an s p i e s . Similarly, for a hole to be injected (45) Cox, R. A.; Darwent, R. G.;Williams, M. R. Enuiron. Sci. Technol. into TiO, an electron would have to be removed from the valence 1980, 14, 57. band with a resulting radical lattice oxygen species f ~ r m e d . ~ ~ . ~ ~(46) Barbeni, M.; Minero, C.; Pelizzetti, E.; Borgarello, E.; Serpone, N.
+
~~~~~~~
(31) Draper,R. B.; Fox, M. A. J. Phys. Chem. 1990,94,4628. (32) DimitrijeviE, N. M.; Kamat, P. V. k n p u i r 1987, 3, 1004. (33) Augustynski, J. Strucr. Bonding 1988, 69, 1 .
Chemosphere 1981. 16, 2225. (47) Jaegar, C. D.; Bard, A. J . Phys. Chem. 1979,83, 3146. (48) Cunningham, J.; Srijaranai. S. J . Photochem. Photobiol. A Chem. 1988.43, 329. (49) Turchi, C. S.;Ollis, D. F. J . Caral. 1990, 122, 178.
5170
J. Phys. Chem. 1991, 95, 5170-5181
is a prerequisite for the complete photodegradation of organic compounds. In summary, the prevailing v i e d 9 seems to be that the hydroxyl radical acts as the primary oxidant. Equally convincing evidence in support of direct hole oxidation as the main step in the photooxidation process has been presented. An early study by Boonstra and Mutsaerss3 concluded that the hydroxyl radical was unlikely to participate in the reaction on Ti02. Modification of the T i 0 2surface with chlorosilicon compounds leads to a decrease in the activity for several photocatalytic reactions, yet the effcct was smaller than the extent of the eliminated hydroxyl groups. The strongest evidence for the direct hole oxidation pathway comes from a recent study" which failed to detect any of the expected hydroxylated intermediate OH' adducts following flash photolysis of several Ti02/substrate combinations. We hasten to point out, however, that experimental difficulties, and the fact that OH' adducts often possess absorption bands in the UV region where Ti02 absorption may interfere, often obstruct observations of such expected hydroxylated species. Our results are consistent with the interpretation that adsorbed OH' (surface-trapped holes) is the major oxidant while free hydroxyl radicals play only a minor role, if any. Since we find that the OH' radical reacts with the Ti02at a diffusionantrolled rate, the reverse reaction, i.e., desorption of OH' to the solution, would seem unlikely. The surface-trapped hole, as defined in eq A l , could then account for most of the observations that have previously led to the suggestion of OH' radical oxidation. The Navio, J. A,; Rives-Arneau. V. Lungmuir 1990,6, 1525. Bickley, R. 1.; Jayanty, R. K. M.;Navio, J. A,; Vishwanatham, V. In Homogenous and Hererogenous Phorocutalysis; Nato-AS1 Series C; Pelizzotti, E., Serpone, N., Eds.; D. Reidel: Dordrecht, 1986; Vol. 174, p 555. (52) Munuera, G.; Gonzalez-Elipc,A. R.; Rives-Ameau, V.; Navio, J. A.; Malet, P.; Soria, J.; Conesa, J. C.; Sanz, J. In Adsorption und Curulysis on Oxide Surfuces; Che, M.,Bonds, G. C., Eds.; Elsevier: Amsterdam, 1985; (50) (51)
p 113. (53) Boonstra, H.; Mutsaers, C. J . Phys. Chem. 1975, 79, 1940. (54) Draper, R. B.; Fox, M. A. Langmuir 1990,6, 1396.
formahn of H202and the observation of hydroxylated intermediates can all occur via surface reaction of this species. While we cannot entirely exclude the remote possibility that a small fraction of the OH' radicals will leak out from the "surface" and mediate the photooxidation process in the diffuse Gouy-Chapman layer, we consider this contribution to the total photooxidative process to be minimal. Conclusions Oxidation of small Ti02 particles by pulse radiolytically generated OH' radicals yields trapped holes on the particle surface. The product shows a spectral maximum at X = 350 nm. The high oxidation potential of the trapped hole is inferred from its observed decay behavior when thiocyanate is added to the Ti02sol. Our observations are in close agreement with the spectral features for a hole trapped on Ti02 reported by Henglein and co-workers,I3J4 but are in clear contrast with more recent assertions." The decay of the lTi02 + OH') product follows a first-order intrinsic path (forming H 2 0 J by collapse of two trapped holes on the same particle. In the presence of adsorbed a SCN-, the same trapped holes may oxidize the adsorbate to yield the oxidized dimer radical anion (SCN)2'-. From the present work, together with earlier evidence, it can be argued that photooxidation reactions occur at the semiconductor particle surface primarily via this trapped hole.
Acknowledgment. Support by the Natural Sciences and Engineering Research Council of Canada and by the Office of Basic Energy Sciences, Division of Chemical Science, US-DOE, under contract no. W-31-109-ENG-38 is gratefully acknowledged. We are grateful to Prof. J. Fendler and Ms. Youxin Yuan (Syracuse University) for their courtesy and assistance in using the lightscattering apparatus, Mr. Liao Youxin (MSD, Argonne) for his assistance with electron microscopy, and to Dr. R. Le Van Mao and Mr. B. Sarjiel (Concordia University) for BET analysis. Registry No. TiO,, 13463-67-7; OH', 3352-57-6.
Inhomogeneous Decay Klnetlcs and Apparent Sdvent Relaxation at Low Temperatures Richard S. Fee, John A. Milsom, and Mark Maroncelli* Department of Chemistry, 152 Davey Laboratory, The Pennsylvania State University, University Park, Pennsylvania 16802 (Received: December 3, 1990)
We have observed timedependent shifts in the fluorescence spectra of solvatochromicprobe molecules in frozen polar solvents. The phenomenon is a general one and is observed in a variety of fluorophoresin both hydrogen-bonding and non-hydrogen-bonding solvents. These shifts are not related to solvent relaxation but rather result from inhomogeneous fluorescence decay kinetics. Molecules in different portions of the fluorescence spectrum decay at different rates and thereby cause the spectrum to evolve in time due to nonuniform loss of excited-state population. We have modeled such kinetics based on the assumption that fluorescence decay rates are solvent sensitive only through the relation between the radiative rate and emission frequency, knd u3. Measurements of the effect of excitation wavelength on steady-state fluorescence spectra are used to quantify the inhomogeneous broadening present in a number of fluorophorelsolvent systems. These results are then used as input to model their time dependence. In most cases, nearly quantitative agreement between the calculated and observed spectral dynamics is achieved.
I. Introduction In the past few years a number of researchers have employed time-resolved fluorescence to probe the dynamics of solvation in polar solvents.' These measurements involve the use of solvatochromic fluorophores, probe molecules that exhibit large frequency shifts in their fluorescence spectra as a function of solvent polarity. Such probes are excited with an ultrafast laser pulse, and their subsequent fluorescence is monitored as a function of
time. Time-dependent fluorescence shifts are then interpreted in terms of relaxation of the solvent environment in response to the change in solute charge distribution produced by electronic excitation. Most current work of this type has focused on the behavior of liquids near room temperature;2 however, we have also been pursuing such studies in low-temperature alcohols, where the dynamics appear to be qualitatively different?,' During the course
(1) See for example the recent reviews: Maroncelli, M.; Maclnnis, J.; Fleming, G. R. Science 1989, 213, 1674. Barbara, P.F.; Jarzeba. W . Adu. Photochem. 1990, IS, 1. Bagchi, B. Annu. Reu. Phys. Chem. 1990,40, 1 1 5 .
(2) Recent exceptions are: Barigelletti, F. J . Phys. Chem. 1988,92,3679. Kim, H.-B.;Kitamura, N.; Tazukc, S . J. Phys. Chem. 1990,94, 1414. Richert, R. Chem. Phys. Lett. 1990, 171, 222.
0022-3654191f 2095-5 170$02.50/0 0 1991 American Chemical Society
