Photoinduced Coloration and Photobleaching of Titanium Dioxide in

Oct 3, 2007 - Fock Research Institute of Physics, St. Petersburg State University, Ulyanovskaya st. 1, St. Petersburg, 198504 Russia, and Dipartimento...
0 downloads 0 Views 322KB Size
Download PDF
J. Phys. Chem. C 2007, 111, 15277-15288

15277

Photoinduced Coloration and Photobleaching of Titanium Dioxide in TiO2/Polymer Compositions upon UV- and Visible-Light Excitation of Color Centers’ Absorption Bands: Direct Experimental Evidence Negating Band-Gap Narrowing in Anion-/Cation-Doped TiO2s Vyacheslav N. Kuznetsov† and Nick Serpone*,‡ Fock Research Institute of Physics, St. Petersburg State UniVersity, UlyanoVskaya st. 1, St. Petersburg, 198504 Russia, and Dipartimento di Chimica Organica, UniVersita di PaVia, Via Taramelli 10, PaVia 27100, Italy ReceiVed: May 8, 2007; In Final Form: August 8, 2007

The present article examines the photocoloration of TiO2/polymer compositions with various polymers and the photobleaching of color centers at selected irradiation wavelengths from the UV region to the nearinfrared region. Photoactivation of color centers was examined by irradiating into the absorption bands (ABs) with maxima at 2.90 eV (427 nm, AB1), 2.55 eV (486 nm, AB2), and 2.05 eV (604 nm, AB3). Two principal types of photostimulated absorbance changes were observed: (i) increases in absorbance and (ii) decreases in absorbance. The latter is a direct experimental manifestation of the photobleaching of colored TiO2/polymer compositions, which demonstrates the presence and photoinduced disappearance/destruction of color centers in these TiO2 systems. The spectral results also demonstrate that photobleaching of colored TiO2/polymer compositions originates from intrinsic absorption of light by the TiO2 (at hν > 3.2 eV) and also from extrinsic absorption of light by the color centers at wavelengths corresponding to their absorption spectral bands (i.e., at hν < 3.2 eV). These bands, therefore, are also active in the photodestruction of the color centers. A photochemical mechanism is proposed for the photobleaching process involving oxygen-assisted annihilation of oxygen vacancies. The unambiguous experimental data reported herein confirm an earlier proposal that the absorption of light by the various TiO2 systems in the visible region originates only from color centers and not from a narrowing of the band gap of pristine TiO2. Unlike the color centers, the valence and conduction bands, which some have suggested as being involved in the observed red shifts of the absorption edges in doped visible-light-active TiO2s because of an apparent narrowing of the band gap of TiO2, cannot be photodestroyed. The competitive photoinduced formation and destruction of color centers were modeled by simple considerations, the results of which are in qualitative agreement with experimental observations.

1. Introduction Second-generation cation- and anion-doped TiO2 photocatalysts, whose absorption edges are red-shifted to lower energies (hν < 3.2 eV; anatase) relative to undoped TiO2, i.e., into the visible spectral region, provide for an overall increased efficiency of surface processes based on the total amount of UVvisible radiation incident on the photoreaction system. Some workers1-10 have proposed that the red shifts originate from a narrowing of the band gap of pristine TiO2 (Ebg of anatase, 3.2 eV; absorption edge, 387 nm), whereas others11-17 have deduced that intragap localized states of the dopants were responsible for the visible-light photoactivity. The past decade has witnessed extensive activity in developing second-generation TiO2 photocatalysts with high visible-light activities in photoinduced reactions. Heat-induced and photoinduced absorption spectra of various compositions involving Degussa P-25 TiO2 and different polymers were recently examined,18 as were the absorption spectra of visible-light-active (VLA) TiO2 photocatalysts.8,19-28 * Corresponding author. E-mail: [email protected] or nickser@ alcor.concordia.ca. † St. Petersburg State University. ‡ Visiting Professor, Universita di Pavia.

In all cases, the absorption spectral envelope consisted of the sum of overlapping absorption bands (ABs) with maxima at 2.90 eV (427 nm, AB1), 2.55 eV (486 nm, AB2), and 2.05 eV (604 nm, AB3). The spectra correlated fully with experimentally observed absorption spectra of reduced TiO2.18 The remarkable overlap and congruence of the absorption spectra, regardless of the nature of the dopant (anion or cation) and the history of how the TiO2 had been reduced, pointed to a common thread among all the doped TiO2 specimens. Accordingly, it was deduced18 that visible-light activation of TiO2 (anion-doped or otherwise) implicates defects associated with oxygen vacancies that give rise to color centers that display light absorption in the visible spectral region and not to a narrowing of the intrinsic band gap of TiO2 through mixing of dopant and oxygen states in the valence band.1,2,6,9 An extensive examination18,29,30 of the various claims of bandgap narrowing in a variety of VLA TiO2 systems, prepared under various experimental conditions, concluded that the red shifts of the absorption edges could originate only from the formation of color centers during the syntheses of the TiO2 specimens. In all instances, syntheses included a form of heat treatment during the doping process. Although not an unknown occurrence in semiconductor physics, band-gap narrowing would, of necessity, require heavy doping of the metal oxide with the resulting

10.1021/jp073511h CCC: $37.00 © 2007 American Chemical Society Published on Web 10/03/2007

15278 J. Phys. Chem. C, Vol. 111, No. 42, 2007 possibility that the material would no longer be TiO2 but some new material having an entirely different chemical composition and, more importantly, a completely different electronic structure. The picture that also emerged from this extensive examination was that density-of-states calculations of the electronic structures of TiO2, through density functional theory (DFT) algorithms, severely underestimated the band-gap energies of undoped anatase and rutile TiO2 by a factor of ca. 30-40%, a well-known shortcoming of DFT. To our knowledge, there is no direct experimental evidence for band-gap narrowing. The present article examines in some detail the photocoloration of TiO2 in various compositions with different polymers and probes the photobleaching of color centers using selected wavelengths (cutoff filters and a combination thereof) spanning the UV to the near-infrared region to photoactivate the centers upon irradiation into bands AB1-AB3.18 Two principal types of photostimulated absorbance changes can occur: (i) increases in absorbance and (ii) decreases in absorbance. In special cases, no absorbance changes are observed. The decrease in absorbance is a direct experimental manifestation of the photobleaching of colored TiO2/polymer compositions, which clearly demonstrates the presence and photoinduced disappearance of color centers in the TiO2 systems. The spectral results demonstrate that photobleaching of colored TiO2/polymer compositions originates both from intrinsic absorption of light (hν < 3.2 eV) by TiO2 and from direct (extrinsic) absorption of light by the color centers at wavelengths corresponding to their individual spectral bands. The bands are also active in the photodestruction of the color centers. The experimental data unambiguously confirm the earlier proposal18,29,30 that absorption of light by the various TiO2 systems in the visible region originates only from color centers and not from a narrowing of the band gap of pristine TiO2. The valence and conduction bands of TiO2 involved in the band-to-band (intrinsic) absorption of light simply cannot be changed photochemically under irradiation. 2. Experimental Section Titanium dioxide (Degussa P-25, ca. 80% anatase, 20% rutile) and the various polymers, poly[(vinylidene fluoride)-cohexafluoropropylene] [P(VDF-HFP)], poly[(vinylidene fluoride)co-tetrafluoroethylene] [P(VDF-TFE)], poly(methylphenylsiloxane)[PMPS],poly(vinylbutyral)(PVB),andpoly(ethylmethacrylateco-methylacrylate) [P(EMA-MA)] were available from an earlier study.18 The TiO2 was calcined in air at ca. 600 K for 5 h to eliminate any adsorbed/absorbed organic impurities, after which the powdered samples (200 mg) were placed in a 30mm-diameter stainless steel dish and impregnated with an acetone solution of the polymer (loading, 10-30 mg mL-1 of solution). After being dried, the compositions contained 1020 wt % of the polymer. Samples were kept in the dark before and between the heat or irradiation treatments and the optical measurements. The polymers used were transparent in the visible spectral region and were thermally and photochemically stable under our conditions. The absence of (photo- or thermal-) treatmentinduced absorption features in control samples of the polymers and TiO2 specimens indicated that the TiO2/polymer compositions undergo oxidative thermal degradation or photodegradation only upon interaction of the polymers with the metal oxide,18,31 and in contrast to ZnO, displacement of spectral bands that might be caused by the heat treatment was insignificant for TiO2.31 Samples of the TiO2/polymer compositions were irradiated in a specially designed stainless steel chamber covered with a glass lid to provide access to the samples. The chamber was

Kuznetsov and Serpone

Figure 1. Graph illustrating the transmittance spectra of glass filters (curves A-H). The convolution of absorption bands AB1-AB3 accords with the experimental absorption spectrum of treatment-reduced TiO218 and with spectra reported herein.

designed as a circulatory system and was equipped with a humidifier, a heating element, and sensors to monitor relative humidity and temperature. Four 20-W fluorescent lamps were the light source. The emitted spectrum of the lamps consisted of a quasicontinuous fluorescent emission in the visible region along with the usual Hg lines at λ g 365 nm that accounted for less than ca. 3% of the total light flux. A compressor pumped air or N2 gas (with ∼1% O2 impurity) over the humidifier and heating element to purge the samples with air or N2. Temperatures in the chamber ranged from 300 to 500 K. The setup allowed for a relative humidity in the chamber of between 2% and 90%; however, under our conditions, relative humidity had no significant effect on the spectral results. The illuminance was ca. 20 klx (i.e., 20 000 lux). Fifteen samples that included TiO2/polymer compositions and TiO2-free polymer films were placed in the chamber and irradiated simultaneously. UV-vis diffuse reflectance spectra (DRS, F) were measured on a Beckman UV-5270 spectrophotometer equipped with an integrating sphere assembly; BaSO4 was the reference standard. Difference diffuse reflectance spectra (∆F) were plotted to demonstrate the spectral changes in reflectivity (absorption enhancement). If F1(hν) and F2(hν) are the DRS values measured before and after the sample treatment (heat or irradiation), respectively, then the difference [F1(hν) - F2(hν)] > 0, that is, ∆F(hν) > 0, corresponds to the treatment-induced absorption spectrum. Details regarding the DRS analysis are available elsewhere.18 The spectral regions of wavelengths active in the photocoloration of TiO2 and photobleaching of color centers (see below) were selected using a set of cutoff glass filters or a combination thereof. The relevant transmittance spectra of the filters are illustrated in Figure 1 and are designated by curves A-H. The letters also identify the corresponding filters. The transmittance of the A and B filters in the infrared region is not shown in Figure 1. Filters C and E are combinations of two other filters whose transmittance spectra are reminiscent of those of interference filters. Table 1 summarizes the photon energies and wavelengths of the various spectral regions available and used for the photoinduced coloration of TiO2 and photobleaching of the color centers’ absorption bands AB1-AB3. These bands are also illustrated in Figure 1. Note the remarkable overlap of bands AB1-AB3, which, when convoluted, constitute the treatment-induced spectra of the TiO2/polymer compositions.

Photoinduced Coloration and Photobleaching of TiO2

Figure 2. Kinetics of the increase in absorbance at λ ) 430 nm (∆Fλ) for compositions of TiO2 with PMPS (curves 1, 1a), PVB (curve 2), [P(EMA-MA)] (curves 3, 3a), and [P(VDF-HFP)] (curves 4, 4a) that had been subjected to fluorescent-lamp irradiation in air (relative humidity of ∼20%). Curves 1-4 were obtained under irradiation by all emitted wavelengths of the lamps; curve 1a was obtained using filter A (λ < 406 nm, hν > 3.05 eV), whereas curves 3a and 4a were obtained with filter B (λ < 463 nm, hν > 2.68 eV).

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15279

Figure 3. Absorption spectra (difference diffuse reflectance spectra, ∆F) for compositions consisting of TiO2 and PMPS (curve 1), PVB (curve 2), [P(EMA-MA)] (curve 3), and [P(VDF-HFP)] (curve 4) that had previously been subjected to fluorescent-lamp irradiation for about 4.5 h in air.

TABLE 1: Cutoff Filters and the Corresponding Available Photon Energies and Wavelengths Used for Irradiation of the Color Centers in TiO2/Polymer Compositions

a

filter

photon energies (eV)

wavelengths (nm)

A B Ca D Ea F G H

>3.05 >2.68 3.23-2.55 3a, and 1 > 1a, as expected from the changes in the spectral range of the incident light for curves 4a, 3a, and 1a (see below). The absorption spectral intensities (Figure 3) are also sensitive to the nature of the polymer and to the incident light irradiance at longer irradiation times. Activation of the color centers’ photoinformation in the region of intrinsic absorption, i.e., band-toband absorption at hν > 3.2 eV, by TiO2 was a rather predictable result. The inset in Figure 4 reports the various absorption spectra calculated (see section 3.2) from diffuse reflectance spectra of anion-doped TiO2 specimens: (i) mechanochemically activated N-doped TiO2 sample;24 (ii) N-doped oxygen-deficient TiO2 sample;19 (iii) N/F-doped TiO2 sample prepared by a spray pyrolytic method;26 (iv) N-doped anatase TiO2 specimen prepared by a solvothermal process;27 (v) N-doped rutile TiO2 sample also prepared by a solvothermal process;27 (vi) yellow N-doped TiO2 system synthesized in a short time at ambient temperatures using a nanoscale-exclusive direct nitridation of TiO2 nanocolloids with alkyl ammonium compounds;8 (vii, viii) N-doped TiO2 samples prepared by evaporation of the sol-gel with N-doping performed under a stream of ammonia at different temperatures;28 and the cation-doped titanium dioxide samples (ix) Cr-implanted TiO2,22 (x) Ce-doped TiO2,23 and (xi) Sr0.95La0.05TiO3+δ treated with HNO3 acid.25 Regardless of the nature of the dopant and the preparative histories, the spectral similarities are remarkable.

Figure 5. Temporal changes in absorbance intensity at λ ) 430 nm, ∆(∆Fλ), for pre-thermally degraded compositions of TiO2 with PMPS (curves 1, 1a), [P(VDF-HFP)] (curve 2), [P(VDF-TFE)] (curve 3), and PVB (curves 4, 4a), followed by fluorescent-lamp irradiation under a N2 atmosphere for the first 490 min and then in air for time > 490 min (relative humidity ca. 40-50%).

3.2. Photobleaching of Color Centers in Heat-Treated TiO2/Polymer Compositions. Figure 5 illustrates the temporal changes in absorbance, ∆(∆Fλ), measured at 430 nm for previously thermally degraded TiO2/polymer compositions followed by fluorescent-lamp irradiation under a N2 atmosphere for times ranging from 0 to 490 min. After this period, air was introduced, and measurements was continued for times from 490 to ca. 800 min. Initial absorbance values (∆Fλ)ini at 430 nm for the compositions (see Figure 5 caption) after the heat treatment but before irradiation were 0.03 (curve 1), 0.09 (curve 1a), 0.18 (curve 2), 0.14 (curve 3), 0.26 (curve 4), and 0.07 (curve 4a). Because Figure 5 depicts photoinduced changes in absorbance, for comparison purposes, we set the values of ∆(∆Fλ) to 0 at time 0 for all TiO2/polymer compositions. Three types of photostimulated absorbance changes are clearly evident: (i) increases in absorbance (curves 1 and 1a; k ) 1.1 × 10-2 and 0.97 × 10-2 min-1, respectively), (ii) absorbance decreases (curves 2-4; k ) 3.2 × 10-2, 4.4 × 10-2, and 1.9 × 10-2 min-1, respectively), and (iii) lack of any absorbance change (curve 4a; k ≈ 0 min-1). The decrease in absorbance is a direct experimental manifestation of photobleaching of the colored TiO2/polymer compositions that give rise to the bands in the visible spectral region. Photobleaching clearly demonstrates the presence and photostimulated disappearance of color centers in titanium dioxide systems.18 Figure 5 also shows that, for a given composition, for example, TiO2/PVB, both a strong decrease (curve 4) and negligible absorbance changes (curve 4a) can take place. For the TiO2/PMPS composition, the initial rate of the absorbance growth differs by about a factor of 3 (curve 1 versus 1a) under the same irradiation conditions as in Figure 2. To interpret the above results, it is useful to compare the following quantities for each composition: (i) the initial absorbance (∆Fλ)ini; (ii) the change in absorbance, ∆(∆Fλ), after prolonged irradiation as in Figure 5; and (iii) the absorbance after prolonged irradiation of untreated (not subjected to heat treatment) TiO2/polymer compositions that provide the equilibrium or quasi-equilibrium absorbance value (∆Fλ)eq. The latter can be determined from the data in Figure 2 at the long irradiation times. The relevant data are summarized in Table 2 (note that column 6 ) column 4 + column 5). Comparison of (∆Fλ)eq obtained in air or under

Photoinduced Coloration and Photobleaching of TiO2

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15281

nitrogen atmosphere with the sum [(∆Fλ)ini + ∆(∆Fλ)] allows a simple regularity to be deduced, namely

(∆F)ini + ∆(∆F) ≈ (∆F)eq

(1)

where the value of ∆(∆F) is taken with its appropriate algebraic sign (+or -); the notation λ has been omitted for simplification. Expressing eq 1 in the form ∆(∆F) ≈ (∆F)eq - (∆F)ini allows for a better understanding of the features of the temporal changes in absorbance and changes in bleaching shown in Figure 5: (i) where (∆F)ini > (∆F)eq, the absorbance decreases (curves 2-4), i.e., ∆(∆F) < 0; (ii) where (∆F)ini < (∆F)eq, an additional absorbance growth occurs (curves 1, 1a), i.e., ∆(∆F) > 0; and (iii) where (∆F)ini ≈ (∆F)eq, the absorbance change is negligible, i.e., ∆(∆F) ≈ 0 (curve 4a). Finally, (∆F)ini ) 0 yields the kinetics of absorbance growth for thermally untreated TiO2/polymer compositions as in Figure 2. In terms of the photostimulated reactions involving color centers, the above results demonstrate that, when the TiO2/ polymer compositions are subjected to irradiation by the full emitted spectrum of the lamps (as done herein), two photoinduced concomitant and competitive processes always occur: (a) formation of color centers and (b) disappearance of color centers. Accordingly, the resulting kinetics reflect changes in the concentration of color centers. In the TiO2/polymer compositions, an increase in temperature and/or heating time provides both the growth of intensities of the three absorption bands AB1-AB3 and the changes of the relative intensities of these bands (see Figure 2 and Table 1 in ref 18). Hence, the thermally degraded TiO2/PVB and TiO2/ PMPS compositions with different initial absorbance (∆Fλ)ini at λ ) 430 nm also differ in the relative concentrations of the color centers responsible for all of the absorption bands. However, the influence of the change on the relative concentration of color centers N is negligible because the sign and value of the difference (Neq - N0) for the AB1 band completely provide both the sign and the kinetics of the photobleaching/ photocoloration at λ ) 430 nm, i.e., the kinetics of the change in the number of centers responsible for AB1. The spectral region of wavelengths active in photobleaching was selected using several cutoff glass filters; see Figure 1, curves A-H. For further study, we selected the TiO2/[P(VDFHFP)] composition because it is a highly active system in the coloration process under low-temperature heating conditions (see Figure 3 in ref 18) and because it is less active under irradiation in air (see Table 2). This implies that irradiation of such heatinduced colored compositions induces mainly the photostimulated disappearance or destruction of the color centers. Temporal changes in the photobleaching of TiO2/polymer compositions, previously heat-treated under otherwise identical conditions, illustrated in Figure 6, were monitored at λ ) 430 nm. Curve 1 displays an absorbance decrease (k ) 5.1 × 10-2 min-1) upon irradiation at wavelengths available with filter A, i.e., at λ < 406 nm (hν > 3.05 eV). Note that points b and b′ in all of the curves denote the termination and continuation of irradiation, respectively (see below and section 3.5). Curve 2 exhibits temporal variations in photobleaching (k ) 8.2 × 10-2 min-1) under irradiation with filter B (λ < 463 nm, hν > 2.68 eV) in the range from 0 to point a, with filter F (λ > 539 nm, hν < 2.30 eV) in the range from a to b, with filter G (λ > 590 nm, hν < 2.10 eV) in the range from b′ to c, and with filter H (λ > 680 nm, hν < 1.82 eV) in the range beyond point c. In the interval 0 to point c, curve 3 was obtained with filter E (460 nm < λ < 620 nm, 2.70 > hν > 2.00 eV); the kinetics of absorbance decrease, k ) 7.7 × 10-2 min-1. For curve 3 in the

Figure 6. Kinetics of the photobleaching (decrease in the absorbance) at λ ) 430 nm [∆(∆Fλ)] for heat-induced TiO2/[P(VDF-HFP)] colored compositions. Points a, b, b′, and c correspond to changes of the irradiation conditions: curve 1 with filter A (λ < 406 nm, hν > 3.05 eV); curve 2 with filters B (λ < 463 nm, hν > 2.68 eV), F (λ > 539 nm, hν < 2.30 eV), G (λ > 590 nm, hν < 2.10 eV), and H (λ > 680 nm, hν < 1.82 eV); and curve 3 with filter E (λ ) 460-620 nm, hν ) 2.70-2.00 eV). Also see text for additional details.

time regime beyond point c, the colored TiO2/[P(VDF-HFP)] composition was kept in the dark. The results in Figures 5 and 6 clearly demonstrate that the photobleaching of the colored TiO2/polymer compositions (i.e., photodestruction of color centers) originates both from the intrinsic (band-to-band) absorption of light by the TiO2 and from the extrinsic absorption of light by the color centers at wavelengths corresponding to their spectral bands. Light absorbed in the three absorption bands (AB1-AB3) is active toward the destruction of the color centers responsible for these bands. Only at wavelengths longer than 680 nm (hν < 1.82 eV) does the rate of absorbance decrease become 0 (k ≈ 0 min-1). It cannot be overemphasized that intrinsic absorption of light by TiO2 not only stimulates the formation of color centers (Figure 2, curves 1a, 3a, and 4a) but also causes the disappearance/destruction of these color centers (Figure 6, curve 1). Figure 6 also exhibits a postirradiation effect, i.e., the response of the absorbance of the compositions to a pause in irradiation (ca. 12 h, dark response). Point b indicates the end of irradiation, whereas point b′ indicates the beginning of the next irradiation period. For convenience, the data presented in Figure 6 at point b′ were slightly displaced relative to point b. The absorbance growth in curve 3 in the interval b-b′ is nearly the same as at that times beyond point c. The time needed for the growth in absorbance between b and b′ is estimated to be ca. 50-60 min. To demonstrate the spectral changes in the diffuse reflectance data from the photobleaching process, the difference DRS spectra were calculated by the method reported earlier (ref 18 and briefly herein) and then multiplied by -1 to facilitate the comparison typified in Figure 7. The average bleaching spectrum of colored TiO2/[P(VDF-HFP)] resulting from different irradiation conditions is denoted curve 3 and is compared to the average heat-induced absorption spectrum (curve 1) and photoinduced absorption spectrum (curve 2) of all the TiO2/polymer compositions. Spectrum 3 of Figure 7 corresponds to the near-complete discoloration of the TiO2/[P(VDF-HFP)] composition under irradiation in different spectral regions. If photobleaching were

15282 J. Phys. Chem. C, Vol. 111, No. 42, 2007

Figure 7. Comparison of the average absorption spectra (1, 2) of various TiO2/polymer compositions normalized by ∆Fmax and the average bleaching spectrum (3) of the TiO2/[P(VDF-HFP)] composition irradiated at different wavelengths (see text) also normalized by the ∆Fmax parameter. Spectrum 1 of the heat-induced absorption is the same as spectrum 2 of Figure 2 in ref 18, whereas spectrum 2 of the photoinduced absorption is the same as spectrum 2 of Figure 4. Note that the TiO2/[P(VDF-HFP)] composition was heat-treated prior to irradiation with all wavelengths emitted by the fluorescent lamps.

Figure 8. Bleaching spectrum (1) of the heat-treated TiO2/[P(VDFHFP)] composition normalized by the factor ∆Fmax. Spectrum 1 was deconvoluted into three Gaussian absorption bands (spectra 2a-2c). Spectrum 2 represents the sum of the three absorption bands 2a-2c. Note that the bleaching was photoinduced by irradiation with wavelengths selected with filter F (λ > 539 nm, hν < 2.30 eV).

induced only by wavelengths longer than 539 nm (filter F, hν < 2.30 eV; note that the decrease in absorbance was ∼20%), the photobleaching spectrum should then differ significantly from the spectra of Figure 7, as observed on comparing the spectra with curve 1 in Figure 8. The latter spectrum is reasonably described by the sum of three Gaussian absorption bands (curves 2a-2c) with hνmax at 2.91 eV (AB1), 2.53 eV (AB2), and 2.08 eV (AB3). The respective δhν1/2 values are 0.34 eV (AB1), 0.56 eV (AB2), and 0.54 (AB3). The characteristics of the Gaussian ABs coincide with the heat-induced and photoinduced absorption spectra reported earlier.18 The AB2 band (curve 2b) dominates in spectrum 1 of Figure 8. For spectra 1-3 reported in Figure 7, the relative band intensities of the

Kuznetsov and Serpone three deconvoluted AB bands (not shown) are AB1/AB2/AB3 ≈ 0.50:0.45:0.05, whereas for curve 1 of Figure 8, the corresponding relative intensities of the deconvoluted bands are AB1/AB2/AB3 ≈ 0.25:0.54:0.21. Two points are worth noting with regard to the results of Figure 8 relative to those of Figure 7. First, Figure 8 presents further experimental evidence that the absorption spectrum of TiO2 in the visible region consists of overlapping individual absorption bands. Considering the present data and the data reported earlier,18 both absorption and bleaching spectra display a common absorption envelope with maxima at 2.90 eV (see Figure 7). However, the same deconvoluted absorption bands for the TiO2/polymer composition irradiated at other visible wavelengths (filter F, λ > 539 nm, hν < 2.30 eV) with different intensity ratios AB1/AB2/AB3 display a spectrum (curve 1 of Figure 8) with a maximum around ∼2.5 eV. Second, Figure 8 depicts the features of the photobleaching phenomenon resulting from light absorption into the AB3 absorption band (see the filter F curve and the AB3 band in Figure 1). We deduce that the AB2 band (curve 2b, Figure 8) is more “sensitive” than the AB1 band (curve 2a, Figure 8) when bleaching originates from photoactivation of the AB3 band (curve 2c, Figure 8). The total overlap of the absorption and bleaching spectra illustrated in Figure 7 demonstrates unambiguously that the same color centers are formed during the treatment that induced the absorption and that these color centers are subsequently destroyed upon irradiation during the photobleaching process. This result also argues against any inference of widening the valence band of TiO2 to account for the red shifts of the absorption edges in the various doped VLA TiO2 systems.3,4 The valence and conduction bands can be neither photodestroyed nor phototransformed, in contrast to the color centers. To demonstrate the similarity of the absorption spectra of the color centers with the spectra of the VLA TiO2 photocatalysts, several absorption spectra of anion- and cation-doped TiO2 samples were selected for numerical analysis. The analysis included digitization (numbering) of the DRS spectra of the sample in non-VLA and VLA specimens, calculation of the difference between the DRS spectra, and normalization of the spectra by the factor ∆Fmax. Results of the analysis show unambiguously that the relatively narrow absorption spectra of VLA TiO2 samples are very similar (inset to Figure 4) and were thus taken to be independent of the preparative methods and of the nature of the dopant. The strong spectral similarities afforded calculation of the average spectrum of these VLA TiO2 materials. To account for any difference(s) in the optical properties near the band-gap absorption edge (3.1-3.2 eV) of different VLA TiO2 photocatalysts and of TiO2/polymer compositions, the average spectrum was multiplied by the mean DRS spectrum of the freshly prepared, thermally untreated compositions. The resulting averaged spectrum of anion- and cationdoped titanias is reported as curve 5 in Figure 4; for hν < 3.2 eV, the standard error was less than 2.5%. Clearly, spectrum 5 of Figure 4 is similar to the spectra of photoinduced (Figure 4, curves 1 and 2) and heat-induced (Figure 7, curve 1) absorption spectra and to the photobleaching spectrum 3 of Figure 7. Differences in the shape of the spectra are due to different relative intensities of the long-wavelength band AB2 (curve 3 in Figure 4) and band AB3. 3.3. Photocoloration and Photobleaching of TiO2/Polymer Compositions. In the previous section, we examined the photocoloration and photobleaching spectral features (Figures 7 and 8) originating from heat-treated (thermally degraded) TiO2/polymer compositions prior to irradiation with the UV/

Photoinduced Coloration and Photobleaching of TiO2

Figure 9. Kinetic behavior of the change in absorbance at λ ) 430 nm [∆(∆Fλ)] under fluorescent-lamp irradiation with different filters for TiO2/polymer compositions previously irradiated under conditions corresponding to those of Figure 2. Filters used were filter F (λ > 539 nm, hν < 2.30 eV) in the interval a-b for curves 1, 1a, and 3; filter D (λ > 433 nm, hν < 2.86 eV) at times above b for curves 1, 1a, and 3a and for the whole curve 3; filter B (λ < 463 nm, hν > 2.68 eV) for curve 4; and filter C (λ ) 384-486 nm, hν ) 3.23-2.55 eV) for curve 2. Curve 4a was obtained without filters. For details, see text and Figure 2 caption for the polymers used in the compositions.

visible wavelengths emitted by the fluorescent lamps. We now examine the photocoloration and photobleaching spectral features of the same but heat-untreated compositions that were photodegraded under irradiation conditions identical to those reported earlier in Figure 2 (see caption). Figure 9 illustrates the effect(s) of light absorbed in the different spectral regions of the absorption spectra of photodegraded TiO2/polymer compositions. In essence, Figure 9 is the extension of Figure 2 (note the change in the time axis), except that the data are presented as the change in absorbance, ∆(∆Fλ), so that ∆(∆Fλ) values at the time of 270 min (point a, Figure 9) were set equal to 0 for all TiO2/polymer compositions to facilitate comparison. The numbers identifying the curves in Figure 9 are the same as those in Figure 2. Two important points are worth noting in the data of Figure 9. First, except for curves 4 and 4a, Figure 9 provides direct experimental evidence for the photobleaching of photoinduced colored compositions. Earlier, the phenomenon of photobleaching was demonstrated only for heat-treated (thermally degraded) compositions (Figures 7 and 8), and the coexistence of the two photoinduced processes, namely, the formation and disappearance of color centers, was deduced from an analysis of eq 1. Second, irradiation into the spectral regions hν < 2.30 eV (filter F, λ > 539 nm), hν < 2.86 eV (filter D, λ > 433 nm), and 2.55 eV < hν < 3.23 eV (filter C, 486 nm > λ > 384 nm) also activate the photobleaching process. Accordingly, the photodestruction of photogenerated color centers results from light absorption in all three visible absorption bands of TiO2. However, there is no complete disappearance of the color centers upon irradiation at wavelengths that induce the photobleaching process. Nonetheless, in comparison to Figure 2, new equilibrium (curves 2, 3, and 3a) or quasiequilibrium (curves 1, 1a, and 4) values of the absorbance, (∆Fλ)eq, are achieved. This suggests that irradiation (at least) into bands AB1 and AB2 induces both the formation and the disappearance of the color centers and that the ratio of the rates of these photoprocesses, which depend on the spectral range and intensity of the incident light, provides the equilibrium value

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15283

Figure 10. Kinetics of the change in absorbance at λ ) 500 nm (∆Fλ) for various TiO2/polymer compositions normalized by the value of ∆Fλ at time ) 270 min (point a). In the interval from 0 min to point a, the absorbance growth (curves 1, 1a, and 2) was induced by fluorescentlamp irradiation without filters in air (relative humidity 20%). For curves 3a-3c and for time > 270 min, filter F (λ > 539 nm, hν < 2.30 eV) was used in interval a-b, whereas at times beyond point b, filter D (λ > 433 nm, hν < 2.86 eV) was employed. Curve 4 was obtained using filter C (λ ) 384-486 nm, hν ) 3.23-2.55 eV). See text for additional details.

of the absorbance, (∆Fλ)eq. Thus, both the decrease and the increase in (∆Fλ)eq can be explained from such a viewpoint. For instance, after irradiation of the TiO2/polymer compositions by the full spectrum of the set of fluorescent lamps, i.e., by both UV and visible-light radiation (Figure 2, curve 4), removal of the visible-light wavelengths in the region λ > 463 nm (hν < 2.68 eV) at point a (time >270 min) suppresses the destruction of the color centers (Figure 9, curve 4) that was induced by the visible-light radiation and thus enhances (∆Fλ)eq. It should also be noted that, upon irradiation of the TiO2/ PVB composition, selecting the spectral range of irradiation (2.55 < hν < 3.23 eV) to correspond to the AB1 band results in a relatively poor response of the (∆Fλ)eq value (Figure 9, curve 2). By contrast, irradiation of the thermally degraded TiO2/ [P(VDF-HFP)] composition in the same spectral range of 2.55-3.23 eV shows pronounced photobleaching (Figure 6, curve 2 in the interval from 0 to point a). We deduce, therefore, that the ratio of the rate of photoinduced formation of the color centers to the rate of their photodestruction, at least in band AB1, depends on the type of polymer. Next, we examine the kinetic behavior of the photoinduced change(s) in absorbance at λ ) 500 nm (in the AB2 absorption band) measured concurrently with the change(s) in absorbance at λ ) 430 nm (AB1 band). Figure 10 shows additional noless-remarkable results. In the first stage of irradiation from 0 to 270 min (point a), the absorbance reaches equilibrium (curves 1 and 1a) or quasi-equilibrium values (curve 2). The shapes of the curves obtained at λ ) 500 and 430 nm are otherwise similar overall (see also Figure 2). In the next stage (interval a-b, Figure 10), the change in absorbance at λ ) 500 nm induced by irradiation at hν < 2.30 eV (λ > 539 nm) with filter F (curves 3a-3c) are also similar, at least in the sign of the change, to those for λ ) 430 nm (Figure 9, curves 1, 1a, and 3 in interval a-b). However, irradiation at photon energies of hν < 2.86 eV (λ > 433 nm, filter D) induces a strong growth in the absorbance at λ ) 500 nm, i.e., a significant increase in the intensity of absorption band AB2 (Figure 10, curves 3a-3c at times above

15284 J. Phys. Chem. C, Vol. 111, No. 42, 2007 point b). Moreover, growth of the intensity of band AB2 is also clearly seen upon irradiation in the range hν ) 3.23-2.55 eV (i.e., λ ) 384-486 nm; Figure 10, curve 4 for times above point a). Figure 10 thus exhibits a clear demonstration of the photoinduced formation of color centers responsible for absorption band AB2, which resulted from absorption of light into the bands AB1 (curve 4) and AB2 (curves 3a-3c). There is no simple explanation as to why only band AB2 yields such a clear manifestation. It is plausible that the color centers responsible for band AB2 are more sensitive to previous photobleaching (in the interval a-b) in the range of irradiation hν < 2.3 eV, i.e., absorption of light into band AB3. An analogous conclusion was reached after analysis of the data of Figure 9. In any case, we conclude that both intrinsic absorption of light by TiO2 and extrinsic light absorption by the color centers responsible for bands AB1 and AB2 activate a competitive process of formation and destruction of color centers associated with oxygen vacancies. This is a key result for unraveling the mechanism(s) of photostimulated processes involving color centers in titanium dioxide. 3.4. Plausible Mechanism(s) for the Photoformation and Photodestruction of Color Centers. All of the spectral results can be interpreted consistently in terms of the formation and destruction of color centers. The data confirm that absorption of light by TiO2 in the visible region originates only from color centers and not from a narrowing of the band gap of pristine TiO2.18,29,30 Moreover, the photobleaching phenomenon attracts attention insofar as the process causes the disappearance of color centers associated with oxygen vacancies. With the presently available data, we can now describe a plausible mechanism(s) for this photoinduced process. As a first step in our working hypothesis, we consider two different approaches. (i) In the first approach, we assume that, under irradiation, the color centers become optically silent. This assumption is based on the known properties of localized defects associated with oxygen vacancies in wide-band-gap metal oxides.32-34 There are three types of defects in the oxygen sublattice related to oxygen vacancies (viz. the so-called F-type centers): electrically neutral (with respect to the lattice) and optically silent oxygen vacancies, F++ centers, and oxygen vacancies with one (F+ centers) or two trapped electrons (F centers). Light absorption into the band of F+ and F centers in the visible region can potentially lead to photoionization of such centers and transform them into neutral F++ centers. The intrinsic light absorption by TiO2 to generate free valence-band holes (and free conduction-band electrons) can lead to recombination of the valence-band holes with the F+ and F centers converting these centers into the optically silent form F++. Also, the existence of several empty traps or other sites that are also capable of trapping the excess electrons are not precluded in this approach. (ii) In the second approach, we consider that a photostimulated process that involves the participation of oxygen can lead to annihilation of the oxygen vacancies and, thus, to the destruction of all the color centers. Earlier studies have shown that oxygen vacancies produced in a vacuum-annealed TiO2 surface can be oxidized in the presence of molecular oxygen.35,36 Note that we refer to oxygen-vacancy-related centers that are stable to interactions with molecular oxygen in the absence of irradiation. The thermal- or phototreatment-induced absorption spectra of TiO2 in TiO2/polymer compositions are not altered in the dark (see ref 18 and herein). Oxidation of such oxygen vacancies necessitates some form of photoactivation. In this

Kuznetsov and Serpone regard, we consider the mechanism of photoactivation to be analogous to photoadsorption (i.e., photoreduction) of molecular oxygen. Light absorption in the spectral region of intrinsic absorption (hν > 3.2 eV) and in the bands corresponding to the color centers (extrinsic absorption, hν < 3.2 eV) generates free electrons that induce the formation of surface oxygen species such as the superoxide radical anion O2-• and possibly also O- species. It is these anionic species that annihilate (oxidize) the oxygen-vacancy-related color centers. Comparing the two approaches, it is important to take into account the long-term ability of visible-light-active TiO2 photocatalysts to absorb light in the visible region, i.e., to consider the photostability of color centers as inferred above. It is not unreasonable to suppose that, in VLA TiO2 systems, the color centers are formed mainly within the volume of the micro-/ nanoparticles’ bulk during the preparation of the specimens. By contrast, in TiO2/polymer compositions, the color centers formed during the low-temperature heat treatment or during the photoinduced treatment are located on the particle surface or, at best, in the near subsurface region of the metal-oxide particles. The first mechanism (photophysical, approach i) can occur both in the bulk and on the surface of the metal-oxide particle, whereas the second mechanism (photochemical, approach ii) can apply only to the surface and subsurface color centers. Consequently, on the basis of the above analysis, the photochemical mechanism (approach ii) of the photobleaching process occurring through annihilation of oxygen vacancies would seem preferable to the photostimulated transformation of F and F+ centers into optically silent F++ centers (approach i). An interesting result from the TiO2/polymer compositions regards the insufficient optical response by these samples when the oxygen content in the surrounding atmosphere increased from ∼1% (N2 atmosphere) to about 20% O2 (air), i.e., the relatively low decrease of the AB1 band intensity (compare columns 2 and 3 in Table 2) and the not-insignificant decrease of the relative intensity of band AB2 (compare curves 1 and 2 in Figure 4). Note that oxygen-assisted photobleaching of heatdegraded compositions is clearly seen upon irradiation even under oxygen-poor conditions (N2 atmosphere; curves 2-4 in Figure 5). The difference in the O2 level in the two cases might provide a possible explanation for this observation. The concentration of O2 (CO2) present in the 1-L reactor chamber under the N2 atmosphere was CO2 ≈ 2.7 × 1020 O2 molecules, kept fairly constant under our conditions; in air, the O2 content was CO2 ≈ 5.3 × 1021 molecules. Considering that the overall surface of the samples (S) was ∼1.5 × 106 cm2, the maximum surface oxygen coverage, CO2/S, under the N2 atmosphere was 2.0 × 1014 molecules cm-2 (ca. 0.1 monolayer), whereas in air, the corresponding coverage was 4.0 × 1015 molecules cm-2 or about 2 monolayers. This significantly exceeded, by several orders of magnitude, the surface concentration of adsorbed O2 on P-25 TiO2 after reduction in H2 or CO atmosphere (CO2/S ) 3.0 × 1011 molecule cm-2) or after UV irradiation (CO2/S ) 6.0 × 1010 molecule cm-2).37,38 Hence, the occurrence of photobleaching under an oxygen-deficient atmosphere (nitrogen used herein) is not inconsistent with photochemical mechanism ii. Another relevant question regards the mechanism of the photoinduced formation of color centers. The present experimental data demonstrate that (a) formation of oxygen-vacancyrelated color centers occurs through both intrinsic absorption of light by TiO2 and extrinsic absorption of light into the 2.90 eV AB1 and 2.55 eV AB2 bands of the color centers and that (b) the photoinduced formation of color centers is a process that competes with the photoelectron-induced destruction of the

Photoinduced Coloration and Photobleaching of TiO2 color centers (i.e., the photobleaching phenomenon). Under intrinsic absorption of light by titanium dioxide and in competition with photoelectron-stimulated reactions, photohole-stimulated processes can also occur.36 It seems reasonable, therefore, to deduce that irradiation of the color centers into bands AB1 and AB2 should also generate free electrons and free holes. The similar manifestation of color centers responsible for absorption bands AB1 and AB2 suggests that these centers are closely related. However, the similarity of these centers does not, in itself, identify the nature of the color centers. Nonetheless, our results provide clear examples of increased sensitivity of band AB2 to the photobleaching phenomenon, which would explain the results of Figure 10. Band AB2 dominates in Figure 8 (curve 2b), whereas band AB2 is appreciably suppressed in the photoinduced absorption spectrum obtained in a relatively oxygen-rich atmosphere (air; spectrum 1 of Figure 4). This is a result of the increase (relative to band AB1) in the rate of the oxygen-assisted photoinduced destruction of the color centers responsible for band AB2. At the same time, we found no evidence of photoinduced formation of color centers resulting from light absorption into band AB3 (irradiation with filter F). In all cases, only the photostimulated destruction of color centers occurred (see, e.g., curve 2 in Figure 2 in interval a-b; Figure 8; curves 1, 1a, and 3 in Figure 9 in interval a-b; and curves 3a-3c in Figure 10 in interval a-b). This implies that the color centers responsible for the 2.05 eV AB3 band (e.g., formed by reduction of TiO2 or otherwise) likely differ in structure from the color centers that give rise to the 2.90 eV AB1 and 2.55 eV AB2 bands. An absorption band at ∼2.0 eV (ca. 620 nm) was reported in colloidal TiO2 following pulsed UV laser irradiation that was attributed to Ti3+ ions.39,40 Elsewhere, the 1.80 eV (∼690 nm) band seen in the absorption spectra of organometallic compounds containing a TiO2 structural group was also attributed to d-d transitions in Ti3+.41 The defect center (Ti3+-VO) with the oxygen vacancy (VO) located nearest the site of the central Ti3+ ion examined theoretically by Lu and co-workers 42 explained the spectral features at 2.9 and 1.7 eV in the absorption spectrum of a reduced rutile TiO2 single crystal. In line with the latter study, Sekiya et al.43 reported various color changes taking place when a single TiO2 anatase crystal was heat-treated in the presence of H2 and O2; the as-grown crystal was transformed from pale blue to dark blue to dark green to yellow to colorless upon annealing first in H2 and then in the presence of oxygen. Polarized crystal spectra43.44 revealed a band at ca. 2.9 eV and a very broad band around 1.8 eV that disappeared upon annealing in O2. The related EPR spectra of these multicolored TiO2 crystal specimens (pale blue to dark blue/ dark green) displayed only a single peak with g components of g(⊥) ) 1.992 and g(|) ) 1.963 that were attributed to the trapped electron Ti3+ (pale blue crystal) and to the presence of both Ti3+ and (Ti3+-VO) species in the dark blue/dark green specimens; the diamagnetic yellow TiO2 crystal was said to contain F centers.43 No EPR signal was observed with g ) 2.003-2.004 in the studies of Sekiya and co-workers,43 in contrast to EPR studies45 on powdered TiO2 nanoparticles that displayed signals at both g ) 1.981 and g ) 2.003-2.004, with the latter assigned to electrons trapped in oxygen vacancies (i.e., to an F+ center). By contrast, a related EPR study 46 on nanocrystalline N-doped TiO2 powder attributed the complex EPR signal at g ) 2003-2.005 to contributions from a paramagnetic N species in the bulk of TiO2, namely, Nb•, and from the superoxide radical anion O2-•. A close examination of several other studies in which absorption band assignments

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15285 correlated with EPR results suggested29,30 that the bands in the ranges 2.9-3.0 and 2.4-2.6 eV might be due to Jahn-Teller split 2T2 f 2E transitions of Ti3+ centers in anion- and cationdoped TiO2, whereas the band at 1.7-2.1 eV might implicate F+ centers. Despite the above suppositions, assignment of the optical properties of color centers that manifest absorption features in the visible spectral region remains a challenging task. We inferred above that photoactivation of color centers can generate free electrons and/or free holes. We hasten to note, however, that optical (diffuse reflectance) spectroscopy provides no direct experimental evidence for the photogeneration of free electrons and/or free holes by extrinsic light absorption by color centers. This inference needs to be confirmed by other experimental techniques, for example, by photoconductivity or other equivalent studies. 3.5. Postirradiation Effect. The results reported in Figure 6 show that termination of irradiation led to an increase in absorbance of the TiO2/polymer samples under dark conditions. To rationalize this postirradiation phenomenon, it is reasonable to assume that, prior to any treatment, intrinsic traps for photogenerated electrons and photogenerated holes pre-exist in the titanium dioxide of the TiO2/polymer compositions. Under irradiation, charge-carrier trapping occurs concomitantly and in competition with the photogeneration and recombination of free charge carriers and with hole-activated formation (oxidation of the polymer) and electron-activated annihilation (through reactive surface oxygen species O2-• and O-) of color centers. Subsequent to termination of irradiation, charge carriers can thermally escape from the traps (i.e., can be thermally detrapped) to partake in recombination processes and in corresponding surface reactions. The existence of intrinsic traps, especially for powdered specimens, is invoked frequently in solid-state physics to rationalize various dark phenomena subsequent to irradiation, e.g., long persistent phosphorescence and thermoluminescence of solids. In this regard, a recent study47 reported slow (minutes to hours) surface charge carrier trapping in P-25 TiO2 samples and their slow annihilation subsequent to UV irradiation in vacuum. Figure 6 exhibits the case where the photoinduced disappearance of color centers dominates, which means that reaction involving photogenerated electrons (and oxygen) dominates, with the consequence that there will be an excess of free and trapped holes. Upon termination of irradiation, the increase in the number of free holes through thermal detrapping provides a path for the growth of the absorbance, i.e., for an increase in the number of color centers. The observed dark response is a rather characteristic feature of the TiO2/polymer compositions examined in this study, as attested by the results of Figure 6. It should be noted, however, that, in general, the dark response can manifest itself either as an increase in absorbance (+ sign, as in Figure 6) or sometimes as a decrease in absorbance (sign, bleaching). Thus, the postirradiation phenomenon confirms, albeit indirectly, the proposed photochemical mechanism (approach ii) for the photocoloration of TiO2 and the photobleaching of color centers. It will be instructive to examine the kinetics of these two competitive processes by modeling the temporal changes in absorption and changes in bleaching features observed in the visible region at 430 nm for TiO2/polymer compositions (Figure 5) and rationalized using the simple regularity eq 1. 3.6. Modeling of the Photoinduced Formation and Destruction of Color Centers. We consider the kinetics of a reaction consisting of two competitive photoinduced processes, namely, the formation and disappearance of the color centers.

15286 J. Phys. Chem. C, Vol. 111, No. 42, 2007

Kuznetsov and Serpone

Denoting the rate of photoinduced formation by V+ and the rate of photoinduced destruction by V-, one can express the rate of change of the number of color centers N by the differential eq 2

dN ) V+ - V dt

(2)

For irradiation of the TiO2/polymer compositions in the UV region, we can assume that the rate of formation of photoholeactivated color centers is a constant with time, i.e., V+ ) k+, where k+ depends on the light intensity, the absorbance of the sample, the nature of the polymer, and the quantum yield of the photoreaction, among other factors. The above assumption means that the degree of photoinduced transformation of the TiO2/polymer system is expected to be relatively low. By contrast, the rate of disappearance of photoelectron-activated color centers is directly proportional to the number of color centers N in accord with the photochemical mechanism ii of photobleaching. Consequently, V- ) k-N, where k- is a constant of the rate of photodestruction of color centers. The constant k- depends on the same factors above in addition to the oxygen content available in the gas-phase environment and on the permeability of the polymer film to oxygen (among others). Accordingly, for the UV region, eq 2 takes the differential form of eq 3a

dN ) k+ - k - N dt

(3a)

We showed earlier that, in the visible region, the absorption of TiO2/polymer compositions originates from the absorption of color centers. As a result, the main feature of the kinetics expressed by eq 2 under irradiation in the visible region is the dependence of the absorbance of the sample on the number of color centers N. If we suppose that k+ and k- are directly proportional to N, then, for the visible-light region, the differential equation takes the form

dN ) k′+N - k′-N 2 dt

(3b)

where k′+ and k′- are constants analogous to k+ and k- but independent of the absorbance of the sample. Equation 3a is a simple first-order linear equation, and eq 3b is of the form of the Bernoulli equation that can be converted into a first-order linear differential equation by simple change of variable, z ) 1/N. Solving eqs 3a and 3b for N for irradiation of the compositions in the UV region (eq 4a) and in the visible region (eq 4b), we obtain

N)

(

)

k+ k+ - N0 exp(-k-t) kk-

(4a)

and

N)

1 k′k′1 - +exp(-k′+t) + N k′ k′ 0

(

)

(4b)

where N0 is the number of color centers that exist prior to irradiation. Note that, in eq 4b, N0 > 0 always because, if N0 were 0, it would contradict the initial premise (for the visible region) that light absorption occurs by the color centers. It follows from eqs 4a and 4b that, as t f ∞, then N f k+/k- or k′+/k′-. That is, the ratio k+/k- (or k′+/k′-) corresponds to the

equilibrium value Neq, which is independent of N0 but dependent on the irradiation conditions and the properties of the polymer macromolecules and polymer films. These inferences are consistent with the reported experimental data. Indeed, the value (∆F)eq (i.e., Neq in the present treatment) is independent of (∆F)ini (i.e., N0) and characterizes a definite TiO2/polymer composition (see Table 2). Furthermore, it follows from eqs 4a and 4b that the difference (Neq - N0) [in our experiments, we used (∆F)eq - (∆F)ini] provides the sign of change for the number of color centers N. Further treatment (differentiation) of eqs 4a and 4b yields the following expressions for irradiation in the UV region (eq 5a) and in the visible region (eq 5b)

dN ) (Neq - N0)k- exp(-k-t) dt

(5a)

and

(

)

Neq - N0 2 + dN ) N k′ exp(-k′+t) dt NeqN0

(5b)

We deduce from eqs 5a and 5b that N0 < Neq leads to an increase in the number of color centers (additional photocoloration in the experiments), whereas N0 > Neq leads to a decrease in the number of color centers N (photobleaching). Finally, if N0 ) Neq, then N would always equal N0 and Neq. These expectations are in full accord with results of the analysis from eq 1 written in the form ∆(∆F) ) (∆F)eq - (∆F)ini. The results reported earlier in Figures 9 and 10 (section 3.3) were explained on the basis that the absorbance growth and absorbance decrease [i.e., (∆F)eq] following a change in the spectral region of irradiation is the result of a change in the ratio of the rates of photoformation and photodisappearance of color centers. The present model confirms this inference unambiguously because only the ratio k+/k- provides the Neq value. A full description of the kinetics of photocoloration/photobleaching of TiO2/polymer compositions would necessitate the construction of a set of differential equations for the UV region and for (at least) absorption bands AB1 and AB2. A solution to that set of equations can be obtained only by numerical methods, which is beyond the scope of the present exercise. However, it is evident that the results would include some uncertainties because of variations in the values of k+ and kin the different spectral regions examined. Nonetheless, if we assume certain reasonable values for k+ (0.0015 min-1), k(0.020 min-1), and N0 (0, 0.040, 0.075, 0.10, and 0.15), we obtain the results illustrated in Figure 11 for the simulated UVlight irradiation of TiO2/polymer compositions. Clearly, the absence of pre-existing defects/color centers (N0 ) 0) in TiO2 yields curve 1, indicating formation of color centers, i.e., photocoloration occurs (compare with curves 3a and 4a in Figure 2). If N0 < Neq () 0.075), additional photocoloration takes place (curve 2; compare with curves 1 and 1a for TiO2/PMPS in Figure 5), whereas N0 ) 0.075 yields curve 3, indicating no changes in N and thus the occurrence of competitive formation and destruction of the color centers (compare to curve 4a in Figure 5 for TiO2/PVB). When N0 > 0.075, photoinduced bleaching of color centers dominates. For reasonable values of the constants k′+ and k′- and for an initial number of color centers N0, Figure 12 demonstrates the three main cases occurring for TiO2/polymer compositions subjected to irradiation in the visible region: photocoloration, photobleaching, and “absence of change”. Curves 4 and 5 are qualitatively comparable to curves 2-4 in Figure 5 and curves

Photoinduced Coloration and Photobleaching of TiO2

Figure 11. Simulated temporal changes in the number of color centers (N - N0) calculated in accord with eq 4a for irradiation of TiO2/polymer compositions in the UV region. k+ ) 0.0015 min-1, k- ) 0.020 min-1. N0 ) 0 (curve 1), 0.040 (curve 2), 0.075 (curve 3), 0.10 (curve 4), and 0.15 (curve 5).

J. Phys. Chem. C, Vol. 111, No. 42, 2007 15287 (band AB2), and 604 nm (band AB3). Coloration could also be induced thermally after a heat treatment. Both the photoinduced and thermally induced color centers are photobleached upon irradiation of the colored compositions by both UV radiation (i.e., in the intrinsic spectral region at energies greater than 3.2 eV) and visible-light irradiation into the absorption bands of the color centers (i.e., in the extrinsic spectral region at hν < 3.2 eV). The latter indicates that the spectral bands of the color centers are photoactive in the destruction of color centers. A plausible photochemical mechanism is proposed for this photobleaching phenomenon that involves oxygen-assisted annihilation of oxygen vacancies. The unambiguous experimental data reported, namely, the observed photobleaching and the commonality of the absorption spectral features with those of anion- and cation-doped titanias (regardless of their historical preparative methods), negate the existence of band-gap narrowing in TiO2 systems proposed by several workers to account for the red shifts of the absorption edge observed in doped titania specimens. Modeling of the competitive formation and destruction of the color centers by simple mechanistic considerations provides further support, albeit qualitative, for the observations of both increases and decreases in absorbance. Acknowledgment. Early work in this area received financial support from the Russian Foundation for Basic Research (to V.N.K.), whereas studies in Pavia, Italy, were supported by a grant from the Ministero dell’Universita e Ricerca (MUR, Rome, Italy). One of us (N.S.) thanks Prof. Angelo Albini and his group at the University of Pavia for their gracious hospitality during the 2007 winter semester. References and Notes

Figure 12. Simulated temporal changes in the number of color centers (N - N0) calculated in accord with eq 4b for irradiation of TiO2/polymer compositions in the visible region. k′+ ) 0.030 min-1, k′- ) 0.50 min-1. N0 ) 0.00010 (curve 1), 0.060 (curve 2), 0.10 (curve 3), and 0.15 (curve 4).

1-3 in Figure 6. It is difficult to obtain experimental evidence for photoinduced processes that might be compared with curve 1 in Figure 12 because no changes in the number of color centers are expected during the initial irradiation period. The above simulations indicate that the principal regularities observed in the photodegradation of TiO2/polymer compositions can be modeled successfully using a relatively simple scheme, in spite of differences in the kinetic equations for irradiation in the UV and visible regions. The qualitative similarities between the experimental results of Figures 5 and 6 and the simulated results of Figures 11 and 12 lend further credence to the formation and destruction of color centers in TiO2/polymer systems. Indeed, the results from the modeling indirectly confirm the notion that photodegradation occurring in these systems includes competitive processes photoactivated by both bandto-band (intrinsic) UV-light absorption and extrinsic visiblelight absorption by color centers. 4. Concluding Remarks We have demonstrated herein that TiO2/polymer compositions with various polymers are photocolored upon irradiation by fluorescent lamps emitting UV/visible radiation. This leads to the formation of color centers that display absorption features in the visible spectral region at 427 nm (band AB1), 486 nm

(1) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269. (2) Morikawa, T.; Asahi, R.; Ohwaki, T.; Aoki, A.; Taga, Y. Jpn. J. Appl. Phys. Part 2 2001, 40, L561. (3) Umebayashi, T.; Yamaki, T.; Itoh, H.; Asai, K. Appl. Phys. Lett. 2002, 81, 454. (4) Umebayashi, T.; Yamaki, T.; Yamamoto, S.; Miyashita, A.; Tanaka, S.; Sumita, T.; Asai, K. J. Appl. Phys. 2003, 93, 5156. (5) Belver, C.; Bellod, R.; Stewart, S. J.; Requejo, F. G.; FernandezGarcia, M. Appl. Catal. B: EnViron. 2006, 65, 309. (6) Thomas, A. G.; Flavel, W. R.; Stockbauer, R.; Patel, S.; Gratzel, M.; Hengerer, R. Phys. ReV. B 2003, 67, 035110. (7) Colon, G.; Hidalgo, M. C.; Munuera, G.; Ferino, I.; Cutrufello, M. G.; Navio, J. A. Appl. Catal. B: EnViron. 2006, 63, 45. (8) Gole, J. L.; Stout, J. D.; Burda, C.; Lou, Y.; Chen, X. J. Phys. Chem. B 2004, 108, 1230. See also: http://www.physics.gatech.edu/people/ faculty/jgole.html#links. (9) Nakamura, I.; Negishi, N.; Kutsuna, S.; Ihara, T.; Sugihara, S.; Takeuchi, K. J. Mol. Catal. A: Chem. 2000, 161, 205. (10) Wang, H.; Lewis, J. P. J. Phys.: Condens. Matter 2006, 18, 421. (11) Lindgren, T.; Mwabora, J. M.; Avendano, E.; Jonsson, J.; Granqvist, C. G.; Lindquist, S. E. J. Phys. Chem. A 2003, 107, 5709. (12) Irie, H.; Watanabe, Y.; Hashimoto, K. J. Phys. Chem. B 2003, 107, 5483. (13) Di Valentin, C.; Pacchioni, G.-F.; Selloni, A. Phys. ReV. B 2004, 70, 085116. (14) Nakamura, R.; Tanaka, T.; Nakato, Y. J. Phys. Chem. B 2004, 108, 10617. (15) Di Valentin, C.; Pacchioni, G.-F.; Selloni, A.; Livraghi, S.; Giamello, E. J. Phys. Chem. B 2005, 109, 11414. (16) Zhao, J.; Chen, C.; Ma, W. Top. Catal. 2005, 35, 267. (17) Li, D.; Haneda, H.; Hishita, S.; Ohashi, N. Res. Chem. Intermed. 2005, 31, 331. (18) Kuznetsov, V. N.; Serpone, N. J. Phys. Chem. B 2006, 110, 25203. (19) Ihara, T.; Miyoshi, M.; Iriyama, Y.; Matsumoto, O.; Sugihara, S. Appl. Catal. B: EnViron. 2003, 42, 403. (20) Wang, X. H.; Li, J.-G.; Kamiyama, H.; Ishigaki, T. Thin Solid Films 2006, 506/507, 278. (21) Cheng, P.; Li, W.; Zhou, T.; Jin, Y.; Gu, M. J. Photochem. Photobiol. A: Chem. 168 (2004) 97. (22) Anpo, M.; Takeuchi, M. J. Catal. 2003, 216, 505.

15288 J. Phys. Chem. C, Vol. 111, No. 42, 2007 (23) Li, F. B.; Li, X. Z.; Hou, M. F.; Cheah, K. W.; Choy, W. C. H. Appl. Catal. A: Gen. 2005, 285, 181. (24) Yin, S.; Yamaki, H.; Komatsu, M.; Zhang, Q.; Wang, J.; Tang, Q.; Saito, F.; Sato, T. J. Mater. Chem. 2003, 13, 2996. (25) Otsuka-Yao-Matsuo, S.; Ueda, M. J. Photochem. Photobiol. A: Chem. 2004, 168, 1. (26) Li, D.; Ohashi, N.; Hishita, S.; Kolodiazhnyi, T.; Haneda, H. J. Solid State Chem. 2005, 178, 3293. (27) Aita, Y.; Komatsu, M.; Yin, S.; Sato, T. J. Solid State Chem. 2004, 177, 3235. (28) Joung, S-K.; Amemiya, T.; Murabayashi, M.; Itoh, K. Appl. Catal. A: Gen. 2006, 312, 20. (29) Serpone, N. J. Phys. Chem. B 2006, 110, 24287. (30) Serpone, N.; Emeline, A. V.; Kuznetsov, V. N.; Ryabchuk, V. K. Second-Generation Visible-Light-Active Photocatalysts: Preparation, Optical Properties and Consequences of Dopants in the Band Gap Energies of TiO2. In EnVironmentally Benign CatalystssApplications of Titanium OxideBased Photocatalysts; Anpo, M., Kamat, P. V., Eds.; Springer: New York, 2008; in press. (31) Kuznetsov, V. N.; Malkin, M. G. J. Appl. Spectrosc. 2000, 67, 763. (32) Stoneham, A. M. Theory of Defects in Solids; Clarendon Press: Oxford, U.K., 1975. (33) Henderson, B.; Werts, J. E. Defects in the Alkaline Earth Oxide and Applications to Radiation Damage and Catalysis; Taylor and Francis: London, 1977. (34) Kotomin, E. A.; Popov, A. I. Nucl. Instrum. Methods Phys. Res. B 1998, 141, 1. (35) Epling, W. S.; Peden, C. H. F.; Henderson, M. A.; Diebold, U. Surf. Sci. 1998, 412/413, 333. (36) Diebold, U. Surf. Sci. Rep. 2003, 48, 53.

Kuznetsov and Serpone (37) Kuznetsov, V. N.; Krutitskaya, T. K. Kinet. Catal. 1996, 37, 446. (38) Lisachenko, A. A.; Kuznetsov, V. N.; Zacharov, M. N.; Michailov, R. V. Kinet. Catal. 2004, 45, 189. (39) Rothenberger, G.; Moser, J.; Gratzel, M.; Serpone, N.; Sharma, D. K. J. Am. Chem. Soc. 1985, 107, 8054. (40) Bahnemann, D.; Henglein, A.; Spanhel, L. Faraday Discuss. 1984, 78, 151. (41) Jeske, P.; Haselhorst, G.; Weyhermuller, T.; Wieghardt, K.; Nuber, B. Inorg. Chem. 1994, 33, 2462. (42) Lu, T.-C.; Wu, S.-Y.; Lin, L.-B.; Zheng, W.-C. Physica B 2001, 304, 147. (43) Sekiya, T.; Yagisawa, T.; Kamiya, N.; Mulmi, D. D.; Kurita, S.; Murakami, Y.; Kodaira, T. J. Phys. Soc. Jpn. 2004, 73, 703. (44) Sekiya, T.; Tasaki, M.; Wakabayashi, K.; Kurita, S. J. Lumin. 2004, 108, 69. (45) (a) Sun, Y.; Egawa, T.; Shao, C.; Zhang, L.; Yao, X. J. Cryst. Growth 2004, 268, 118. (b) Sun, Y.; Egawa, T.; Shao, C.; Zhang, L.; Yao, X. J. Phys. Chem. Solids 2004, 75, 1793. (c) Nakamura, I.; Negishi, N.; Kutsuma, S.; Ihara, T.; Sugihara, S.; Takeuchi, K. J. Mol. Catal. A: Chem. 2000, 161, 205. (d) Ihara, T.; Miyoshi, M.; Ando, M.; Sugihara, S.; Iriyama, Y. J. Mater. Sci. 2001, 36, 4201. (e) Serwicka, E.; Schlierkamp, M. W.; Schindler, R. N. Z. Naturforsch. A 1981, 36, 226. (f) Serwicka, E. Colloids Surf. 1985, 13, 287. (g) Volodin, A. M.; Cherkashin, A. E.; Zakharenko, V. S. React. Kinet. Catal. Lett. 1979, 11, 103; 1979, 11,107; 1979, 11, 221. (h) Cornaz, P. F.; Van Hoof, J. H. C.; Pluijim, F. J.; Schuit, G. C. A. Discuss. Faraday Soc. 1966, 41, 290. (46) Livraghi, S.; Paganini, M. C.; Giamello, E.; Selloni, A.; Di Valentin, C.; Pacchioni, G.-F. J. Am. Chem. Soc. 2006, 128, 15666. (47) Szczepankiewicz, S. H.; Moss, J. A.; Hoffman, M. R. J. Phys. Chem. B 2002, 106, 2922.