Degradation of phenol and benzoic acid in the presence of a TiO2-based heterogeneous photocatalyst

Applied Catalysis B: Environmental 58 (2005) 79–88 www.elsevier.com/locate/apcatb

Degradation of phenol and benzoic acid in the presence of a TiO2-based heterogeneous photocatalyst Davide Vionea,*, Claudio Mineroa, Valter Maurinoa, M. Eugenia Carlottib, Tatiana Picatonottoa, Ezio Pelizzettia a

b

Dipartimento di Chimica Analitica, Universita` di Torino, Via Pietro Giuria 5, 10125 Torino, Italy Dipartimento di Scienza e Tecnologia del Farmaco, Universita` di Torino, Via P. Giuria 9, 10125 Torino, Italy Received 5 October 2004; received in revised form 24 November 2004; accepted 29 November 2004 Available online 6 January 2005

Abstract This paper reports the results of a study on the titanium dioxide Wackherr’s ‘‘Oxyde de titane standard’’, which shows very interesting photocatalytic activity. Produced for cosmetic purposes as a white pigment, its features make it very interesting in the field of heterogeneous photocatalysis. The results obtained with this photocatalyst are compared with the behaviour of the well-studied and widely used TiO2 Degussa P25 under the same conditions. In particular, the TiO2 Wackherr induces relevantly faster phenol degradation than P25 when high photocatalyst loading is used (up to 2.00 g l
1), as phenol degradation rate in the presence of TiO2 Wackherr continues to increase with increasing photocatalyst loading. This is most likely due to the lower radiation scattering in the UV region of TiO2 Wackherr when compared with Degussa P25, which is linked with the higher particle size of TiO2 Wackherr. The initial phenol degradation rate by TiO2 Wackherr as a function of phenol concentration has a maximum for [phenol]  3  10
4 M, and decreases for higher concentration values. The addition of fluoride ions to TiO2 Wackherr at pH 3.7 increases phenol degradation rate, as already found for Degussa P25. The increase is more relevant for higher phenol concentration values and makes the maximum as a function of phenol concentration to disappear. Comparable results are also obtained when benzoic acid is used as a substrate, with some differences from phenol that can be accounted for by benzoic acid more strongly interacting with the surface. The use of TiO2 Wackherr in heterogeneous photocatalysis can be desirable when high photocatalyst loading is required. # 2004 Elsevier B.V. All rights reserved. Keywords: Pollutant photodegradation; Photocatalysis; Laser light scattering; Kubelka-Munk equation; Diffuse reflectance; Surface recombination processes

1. Introduction The availability of non-polluted water for human use is a major problem, which is going to increase in the future. The improvement of methods for sewage treatment is of capital importance in this context. Efforts on cost abatement would allow a larger fraction of wastewater to be properly treated. In addition, as good-quality spring water is less and less available, the use of contaminated surface waters as a source of drinking water is steadily increasing, and their treatment * Corresponding author. Tel.: +39 011 6707633; fax: +39 011 6707615. E-mail address: [email protected] (D. Vione) URL: http://www.environmentalchemistry.unito.it, http://www.abcrg.it. 0926-3373/$ – see front matter # 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2004.11.018

is becoming crucial. The growing concern about the problem of water decontamination from organic pollutants (both wastewater and water intended for human or industrial use) has led to the research of methods being able to obtain better results than the traditional ones, possibly with lower consumption of chemical reagents. A promising solution is the oxidation of the pollutants, often up to the complete mineralisation [1]. Among the so-called advanced oxidation processes (AOPs), titanium dioxide heterogeneous photocatalysis is one of the most promising. In this process, a suspension or a supported layer of water-insoluble semiconductor oxide is exposed to UV radiation, or to sunlight. Irradiation induces the formation of reactive species on the surface of the photocatalyst (h+, surface-adsorbed OH, and e
), which

80

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

results in the degradation, and in most cases in the complete mineralisation, of a large variety of organic contaminants [2,3]. However, one of the main problems that have limited the practical applications of heterogeneous photocatalysis so far is either the relative slowness of the process, or the limited efficiency in the use of irradiated energy [1,4]. A possible way to solve the problem is the exploitation of low-cost radiation (solar energy [5,6]), but this approach finds an obstacle in the limited UV radiation intensity in the solar spectrum. For this reason, use of metal doping to extend TiO2 absorption in the visible is currently under study [7,8]. Another approach is the search for new photocatalysts showing higher activity, or for new methods to increase the activity of known photocatalysts, which would allow a better use of the radiation energy. For instance, it has recently been found that the addition of fluoride to TiO2 Degussa P25 relevantly enhances the degradation rate of phenol [9,10]. The titanium dioxide Degussa P25 is a widely studied material, particularly suitable for use as a standard of photocatalytic reactivity [4]. In recent studies into the photocatalytic activity of inorganic sunscreens, we have found that some cosmetic pigments are more active than TiO2 Degussa P25 towards the degradation of phenol [11,12]. This finding confirms the concerns about the photocatalytic activity of inorganic pigments used as sunscreens [13–16], but the relevant activity of some of the studied pigments is also interesting for the environmental applications of heterogeneous photocatalysis as well as for mechanistic implications. The most active pigment among the ones we have studied was adopted in the present work, and its photocatalytic behaviour compared with the one of Degussa P25 under a wider variety of conditions than was originally done. The effect of fluoride addition was studied for the new photocatalyst, as well as the degradation of benzoic acid in the presence and in the absence of fluoride.

2. Experimental The studied pigment was Wackherr’s ‘‘Oxyde de titane standard’’ (anatase), produced with the sulphate process. TiO2 P25 (anatase around 80%, rutile around 20% [7,17]) was a gift from Degussa. Table 1 reports as a comparison

various parameters of the two compounds. Phenol (purity grade >99%) was purchased from Aldrich, benzoic acid (99%) from Carlo Erba, acetonitrile (Gradient grade), 2propanol (Gradient grade), NaF (>99%), H3PO4 (85%), NaH2PO4H2O (>99%), and HClO4 (70%) from Merck. The suspension pH was 3.7 for all the experiments. This pH value was chosen to make direct comparisons possible between the experiments in the presence and in the absence of fluoride. In the presence of fluoride, the interaction of the ion with the TiO2 surface is maximised around pH 3.7 [9]. The pH value was adjusted with the addition of HClO4, since this acid does not interfere with the photocatalytic processes [3,18]. It is to be observed that TiO2 Wackherr was originally compared with Degussa P25 towards the degradation of phenol in neutral solution, without HClO4, and proved to be more active under such conditions [11]. The semiconductor suspensions were prepared using a 200 W Branson 2200 sonifier. Irradiation was carried out in cylindrical Pyrex glass cells (4 cm diameter, 2.5 cm height, 5 ml suspension volume) under a Solarbox (CO.FO.ME.GRA., Milan, Italy), equipped with a 1500 W Xenon lamp. This irradiation device simulates the solar spectrum. The incident radiation in solution in the UV region was 24 W m
2, measured with a CO.FO.ME.GRA. Multimeter. The photon flux in solution was also determined by ferrioxalate actinometry [19] in the wavelength interval 300–510 nm, and calculated taking into account the absorption spectrum and quantum yield of potassium ferrioxalate over the considered wavelength range. We measured a photon flux in solution of 5.5  10
7 Ein s
1. The transformation of phenol and benzoic acid was followed by HPLC analysis. Isocratic elution was carried out with a mixture of 30/70 acetonitrile/phosphate buffer (0.050 M total phosphate, pH 2.8), flow rate 1.00 ml min
1. The column used was a Merck LiChroCART 125-4 (125 mm length, 4 mm diameter), packed with LiChrospher 100 RP-18 (particle diameter 5 mm). Retention times were 3.65 min for phenol and 3.90 min for benzoic acid, the column dead time being 0.90 min. Detection wavelength was 210 nm. Before analysis, the suspensions were filtered through Millipore HA membranes. Absorbance measures for actinometry and UV–vis spectra were carried out with a Varian Cary ‘‘100 Scan’’ spectrophotometer. The extinction of the photocatalysts was measured on freshly prepared, previously sonicated suspen-

Table 1 Comparison between TiO2 Degussa P25 and Wackherr’s ‘‘Oxyde de titane standard’’ TiO2 Degussa P25a Phase composition Crystallite size (nm) BET specific surface area (m2 g
1) Pore diameter a b c d

Data reported in Ref. [17]. Data provided by the manufacturer. Calculated from X-ray diffraction patterns. Derived from scanning electron microscopy data.

Wackherr’s ‘‘Oxyde de titane standard’’b c

79% anatase, 21% rutile 22c 51.0 ˚ (sic) 315 A

Anatase 300d 8.5  1.0 Not relevant

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

sions, at low loading values (around 10 mg l
1), and 1 cm optical path length. Under low loading conditions the suspensions were rather stable at the time scales necessary for the measurements, and the results of repeated measures were fairly reproducible. Suspensions were not stirred during the measures. Absorption and scattering coefficients for both photocatalysts at 360 nm were also evaluated by fitting the extinction data at different loading values with the KubelkaMunk equation. Such an equation has proved suitable for the description of the optical properties of rutile titania [20]. Particle size distribution was obtained with an ALV-NIBS High Performance Particle Sizer by ALV GmbH, Landen, Germany.

3. Results and discussion 3.1. Comparison between TiO2 Wackherr and TiO2 Degussa P25 The comparison between the two photocatalysts was carried out using the degradation of phenol as a probe. Actually, the behaviour of this compound under photocatalytic conditions has been extensively studied [2,3] and it can be very effectively used to get insight into the photocatalytic activity of a given oxide [9,10]. The initial degradation rate of 2.1  10
4 M phenol at pH 3.7 was studied as a function of TiO2 loading for both photocatalysts (Wackherr and Degussa). The results are reported in Fig. 1. In the case of Degussa P25 the plot deviates very soon from linearity, and phenol degradation rate reaches a roughly constant value above 0.50 g l
1. This result is consistent with previous findings, and is most likely due to the combination of light absorption and scattering by the semiconductor oxide [21,22]. TiO2 Wackherr shows an increasing phenol degradation rate versus TiO2 loading in

Fig. 1. Phenol initial degradation rate as a function of photocatalyst loading, for TiO2 Wackherr (&) and for Degussa P25 (~). Initial phenol concentration 2.1  10
4 M, suspension pH 3.7, adjusted with the addition of HClO4.

81

the interval 0–2.00 g l
1. Although P25 is slightly more active at low photocatalyst loading, TiO2 Wackherr becomes relevantly more active at high loading. This behaviour can be attributed to lower scattering of radiation by TiO2 Wackherr when compared with P25. Differences in substrate degradation rates under photocatalytic conditions can also be due to different charge-carrier dynamics [23–25]. In this scenario, however, one would expect a similar systematic effect (in our case, higher degradation rate for TiO2 Wackherr) over the whole range of TiO2 loading values. When considering Fig. 1, it is apparent that the differences between TiO2 Wackherr and Degussa P25 strongly depend on the loading. This dependence on the loading points to the photocatalyst optical properties as the main cause for the observed differences. Differences in charge-carrier dynamics cannot be ruled out, but are likely to be not so high as to play the main role in the studied system. Due to the interest of the described effect of photocatalyst loading for the practical applications of heterogeneous photocatalysis, we carried out further investigation on the subject in order to test our hypothesis. Fig. 2 shows the spectra of the two photocatalysts (0.010 g l
1 = 1.0  10
5 g cm
3) in the UV–vis region at pH 3.7. In both cases the spectrophotometrically measured extinction is the combination of absorption and scattering by the particles [26]. For instance, the light extinction of TiO2 Wackherr in the visible region can most likely be attributed to scattering, as titanium dioxide does not absorb in the visible. For the same reason, P25 is not transparent above 400 nm. The wavelength interval useful for photocatalytic purposes is the near UV, as TiO2 absorbs radiation below 400 nm [26]. In this region there is competition between absorption and scattering. It is worth recalling that the adopted irradiation device emits radiation above 300 nm, simulating the solar spectrum, which sets the useful

Fig. 2. UV–vis extinction spectra for TiO2 Wackherr and Degussa P25. Photocatalyst loading 0.010 g l
1 = 1.0  10
5 g cm
3, optical path length 1 cm, pH 3.7. Rayleigh scattering for Degussa P25 (*). The extinction spectrum of a 1.0  10
5 g cm
3 colloidal TiO2 sample is also reported for a comparison.

82

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

wavelength range for photocatalysis in the interval 300– 400 nm. Fig. 2 also reports the extinction spectrum of a colloidal TiO2 sample at the same loading value as TiO2 Wackherr and Degussa P25 (1.0  10
5 g cm
3). Colloidal TiO2 has been synthesised by one of us (VM) via TiCl4 forced hydrolysis [9,10]. A suspension of the cited colloidal TiO2 sample looks transparent: indeed, it has negligible extinction in the visible region (see Fig. 2), because of negligible scattering and negligible absorption of visible light. Actually, the extinction spectrum of colloidal TiO2 is the nearest thing to a genuine absorption spectrum of titanium dioxide that can be obtained by direct spectrophotometric measures [2,3]. The comparison between the extinction spectra of colloidal TiO2 and of Degussa P25 indicates that most of the extinction of Degussa P25 in the wavelength interval 300–400 nm is due to scattering. The behaviour of Degussa P25 and TiO2 Wackherr towards scattering is very different. TiO2 Wackherr shows a very relevant scattering in the visible, with a maximum around 500 nm, while the scattering by Degussa P25 in the visible follows with very good approximation the Rayleigh law: for limited values of the scattering, its intensity is / l
4 [27]. The trend of the Rayleigh scattering for Degussa P25, shown in Fig. 2, has been determined by fitting of the extinction data above 400 nm, where absorption by TiO2 is negligible, with the equation. Extinction = kl
4, and by extrapolation of the curve until 350 nm. The comparison between the experimental extinction by Degussa P25 and the Rayleigh scattering curve gives a rough indication of the relative weight of absorption and scattering. As already done with a completely different approach, it can again be inferred that most of the extinction of Degussa P25 around, e.g. 360 nm is due to scattering. At l < 360 nm absorption by Degussa P25 can be expected to increase [26], but such an increase is compensated by the lower emission intensity of the adopted lamp at the shorter wavelengths [11,12]. Furthermore, the evaluated Rayleigh scattering of Degussa P25 below 390 nm is higher than the total extinction of TiO2 Wackherr. It can therefore be inferred that UV radiation scattering by TiO2 Degussa P25 is higher than scattering by TiO2 Wackherr. The main drawback of the described procedures is the difficulty to evaluate the relative contributions of absorption and scattering to extinction: while a rough indication could be obtained for Degussa P25, this was not possible for TiO2 Wackherr. Furthermore, measurement of the extinction with the experimental set-up described above only gives a lower limit: the instrument measures higher transmitted radiation intensity than the one expected from the ‘‘true’’ extinction coefficient. This is due to the combined effects of scatteringout and of scattering-in (multiple scattering) [4]. The instrument is likely to measure some of the scattered-out radiation because the slit of the spectrophotometer is finite in width, thus collecting some radiation that has actually been scattered by a small angle. The other problem is the

scattering-in or multiple scattering. While a single scattering event modifies the direction of the photon, the original direction can be restored in case of a second scattering event, or a series of events. As a consequence of both effects, the extinction coefficients that can be obtained from Fig. 2 are underestimated [26]. It is possible to account for multiple scattering in an approximate but simple way applying the Kubelka-Munk model of diffuse reflectance and transmission, where scattered light consists of two streams, one directed forward and the other backward. Moreover, the model allows one to estimate the values of the absorption and of the scattering coefficients by fitting of the experimental data. The following equation is obtained for the extinction E (E =
log10(T), T being transmittance) [20]: 0 1 4k R B C E ¼
log10 @ðk þ RÞ2 expð2Ccat bRÞ (1) A
ðk
RÞ2 expð
2Ccat bRÞ

where k* is the ffinatural specific absorption coefficient, pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R ¼ k ðk þ s Þ, s* the natural specific scattering coefficient, Ccat the photocatalyst loading and b the optical path. It is quite convenient to express b in cm and Ccat in g cm
3, so that k* and s* will be expressed in cm2 g
1. It is also more common to use the decadic absorption and scattering coefficients, eabs and escat, in place of the natural ones. It is k* = 2.303 eabs and s* = 2.303 escat. It is now possible to measure spectrophotometrically the extinction E of a semiconductor suspension for different values of the product Ccatb, expressed in g cm
2. The experimental data can then be fitted with Eq. (1) to evaluate k*, s* and, as a consequence, eabs, escat. Fig. 3 reports the results of the application of such a procedure to the cases of TiO2 Degussa P25 and TiO2 Wackherr at 360 nm. It also indicates the level of reproducibility of repeated extinction measures on the semiconductor suspensions.

Fig. 3. Extinction at 360 nm of TiO2 Degussa P25 (~) and TiO2 Wackherr (&) as a function of the product between the optical path length and the photocatalyst loading (bCcat). pH 3.7 by HClO4. Data fitting with Eq. (1) yielded the values of k*, s*, eabs and escat that are reported on the graph.

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

Making use of the Kubelka-Munk equation applied to Fig. 3, one obtains that the scattering coefficient escat of TiO2 Wackherr (2.4  104 cm2 g
1) is little more than one half the one of Degussa P25 (4.0  104 cm2 g
1). Conversely, the absorption coefficient eabs of TiO2 Wackherr (1.2  103 cm2 g
1) is only slightly lower than the one of Degussa P25 (1.5  103 cm2 g
1). The presence of similar absorption coefficients, combined with a higher scattering coefficient for TiO2 Degussa P25, is the expected result if the interpretation of phenol degradation rates as a function of TiO2 loading (Fig. 1), based on the photocatalyst optical properties, is correct. Quite interestingly, the absorption coefficient eabs obtained for TiO2 Degussa P25 by application of the Kubelka-Munk equation is comparable to the one measured with a completely different experimental set-up (integrating sphere spectrophotometer, [26]). The data of eabs, escat of both photocatalysts also indicate that at 360 nm scattering accounts for the vast majority of extinction, as already evaluated for Degussa P25 in Fig. 2. The fact that the extinction coefficient that can be obtained by direct spectrophotometric measurements, as shown in Fig. 2, is only a lower limit of the actual value can be further appreciated upon comparison between Figs. 2 and 3. The extinction coefficients at 360 nm can be obtained from the extinction data of Fig. 2 upon generalisation of the Lambert-Beer law, yielding E = (eabs + escat)bCcat. Following this procedure one obtains (eabs + escat)360 nm = 2.8  104 cm2 g
1 for TiO2 Degussa P25 and (eabs + escat)360 nm = 1.7  104 cm2 g
1 for TiO2 Wackherr, while data from Fig. 3 yield relevantly higher values: (eabs + escat)360 nm = 4.2  104 cm2 g
1 for TiO2 Degussa P25 and (eabs + escat)360 nm = 2.5  104 cm2 g
1 for TiO2 Wackherr. The actual values are likely to be even slightly higher than the ones obtained with the Kubelka-Munk equation due to the unaccounted effect of scattering-out [26], but their ratio would not change very much. The presence of an absorption coefficient, which is more than 10 times lower than the scattering one has an influence on the optical behaviour of the semiconductor suspension, which depends on the photocatalyst loading. Intuitively, scattering is expected to inhibit absorption at high loading. Absorption inhibition can be treated in an exact and simple way in the case of two absorbing species, 1 and 2, simultaneously present in solution and irradiated by monochromatic light. The treatment that will follow is not intended as a rigorous modelling of the behaviour of an irradiated suspension of titanium dioxide because the effects of scattering-in and scattering-out are not taken into account. The effect of scattering-out can be expected to be particularly relevant in the 4 cm diameter photoreactor we used. Indeed, much of the scattered-out radiation will be still available for photocatalytic reactions before going out of the solution, while in spectrophotometric measurements the narrow slit would hopefully exclude much of the scatteredout radiation. Furthermore, the results reported in Fig. 1 have

83

been obtained upon irradiation over a wide range of wavelengths. Without aims of exactitude, the following discussion is just intended to show that absorption inhibition can qualitatively reproduce the trends reported in Fig. 1 for TiO2 Wackherr and Degussa P25, thus showing that the different behaviour of the two photocatalysts can be ascribed to the differences in radiation absorption and scattering. By analogy with the case of absorption and scattering by a single compound, assume that the two absorbing species 1 and 2 are present in equal concentration in solution, thus c = c1 = c2. Absorption by the species 2 would be the analogous term for scattering. According to the LambertBeer law, the total absorbance of the solution is equal to the sum of the absorbances of the single species when present separately ðA 1 ; A 2 Þ, thus Atot ¼ A 1 þ A 2 . Radiation absorption intensity (expressed for instance in Ein s
1) does not share with absorbance such a property: a species does not absorb the same radiation intensity when alone in solution or abs the radiation intensity when present in a mixture. Be Itot absorbed by the mixture 1 + 2, which is the sum of the contribution of absorbed radiation by each species in the abs ¼ I abs þ I abs . Absorbed intensity by 1 and by 2 mixture: Itot 1 2 can easily be evaluated for each species alone in solution, obtaining I1abs and I2abs , but it is I1abs < I1abs and I2abs < I2abs (on the contrary, it is A1 ¼ A 1 and A2 ¼ A 2 ). This is due to the fact that radiation absorption by one species in the mixture decreases the radiation intensity available for the other one. It is possible to obtain the radiation absorption intensities in the mixture ðI1abs ; I2abs Þ when considering that the ratio of such intensities is equal to the ratio of the absorbances: I1abs =I2abs ¼ A1 =A2 [28,29]. The obvious difference is that there is no theoretical upper limit for the value of the total absorbance, Atot, while total abs cannot be higher than incident absorbed intensity Itot radiation intensity, I0. Given these premises, it is possible to calculate the radiation intensity absorbed by the species 1 in the presence of 2 (I1abs ), or better the absorptance a1 of the species 1, that is the fraction of incident radiation absorbed by 1 (a1 ¼ I1abs =I0 ) [28]. In Eq. (2), e1 and e2 are the decadic absorption coefficients of the species 1 and 2, b the optical path, and c = c1 = c2 the concentration of the two species. a1 ¼

e1 ð1
10
ðe1 þe2 Þbc Þ e1 þ e2

(2)

Fig. 4 reports the absorptance a1 of the species 1 as a function of the product b  c in two cases, which differ for the couple of values (e1, e2). A couple has been chosen in analogy with the measured values of the absorption and scattering coefficients of TiO2 Degussa P25 at 360 nm (case (D)), the other in analogy with the corresponding coefficients of TiO2 Wackherr at 360 nm (case (W)). The term e1 corresponds to the absorption coefficient, the term e2 to the scattering one. Comparison between Figs. 1 and 4 shows a qualitative agreement. In Fig. 4, case (W) has slightly lower absorptance than case (D) at low b  c, but relevantly higher absorptance at high b  c. Referring to Fig. 1, phenol

84

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

Fig. 4. Absorptance a1 of a species 1 in solution in the presence of a second absorbing species (calculated according to Eq. (2)). Case (D) corresponds to the absorption and scattering coefficients of TiO2 Degussa P25 at 360 nm, case (W) to the coefficients of TiO2 Wackherr. The trend is in qualitative agreement with the experimental data reported in Fig. 1.

degradation rate by TiO2 Wackherr is slightly lower than the one by Degussa P25 at low loading, and relevantly higher at high loading. The qualitative agreement between absorptance values and initial rate ones is motivated by the fact that photocatalytic degradation processes are initiated by radiation absorption by the semiconductor oxide and are therefore dependent on the absorbed radiation intensity. Comparison between Figs. 1 and 4 suggests that the different optical properties of TiO2 Wackherr and Degussa P25 can account for the different trend of photocatalytic activity versus photocatalyst loading. Size has an important role in determining the scattering behaviour of small particles [4,26,27]. It can therefore be hypothesised that the difference in the optical behaviour between TiO2 Wackherr and TiO2 Degussa P25 is due to the different size distribution of the particles. The data reported in Table 1 suggest that the Degussa P25 powder is made up of finer particles than TiO2 Wackherr. The interpretation of the photocatalytic activity data obtained in aqueous systems does however require the measure of the particle size in aqueous suspension. Particle size distributions for Degussa P25 and TiO2 Wackherr in water at pH 3.7, measured by laser light scattering, are reported in Fig. 5. The distribution of Degussa P25 particle radii at pH 3.7 has a maximum around 100 nm, indicating that some aggregation of the crystallites in water does take place (Degussa P25 crystallites are smaller than 100 nm, see Table 1). The particles of TiO2 Wackherr in aqueous suspension are relevantly larger than 100 nm, with a maximum in the distribution around 350–400 nm. Accordingly, the particles of TiO2 Wackherr in aqueous suspension are relevantly larger than the ones of Degussa P25. Furthermore, the extinction spectrum of TiO2 Wackherr (see Fig. 2) is analogous to the ones of TiO2 samples with similar size distribution [26]. From the data presented in this section it can be concluded that the radiation scattering by TiO2 Wackherr

Fig. 5. Particle size distribution of TiO2 Degussa P25 and TiO2 Wackherr at pH 3.7.

is much higher than the one of Degussa P25 in the visible region (see Fig. 2), where it has little effect on the photocatalytic activity of TiO2 since no absorption takes place in this spectral region. On the contrary, scattering by Degussa P25 is relevantly higher below 400 nm, where it interferes with radiation absorption. The higher interference between absorption and scattering for Degussa P25 when compared with TiO2 Wackherr accounts for the trends of phenol degradation rate as a function of the photocatalyst loading shown in Fig. 1. It is also interesting to observe that higher particle size in the case of TiO2 Wackherr, which is usually expected to be detrimental to reactivity, results in more favourable optical properties due to lower scattering in the UV region and, as a consequence, in better performance at high photocatalyst loading. 3.2. Phenol degradation with TiO2 Wackherr Phenol degradation in the presence of TiO2 Wackherr mainly occurs upon reaction with surface-adsorbed hydroxyl, OHads, as already inferred from the initial formation rates of the hydroxylated products catechol and hydroquinone [11]. It is possible to make a further check upon addition of 2-propanol and t-butanol to the system. These alcohols preferentially react with OHads rather than with h+, and selectively inhibit the OHads – initiated degradation processes [30–32]. Fig. 6 shows that 2-propanol and tbutanol relevantly inhibit phenol degradation in the presence of TiO2 Wackherr. The pseudo-first order degradation rate constant of 2.1  10
4 M phenol in the presence of 0.50 g l
1 TiO2 Wackherr is 1.6  10
3 s
1, and becomes 1.4  10
4 s
1 upon addition of 0.20 M 2-propanol, and 1.7  10
4 s
1 upon addition of 0.20 M t-butanol. The inhibition performed by an excess of both 2-propanol and tbutanol indicates that about 90% of phenol degradation in

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

Fig. 6. Time evolution of 2.1  10
4 M phenol in the presence of 0.50 g l
1 TiO2 Wackherr (&) and (when present) 0.20 M 2-propanol (~) and tbutanol (^). Suspension pH 3.7, adjusted by addition of HClO4. The time evolution of 2.1  10
4 M phenol in the presence of 0.50 g l
1 TiO2 Degussa P25 (&) is also reported for a comparison.

the presence of TiO2 Wackherr occurs via reaction with  OHads. Fig. 6 also reports as a comparison the corresponding time evolution curve of phenol in the presence of 0.50 g l
1 TiO2 Degussa P25, without alcohols. The curves corresponding to TiO2 Wackherr and to Degussa P25 are almost coincident, which is fully consistent with the initial rate data reported in Fig. 1. Actually, the initial degradation rate of 2.1  10
4 M phenol is almost equal in the presence of TiO2 Wackherr and of Degussa P25 if the photocatalyst loading is 0.50 g l
1. The fact that very similar initial rates are reflected into very similar time evolution curves implies that the kinetic law of phenol degradation is the same for both TiO2 Wackherr and Degussa P25. The initial degradation rate of phenol was also studied as a function of the initial phenol concentration in the range 1.0  10
4 to 2.1  10
3 M, for a fixed loading of TiO2 Wackherr (0.50 g l
1). The results are reported in Fig. 7. The

Fig. 7. Initial phenol degradation rate as a function of initial phenol concentration for TiO2 Wackherr. Absence of fluoride (&), and presence of 0.01 M NaF (^). TiO2 loading 0.50 g l
1, suspension pH 3.7 by HClO4.

85

behaviour in the absence of added fluoride will be discussed first. The initial phenol degradation rate versus initial phenol concentration shows a maximum for [phenol]  3  10
4 M. This behaviour is very similar to the one observed for TiO2 Degussa P25 under comparable conditions [9,33]. The data of the initial degradation rate of a substrate versus substrate concentration, under photocatalytic conditions, are often fitted with a Langmuir–Hinshelwood curve [34]. This curve is often suitable for a phenomenological fitting to the data, although its use to obtain physical parameters has been questioned [3]. However, it is very difficult to account for the phenol rate data reported in Fig. 7 in the context of the Langmuir–Hinshelwood kinetic model, in which case a plateau in the initial degradation rate would be expected at high substrate concentration, not a maximum as in Fig. 7. The maximum as a function of substrate concentration can be interpreted by means of a theoretical analysis of the kinetics in photocatalytic systems [35]. Given an organic substrate S (in our case S is phenol) in the presence of dissolved oxygen, the following reactions are considered (Fi = quantum yield, kn = rate constant): TiO2 þ hn ! e
þ hþ

½F3 

(3)

þ

e þ h ! heat

½k4 

(4)

hþ þ S ! Sþ

½k5 

(5)

½k6 

(6)

½k7 

(7)

½k8 

(8)

½k9 

(9)

þ e þ 2H ! H2 O2

½k10 

(10)

 þ O2 ! S
Oþ 2

½k11 

(11)







 e þ O2 ! O
2  þ O
2 þ h ! O2 þ


S

þ e !S

þ

þ h !P

 O
2 þ




S

S

þ

þ

P is an oxidation product. Note that the kinetic model does not differentiate between the oxidation reactions of S with h+ and with OHads (both included in reaction (5)), or between S+ and S–OH (both indicated as S+). However, this fact does not alter the conclusions that can be reached [35]. The initial degradation rate of the substrate can have a maximum for a certain value of the parameter a = k5k8{S}, where {S} is the concentration of S on the surface of the photocatalyst. The curve (d{S}/dt)t=0 versus a reaches a maximum instead of showing a plateau if the recombination reaction (8) between S+ and e
is kinetically important. The maximum will be reached at a lower {S} as k8 will be higher. Moreover, the maximum is possible only if reaction (11) between S+ and O2 is negligible when compared with reactions (8) and (9), so that oxygen acts as a mere electron scavenger (reactions (6) and (10), [35]). The presence of the maximum is tightly connected with the relevance of the recombination reaction (8) between S+ and e
, yielding back S from the partially oxidised intermediate S+. This reaction slows down the degradation kinetics of the substrate S, and its importance increases by

86

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

increasing the surface concentration {S+}, which is proportional to {S}. As there is also proportionality between {S} and the bulk concentration [S], the maximum is reflected in the plots of (d[S]/dt)t=0 versus [S], as in the case of Fig. 7 (phenol degradation rate in the presence of TiO2 Wackherr without fluoride). The presence of a maximum in Fig. 7 would imply that, for phenol, reaction (11) plays a minor role when compared with reactions (8) and (9), and that the recombination reaction (8) plays an important role for [phenol] > 3  10
4 M. Reaction (11), when occurring between the substrate radical cation, S+, and oxygen (or superoxide), has clearly been identified in the case of the photocatalytic degradation of quinoline [36] and 1,2-dimethoxybenzene [37]. Clear identification was possible because such a pathway gives peculiar transformation intermediates. In the case of phenol degradation on TiO2 Wackherr, both the initial formation rates of catechol and hydroquinone [11], and the effect of the addition of 2-propanol and t-butanol (Fig. 6) indicate that the main pathway is initiated by OHads. The limited role of the reaction between phenol radical cation and oxygen/superoxide can be motivated by the fact that the cation is a very strong acid (pKa =
1.95 [38]). In aqueous solution it can be expected to undergo very fast deprotonation to the phenoxy radical, which is rather unreactive towards oxygen and, in the presence of superoxide, is reduced back to phenol instead of giving further transformation products [39]. The addition of fluoride increases the degradation rate of phenol, and the corresponding curve (d[phenol]/dt)t=0 versus [phenol] shows no more a maximum (see Fig. 7). The addition of fluoride to TiO2 in the pH range 2–5 induces an exchange with the hydroxyl groups on the surface of the photocatalyst [40]: BBTi
OH þ F
? BBTi
F þ OH


(12)

The exchange is almost complete at pH 3.6–3.7. The BBTi–F groups are not able to generate the surface species OHads, but fluorinated titania can induce the production of homogeneous OH in the solution bulk [9,10]. þ  hþ ðTiO2 Þ þ H2 O ! HðaqÞ þ OHðaqÞ

The minor role of recombination reaction (8) accounts for the increased phenol degradation rate in the presence of fluoride. For the same reason, the maximum in the curve (d[phenol]/dt)t=0 versus [phenol] disappears upon addition of fluoride at pH 3.7. A practical consequence is that the enhancement effect of fluoride on the degradation rate of phenol is more marked for higher values of phenol concentration. 3.3. Degradation of benzoic acid with TiO2 Wackherr Benzoic acid, too, was studied as a substrate in the presence of TiO2 Wackherr. The use of benzoic acid is interesting since this compound, in a similar way as salicylic acid, adsorbs on the surface of the photocatalyst and its photocatalytic degradation in the absence of fluoride mainly occurs via electron-transfer reactions involving surfaceadsorbed species [32,42]. Fig. 8 shows the initial degradation rate of benzoic acid (HBz) as a function of its initial concentration. In the absence of fluoride the curve shows a broad maximum for [HBz]  2–3  10
4 M, comparable to the one of phenol. The addition of fluoride to TiO2 Wackherr increases the initial degradation rate of benzoic acid, while the curve shows no maximum in the concentration range 0– 1.6  10
3 M (see Fig. 8). Considering that adsorption of organic molecules onto fluorinated titania is strongly suppressed [9,43], and that homogeneous OH generation takes place in the system (reaction (13) [9,10]), the degradation of benzoic acid is most likely shifted from the surface of the photocatalyst to the solution bulk. As in the case of phenol, one can infer that the recombination reaction (8) plays a minor role in the presence of fluoride, which causes the maximum to disappear. A comparison between Figs. 7 and 8 shows that the effect of fluoride is more marked for benzoic acid than for phenol. In the case of phenol the addition of fluoride shifts the reactive species from surface-adsorbed to bulk hydroxyl,

(13)

As a consequence, the degradation processes in the presence of fluoride mainly occur in the solution bulk, where the reactions involving oxygen (e.g. reaction (11)) can be expected to prevail over the recombination reaction (8). Indeed, reaction between phenol and homogeneous OH mainly yields the dihydroxycyclohexadienyl radicals upon addition to the ring. In the solution bulk these intermediates mainly react with oxygen by hydrogen abstraction to yield the corresponding dihydroxybenzenes (mainly catechol and hydroquinone) [10,41]. In the following reaction scheme, H–Ph–OH is phenol, HO(H)–Ph–OH dihydroxycyclohexadienyl, HO–Ph–OH a dihydroxybenzene. H
Ph
OH þ  OH ! HOðHÞ
Ph
OH

(14)

HOðHÞ
Ph
OH þ O2 ! HO
Ph
OH þ HO2 

(15)

Fig. 8. Initial degradation rate of benzoic acid as a function of its initial concentration in the presence of TiO2 Wackherr. Absence of fluoride (&), and presence of 0.01 M NaF (^). TiO2 loading 0.50 g l
1, suspension pH 3.7 by HClO4.

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

and the recombination reaction (8) plays a minor role in the solution bulk. In the absence of fluoride, the photocatalytic degradation of benzoic acid mainly occurs via electrontransfer processes involving surface species [32,42]. The interaction with the surface can be expected to make the recombination reaction (8) quite important in the case of benzoic acid. The addition of fluoride most likely induces an abrupt change in the degradation pathway, but reaction with bulk hydroxyl is quite effective for both benzoic acid (bimolecular rate constant 4.3  109 M
1 s
1) and the benzoate anion (5.9  109 M
1 s
1) [44]. The fact that the degradation process mainly occurs away from the photocatalyst surface, and from reaction (8), results in strongly enhanced degradation rate in the presence of fluoride.

87

Similar results were obtained in the case of benzoic acid, as the addition of fluoride increases the substrate degradation rate and makes the maximum to disappear. The increase in substrate degradation rate is larger for high concentration of the substrate. This is an important finding since benzoic acid and phenol undergo very different photocatalytic degradation pathways in the absence of fluoride [42]. Thus, fluoride addition to TiO2 Wackherr may enhance the photocatalytic degradation of many organic compounds, transforming via different pathways, provided that such compounds are able to react with OH in the solution bulk at a relevant rate.

Acknowledgements 4. Conclusions The TiO2 Wackherr ‘‘oxyde de titane standard’’ has some interesting features for photocatalytic applications and its use can be desirable when high photocatalyst loading is required to obtain faster degradation of the substrate. The main feature of this photocatalyst when compared with Degussa P25 is the lower scattering of radiation in the UV region, which is most likely the consequence of higher particle radii. The larger particle size of TiO2 Wackherr when compared with Degussa P25 results in lower specific surface area (see Table 1). Particles with low surface area are usually expected to show poor catalytic reactivity, but they can still show better performance in the field of photocatalysis, where the optical properties are very important and low scattering of radiation plays a key role. As a consequence of more favourable optical properties, in the case of TiO2 Wackherr the degradation rate of phenol continues to increase with increasing photocatalyst loading (at least up to 2.00 g l
1), while in the case of TiO2 Degussa P25 the rate does not increase above 0.50 g l
1. The degradation rate of phenol in the presence of TiO2 Wackherr has a maximum as a function of phenol concentration, because the importance of the recombination processes (reaction (8)) increases at high substrate concentration [35]. The addition of fluoride at pH 3.7 increases the degradation rate of phenol, and the corresponding curve (d[phenol]/dt)t=0 versus [phenol] shows no maximum in the considered concentration range ([phenol]  2.0  10
3 M). A reasonable explanation for this phenomenon is the minor role that the surface recombination reactions play in the TiO2–fluoride system, in which the reactive species are homogeneous OH radicals in the solution bulk. The addition of fluoride is thus a way to increase the efficiency of the photocatalyst. The effect of added fluoride on the degradation rate of phenol is very similar for both TiO2 Degussa P25 [9,10] and TiO2 Wackherr (this work), indicating that very similar processes take place in both cases, and that these processes can be described in the frame of the same theoretical model.

Thanks are due to Dr. Daniela Mondelli for assistance with the light-scattering measures and to Dr. Serge Grizzo for providing information on TiO2 Wackherr. Financial support from MIUR (PRIN 2003, contract no. 2003035534_001), Interuniversity Consortium ‘‘Chemistry for the Environment’’ (INCA) and Universita` di Torino – Ricerca Locale is gratefully acknowledged.

References [1] O. Legrini, E. Oliveros, A.M. Braun, Chem. Rev. 93 (1993) 671. [2] E. Pelizzetti, N. Serpone (Eds.), Homogeneous and Heterogeneous Photocatalysis, Reidel, Dordrecht, 1986. [3] D. Bahnemann, J. Cunningham, M.A. Fox, E. Pelizzetti, P. Pichat, N. Serpone, in: G.R. Helz, R.G. Zepp, D.G. Crosby (Eds.), Aquatic and Surface Photochemistry, Lewis, London, 1994, and references therein. [4] A.E. Cassano, O.M. Alfano, Catal. Today 58 (2000) 167. [5] J.M. Herrmann, J. Disdier, P. Pichat, S. Malato, J. Blanco, Appl. Catal. B: Environ. 17 (1998) 15. [6] J.M. Herrmann, C. Guillard, J. Disdier, C. Lehaut, S. Malato, J. Blanco, Appl. Catal. B: Environ. 35 (2002) 281. [7] A. Di Paola, E. Garcı´a-Lo´ pez, G. Marcı`, C. Marti, L. Palmisano, V. Rives, A.M. Venezia, Appl. Catal. B: Environ. 48 (2004) 223. [8] G. Mele, R. Del Sole, G. Vasapollo, E. Garcı´a-Lo´ pez, L. Palmisano, M. Schiavello, J. Catal. 217 (2003) 334. [9] C. Minero, G. Mariella, V. Maurino, E. Pelizzetti, Langmuir 16 (2000) 2632. [10] C. Minero, G. Mariella, V. Maurino, D. Vione, E. Pelizzetti, Langmuir 16 (2000) 8964. [11] T. Picatonotto, D. Vione, M.E. Carlotti, M. Gallarate, J. Disp. Sci. Technol. 22 (2001) 381. [12] V. Rossatto, T. Picatonotto, D. Vione, M.E. Carlotti, J. Disp. Sci. Technol. 24 (2003) 259. [13] R. Dunford, A. Salinaro, L.Z. Cai, N. Serpone, S. Horikoshi, H. Hidaka, J. Knowland, FEBS Lett. 418 (1997) 87. [14] M.E. Carlotti, S. Sapino, D. Vione, E. Pelizzetti, M. Trotta, J. Disp. Sci. Technol. 25 (2004) 193. [15] T. Hancock-Chen, J.C. Scaiano, J. Photochem. Photobiol. B: Biol. 57 (2000) 193. [16] D. Vione, T. Picatonotto, M.E. Carlotti, J. Cosmet. Sci. 54 (2003) 513. [17] C.A. Emilio, J.J. Testa, D. Hufschmidt, G. Colon, J.A. Navio, D.W. Bahnemann, M.I. Litter, J. Ind. Eng. Chem. 10 (2004) 129. [18] G.K-C. Low, S.R. McEvoy, R.W. Matthews, Environ. Sci. Technol. 25 (1991) 460.

88

D. Vione et al. / Applied Catalysis B: Environmental 58 (2005) 79–88

[19] J.G. Calvert, J.N. Pitts, Photochemistry, Wiley, New York, 1966, pp. 780–786. [20] L.E. McNeil, R.H. French, Acta Mater. 48 (2000) 4571. [21] J.M. Herrmann, Catal. Today 53 (1999) 115. [22] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69. [23] K.-M. Schindler, M. Kunst, J. Phys. Chem. 94 (1990) 8222. [24] C. Colbeau-Justin, M. Kunst, D. Huguenin, J. Mater. Sci. 38 (2003) 2429. [25] Y.V. Kolen’ko, B.R. Churagulov, M. Kunst, L. Mazerolles, C. Colbeau-Justin, Appl. Catal. B: Environ. 54 (2004) 51. [26] M.I. Cabrera, O.M. Alfano, A.E. Cassano, J. Phys. Chem. 100 (1996) 20043. [27] M. Wolff, A. Dollfus, Appl. Opt. 29 (1990) 1496. [28] V. Balzani, I.P. Beletskaya, J.R. Bolton, M. Chanon, A.K. Chibisov, L. Eberson, M.A. Fox, A.B.P. Lever, N.A. Lewis, R.A. Marcus, N. Mataga, K.A. Zchariasse, Pure Appl. Chem. 68 (1996) 2223. [29] D. Vione, V. Maurino, C. Minero, E. Pelizzetti, Environ. Sci. Technol. 36 (2002) 669. [30] C. Richard, P. Boule, New J. Chem. 18 (1994) 547.

[31] Y. Sun, J.J. Pignatello, Environ. Sci. Technol. 29 (1995) 2065. [32] S. Tunesi, M.A. Anderson, J. Phys. Chem. 95 (1991) 3399. [33] J. Cunningham, G. Al-Sayyed, P. Sedlak, J. Caffrey, Catal. Today 53 (1999) 145. [34] V.A. Sakkas, T.A. Albanis, Appl. Catal. B: Environ. 46 (2003) 175. [35] C. Minero, Catal. Today 54 (1999) 205. [36] L. Cermenati, P. Pichat, C. Guillard, A. Albini, J. Phys. Chem. B 101 (1997) 2650. [37] L. Amalric, C. Guillard, P. Pichat, Res. Chem. Intermed. 20 (1994) 579. [38] W.T. Dixon, D. Murphy, J. Chem. Soc., Faraday Trans. II (1976) 1221. [39] E.P.L. Hunter, M.F. Desrosiers, M.G. Simic, Free Rad. Biol. Med. 6 (1989) 581. [40] H.P. Bohem, J. Chem. Soc., Faraday Discuss 66 (1972) 264. [41] P. Machado, Boule, J. Photochem. Photobiol. A: Chem. 86 (1995) 73. [42] T. Picatonotto, D. Vione, M.E. Carlotti, J. Disp. Sci. Technol. 23 (2002) 845. [43] D. Vasudevan, A.T. Stone, Environ. Sci. Technol. 30 (1996) 1604. [44] G.V. Buxton, C.L. Greenstock, W.P. Helman, A.B. Ross, J. Phys. Chem. Ref. Data 17 (1988) 513.