- Email: [email protected]
Surface photovoltage measurements: A quick assessment of the photocatalytic activity?
Catalysis Today 209 (2013) 215–220
Contents lists available at SciVerse ScienceDirect
Catalysis Today journal homepage: www.elsevier.com/locate/cattod
Surface photovoltage measurements: A quick assessment of the photocatalytic activity? S.W. Verbruggen a,b,∗ , J.J.J. Dirckx c , J.A. Martens b , S. Lenaerts a a b c
Sustainable Energy and Air Purification, Department of Bioscience Engineering, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium Center for Surface Science and Catalysis, Department of Microbial and Molecular Systems (M2 S), Catholic University of Leuven, Kasteelpark Arenberg 23, 3001 Heverlee, Belgium The Laboratory of BioMedical Physics (BIMEF), Department of Physics, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium
a r t i c l e
i n f o
Article history: Received 25 June 2012 Received in revised form 3 September 2012 Accepted 20 November 2012 Available online 16 January 2013 Keywords: Surface photovoltage Photocatalysis Titanium dioxide Silver Methylene blue Acetaldehyde
a b s t r a c t Surface photovoltage (SPV) measurements can contribute to a better understanding of electronic properties of photocatalysts under illumination. Direct linking of SPV data to the actual photocatalytic activity remains troublesome. This work aims to discuss SPV measurements from a photocatalytic point of view. By means of several application-based scenarios we illustrate that the trend between SPV and photocatalysis strongly depends on parameters such as the crystal structure, surface modifications, morphology and humidity. This makes the interpretation far from straightforward. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Surface photovoltage (SPV) measurement is a well-established contactless technique for the characterization of semiconductors, including photocatalysts, based on monitoring the changes in surface voltage before and during illumination. Several excellent review articles have been published on the physical and electrochemical fundamentals [1,2]. The electronic band structure of n-type semiconductors such as TiO2 is characterized by upward band bending at the semiconductor–conductor interface, leading to a Shottky-type barrier. Surface-localized electronic states induce charge transfer between the bulk and the surface of the semiconductor in order to establish thermodynamic equilibrium. This charge transfer results in a non-neutral surface space charge region (SCR) near the surface [3]. This in turn results in a builtin electric field, Vs , commonly referred to as the surface potential barrier. When the photocatalyst is irradiated by light with a photon energy content greater than the bandgap, free charge carriers are generated by band-to-band transitions induced by the incident photons. Trap-to-band transitions are also possible, but since these are related to photons with a lower energy than the bandgap,
∗ Corresponding author at: Sustainable Energy and Air Purification, Department of Bioscience Engineering, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium. Tel.: +32 3 265 35 17; fax: +32 3 265 32 25. E-mail address: [email protected] (S.W. Verbruggen). 0920-5861/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cattod.2012.11.010
their effect becomes negligible and will therefore not be considered here [4]. The generated charge carriers will redistribute within the surface and/or the bulk as a result of the built-in electric field. Hence, the surface potential barrier will be altered after illumination. Exactly this difference between the surface potential barrier in dark (Vs ) and the one under illumination (Vs ), is defined as the SPV signal (Fig. 1) and is used to characterize the photocatalyst. SPV spectroscopy could be further expanded to electric field induced SPV spectroscopy (EFISPS). In EFISPS an external electric field directly affects the response intensity of the SPV signal and the photovoltaic characteristics [5]. Using SPV measurements, information can be obtained concerning the (photo)electronic properties of semiconductors, such as the band bending, surface and bulk carrier recombination, type of defect states and magnitude of the bandgap [6–8]. One of the most interesting features of the SPV technique is that it is directly applicable to powders and therefore appears to be a valuable tool in mechanistic photocatalytic studies. It would indeed be wonderful to have a very fast, easy-to-use technique available that does not require tedious sample preparations and can be employed for assessing the photocatalytic potential of a great number of photocatalyst powders, without time-consuming catalytic experiments. Comparable information could also be acquired by means of electrochemical photocurrent measurements, in which the photocatalyst has to be attached to an electrode and is then immersed in an electrolyte solution [9,10]. However, by circumventing the necessity to fixate the catalyst to an electrode and excluding the
216
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
Vs eEC
V’s
SPV
The change in crystallite size due to the heat treatment was calculated using the Scherrer equation: D = 0.9
EF hν
EV
h+
Fig. 1. Scheme on the origin of the SPV signal. The change in band bending is represented by dashed lines. The redistribution of generated charges is indicated with the arrows.
presence (and the influence) of the electrolyte, SPV measurements present a very attractive alternative. The objective of this work is to discuss the concept of SPV measurements from a photocatalytic point of view. The thus far reported trends in literature indicate that directly relating SPV results to actual photocatalytic activity is not unambiguous [3,6,11–13]. Several situations where care should be taken when interpreting SPV measurements with regard to photocatalytic activity are discussed. It has to be stressed that in this work it was not the intention to use SPV for a detailed study of the (photo)electronic properties of photocatalysts, but rather to apply SPV as a simple, quick screening method for assessing photocatalytic performance. We show that photocatalytic activity is not merely predictable based on (photo)electronic properties as measured by SPV, but is also strongly influenced by the catalyst crystal phase composition, type of catalyst modification, morphology and general reaction conditions (type of fluid phase and the catalyst humidity).
ˇ × cos
with D, crystallite size [nm]; , X-ray wavelength [nm] (CuK␣ Xrays: 0.1542 nm); ˇ, full width at half maximum (FWHM) [radians]; , Bragg angle [radians]. 2.1.2. TiO2 modified with silver nanoparticles Silver was deposited on the TiO2 P90 (Evonik) surface by the photodeposition method using an adapted protocol based on Iliev et al. [15,16]. For 0.5 g of TiO2 P90, AgNO3 (Sigma–Aldrich, >99.0% purity) was used as the silver precursor and dissolved in 50 mL of bidistilled water, so that samples with 0.1, 0.3, 0.6 and 1 at% of silver (TiO2 basis) were obtained. Methanol (Merck, >99.5% purity) was used a sacrificial hole scavenger and added to the solution in a molar ratio methanol:AgNO3 of 500:1. The P90 powder was added to the solution and brought to pH 2 with nitric acid (Merck, 65%). The mixture was stirred for 30 min at 750 rpm in the dark and consecutively illuminated with UV light (Philips Cleo, 25 W, 205 W cm−2 at 365 nm) for 30 min while stirring. The resulting colored suspension was centrifuged and washed 2 times with bidistilled water. The cake was further dried overnight at 110 ◦ C and grinded in an agate mortar. 2.1.3. TiO2 with different intrinsic properties or water content Three commercial photocatalysts were selected for this study. P90 (Evonik), P25 (Evonik) and PC500 (Cristal global), were used to determine the SPV response and the photocatalytic activity in the aqueous as well as in the gas phase. In order to study the qualitative effect of the humidity of the catalyst powder, P90 was dried overnight at 300 ◦ C. While handling this dried P90, it was continuously kept at 110 ◦ C to avoid unwanted rehydration of the catalyst. Controlled rehydration of the catalyst was achieved by the deposition of 2.5 or 5.0 L H2 O onto 4 mg dried P90 powder on the ITO glass slide of the SPV setup.
2. Materials and methods 2.1. Sample preparation For all tests in this work, TiO2 P90 (Evonik) was selected as the reference material. It consists predominantly of crystalline anatase particles and a small fraction of rutile (∼10%). Therefore, it has proven to be an excellent starting material for further modifications such as the introduction of higher rutile content or the deposition of silver particles on the surface. P90 also offered a good basis for comparative experiments with other (commercial) photocatalysts with different material characteristics. The other commercial photocatalysts used in this work were P25 (Evonik) and PC500 (Cristal Global). Both were used as received. 2.1.1. TiO2 with variable rutile content In order to obtain TiO2 samples with different anatase/rutile ratios, commercial P90 (Evonik) was used as such or calcined at 450 ◦ C, 525 ◦ C, 600 ◦ C, 675 ◦ C, 750 ◦ C or 900 ◦ C for 3 h. The precise rutile content was calculated based on the XRD crystal spectra of the resulting powders. The weight fraction of rutile was calculated using the formula originally introduced by Spurr and Myers [14]: Wrutile % =
1 1 + 0.8 × (Ianatase /Irutile )
with Ianatase and Irutile the peak intensities of the strongest anatase and strongest rutile reflection peak respectively.
2.2. SPV measurements This work aims to study the possible use of SPV for quickly assessing the photocatalytic performance, rather than carrying out a detailed study of the (photo)electronic properties. Therefore, no classical SPV ‘sweep’ spectra were recorded, but a single SPV signal was measured after illumination with a low-intensity UVA lamp (Philips Cleo, 25 W, 205 W cm−2 at 365 nm). This was the same light source as used in all photocatalytic tests, and therefore best suited for a realistic study of the relation between SPV and photocatalysis. The set-up itself was a custom-made apparatus, in which the catalyst powder was sandwiched between two ITO electrodes (Sigma–Aldrich, d = 1.2 mm, resistivity: 8–12 cm−2 ), connected to a photo-amplifier (1 × 106 voltage amplification). No external bias was applied. A schematic representation of the SPV set-up is depicted in Fig. 2. In order to perform highly reproducible measurements, only a controlled square section of 5 mm × 5 mm was illuminated. A constant amount of catalyst was spread out on this marked area and sandwiched. For all tests on TiO2 based samples, the ideal amount of catalyst that covered this area and resulted in a good contacting layer between both ITOs was determined to be 4.0 ± 0.1 mg. Since the Ag-P90 samples were considerably coarser due to the photodeposition treatment, 11.0 ± 0.1 mg was used for these specific tests. For the exact determination of the SPV value, the difference was taken between the steady voltage readout after 1.5 min of UVA illumination and the steady voltage readout in dark.
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
217
Fig. 2. Schematic representation of the SPV set-up.
For all samples, three independent measurements were conducted and averaged. 2.3. Photocatalytic activity tests 2.3.1. Aqueous phase photocatalytic test For the assessment of the photocatalytic activity of all samples, methylene blue (MB) discoloration batch experiments in aqueous phase were performed. A typical experiment involved suspending 20 mg of catalyst in 20 mL of MB solution (9.4 mol L−1 ) in a UV-transparent screw cap, stirred at 500 rpm. The vessel was placed 5 cm from the UVA lamp (same as in SPV Measurements), which was placed halfway the liquid level. The MB discoloration is monitored by means of UV/VIS absorption spectroscopy (Shimadzu UV-2501PC double beam spectrophotometer), using the peak absorption intensity of the MB peak at 663.5 nm. Before turning on the lamp, the samples were allowed to establish adsorption–desorption equilibrium by stirring in dark for 30 min. Sampling again after 40 min in dark indicated that the absorption level did not change anymore and the equilibrium was reached. Samples were taken after 1, 2, 4, 7, 13, 25 and 45 min of illumination. The MB concentration was determined after centrifugation of the sample. After the spectrum collection, the MB solution and residual catalyst cake were re-added to the reaction vessel. At every sampling interval the residual MB concentration (C/C0 ) was plotted as a function of illumination time. The apparent rate constant k is taken as the characteristic parameter for the activity of each sample. It is calculated as the slope of the first (linear) part of the C/C0 vs. time (in min) plot. A blank test was performed in which 20 mL of MB solution was illuminated in the absence of any catalyst. A slight decrease in the MB concentration was detected, resulting in an apparent rate constant k = 0.002 min−1 , which is negligible compared to the rate constants obtained over all catalyst samples. It is important to note that we did not monitor the formation of MB degradation products. Therefore we cannot distinguish between dye oxidation and the possible photoreduction to the colorless leuco methylene blue (LMB) form. The latter is preferentially formed under conditions of low dissolved oxygen levels and low pH values [17]. In the present study, however, the MB solution was well aerated prior to the measurements and reactions were carried out at neutral pH. Hence, formation of LMB was not to be expected. 2.3.2. Gas phase photocatalytic test The photocatalytic gas phase tests involved the degradation of acetaldehyde in air, using a single-pass flow through photoreactor, as described in previous work [18]. In short, 50 mg of catalyst was deposited onto 108 g 2 mm glass beads by means of suspension coating and loaded into the photoreactor. Acetaldehyde (1% in N2 , Air Liquide) and compressed air (Air Liquide Alphagaz) were mixed using mass flow controllers (Brooks), establishing 170 ppmv inlet concentration at a total flow rate of 2000 cm3 min−1 . The concentrations of (un)degraded acetaldehyde and formed CO2 were monitored using online FTIR spectroscopy of the outlet gas stream
(NicoletTM 380 Thermo Fisher Scientific, ZnSe windows and 2 m heated gas cell). The steady-state conversion of acetaldehyde is determined as the part of the acetaldehyde flow that is continuously degraded over time, divided by the total acetaldehyde inlet flow (after establishing adsorption/desorption equilibrium). By simultaneous comparison with the amount of CO2 formed over time, a carbon balance very close to 100% indicates full mineralization of the pollutant [10]. 2.4. Characterization Determination of the Brunauer–Emmett–Teller (BET) surface area was performed with a Micrometrics Tristar Surface Area and Porosity Analyzer. The samples were degassed overnight at 250 ◦ C. X-ray powder diffraction (XRD) was carried out using a STOE StadiP apparatus with CuK␣ radiation and an image plate detector. UV–VIS diffuse reflectance spectra were collected in the range of 300–700 nm using a Shimadzu UV-2501PC double beam spectrophotometer equipped with a 60 mm BaSO4 coated integrating sphere and a Photomultiplier R-446U detector. 3. Results and discussion 3.1. Variation in rutile content It has been stated many times that a certain fraction of rutile in an excess of anatase TiO2 improves charge separation by the transfer of photogenerated electrons from anatase to rutile [19]. This physically separates them from the photoholes. This is beneficial for the photocatalytic activity, as long as the rutile fraction does not become too high [20]. The material characteristics of the P90 samples calcined at different temperatures are listed in Table 1. It is clear that with increasing temperature the particles start to sinter, which is evidenced by the decrease in specific surface area and the increase in crystallite size. Furthermore a phase transformation occurs from predominantly anatase to exclusively rutile. Fig. 3a shows the MB discoloration of the P90 samples over time and Fig. 3b the SPV signal of these samples, together with the photocatalytic activity determined by the rate of MB discoloration. It is immediately striking that the SPV signals and the apparent rate constants are very closely related. The SPV signal is strongest for the mixed Table 1 Photocatalyst characteristics after calcination of TiO2 P90 at different temperatures. Calcination T
Surface area [m2 g−1 ]
Rutile fraction [%]
Anatase crystallite size [nm]
Rutile crystallite size [nm]
RT 450 ◦ C 525 ◦ C 600 ◦ C 675 ◦ C 750 ◦ C 900 ◦ C
126 123 114 86 53 11 2
10 10 11 26 58 100 100
10.9 11.2 11.8 13.6 17.5 NA NA
13.4 15.3 17.4 21.1 27.8 23.2 35.9
218
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
Fig. 3. P90 samples calcined at different temperatures: (a) MB discoloration (C/C0 ) as a function of illumination time and (b) SPV signal (䊉, left axis) and apparent rate constant for MB discoloration (, right axis) under UVA illumination as a function of the calcination temperature.
phase anatase–rutile samples obtained at calcination temperatures between 600 and 675 ◦ C. Under these conditions, charge separation seems very efficient, resulting in high photocatalytic activity. At higher calcinations temperatures, the rutile content becomes excessive, leading to lower photocatalytic reaction rates. It is noteworthy to mention that in this experiment the role of the specific surface area seems of no vital importance, which confirms the fact that SPV mainly probes electronic properties rather than morphological properties. Furthermore, we found evidence that mainly the electronic properties dominate the photocatalytic activity of photocatalysts with nanosized pores in aqueous phase applications [10]. Our observations suggest that for purely TiO2 based photocatalysts, a stronger SPV signal corresponds to a higher photocatalytic activity in aqueous phase. These findings support earlier contributions in literature [11,13]. 3.2. Modification with noble metals The most distinct case in which SPV is used in photocatalytic studies is the one in which the material is modified by depositing noble metals on the surface. In this context Xie et al. report that after Pd and Ru deposition on the surface, a stronger SPV response corresponds to a higher photocatalytic activity [12]. This is in contrast to Jing et al. who state that after Pd and Ag deposition on the surface of ZnO, a weaker SPV signal corresponds to a higher photocatalytic activity [3]. Generally speaking, a more intense SPV signal should correspond to a better charge separation efficiency and thus a higher photocatalytic activity is expected. In the case of
noble metal depositions, however, charge separation is based on the metals acting as electron traps under UV illumination. Hence, when the noble metals are effectively trapping the photogenerated electrons, less charge is available for charge transport between the ITO electrodes of the SPV apparatus. This rationalizes why a weaker SPV signal is correlated to higher photocatalytic activity for this type of modified materials. This theory is also validated by the results of Xin et al. who used Fe3+ dopants as electron traps [6]. They also come to the conclusion that when the photogenerated electrons are effectively trapped in the Fe3+ states, the SPV signal weakens, whereas the photocatalytic activity improves. In order to evaluate these theories, a set of Ag–TiO2 samples was prepared. Silver was deposited onto TiO2 P90 using the photodeposition method, resulting in samples with 0.1, 0.3, 0.6 and 1.0 at% of silver (TiO2 basis). No significant change was measured in surface area, porosity, crystal phase composition or crystallite size after deposition. There is a clear color change from almost white (0.1 at% Ag) to gray-red (1 at% Ag), indicating that increasing amounts of silver are effectively deposited on the surface. This is also plain from the diffuse reflectance spectra (Fig. 4a), in which the surface plasmon peak of the silver nanoparticles becomes more pronounced with increasing silver content. Fig. 4c shows the SPV signal and the apparent photocatalytic reaction rate for MB discoloration (Fig. 4b) measured on these samples. In contrast to the findings in Section 3.1 and Fig. 3b, the SPV signal and photocatalytic activity show opposite trends in this case: the lower the SPV signal, the higher the photocatalytic activity. This supports the findings of Xin et al. [6] and Jing et al. [3]. The remark has to be made that the observed trend is only true in the studied range of silver content. Unfortunately it is impossible to observe experimentally what will happen at even larger silver loadings using the current set-up. As a consequence of the better conductive properties of the metal-modified catalysts, silver loadings higher than 1 at% result in a SPV signal that exceeds the operating window of the apparatus. It is to be expected that the photocatalytic activity will reach an optimum, after which it will decline as a result of the increasing amount of recombination centers and TiO2 coverage with increasing silver loading [21]. The experimental boundaries of the set-up, however, do not permit us to evaluate whether the SPV signal will show a similar trend. Therefore, based on these experiments it can only be concluded that in the range between 0 and 1 at% of silver, a higher SPV signal corresponds to a lower photocatalytic activity. 3.3. Catalyst properties vs. reaction conditions Another important parameter that is often overlooked when dealing with SPV measurements and their link with the photocatalytic activity, is the catalyst morphology. This is illustrated by means of three widely used commercial photocatalysts: P90 (Evonik), P25 (Evonik) and PC500 (Cristal Global). P25 is known for its high quantum efficiency, because of its fully crystalline nature and well formed crystallites of ca. 30 nm. PC500, on the other hand, is characterized by a much lower quantum efficiency, due to a significant fraction of amorphous material (∼20–25%). As high quantum efficiency implies the efficient generation of charge carriers under illumination, it can be stated in other words that the (photo)electronic properties as measured by SPV will be a lot better for P25 compared to PC500. P90 lies somewhere in-between. However, quantum efficiency is obviously not the sole important parameter in photocatalysis! For instance, PC500’s surface area is almost three times higher than that of P90 and about six times higher than that of P25. It also contains mesopores of ca. 4 nm that can give access to (small) gaseous pollutants, but can present a significant diffusion resistance for aqueous pollutants. Hence, a large surface area that is predominantly composed of small nanosized pores is only valuable in reaction systems where diffusion
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
219
Fig. 5. SPV response of (from left to right): P90 (as received) (gray bar), P90 dried overnight at 300 ◦ C (white bar), dried P90 rehydrated with 2.5 L H2 O (first striped bar) and dried P90 rehydrated with 5 L H2 O (second striped bar).
Fig. 4. Silver on P90 samples: (a) UV/VIS diffuse reflectance spectra of Ag–TiO2 (P90) (recalculated into absorbance), (b) MB discoloration (C/C0 ) as a function of illumination time and (c) SPV signal (䊉, left axis) and apparent rate constant for MB discoloration (, right axis) under UVA illumination as a function of the silver content (at%).
limitations are negligible. For this reason PC500 performs much better when it is used in the gas phase compared to the aqueous phase (Table 2). Table 2 compares the recorded SPV signals and the corresponding photocatalytic activities in gas and aqueous phase obtained for identical amounts of P25, P90 and PC500. In the aqueous phase, the higher the SPV response (thus the better the photo-electrical properties), the higher the reaction rate. This is in accordance with the observations from Section 3.1. However, when considering the photocatalytic activities in the gas phase, the reactivity order is completely reversed. Although no proper SPV signal could be measured for PC500 (presumably due to the lack of a well defined crystal structure), the acetaldehyde conversion in the gas phase is significantly better than the one obtained over P90, which is in turn better than P25. As discussed above, this is mainly attributed to the high available surface area, which appears to be of vital importance in gas phase applications [10]. Again, all of this indicates that care should be taken when interpreting SPV data that only take into account (photo)electronic properties, but not structural properties, since the latter can have a massive impact on the photocatalytic activity as well. A final remark that has to be made concerns the humidity of the catalyst. Water provides a conductive layer between the catalyst powder particles and the ITO electrodes. This way electron transfer to the electrodes can take place and a SPV signal can be recorded more easily. From Fig. 5 it is clear that the SPV response of P90 (as received) significantly decreases when the powder is completely dried. Rehydrating the powder results in restoring and even improving the SPV signal. Despite the fact that most photocatalytic reactions are carried out in the aqueous phase, or in the gas phase in which the air humidity level plays an important role on catalyst activity [22,23], it is rather surprising that the humidity of the catalyst powder has (almost) never been the subject of investigation
Table 2 SPV signal and photocatalytic activity of TiO2 P25, P90 and PC500 in aqueous and gas phase. Catalyst
Surface area [m2 g−1 ]
SPV [V]
Apparent rate constant (10−3 min−1 ) aqueous phase
Steady state conversion gas phase
P25 P90 PC500
55 126 350
1.7 ± 0.4 0.4 ± 0.1 0.01 ± 0.01
219 ± 10 155 ± 5 84 ± 4
(76 ± 2)% (90 ± 2)% (99 ± 2)%
220
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
in SPV measurements. This again evidences that relating the SPV signal to the photocatalytic performance can be cumbersome.
in programming the SPV software. B. Lenaerts, M. Keulemans and T. Lathouwers are thanked for their help in the methodology optimization.
4. Conclusions SPV measurements are a valuable tool in photocatalytic research. They are especially useful in mechanistic or qualitative studies for determining different surface states. Directly relating the SPV intensity to the photocatalytic activity, however, is not straightforward. We demonstrated that for the photocatalytic discoloration of methylene blue, the apparent rate constant is proportional to the SPV signal for TiO2 P90 samples calcined at different temperatures, possessing a variable rutile content. An optimum SPV response and reaction rate were obtained for the samples calcined at 600–675 ◦ C showing a mixed anatase–rutile crystal phase composition. When improving charge separation by introducing silver nanoparticles, acting as electron traps on the surface of TiO2 the reaction rate becomes inversely proportional to the SPV signal in the range of 0–1 at% of silver. For commercial photocatalysts P25, P90 and PC500, the reactivity order for MB discoloration in aqueous phase is the same as the order in SPV response. In the gas phase, however, the activity order follows the trend of increasing surface area, which is completely opposite to the trend in SPV signal. The general conclusion is that SPV can only account for (photo)electronic properties, whereas structural and morphological properties, as well as reaction conditions, play a crucial role as well. Therefore, care should be taken in the interpretation of SPV data with regard to photocatalytic test results. Acknowledgements S.W. Verbruggen wishes to acknowledge the Research Foundation of Flanders (FWO) for the financial support and Evonik for kindly supplying the P90 powder for our experiments. W. Deblauwe is sincerely thanked for all his help and effort in the construction of the SPV set-up. T. Tytgat is greatly acknowledged for his help
References [1] L. Kronik, Y. Shapira, Surface Science Reports 37 (1999) 1–206. [2] L. Kronik, Y. Shapira, Surface and Interface Analysis 31 (2001) 954–965. [3] L.Q. Jing, X.J. Sun, J. Shang, W.M. Cai, Z.L. Xu, Y.G. Du, H.G. Fu, Solar Energy Materials and Solar Cells 79 (2003) 133–151. [4] Y.K. Hsieh, H.C. Card, Journal of Applied Physics 65 (1989) 2409–2415. [5] D. Wang, J. Zhang, T. Shi, B. Wang, X. Cao, T. Li, Journal of Photochemistry and Photobiology A 93 (1996) 21–25. [6] B.F. Xin, Z.Y. Ren, P. Wang, J. Liu, L.Q. Jing, H.G. Fu, Applied Surface Science 253 (2007) 4390–4395. [7] D. Gal, Y. Mastai, G. Hodes, L. Kronik, Journal of Applied Physics 86 (1999) 5573–5577. [8] E. Fefer, L. Kronik, M. Leibovitch, Y. Shapira, W. Riedl, Applied Surface Science 104–105 (1996) 61–67. [9] M. Gratzel, Nature 414 (2001) 338–344. [10] S.W. Verbruggen, K. Masschaele, E. Moortgat, T.E. Korany, B. Hauchecorne, J.A. Martens, S. Lenaerts, Catalysis Science & Technology 2 (2012) 2311–2318. [11] L. Jing, S. Li, S. Song, L. Xue, H. Fu, Solar Energy Materials and Solar Cells 92 (2008) 1030–1036. [12] T.F. Xie, D.J. Wang, L.J. Zhu, T.J. Li, Y.J. Xu, Materials Chemistry and Physics 70 (2001) 103–106. [13] X.R. Zhang, Y.H. Lin, D.Q. He, J.F. Zhang, Z.Y. Fan, T.F. Xie, Chemical Physics Letters 504 (2011) 71–75. [14] R.A. Spurr, H. Myers, Analytical Chemistry 29 (1957) 760–762. [15] V. Iliev, D. Tomova, L. Bilyarska, A. Eliyas, L. Petrov, Applied Catalysis B 63 (2006) 266–271. [16] V. Iliev, D. Tomova, L. Bilyarska, G. Tyuliev, Journal of Molecular Catalysis A: Chemical 263 (2007) 32–38. [17] A. Mills, J. Wang, Journal of Photochemistry and Photobiology A 127 (1999) 123–134. [18] S.W. Verbruggen, S. Ribbens, T. Tytgat, B. Hauchecorne, M. Smits, V. Meynen, P. Cool, J.A. Martens, S. Lenaerts, Chemical Engineering Journal 174 (2011) 318–325. [19] T. Kawahara, Y. Konishi, H. Tada, N. Tohge, J. Nishii, S. Ito, Angewandte Chemie International Edition 41 (2002) 2811–2813. [20] R.R. Bacsa, J. Kiwi, Applied Catalysis B 16 (1998) 19–29. [21] S.X. Liu, Z.P. Qu, X.W. Han, C.L. Sun, Catalysis Today 93–95 (2004) 877–884. [22] C. Akly, P.A. Chadik, D.W. Mazyck, Applied Catalysis B 99 (2010) 329–335. [23] H.H. Ou, S.L. Lo, Journal of Hazardous Materials 146 (2007) 302–308.
Contents lists available at SciVerse ScienceDirect
Catalysis Today journal homepage: www.elsevier.com/locate/cattod
Surface photovoltage measurements: A quick assessment of the photocatalytic activity? S.W. Verbruggen a,b,∗ , J.J.J. Dirckx c , J.A. Martens b , S. Lenaerts a a b c
Sustainable Energy and Air Purification, Department of Bioscience Engineering, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium Center for Surface Science and Catalysis, Department of Microbial and Molecular Systems (M2 S), Catholic University of Leuven, Kasteelpark Arenberg 23, 3001 Heverlee, Belgium The Laboratory of BioMedical Physics (BIMEF), Department of Physics, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium
a r t i c l e
i n f o
Article history: Received 25 June 2012 Received in revised form 3 September 2012 Accepted 20 November 2012 Available online 16 January 2013 Keywords: Surface photovoltage Photocatalysis Titanium dioxide Silver Methylene blue Acetaldehyde
a b s t r a c t Surface photovoltage (SPV) measurements can contribute to a better understanding of electronic properties of photocatalysts under illumination. Direct linking of SPV data to the actual photocatalytic activity remains troublesome. This work aims to discuss SPV measurements from a photocatalytic point of view. By means of several application-based scenarios we illustrate that the trend between SPV and photocatalysis strongly depends on parameters such as the crystal structure, surface modifications, morphology and humidity. This makes the interpretation far from straightforward. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Surface photovoltage (SPV) measurement is a well-established contactless technique for the characterization of semiconductors, including photocatalysts, based on monitoring the changes in surface voltage before and during illumination. Several excellent review articles have been published on the physical and electrochemical fundamentals [1,2]. The electronic band structure of n-type semiconductors such as TiO2 is characterized by upward band bending at the semiconductor–conductor interface, leading to a Shottky-type barrier. Surface-localized electronic states induce charge transfer between the bulk and the surface of the semiconductor in order to establish thermodynamic equilibrium. This charge transfer results in a non-neutral surface space charge region (SCR) near the surface [3]. This in turn results in a builtin electric field, Vs , commonly referred to as the surface potential barrier. When the photocatalyst is irradiated by light with a photon energy content greater than the bandgap, free charge carriers are generated by band-to-band transitions induced by the incident photons. Trap-to-band transitions are also possible, but since these are related to photons with a lower energy than the bandgap,
∗ Corresponding author at: Sustainable Energy and Air Purification, Department of Bioscience Engineering, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium. Tel.: +32 3 265 35 17; fax: +32 3 265 32 25. E-mail address: [email protected] (S.W. Verbruggen). 0920-5861/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cattod.2012.11.010
their effect becomes negligible and will therefore not be considered here [4]. The generated charge carriers will redistribute within the surface and/or the bulk as a result of the built-in electric field. Hence, the surface potential barrier will be altered after illumination. Exactly this difference between the surface potential barrier in dark (Vs ) and the one under illumination (Vs ), is defined as the SPV signal (Fig. 1) and is used to characterize the photocatalyst. SPV spectroscopy could be further expanded to electric field induced SPV spectroscopy (EFISPS). In EFISPS an external electric field directly affects the response intensity of the SPV signal and the photovoltaic characteristics [5]. Using SPV measurements, information can be obtained concerning the (photo)electronic properties of semiconductors, such as the band bending, surface and bulk carrier recombination, type of defect states and magnitude of the bandgap [6–8]. One of the most interesting features of the SPV technique is that it is directly applicable to powders and therefore appears to be a valuable tool in mechanistic photocatalytic studies. It would indeed be wonderful to have a very fast, easy-to-use technique available that does not require tedious sample preparations and can be employed for assessing the photocatalytic potential of a great number of photocatalyst powders, without time-consuming catalytic experiments. Comparable information could also be acquired by means of electrochemical photocurrent measurements, in which the photocatalyst has to be attached to an electrode and is then immersed in an electrolyte solution [9,10]. However, by circumventing the necessity to fixate the catalyst to an electrode and excluding the
216
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
Vs eEC
V’s
SPV
The change in crystallite size due to the heat treatment was calculated using the Scherrer equation: D = 0.9
EF hν
EV
h+
Fig. 1. Scheme on the origin of the SPV signal. The change in band bending is represented by dashed lines. The redistribution of generated charges is indicated with the arrows.
presence (and the influence) of the electrolyte, SPV measurements present a very attractive alternative. The objective of this work is to discuss the concept of SPV measurements from a photocatalytic point of view. The thus far reported trends in literature indicate that directly relating SPV results to actual photocatalytic activity is not unambiguous [3,6,11–13]. Several situations where care should be taken when interpreting SPV measurements with regard to photocatalytic activity are discussed. It has to be stressed that in this work it was not the intention to use SPV for a detailed study of the (photo)electronic properties of photocatalysts, but rather to apply SPV as a simple, quick screening method for assessing photocatalytic performance. We show that photocatalytic activity is not merely predictable based on (photo)electronic properties as measured by SPV, but is also strongly influenced by the catalyst crystal phase composition, type of catalyst modification, morphology and general reaction conditions (type of fluid phase and the catalyst humidity).
ˇ × cos
with D, crystallite size [nm]; , X-ray wavelength [nm] (CuK␣ Xrays: 0.1542 nm); ˇ, full width at half maximum (FWHM) [radians]; , Bragg angle [radians]. 2.1.2. TiO2 modified with silver nanoparticles Silver was deposited on the TiO2 P90 (Evonik) surface by the photodeposition method using an adapted protocol based on Iliev et al. [15,16]. For 0.5 g of TiO2 P90, AgNO3 (Sigma–Aldrich, >99.0% purity) was used as the silver precursor and dissolved in 50 mL of bidistilled water, so that samples with 0.1, 0.3, 0.6 and 1 at% of silver (TiO2 basis) were obtained. Methanol (Merck, >99.5% purity) was used a sacrificial hole scavenger and added to the solution in a molar ratio methanol:AgNO3 of 500:1. The P90 powder was added to the solution and brought to pH 2 with nitric acid (Merck, 65%). The mixture was stirred for 30 min at 750 rpm in the dark and consecutively illuminated with UV light (Philips Cleo, 25 W, 205 W cm−2 at 365 nm) for 30 min while stirring. The resulting colored suspension was centrifuged and washed 2 times with bidistilled water. The cake was further dried overnight at 110 ◦ C and grinded in an agate mortar. 2.1.3. TiO2 with different intrinsic properties or water content Three commercial photocatalysts were selected for this study. P90 (Evonik), P25 (Evonik) and PC500 (Cristal global), were used to determine the SPV response and the photocatalytic activity in the aqueous as well as in the gas phase. In order to study the qualitative effect of the humidity of the catalyst powder, P90 was dried overnight at 300 ◦ C. While handling this dried P90, it was continuously kept at 110 ◦ C to avoid unwanted rehydration of the catalyst. Controlled rehydration of the catalyst was achieved by the deposition of 2.5 or 5.0 L H2 O onto 4 mg dried P90 powder on the ITO glass slide of the SPV setup.
2. Materials and methods 2.1. Sample preparation For all tests in this work, TiO2 P90 (Evonik) was selected as the reference material. It consists predominantly of crystalline anatase particles and a small fraction of rutile (∼10%). Therefore, it has proven to be an excellent starting material for further modifications such as the introduction of higher rutile content or the deposition of silver particles on the surface. P90 also offered a good basis for comparative experiments with other (commercial) photocatalysts with different material characteristics. The other commercial photocatalysts used in this work were P25 (Evonik) and PC500 (Cristal Global). Both were used as received. 2.1.1. TiO2 with variable rutile content In order to obtain TiO2 samples with different anatase/rutile ratios, commercial P90 (Evonik) was used as such or calcined at 450 ◦ C, 525 ◦ C, 600 ◦ C, 675 ◦ C, 750 ◦ C or 900 ◦ C for 3 h. The precise rutile content was calculated based on the XRD crystal spectra of the resulting powders. The weight fraction of rutile was calculated using the formula originally introduced by Spurr and Myers [14]: Wrutile % =
1 1 + 0.8 × (Ianatase /Irutile )
with Ianatase and Irutile the peak intensities of the strongest anatase and strongest rutile reflection peak respectively.
2.2. SPV measurements This work aims to study the possible use of SPV for quickly assessing the photocatalytic performance, rather than carrying out a detailed study of the (photo)electronic properties. Therefore, no classical SPV ‘sweep’ spectra were recorded, but a single SPV signal was measured after illumination with a low-intensity UVA lamp (Philips Cleo, 25 W, 205 W cm−2 at 365 nm). This was the same light source as used in all photocatalytic tests, and therefore best suited for a realistic study of the relation between SPV and photocatalysis. The set-up itself was a custom-made apparatus, in which the catalyst powder was sandwiched between two ITO electrodes (Sigma–Aldrich, d = 1.2 mm, resistivity: 8–12 cm−2 ), connected to a photo-amplifier (1 × 106 voltage amplification). No external bias was applied. A schematic representation of the SPV set-up is depicted in Fig. 2. In order to perform highly reproducible measurements, only a controlled square section of 5 mm × 5 mm was illuminated. A constant amount of catalyst was spread out on this marked area and sandwiched. For all tests on TiO2 based samples, the ideal amount of catalyst that covered this area and resulted in a good contacting layer between both ITOs was determined to be 4.0 ± 0.1 mg. Since the Ag-P90 samples were considerably coarser due to the photodeposition treatment, 11.0 ± 0.1 mg was used for these specific tests. For the exact determination of the SPV value, the difference was taken between the steady voltage readout after 1.5 min of UVA illumination and the steady voltage readout in dark.
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
217
Fig. 2. Schematic representation of the SPV set-up.
For all samples, three independent measurements were conducted and averaged. 2.3. Photocatalytic activity tests 2.3.1. Aqueous phase photocatalytic test For the assessment of the photocatalytic activity of all samples, methylene blue (MB) discoloration batch experiments in aqueous phase were performed. A typical experiment involved suspending 20 mg of catalyst in 20 mL of MB solution (9.4 mol L−1 ) in a UV-transparent screw cap, stirred at 500 rpm. The vessel was placed 5 cm from the UVA lamp (same as in SPV Measurements), which was placed halfway the liquid level. The MB discoloration is monitored by means of UV/VIS absorption spectroscopy (Shimadzu UV-2501PC double beam spectrophotometer), using the peak absorption intensity of the MB peak at 663.5 nm. Before turning on the lamp, the samples were allowed to establish adsorption–desorption equilibrium by stirring in dark for 30 min. Sampling again after 40 min in dark indicated that the absorption level did not change anymore and the equilibrium was reached. Samples were taken after 1, 2, 4, 7, 13, 25 and 45 min of illumination. The MB concentration was determined after centrifugation of the sample. After the spectrum collection, the MB solution and residual catalyst cake were re-added to the reaction vessel. At every sampling interval the residual MB concentration (C/C0 ) was plotted as a function of illumination time. The apparent rate constant k is taken as the characteristic parameter for the activity of each sample. It is calculated as the slope of the first (linear) part of the C/C0 vs. time (in min) plot. A blank test was performed in which 20 mL of MB solution was illuminated in the absence of any catalyst. A slight decrease in the MB concentration was detected, resulting in an apparent rate constant k = 0.002 min−1 , which is negligible compared to the rate constants obtained over all catalyst samples. It is important to note that we did not monitor the formation of MB degradation products. Therefore we cannot distinguish between dye oxidation and the possible photoreduction to the colorless leuco methylene blue (LMB) form. The latter is preferentially formed under conditions of low dissolved oxygen levels and low pH values [17]. In the present study, however, the MB solution was well aerated prior to the measurements and reactions were carried out at neutral pH. Hence, formation of LMB was not to be expected. 2.3.2. Gas phase photocatalytic test The photocatalytic gas phase tests involved the degradation of acetaldehyde in air, using a single-pass flow through photoreactor, as described in previous work [18]. In short, 50 mg of catalyst was deposited onto 108 g 2 mm glass beads by means of suspension coating and loaded into the photoreactor. Acetaldehyde (1% in N2 , Air Liquide) and compressed air (Air Liquide Alphagaz) were mixed using mass flow controllers (Brooks), establishing 170 ppmv inlet concentration at a total flow rate of 2000 cm3 min−1 . The concentrations of (un)degraded acetaldehyde and formed CO2 were monitored using online FTIR spectroscopy of the outlet gas stream
(NicoletTM 380 Thermo Fisher Scientific, ZnSe windows and 2 m heated gas cell). The steady-state conversion of acetaldehyde is determined as the part of the acetaldehyde flow that is continuously degraded over time, divided by the total acetaldehyde inlet flow (after establishing adsorption/desorption equilibrium). By simultaneous comparison with the amount of CO2 formed over time, a carbon balance very close to 100% indicates full mineralization of the pollutant [10]. 2.4. Characterization Determination of the Brunauer–Emmett–Teller (BET) surface area was performed with a Micrometrics Tristar Surface Area and Porosity Analyzer. The samples were degassed overnight at 250 ◦ C. X-ray powder diffraction (XRD) was carried out using a STOE StadiP apparatus with CuK␣ radiation and an image plate detector. UV–VIS diffuse reflectance spectra were collected in the range of 300–700 nm using a Shimadzu UV-2501PC double beam spectrophotometer equipped with a 60 mm BaSO4 coated integrating sphere and a Photomultiplier R-446U detector. 3. Results and discussion 3.1. Variation in rutile content It has been stated many times that a certain fraction of rutile in an excess of anatase TiO2 improves charge separation by the transfer of photogenerated electrons from anatase to rutile [19]. This physically separates them from the photoholes. This is beneficial for the photocatalytic activity, as long as the rutile fraction does not become too high [20]. The material characteristics of the P90 samples calcined at different temperatures are listed in Table 1. It is clear that with increasing temperature the particles start to sinter, which is evidenced by the decrease in specific surface area and the increase in crystallite size. Furthermore a phase transformation occurs from predominantly anatase to exclusively rutile. Fig. 3a shows the MB discoloration of the P90 samples over time and Fig. 3b the SPV signal of these samples, together with the photocatalytic activity determined by the rate of MB discoloration. It is immediately striking that the SPV signals and the apparent rate constants are very closely related. The SPV signal is strongest for the mixed Table 1 Photocatalyst characteristics after calcination of TiO2 P90 at different temperatures. Calcination T
Surface area [m2 g−1 ]
Rutile fraction [%]
Anatase crystallite size [nm]
Rutile crystallite size [nm]
RT 450 ◦ C 525 ◦ C 600 ◦ C 675 ◦ C 750 ◦ C 900 ◦ C
126 123 114 86 53 11 2
10 10 11 26 58 100 100
10.9 11.2 11.8 13.6 17.5 NA NA
13.4 15.3 17.4 21.1 27.8 23.2 35.9
218
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
Fig. 3. P90 samples calcined at different temperatures: (a) MB discoloration (C/C0 ) as a function of illumination time and (b) SPV signal (䊉, left axis) and apparent rate constant for MB discoloration (, right axis) under UVA illumination as a function of the calcination temperature.
phase anatase–rutile samples obtained at calcination temperatures between 600 and 675 ◦ C. Under these conditions, charge separation seems very efficient, resulting in high photocatalytic activity. At higher calcinations temperatures, the rutile content becomes excessive, leading to lower photocatalytic reaction rates. It is noteworthy to mention that in this experiment the role of the specific surface area seems of no vital importance, which confirms the fact that SPV mainly probes electronic properties rather than morphological properties. Furthermore, we found evidence that mainly the electronic properties dominate the photocatalytic activity of photocatalysts with nanosized pores in aqueous phase applications [10]. Our observations suggest that for purely TiO2 based photocatalysts, a stronger SPV signal corresponds to a higher photocatalytic activity in aqueous phase. These findings support earlier contributions in literature [11,13]. 3.2. Modification with noble metals The most distinct case in which SPV is used in photocatalytic studies is the one in which the material is modified by depositing noble metals on the surface. In this context Xie et al. report that after Pd and Ru deposition on the surface, a stronger SPV response corresponds to a higher photocatalytic activity [12]. This is in contrast to Jing et al. who state that after Pd and Ag deposition on the surface of ZnO, a weaker SPV signal corresponds to a higher photocatalytic activity [3]. Generally speaking, a more intense SPV signal should correspond to a better charge separation efficiency and thus a higher photocatalytic activity is expected. In the case of
noble metal depositions, however, charge separation is based on the metals acting as electron traps under UV illumination. Hence, when the noble metals are effectively trapping the photogenerated electrons, less charge is available for charge transport between the ITO electrodes of the SPV apparatus. This rationalizes why a weaker SPV signal is correlated to higher photocatalytic activity for this type of modified materials. This theory is also validated by the results of Xin et al. who used Fe3+ dopants as electron traps [6]. They also come to the conclusion that when the photogenerated electrons are effectively trapped in the Fe3+ states, the SPV signal weakens, whereas the photocatalytic activity improves. In order to evaluate these theories, a set of Ag–TiO2 samples was prepared. Silver was deposited onto TiO2 P90 using the photodeposition method, resulting in samples with 0.1, 0.3, 0.6 and 1.0 at% of silver (TiO2 basis). No significant change was measured in surface area, porosity, crystal phase composition or crystallite size after deposition. There is a clear color change from almost white (0.1 at% Ag) to gray-red (1 at% Ag), indicating that increasing amounts of silver are effectively deposited on the surface. This is also plain from the diffuse reflectance spectra (Fig. 4a), in which the surface plasmon peak of the silver nanoparticles becomes more pronounced with increasing silver content. Fig. 4c shows the SPV signal and the apparent photocatalytic reaction rate for MB discoloration (Fig. 4b) measured on these samples. In contrast to the findings in Section 3.1 and Fig. 3b, the SPV signal and photocatalytic activity show opposite trends in this case: the lower the SPV signal, the higher the photocatalytic activity. This supports the findings of Xin et al. [6] and Jing et al. [3]. The remark has to be made that the observed trend is only true in the studied range of silver content. Unfortunately it is impossible to observe experimentally what will happen at even larger silver loadings using the current set-up. As a consequence of the better conductive properties of the metal-modified catalysts, silver loadings higher than 1 at% result in a SPV signal that exceeds the operating window of the apparatus. It is to be expected that the photocatalytic activity will reach an optimum, after which it will decline as a result of the increasing amount of recombination centers and TiO2 coverage with increasing silver loading [21]. The experimental boundaries of the set-up, however, do not permit us to evaluate whether the SPV signal will show a similar trend. Therefore, based on these experiments it can only be concluded that in the range between 0 and 1 at% of silver, a higher SPV signal corresponds to a lower photocatalytic activity. 3.3. Catalyst properties vs. reaction conditions Another important parameter that is often overlooked when dealing with SPV measurements and their link with the photocatalytic activity, is the catalyst morphology. This is illustrated by means of three widely used commercial photocatalysts: P90 (Evonik), P25 (Evonik) and PC500 (Cristal Global). P25 is known for its high quantum efficiency, because of its fully crystalline nature and well formed crystallites of ca. 30 nm. PC500, on the other hand, is characterized by a much lower quantum efficiency, due to a significant fraction of amorphous material (∼20–25%). As high quantum efficiency implies the efficient generation of charge carriers under illumination, it can be stated in other words that the (photo)electronic properties as measured by SPV will be a lot better for P25 compared to PC500. P90 lies somewhere in-between. However, quantum efficiency is obviously not the sole important parameter in photocatalysis! For instance, PC500’s surface area is almost three times higher than that of P90 and about six times higher than that of P25. It also contains mesopores of ca. 4 nm that can give access to (small) gaseous pollutants, but can present a significant diffusion resistance for aqueous pollutants. Hence, a large surface area that is predominantly composed of small nanosized pores is only valuable in reaction systems where diffusion
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
219
Fig. 5. SPV response of (from left to right): P90 (as received) (gray bar), P90 dried overnight at 300 ◦ C (white bar), dried P90 rehydrated with 2.5 L H2 O (first striped bar) and dried P90 rehydrated with 5 L H2 O (second striped bar).
Fig. 4. Silver on P90 samples: (a) UV/VIS diffuse reflectance spectra of Ag–TiO2 (P90) (recalculated into absorbance), (b) MB discoloration (C/C0 ) as a function of illumination time and (c) SPV signal (䊉, left axis) and apparent rate constant for MB discoloration (, right axis) under UVA illumination as a function of the silver content (at%).
limitations are negligible. For this reason PC500 performs much better when it is used in the gas phase compared to the aqueous phase (Table 2). Table 2 compares the recorded SPV signals and the corresponding photocatalytic activities in gas and aqueous phase obtained for identical amounts of P25, P90 and PC500. In the aqueous phase, the higher the SPV response (thus the better the photo-electrical properties), the higher the reaction rate. This is in accordance with the observations from Section 3.1. However, when considering the photocatalytic activities in the gas phase, the reactivity order is completely reversed. Although no proper SPV signal could be measured for PC500 (presumably due to the lack of a well defined crystal structure), the acetaldehyde conversion in the gas phase is significantly better than the one obtained over P90, which is in turn better than P25. As discussed above, this is mainly attributed to the high available surface area, which appears to be of vital importance in gas phase applications [10]. Again, all of this indicates that care should be taken when interpreting SPV data that only take into account (photo)electronic properties, but not structural properties, since the latter can have a massive impact on the photocatalytic activity as well. A final remark that has to be made concerns the humidity of the catalyst. Water provides a conductive layer between the catalyst powder particles and the ITO electrodes. This way electron transfer to the electrodes can take place and a SPV signal can be recorded more easily. From Fig. 5 it is clear that the SPV response of P90 (as received) significantly decreases when the powder is completely dried. Rehydrating the powder results in restoring and even improving the SPV signal. Despite the fact that most photocatalytic reactions are carried out in the aqueous phase, or in the gas phase in which the air humidity level plays an important role on catalyst activity [22,23], it is rather surprising that the humidity of the catalyst powder has (almost) never been the subject of investigation
Table 2 SPV signal and photocatalytic activity of TiO2 P25, P90 and PC500 in aqueous and gas phase. Catalyst
Surface area [m2 g−1 ]
SPV [V]
Apparent rate constant (10−3 min−1 ) aqueous phase
Steady state conversion gas phase
P25 P90 PC500
55 126 350
1.7 ± 0.4 0.4 ± 0.1 0.01 ± 0.01
219 ± 10 155 ± 5 84 ± 4
(76 ± 2)% (90 ± 2)% (99 ± 2)%
220
S.W. Verbruggen et al. / Catalysis Today 209 (2013) 215–220
in SPV measurements. This again evidences that relating the SPV signal to the photocatalytic performance can be cumbersome.
in programming the SPV software. B. Lenaerts, M. Keulemans and T. Lathouwers are thanked for their help in the methodology optimization.
4. Conclusions SPV measurements are a valuable tool in photocatalytic research. They are especially useful in mechanistic or qualitative studies for determining different surface states. Directly relating the SPV intensity to the photocatalytic activity, however, is not straightforward. We demonstrated that for the photocatalytic discoloration of methylene blue, the apparent rate constant is proportional to the SPV signal for TiO2 P90 samples calcined at different temperatures, possessing a variable rutile content. An optimum SPV response and reaction rate were obtained for the samples calcined at 600–675 ◦ C showing a mixed anatase–rutile crystal phase composition. When improving charge separation by introducing silver nanoparticles, acting as electron traps on the surface of TiO2 the reaction rate becomes inversely proportional to the SPV signal in the range of 0–1 at% of silver. For commercial photocatalysts P25, P90 and PC500, the reactivity order for MB discoloration in aqueous phase is the same as the order in SPV response. In the gas phase, however, the activity order follows the trend of increasing surface area, which is completely opposite to the trend in SPV signal. The general conclusion is that SPV can only account for (photo)electronic properties, whereas structural and morphological properties, as well as reaction conditions, play a crucial role as well. Therefore, care should be taken in the interpretation of SPV data with regard to photocatalytic test results. Acknowledgements S.W. Verbruggen wishes to acknowledge the Research Foundation of Flanders (FWO) for the financial support and Evonik for kindly supplying the P90 powder for our experiments. W. Deblauwe is sincerely thanked for all his help and effort in the construction of the SPV set-up. T. Tytgat is greatly acknowledged for his help
References [1] L. Kronik, Y. Shapira, Surface Science Reports 37 (1999) 1–206. [2] L. Kronik, Y. Shapira, Surface and Interface Analysis 31 (2001) 954–965. [3] L.Q. Jing, X.J. Sun, J. Shang, W.M. Cai, Z.L. Xu, Y.G. Du, H.G. Fu, Solar Energy Materials and Solar Cells 79 (2003) 133–151. [4] Y.K. Hsieh, H.C. Card, Journal of Applied Physics 65 (1989) 2409–2415. [5] D. Wang, J. Zhang, T. Shi, B. Wang, X. Cao, T. Li, Journal of Photochemistry and Photobiology A 93 (1996) 21–25. [6] B.F. Xin, Z.Y. Ren, P. Wang, J. Liu, L.Q. Jing, H.G. Fu, Applied Surface Science 253 (2007) 4390–4395. [7] D. Gal, Y. Mastai, G. Hodes, L. Kronik, Journal of Applied Physics 86 (1999) 5573–5577. [8] E. Fefer, L. Kronik, M. Leibovitch, Y. Shapira, W. Riedl, Applied Surface Science 104–105 (1996) 61–67. [9] M. Gratzel, Nature 414 (2001) 338–344. [10] S.W. Verbruggen, K. Masschaele, E. Moortgat, T.E. Korany, B. Hauchecorne, J.A. Martens, S. Lenaerts, Catalysis Science & Technology 2 (2012) 2311–2318. [11] L. Jing, S. Li, S. Song, L. Xue, H. Fu, Solar Energy Materials and Solar Cells 92 (2008) 1030–1036. [12] T.F. Xie, D.J. Wang, L.J. Zhu, T.J. Li, Y.J. Xu, Materials Chemistry and Physics 70 (2001) 103–106. [13] X.R. Zhang, Y.H. Lin, D.Q. He, J.F. Zhang, Z.Y. Fan, T.F. Xie, Chemical Physics Letters 504 (2011) 71–75. [14] R.A. Spurr, H. Myers, Analytical Chemistry 29 (1957) 760–762. [15] V. Iliev, D. Tomova, L. Bilyarska, A. Eliyas, L. Petrov, Applied Catalysis B 63 (2006) 266–271. [16] V. Iliev, D. Tomova, L. Bilyarska, G. Tyuliev, Journal of Molecular Catalysis A: Chemical 263 (2007) 32–38. [17] A. Mills, J. Wang, Journal of Photochemistry and Photobiology A 127 (1999) 123–134. [18] S.W. Verbruggen, S. Ribbens, T. Tytgat, B. Hauchecorne, M. Smits, V. Meynen, P. Cool, J.A. Martens, S. Lenaerts, Chemical Engineering Journal 174 (2011) 318–325. [19] T. Kawahara, Y. Konishi, H. Tada, N. Tohge, J. Nishii, S. Ito, Angewandte Chemie International Edition 41 (2002) 2811–2813. [20] R.R. Bacsa, J. Kiwi, Applied Catalysis B 16 (1998) 19–29. [21] S.X. Liu, Z.P. Qu, X.W. Han, C.L. Sun, Catalysis Today 93–95 (2004) 877–884. [22] C. Akly, P.A. Chadik, D.W. Mazyck, Applied Catalysis B 99 (2010) 329–335. [23] H.H. Ou, S.L. Lo, Journal of Hazardous Materials 146 (2007) 302–308.