A novel route to estimate the photocatalytic activity of titanium dioxide via voltage measurements

Titanium dioxide is known as a photocatalyst, that may be activated by UV-A light and thus be able to lead to chemical reactions, to mineralize organic compounds, to inactivate biomolecules and to kill microorganisms, respectively. To estimate the capability of the photocatalytic activity a novel experimental setup using photovoltage measurements was studied. A distinct correlation between the photocatalytic activity of the titanium dioxide samples and the measurable photovoltages could be demonstrated. The experimental setups included the construction of different electrochemical cells based on TiO2 and using different liquid and gel electrolytes. The investigations were carried out on titanium dioxide layers as well as on TiO2 particle materials. The electrochemical measurement of the resulting voltage was optimised and the results were compared with conventionally used methods for the determination of the photocatalytic activity like the decolourization of methylene blue solutions. Additionally, the increase of carbon dioxide concentration in the gas phase was taken as a measure for total mineralization. Results indicate strong correlations between the different methods.


Introduction
Certain modifications of titanium dioxide (TiO 2 ) show photocatalytic activity during irradiation [3,6,16]. Photocatalytic reactions may lead to mineralization of organic compounds, inactivation of biomolecules and killing of microorganisms [9,10,17]. However, the activity may differ widely depending on the type of modification and on its purity [4,13]. Activity may further be influenced by doping with other elements and a variety of other strategies [1,12,20]. In any case, the unambiguous definition of the activity is mandatory. Most methods to estimate the photocatalytic activity of Titanium dioxide (TiO 2 ) powders or coatings are based on the tracking of the decolourization of a dye using spectrometric measurements. To this end, the redox indicator Methylene Blue (MB) is used frequently. However, there is a number of uncertainties that may lead to inaccuracies in the estimation of the photocatalytic activity. The main and most important issues are, amongst others, the capability of MB to form colourless leuco MB under reductive conditions and the photolysis of the dye by visible light without the action of the photocatalyst [11,19]. The latter can occur if modified TiO 2 -catalysts like doped TiO 2 specimen are tested using visible light [5]. In addition, there are procedural sources of error like inconsistent purity of the used MB, pH-dependent adsorption of the dye to the TiO 2 surface and effects due to inhomogeneous mixing of the dye solution during the test [14]. All these issues may impede a direct comparison of different test results and will lead to increased time and money consuming testing efforts.
To find a reproducible way for the measurement of the photocatalytic activity and to overcome the aforementioned uncertainties, a physical measurement instead of the commonly used chemical effect monitoring was studied. Electrons and "holes" are involved in all photocatalytic activities [8,15]. The measurement of a related voltage should be a promising strategy. In this context, differences concerning the use of semiconductors for photovoltaic processes and for photo-oxidations, respectively, should be mentioned. Solar cells and systems for photo-oxidation processes show some common properties. However, they differ in significant details. Solar cells are based on semiconductors with a forbidden band gap (E g ) that may be overcome by irradiation with light of certain wavelengths. The maximum current of solar cells is limited by E g and is connected to the maximum achievable photovoltage. During irradiation of solar cells, the resulting current is harvested and the maximum achievable photocurrent may be taken as a measure for the efficiency of the photovoltaic module. In oxidation processes, however, the extend of the desired chemical reaction is generally defined by the normal potential of the oxidant. The oxidation (or reduction) potential of a certain reaction may be calculated using the Nernst equation. The extend of a particular oxidation reaction is defined by the potential differences. Thus, in case of semiconductors for photo-oxidation processes, the maximum achievable photovoltage of the system under certain irradiation conditions will be the critical factor whether or not a certain oxidation reaction is generally feasible at all.
The electrical potential differences arising from reductive and oxidative processes during the irradiation of TiO 2 were analysed and included in a new measurement strategy. The aim of this study was to develop a simple and reproducible method for the quantitative evaluation of voltage phenomena related to the electrochemical potential. To this end, a variety of experimental setups were analysed and evaluated and ongoing optimizations were carried out. To prove the feasibility of the concept, photovoltages resulting from the new setup were finally correlated with data from chemical and microbiological monitoring systems.

Experimental Setup A.1: proof of principle
For the proof of principle, a first setup was assembled using two basically identical half-cells made of Platinum cups (diameter = 100 mm, height = 50 mm). Both vessels were filled each with 50 mL aqueous potassium chloride electrolyte (c = 1 gL −1 ) and the solutions were electrically connected via a salt bridge. TiO 2 powder dispersions (P25, Evonik Industries AG, Essen, Germany; Hombikat UV100 Sachtleben Chemie GmbH, Duisburg, Germany) were placed in equal amounts one at a time into both halfcells. The basic setup is illustrated in Fig. 1. During the experiment, one half-cell was illuminated using light with a wavelength maximum of 370 nm (Philips F8T5 BLB 8 W), while the second was shielded from any light. The resulting photovoltage was measured as a potential difference between the half-cells using a high impedance voltmeter (yt writer Goerz Servogor 120). Different tests were performed under variation of the irradiance and the amounts of the applied TiO 2 dispersion.

Experimental Setup A.2: coated glass plates in platinum vessels
With an adjusted setup, TiO 2 -coated glass plates were used instead of the powder dispersions. All specimen were made by IST Fraunhofer, Braunschweig, Germany, by Pulse Magnetron Sputtering (PMS), a common variation of Physical Vapor Deposition (PVD). Pre-sputtering was carried out in argon atmosphere for 5 min to remove contaminations from the target surface. Subsequently, the targets were operated in the oxide mode with a reactive gas flow of 20 cm 3 STP / min O 2 for 15 min. During the main coating process, the TiO 2 thin-films were deposited on unheated 50 × 50 mm 2 non-alkali glass substrates by reactive PMS using a dual magnetron source equipped with two titanium PK 500 cathodes. [18]. The glass plates were inserted into the platinum cells and completely submerged in electrolyte so that the TiO 2 -coated side faced the platinum bottom. Irradiation was done as described before.

Experimental Setup B: use of ITO electrodes
Whilst Experimental Setups A.1 and A.2 were necessary first steps for the proof of principle and as a prerequisite to further optimize the method, the next setup aimed on the optimization for coated TiO 2 samples. To this end, the main progress was the use of a translucent electrode material instead of platinum vessels. For this purpose, Indium-Tin-Oxide (ITO) glass electrodes initially developed for the use in photovoltaic systems were used. ITO shows a fairly good UV-A transmission of about 70% and is electrically conductive. This allows a more compact and cheaper setup with better defined distances between the test specimen and the electrode. Additionally, less electrolyte is necessary. The voltages were measured and stored with a digital multimeter (Agilent, model 34401A).

Preparation of ITO electrodes
The ITO electrodes were manufactured in our laboratory from an Indium-Tin-coated float glass plate as a precursor (www. flues sigkr istal le. com, Dr. Feodor Oestreicher). The precursor plate was cut to the appropriate size and the layout of the electrode surfaces was designed. Subsequently, the desired shape of the electrode was masked with adhesive tape. Using this method, a pair of electrodes was created on one shared plate by removing the indium-tin layer between the covered areas by etching with diluted hydrochloric acid (c = 1 mol L −1 ). The transparent, electrically conductive ITO was formed from the metallic precursor layer by annealing in a muffle oven (Uhlig, U24) at 400 °C for 30 min. Integrity of the electrode surface was finally assessed using conductivity measurements. Figure 2 shows a photo of the structured ITO glass electrode (left). The two conductive ITO layers have a size of 70 × 20 mm 2 each. On the right, the side-view of the Experimental Setup B is shown. The ITO electrode was placed above the TiO 2 -coated sample with a liquid potassium chloride electrolyte (c = 1 g L −1 ) layer in between. One side of the ITO electrode was covered while the other side was illuminated with UV-A light (Philips F8T5 BLB 8 W). Potential differences were tracked by measuring the voltage between the both ITO electrodes.

Experimental Setup C: assessment of TiO 2 particle material using stainless steel electrodes
This setup was designed for voltage measurements of TiO 2 particle materials. Three different TiO 2 materials were used which are pure rutil material (Kronos Worldwide Inc., Leverkusen, Germany), pure anatase material (Hombikat UV100 Sachtleben Chemie GmbH, Duisburg, Germany) and the rutil/anatase mixed material P25 (Evonik Industries AG, Essen, Germany). TiO 2 powder suspensions were applied onto stainless steel plates (50 × 50 mm 2 ) and air-dried under ambient room conditions. Afterwards, the TiO 2 -coated plate was linked with an uncoated plate via an agarose gel electrolyte to set up a connection between these two half-cells. Figure 3 shows this Experimental Setup C. The assembly was illuminated with UV-A light and the resulting photovoltage was measured between the two stainless steel halfcells. TiO 2 powders of different modifications were tested and data were compared with results from other methods for the evaluation of the photocatalytic activity.

Test methods for comparison
For comparison of the voltage measurements the commonly used standardised Methylene Blue test was used [2].

Experimental Setup A.1: proof of principle
For the proof of principle, both half-cells were filled with 50 mL electrolyte (KCl; c = 1 gL −1 ). TiO 2 powder dispersions (2 g suspended in 50 mL electrolyte) containing 0.04 g ±2% TiO 2 per mL were used. Subsequently, in steps of 1 mL of the dispersion was added to the electrolyte in both cells. Two different TiO 2 powder dispersions (P25, Evonik Industries AG, Essen, Germany; Hombikat UV100 Sachtleben Chemie GmbH, Duisburg, Germany) were analysed. The lamp was positioned above one half-cell and the other half-cell was carefully shaded. The voltage was recorded until a constant value was obtained. To vary the irradiance, the distance between the lamp and the electrolyte surface was adjusted. Experiments were repeated three times, using slightly different distances. Figure 4 shows the resulting voltages for P25 and UV100 suspensions in one diagram for different light intensities. Results indicate first that a potential is measurable between both cells and second that the voltage decreases with an increasing distance of the lamp resulting in a lower irradiance of the irradiated half-cell.
Using the same setup, the P25 TiO 2 concentration in both vessels was varied under constant light conditions. As in the previous approach, one half-cell was illuminated, while the other was covered. The distance between the lamp and the electrolyte surface was 83 mm. To increase the amount of nanoparticles in the vessel, repeatedly one mL of TiO 2 suspension was added. Before each addition, the stabilisation of the voltage reading was awaited. Values were accepted as soon as the voltage readout was constant (± 5 mV within 30 s; average time of measurement: 5 min). Mean and standard deviations were depicted in Fig. 5.

Experimental Setup A.2: coated glass plates in platinum vessels
Photovoltage of TiO 2 -coated glass plates (IST Fraunhofer, Braunschweig, Germany) with various layer thicknesses were studied using Experimental Setup A.2. Using this approach, the applicability of the method for the evaluation of surface coatings is to be tested. Basically, the same setup as described for Experimental Setup A.1 was employed. Whilst for the first setup, TiO 2 suspensions in water were used, this approach focusses on immobilized TiO 2 with eminent practical significance. All measurements were done as described before and in Table 1, results are summarised. TiO 2 layer thickness is indicated in the top row (nm).

Comparison with the decolourization of MB
All test samples were subsequently tested for photocatalytic activity using the common standardised methylene blue test method. Comparison of test results indicates similar trends for both methods. Increasing thickness of the TiO 2 layer leads to an increase of photovoltage as well as to an increased MB decolourization capacity (Fig. 6). Table 2 shows the measured absorbance of the MB solutions. With these data, the MB transmission was calculated with a set starting value of 0. The table also shows the residual colouration as a percentage of the starting point and the decrease in colouration as an indicator of the relative decrease in MB concentration.

Experimental Setup C: measurements on TiO 2 particle materials
With Experimental Setup C voltage measurements on TiO 2 particle material were investigated applying ascending TiO 2 powder loads of P25. These measurements resulted in a linear increasing correlation of the maximum voltages with higher powder loads (Fig. 7). The y-axis intercept in Fig. 7 (150-500 mV) was chosen to visualize the trend of the response of the system. Experimental Setup C was also used to perform measurements with particle material of different TiO 2 modifications. The comparison of different TiO 2 modifications indicates different voltages for each material. Furthermore, a strong correlation between the measured photovoltage and the CO 2 release from photocatalytic mineralization of D-Glucose for the different TiO 2 modifications was found. Figure 8 displays the results and the correlation between the two methods for the three different TiO 2 modifications.

Mechanistic considerations under use of Experimental Setup B with ITO electrodes
Using the optimised Experimental Setup B with ITO electrodes voltage measurements over time where carried out. The coated TiO 2 specimen (Fraunhofer IST; Braunschweig) were used as described above and shown in Fig. 2. An exemplary voltage-time curve is depicted in Fig. 9. The curve is comparable to the charging and discharging curves of a plate capacitor and follows the exponential functions for the charging U(t)=U max (1−e −t/τ ) and discharging processes U(t)=U max e -t/τ . The time constant τ = R⋅C describes  the required time until the capacitor is charged to 63%. In the exemplary voltage curve, the exponential fit functions with the above equations are drawn for the charging and discharging processes. The regression values of both curves were R 2 > 0.96. The voltage increase would thus correspond to the charging of a capacitor. The hypothesis for explanation is that a Helmholtz double layer is formed on the illuminated side of the sample due to oxidation reactions which leads to the measurable potential difference compared to the non-illuminated side. As reactive species hydroxyl radicals, which are formed by photocatalysis under the influence of UV-A light, can be considered. The discharging process corresponds to the degradation of the Helmholtz double layer after switching off the UV-A light source. Without light, no further hydroxyl radicals are generated to maintain the voltage, following charges gradually flow away until a state of equilibrium is reached again. This can mainly take place through the electrolyte, since the conductivity within the TiO 2 layer is very low at 10 -9 Ω cm −1 [7].

Conclusions
The results presented here open up a new way to determine the photocatalytic activity of TiO 2 . It was shown that photocatalytically activated TiO 2 sampled generate a measurable voltage. A correlation between the photocatalytic activity and the measurable voltages was shown. This procedure can be a possible new route to estimate the photocatalytic activity via a physical method. This study also lays a groundwork on which a physical measuring device for the objectifiable determination of the photocatalytic activity can possibly be developed.