TiO2 Coated as a

Environ. Sci. Technol. 2009, 43, 7496–7502

Photoelectrocatalytic Removal of Bromate Using Ti/TiO2 Coated as a Photocathode FABIANA MARIA MONTEIRO PASCHOAL,† GREG PEPPING,‡ M A R I A V A L N I C E B O L D R I N Z A N O N I , * ,† A N D M A R C A . A N D E R S O N * ,‡ Departamento de Quı´mica Analı´tica - Instituto de Quı´mica Universidade Estadual Paulista, Av. Prof. Francisco Degni, s/n, C. P. 355, 14801-970 Araraquara, SP, Brazil, and Environmental Chemistry and Technology Program, University of Wisconsin-Madison, 660 N. Park Street, Madison, Wisconsin 53706

Received December 1, 2008. Revised manuscript received May 23, 2009. Accepted July 14, 2009.

These studies represent the rare use of a TiO2 material as a photocathode and the first application of photoelectrocatalysis for BrO3- removal. Photoelectrocatalytic reduction of BrO3to Br- can reach 70% at neutral pH under an applied potential of -0.20 V versus SCE (saturated calomel electrode) after 75 min on the irradiated nanoporous thin-film TiO2-coated working electrode, which presented a flat band potential of -0.0274 V versus SCE. Regardless of the potential applied in these experiments, no BrO3- removal was observed in the counter electrode compartment or during electrolysis or photocatalysis, confirming that reduction of BrO3- to Br- requires the combination of a negative potential (ideally near -0.20 V) and ultraviolet irradiation of the Ti/TiO2 electrode. The process was selective for BrO3- removal in that this process did not significantly reduce levels of Ca2+ and Mg2+ in drinking waters.

Introduction Concerns relating to the formation of trihalomethanes and other chlorine-related disinfection byproducts are driving a trend in municipal drinking water treatment away from chlorination and toward ozonation. Ozonation is also widely used by the fast-growing bottled water industry. While ozonation effectively inactivates potentially harmful pathogens, including Cryptosporidium parvum, Giardia, and Legionella, it also introduces its own health concerns (1). This is because Br-, which is common in natural waters and especially prevalent in coastal areas, reacts with ozone directly to form BrO3- and indirectly, via a free radical pathway (2). While much of the current literature links elevated BrO3levels to ozonation processes, in fact, BrO3- from chlorination is also a concern because of elevated BrO3- levels in some hypochlorite stocks (3, 4). The World Health Organization classifies BrO3- as a category I group B2 or “possible human” carcinogen (5-8). Accordingly, regulatory agencies in the United Kingdom and United States have established a maximum BrO3-concen* Corresponding author phone: (M.A.A) (1) 608 262 2674, (M.V.B.) (016) 33 016619; e-mail: (M.A.A.) [email protected], (M.V.B.) [email protected]. † Universidade Estadual Paulista. ‡ University of Wisconsin-Madison. 7496

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009

tration in drinking waters of 10 µg L-1 (9, 10). Approaches to address elevated BrO3- levels usually fall into three main categories: (1) removing Br- prior to the formation of BrO3-, (2) optimizing the ozonation process to minimize BrO3formation, and (3) removing BrO3- from postozonation waters. The first category includes membrane filtration, granulated activated carbon, and ion exchange (2, 11, 12). The second category includes sequentially adding Cl2 followed by NH3, lowering the pH of the water, controlling the O3 dose and/or contact time, or adding H2O2 or NH3 (2, 11, 13-15). While efforts using these first two categories of treatment have enjoyed some limited success, it appears that BrO3will continue to be a concern in drinking water whether the treatment process is ozonation or chlorination. In the third category of BrO3-treatment, advanced oxidation and reduction methods, including ultraviolet (UV) irradiation, highelectron beam irradiation, and chemical reduction using activated carbon or Fe2+, are also employed (2, 14, 16). The use of photocatalysis as an advanced oxidation process for BrO3- removal, using irradiated TiO2, is also found in the literature. Mills et al. (14) reported successful removal of BrO3- using a platinized TiO2 photocatalyst, and Noguchi et al. (17, 18) reported successful BrO3- removal using a pure TiO2 photocatalyst. When TiO2 is irradiated with light less than 380 nm, electron-hole pairs are generated when electrons are excited from their resting valence band to the conduction band according to the following scheme, TiO2 + hν f ecb + hvb+ (19, 20). The photocatalytically generated electrons and holes can then reduce or oxidize other species, respectively. However, if these electrons and holes immediately recombine, they simply generate heat, and any opportunity to initiate desired oxidation-reduction reactions is lost. Electron-hole recombination is reduced by charge separation, which can be achieved by applying a potential to an irradiated TiO2 electrode. In addition, the movement of the electrons along the external circuit increases the likelihood of reactions occurring at either the working or counter electrode. This more efficient use of electrons and holes is the added benefit of photoelectrocatalysis over photocatalysis (21-31) and the reason for the applied potential in this study. In these previous studies, the use of TiO2 as a suitable anode material to promote oxidation reactions at the irradiated working electrode has drawn the attention of several researchers. Interestingly, some authors (32, 33) have indicated that irradiated TiO2 electrodes can operate in a polarization region where electrons are transferred from the n-type semiconductor to the electrolyte (cathodic current). However, most of these studies are based on doping processes and little further research has been performed on these systems. In most n-type semiconductors, a potential barrier forms at the nonbiased semiconductor-solution interface preventing bulk electrons from reaching the surface and reacting with electrochemical species at the expected potential. However, at a potential slightly more positive than the flat band potential of the semicondutor one might initiate the reduction of adsorbed oxygen and at a more negative potential value the electrode behavior would become metallike with respect to charge transfer to the electrolyte. Therefore, as presented in this paper the reaction occurring at the irradiated working electrode is the reduction of BrO3to Br-, and therefore Ti/TiO2 functions as a photocathode. While photoelectrocatalysis has previously been used for water treatment to oxidize various contaminants (21, 22, 30, 34-37), the results described here represent to our knowledge 10.1021/es803366d CCC: $40.75

 2009 American Chemical Society

Published on Web 08/27/2009

the first evidence of employing a photocathode for water treatment and the first to achieve BrO3- reduction using photoelectrocatalysis of any sort.

Experimental Details Photoelectrochemical experiments were conducted using a two-compartment cell, enabling isolation of the separate oxidation and reduction reactions occurring at the anode and photocathode, respectively. Solutions of NaBrO3 at a pH of 7 in a NaCl electrolyte were used in both compartment cell and BrO3-, and resulting Br- levels were monitored. BrO3reduction to Br- was seen at the TiO2-coated Ti photocathode, with minimal BrO3- removal in the compartment containing a platinum counter electrode. Chemicals. TiO2 sol-gel suspensions used to coat working electrodes were made from titanium isopropoxide (Aldrich Chemical, 97%) and nitric acid (Aldrich Chemical, ACS reagent grade). Solutions and standards of NaBrO3 and NaBr were prepared from stock solutions obtained from SPEX Certiprep (Metuchen, NJ). NaCl and NaNO3 were obtained from Fisher Scientific (Fairlawn, NJ). All chemicals were used without further purification. MgCl2 was obtained from Mallinckrodt (Hazleton, MO) and CaCl2 from Acros (Morris Plains, NJ). All solutions were prepared using ultrapure water (18.1 MΩ cm) from a NANOpure UV system (model 07331, Barnstead/Thermolyne, Dubuque, IA). Working Electrode Preparation. The photocathode substrate material was an annealed Ti foil 0.05 mm thick, with 99.6+% purity (Goodfellow Cambridge, Ltd.). Foils were cut to size for the experimental cell and preheated to remove organic contaminants by firing at 300 °C for 3 h. A suspension of TiO2 was prepared using sol-gel processing methods previously developed in this lab (38). Working electrodes were dip-coated by inserting and then withdrawing the titanium foil from the TiO2 suspension to achieve homogeneous coatings of TiO2 on the titanium metal support (38). Two additional dip-coatings were applied, and the resulting working electrodes were fired at 400 °C for 3 h to sinter the TiO2 coating onto the Ti support. Reactor Design. The reaction vessel was a rectangular Teflon block measuring 13 cm in height, 5.2 cm in width, and 10.4 cm in length, into which two cylindrical cavities were bored, thus creating a single-compartment cell. A 3.5 cm diameter hole was cut through one side of the cell and covered with a quartz window, through which light was passed to irradiate the working electrode. A Nafion 117 membrane was used to separate both compartments, while allowing electrolyte contact. All experiments were conducted in this two-compartment electrochemical cell. For all experiments using light, the working electrode was irradiated with full spectrum light at an intensity of 1.1 W cm-2. The light was generated by a 500W Oriel Hg(Xe) lamp (lamp housing model number 66021, power supply model 8540, lamp model number 66142, Newport Stratford, CT). Light intensities were measured with an International Light IL 1700 research photometer with a SED033/QNDS2/W detector. Wavelength distribution of light (200-900 nm) was measured using an Ocean Optics USB2000TM probe employing OOI Base31TM software, version 2.0.1.4. To prevent heating of the electrolyte, we employed a water filter to absorb infrared irradiation. Electrolyte in both compartments of the cell was mixed by a stir bar and magnetic stir plate. As there was no observed loss of electrolyte during the experiments, the compartments were left unsealed and exposed to room air. All experiments used a saturated calomel reference electrode (SCE) with a Vycor frit and a bridge tube filled with 3 mol L-1 KCl filling solution (Princeton Applied Research-Ametek, Oak Ridge, TN). Hereafter, all applied potential voltages are stated versus SCE.

FIGURE 1. Photocurrent generation as a function of applied potential obtained for a TiO2 thin-film electrode in 426 mg L-1 NaCl under dark conditions (curve A) and UV irradiation (curve B) before and after addition of BrO3- at concentrations of (C) 0.10 mol L-1, and (D) 0.25 mol L-1, with a scan rate of 10 mV/s and light intensity of 1.1 W/cm2. To control the potential during these studies and to measure the generated photocurrent, we used an electrochemical impedance analyzer (Princeton Applied ResearchAmetek, Oak Ridge, TN) with PerkinElmer model 250 research electrochemistry software (PerkinElmer, Waltham, MA). Analytical Methods. The pH was monitored using a pH electrode (model 8272BN) and pH meter (model 370) from Thermo Orion, Beverly, MA. BrO3-, Br-, and NO3- concentrations were measured using a Dionex ion chromatograph (IC) with an Ion Pac AG9-HC guard column (4 × 50 mm) and an Ion Pac AS9-HC analytical column (4 × 250 mm) connected to an ED 50 conductivity detector. An AS 40 autosampler and GP 50 gradient pump were also employed, with a sample loop volume of 200 µL. Mobile phase eluent for the IC was 9 × 10-3 mol L-1 Na2CO3. Solutions of 40 mg L-1of magnesium (MgCl2) and 80 mg L-1 of calcium (CaCl2) were studied together with the BrO3ion, and Ca2+ and Mg2+ concentrations in samples resembling drinking water were measured using inductively coupled plasma optical emission spectroscopy (ICP-OES). The ICPOES instrument was a PerkinElmer ICP-OES Optima 4300 DV, equipped with a PerkinElmer AS90plus autosampler (PerkinElmer, Waltham, MA). Test Cell. Photocurrent measurements and flat band potential were investigated using linear sweep voltammetry (LSV). These determintations were carried out using a 200 mL single-compartment Teflon cell having a circular quartz window of 50 mm diameter. A circular opening in the cell opposite the quartz window allowed exposure of 4.5 cm2 of the working electrode (Ti/TiO2) to UV illumination. The platinum foil counter electrode, having a circular opening of the same size as the light pathway, was placed just in front of the working electrode, 1 cm away. A saturated calomel electrode (SCE), used as a reference, was placed close to the working electrode through a bridge tube with a Vycor frit tip.

Results and Discussion

Effect of BrO3- on the Photocurrent. The dependence of BrO3- reduction on the photocurrent versus applied potential was investigated using linear sweep voltammetry (LSV) experiments for a Ti/TiO2 electrode in an NaCl 426 mg L-1 electrolyte, with a scan rate of 10 mV/s, scan increment of 2.0 mV, and step time of 0.2 s. Figure 1B shows a sharp increase in current once the applied potentials exceeded approximately -0.55 V in relation to dark (Figure 1, curve A), confirming that the sharp increase in current is due to the photocurrent generated from the irradiated working elecVOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7497

FIGURE 2. Langmuir-Hinshelwood model (45). Plot of reciprocal saturated photocurrent versus reciprocal bromated concentration (Iph1- versus C(BrO3- )-1). Data derived from Figure 2. trode. Note that when bromate is added to the electrolyte, the characteristic curves of Iph versus E for the TiO2 film are similar in shape, but photocurrent increases successively as shown curves C and D of Figure 1. Upon UV irradiation (λ< 380 nm) of TiO2 (n-type semiconductor), large anodic photocurrents appear, arising from the flux of holes (minority carriers) arriving at the electrode surface, which have an oxidizing power in relation to the molecule, whose formal potential is more negative than the valence band edge (39). However, the electron in the conduction band flows via an external circuit to the counter electrode where reduction takes place. Therefore, in agreement with the literature (4041), the higher photocurrent observed in the bromate medium could be indicative that this species could be adsorbed on the electrode surface and participate in direct capture of charge. This pathway is thermodynamically feasible because the redox potential of bromate [E0 ) 1.44 V versus normal hydrogen electrode (NHE)] lies above the semiconductor TiO2 valence band edge. Nevertheless, reductions on the TiO2 electrode surfaces are more complex and have received little attention. Considering that the photoactivity of the catalyst depends on the adsorption process of the substrate, we investigated the effect of the initial concentration of BrO3- on this reaction using the Langmuir-Hinshelwood model (45). The photocurrent values at E ) -0.20 V increase markedly when the initial BrO3- concentration is increased from 0 to 0.3 mg L-1 BrO3-. Figure 2 shows the data fit by plotting Iph-1 versus C1(45) for BrO3-. It is possible to observe that at low concentrations (