Effect of NOM on arsenic adsorption by TiO 2 in simulated As(III)-contaminated raw waters

Transcription

1 WATER RESEARCH 42 (28) Available at journal homepage: Effect of NM on arsenic adsorption by Ti 2 in simulated -contaminated raw waters Guojing Liu a, Xiangru Zhang b, Jeffrey W. Talley a,, Clive R. Neal a, Hangyao Wang c a Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA b Department of Civil Engineering, Hong Kong University of Science and Technology, Hong Kong, China c Department of Chemical and Biomolecular Engineering, University of Notre Dame, IN 46556, USA article info Article history: Received 24 August 27 Received in revised form 6 December 27 Accepted 12 December 27 Available online 17 January 28 Keywords: NM Ti 2 Adsorption xidation Mechanism abstract The effect of natural organic matter (NM) on arsenic adsorption by a commercial available Ti 2 (Degussa P) in various simulated -contaminated raw waters was examined. Five types of NM that represent different environmental origins were tested. Batch adsorption experiments were conducted under anaerobic conditions and in the absence of light. Either with or without the presence of NM, the arsenic adsorption reached steadystate within 1 h. The presence of 8 mg/l NM as C in the simulated raw water, however, significantly reduced the amount of arsenic adsorbed at the steady-state. Without NM, the arsenic adsorption increased with increasing solution within the range of With four of the NMs tested, the arsenic adsorption firstly increased with increasing and then decreased after the adsorption reached the maximum at An appreciable amount of arsenate () was detected in the filtrate after the Ti 2 adsorption in the simulated raw waters that contained NM. The absolute amount of in the filtrate after Ti 2 adsorption was dependent: more was presented at 47 than that at o7. The arsenic adsorption in the simulated raw waters with and without NM were modelled by both Langmuir and Frendlich adsorption equations, with Frendlich adsorption equation giving a better fit for the water without NM and Langmuir adsorption equation giving a better fit for the waters with NM. The modelling implies that NM can occupy some available binding sites for arsenic adsorption on Ti 2 surface. This study suggests that in an -contaminated raw water, NM can hinder the uptake of arsenic by Ti 2, but can facilitate the oxidation to at Ti 2 surface under alkaline conditions and in the absence of 2 and light. Ti 2 thus can be used in situ to convert to the less toxic in NM-rich groundwaters. Published by Elsevier Ltd. 1. Introduction Arsenic is a common contaminant in drinking water supplies, with a contamination level ranging from o1: to 41 mg=l (o:13 to 413:3 mm) (Welch et al., 2; Smedley et al., 22; Mandal and Suzuki, 22; Bissen and Frimmel, 23). Longterm exposure to arsenic can cause various cancers (Smith et al., 2). The World Health rganization (WH) recommended the maximum concentration of arsenic of 1 mg=l (1:33 mm) in drinking water in order to reduce its potential harm to human health (WH, 1993). The WH-recommended regulation for arsenic in drinking water has recently been Corresponding author. Tel.: ; fax: addresses: gliu@nd.edu (G. Liu), xiangru@ust.hk (X. Zhang), jtalley1@nd.edu (J.W. Talley), neal.1@nd.edu (C.R. Neal), hwang4@nd.edu (H. Wang) /$ - see front matter Published by Elsevier Ltd. doi:1.116/j.watres

2 231 WATER RESEARCH 42 (28) adopted by the New European Community and the U.S. Environmental Protection Agency (USEPA) (EC, 1998; USEPA, 21a). The USEPA estimates that about 5% of the total water systems in the United States have arsenic levels in source water greater than this updated regulation (USEPA, 21b). The discrepancy between the high arsenic concentrations in raw water and the stringent standard calls for an emergent modification of the current treatment technology and a development of new treatment technologies. Arsenic exists in water primarily as the inorganic oxyanions of arsenite () and arsenate (), with predominating in anaerobic waters and prevailing in oxic waters (Masscheleyn et al., 1991; Smedley et al., 22; Katsoyiannis et al., 27). Recent surveys about arsenic redox speciation in groundwater, which is a major source of drinking water throughout the world (Smedley et al., 22), suggest that can represent up to 67 99% of total arsenic in groundwater (Mukherjee and Bhattacharya, 21; Bednar et al., 22). As a contaminant, is more problematic than because is more toxic and more difficult to remove from water. Commonly used arsenic treatment technologies including coagulation/filtration, ion exchange, adsorption on activated alumina, etc. usually require a preoxidation of to in order to achieve a satisfactory total arsenic removal (USEPA, 21a). Many oxidation technologies, including the addition of conventional oxidants, solar oxidation, and biological oxidation have been developed (Katsoyiannis and Zouboulis, 26). As an emerging oxidation technology for transformation to, Ti 2 photocatalytic oxidation (TP) have attracted substantial attentions as this method is environmentally benign (Yang et al., 1999; Bissen et al., 21; Lee and Choi, 22; Dutta et al., ; Ferguson et al., ; Xu et al., ). During the TP process, is oxidized by 2 or air on the surface of the light-irradiated Ti 2.Ti 2 is a widely used photocatalyst. Upon the absorption of sufficient energetic light, Ti 2 can generate valence-band holes and conductionband electrons. The valence-band holes are powerful oxidants that can convert the adsorbed to (Lee and Choi, 22). n the other hand, the conduction-band electrons can react with the adsorbed 2 or H 2 to form H 2 2 and/or reactive radicals such as H. and. 2, which are also capable of oxidizing (Lee and Choi, 22; Dutta et al., ; Xu et al., ). Because all reactions involved in oxidation through the TP occur on Ti 2 surface, the adsorption of the reacting by Ti 2 plays an important role for the overall reaction. Several studies of adsorption by Ti 2 are available, with most of them considering the influence of, initial arsenic concentration, and the presence of common anions such as phosphate and bicarbonate (Dutta et al., 24; Ferguson et al., ; Bang et al., ; Pena et al.,, 26), and relatively few considering the effect of natural organic matter (NM) on adsorption (Lee and Choi, 22). NM is a complex mixture of acidic organic molecules that originates from a variety of natural sources (e.g., soil, sediment, water, etc.) (Letterman et al., 1999). NM is ubiquitous in aquatic environment. A recent survey about levels in the groundwaters of the Bengal Delta Plain aquifers in Bangladesh suggests that the groundwaters with high concentrations tend to have high concentration of NM (Mukherjee and Bhattacharya, 21). NM has been reported to be an essential component controlling arsenic mobilization at aluminum or iron (hydro)oxide water interface (Grafe et al., 21, 22; Redman et al., 22; Ko et al., 24; Bauer and Blodau, 26). However, few studies have been conducted specifically for NM effects on arsenic adsorption by Ti 2 (Lee and Choi, 22). In studying the effect of a commercially available humic acid (HA) on oxidation rate through the TP, Lee and Choi (22) observed that the presence of HA had an insignificant effect on the adsorption by Ti 2. In their study, however, an extremely high concentration of ð5 mmþ was used, so the effect of NM on adsorption could be different for real raw waters which usually contain less than 13:3 mm of. In this study, the effect of NM on arsenic adsorption by Ti 2 in simulated -contaminated raw waters was examined. Batch adsorption experiments were performed as a function of contact time,, initial DC concentration, and initial concentration. The experiments were performed under anaerobic conditions and in darkness to avoid oxidation caused by Ti 2 photocatalysis. Five types of NM from different environmental origins were tested. Arsenic redox speciation in the filtrates after the adsorption were examined. Mechanisms behind the observed effects were discussed. 2. Materials and method 2.1. Materials and chemicals All chemicals were of analytical grade. Milli-Q water, which was supplied by the Millipore MR3 water purifier system, was used to prepare all solutions. The Milli-Q water was passed through a.2-mm membrane (Gelman FP-Verical) in order to remove the existing microbes or other reacting colloidal substances that may affect the arsenic redox speciation (Liu et al., 26). Sodium arsenite (NaAs 2, 99%) was obtained from Sigma. Stock solution (13.3 mm) was prepared by dissolving.866 g NaAs 2 into 5 ml Milli-Q water. The stock solution was stored in a high density polyethylene bottle, and was kept in darkness at 4 1C. The Ti 2 used in this study was a commercially available titanium dioxide, Degussa P (Germany). The Degussa P Ti 2 contains 8% of anatase and 2% of rutile (Hoffmann et al., 1995). It has a Brunauer Emmett Teller surface area of 55 m 2 =g and a pzc (the at the point of zero charge) of 6.7 in.1 M NaCl. A stock Ti 2 suspension was prepared by mixing 1 g of Degussa P Ti 2 with 1 L of background electrolyte (.1 M NaCl). The Ti 2 stock suspension was sonicated for 2 min each time before use in order to resuspend the precipitated Ti 2 particles. Five different types of NM were used in this study. They are Suwannee River NM (SRNM), Gohy-573 fulvic acid (GFA), Elliott soil humic acid (EHA), Pahokee peat humic acid (PHA), and Aldrich humic acid (AHA). SRNM, EHA, and PHA are wellcharacterized NMs. SRNM originated from Suwannee River of

3 WATER RESEARCH 42 (28) Georgia, USA; EHA was from prairie soils of Indiana, Illinois, and Iowa, USA; PHA was from peat soil of the Florida Everglades, USA. The three types of NM were purchased from the International Humic Substances Society. GFA is a fulvic acid extracted from the Gorleben groundwater, Germany (Kim et al., 199). AHA is a commercially available HA from Aldrich. SRNM and GFA were used without further modification. The stock solution of SRNM or GFA was prepared by dissolving the NM powder standard into.1 M NaCl to reach a concentration of 9 1 mg/l. For EHA, PHA, and AHA, the HA standard was dissolved in.1 M NaH ( 11 12) at 85 1 mg/l; the solution was then acidified to 3:8 :1 with2mhcl,and filtered through a.45-mm pore-size membrane (Gelman Supor) to remove the undissolved part of HAs; the filtrate was collected as the HA stock solution; the HA was modified in order to ensure no NM was precipitated during the arsenic adsorption on Ti 2. The dissolved organic carbon (DC) concentration in the HA stock solution was determined using a Shimazu 55 TC analyzer. The modified EHA, PHA, and AHA are, respectively, denoted as MEHA, MPHA, and MAHA in the rest of the manuscript. All the NM stock solutions were kept in glass bottles in darkness at 4 1C. Adsorption experiments were preformed within 6 weeks after the stock solutions were prepared Adsorption studies Simulated -contaminated raw waters were prepared by mixing.1 M of NaCl with various concentrations of and NM. Batch adsorption experiments were performed to determine the adsorption of arsenic by Ti 2 in these simulated raw waters. All adsorption experiments were performed in an anaerobic chamber in order to avoid oxidation. The anaerobic chamber was maintained with a purified makeup gas consisting of N 2 (95%) and H 2 (5%). All solutions together with the stock Ti 2 suspension were purged with high purity N 2 gas for at least 3 min and then immediately transferred into the anaerobic chamber. The adsorption experiments were carried out in 15-mL acid washed polyethylene bottles (amber). Adsorption of arsenic onto Ti 2 was initiated by adding. ml of Ti 2 stock suspension (1 g/l) into 14. ml of the simulated raw water, resulting in a Ti 2 concentration of.5 g/l. After the addition of Ti 2, the of the suspension was adjusted to a desired value using HCl and NaH solutions (.5.1 M). The suspension was then sealed and mixed on a rotator at room temperature for until the adsorption reached steady-state. At the end of mixing, the of the suspension was re-measured and recorded. The value thus measured was reported as the adsorption. The suspension was then filtered through a.22-mm PTEF syringe filter (Fisher), and the filtrate was immediately analyzed for the concentrations of total arsenic,,, and NM Instrument analysis The NM concentration (expressed in UV 4 ) remaining in the filtrate after adsorption was determined using a Varian Cary 3 Bio UV Visible spectrophotometer at the wavelength of 4 nm. Total arsenic concentration in solution was measured using a Perkin Elmer optima 2DV inductively coupled plasma-optical emission spectrometry (ICP-ES). A Finnigan Element 2 sector field high resolution inductively coupled plasma-mass spectrometry (ICP-MS) was used when the arsenic concentration was lower than :67 mm (5 mg=l). A medium resolution mode was chosen for the ICP-MS analysis to avoid the interference from 4 Ar 35 Cl þ dimmer. The detection limits of arsenic using the ICP-ES and the ICP-MS were determined to be mg=l (:33 mm) and 5 ng/l (6:7 1 5 mm), respectively. The instruments were calibrated each time before use. The concentrations of the calibration standards ranged from.4 to 1:7 mm for ICP-ES and from.67 to :53 mm for ICP-MS. Each sample was injected three times, and the relative standard deviation (RSD) for the triplicate analysis was within 5%. The concentrations of and remaining in solution after adsorption were simultaneously determined using ion chromatography (Dionex LC) coupled with the ICP-MS (IC ICP-MS). The detection limit for both and by the IC ICP-MS was 2:7 1 3 mm. The IC ICP-MS was calibrated with standards varying from.67 to :53 mm. The RSD for three injections of the sample with a concentration above :13 mm was usually less 7%. 3. Results and discussion 3.1. Arsenic adsorption onto Ti 2 as a function of contact time The adsorption of arsenic by Ti 2 in the simulated raw waters containing or 8 mg/l as C of NM (i.e., SRNM or MAHA) was examined as a function of time. The initial ratio of to Ti 2 was 53:4 mmol=g-ti 2 for all the waters tested. No was detected in the filtrates after adsorption in the waters either with or without NM. The experimental results are presented in Fig. 1. Without NM, the arsenic adsorption onto Ti 2 was rapid and reached steady-state within 1 h. The time required for the adsorption to reach steady-state is similar to the reported value that was obtained in the presence of air and light (Pena et al., ), although a conversion of to may occur on the Ti 2 surface with air and light present. At the steady-state, approximately 13:4 mmol=g-ti 2 or % of the total arsenic was adsorbed onto the Ti 2 surface. With the presence of 8 mg/l as C of SRNM or MAHA, the arsenic adsorption onto Ti 2 also reached steady-state within approximately 1 h. The amounts of arsenic adsorbed by Ti 2 at steady-state were 8.12 and 7:59 mmol=g-ti 2 for the waters containing SRNM and MAHA, respectively. The uptake of arsenic was decreased by 39.4% or 43.3%, respectively, due to the presence of 8 mg/l as C of SRNM or MAHA in the simulated raw water. The experimental results suggest that the tested NMs did not change the time required for the arsenic adsorption onto Ti 2 to reach steady-state, but significantly reduced the amounts of arsenic adsorbed on Ti Effect of initial DC concentration on arsenic adsorption onto Ti 2 Fig. 2 presents the effect of initial DC and concentrations on arsenic adsorption by Ti 2 in simulated raw waters

4 2312 WATER RESEARCH 42 (28) Arsenic adsorbed (µmol/g Ti 2 ) No NM With SRNM With MAHA efficiency, however, decreased from 31.% to 12.2% when the initial concentration increased from 1 to 15 mm. When the initial concentration was fixed, the amount of arsenic adsorbed by Ti 2 decreased with increasing initial DC concentration (Table 1). The decrease of arsenic adsorption with increasing DC was more rapid at low DC concentrations than at high DC concentrations. When the initial DC concentrations were below 2 mg/l as C, the reductions in arsenic adsorption were 1.4 7:5 mmol=g-ti 2 per 1 mg/l SRNM as C increased, while the reductions were.3 1:3 mmol=g-ti 2 per 1 mg/l SRNM as C increased when the initial DC concentrations approached 15 mg/l as C. The experimental results suggest that NM can be an important factor influencing adsorption on Ti 2. Arsenic adsorbed (µmol/g Ti 2 ) Time (h) Fig. 1 Adsorption of arsenic onto Ti 2 as a function of contact time in the simulated -contaminated raw waters with or without NM. Ti 2 concentration,.5 g/l; initial As (III) concentration, 2:67 lm; initial DC, or 8 mg/l as C;, 6: :1; ionic strength,.1 M NaCl; error bars show the range of data for duplicate samples. Initial con.=1 µm Initial con.=8 µm Initial con.=15 µm Initial DC concentration (mg/l as C) Fig. 2 Effect of initial DC concentration on arsenic adsorption by Ti 2.Ti 2 concentration,.5 g/l; initial concentration, 1; 8, and 15 lm;, 6: :1; ionic strength,.1 M NaCl; mixing time, 2 h; error bars show the range of data for duplicate samples. containing 14 mg/l SRNM as C and 1 15 mm. The simulated raw water and Ti 2 were mixed for 2 h to ensure the arsenic adsorption to reach steady-state. Either with or without the presence of SRNM, arsenic adsorption increased as the initial concentration increased from 1 to 15 mm, indicating that the available arsenic binding sites on Ti 2 surface have not been saturated even when the initial concentration was as high as 15 mm. The arsenic removal 3.3. Arsenic adsorption edges/envelopes The adsorption of arsenic on Ti 2 in the simulated raw waters with or without the presence of 8 mg/l as C of NM was performed as a function of. Five types of NM from different environmental origins were used. The initial and NM concentrations in the simulated raw waters were 1 mm and 8 mg/l as C, respectively. The simulated raw water and Ti 2 were mixed for 2 h to ensure the adsorption to reach steady-state. The amount of arsenic adsorbed by Ti 2 in the presence and absence of NM was plotted versus. The results are presented in Fig. 3. In the absence of NM, the amount of arsenic adsorbed by Ti 2 increased with increasing within the range of The amount of arsenic adsorbed by Ti 2 increased by 86.7% as the solution varied from 4. to 9.4. The effect of on the arsenic adsorption by Ti 2 was more significant at low values (i.e., ) than at high values (i.e., ). In the presence of 8 mg/l as C of GFA, MEHA, MPHA, or SRNM, the arsenic adsorption firstly increased with the increase of and then decreased with increasing, and the maximum arsenic adsorption occurred at , depending on the type of NM in the simulated raw water. With MAHA, the arsenic adsorption by Ti 2 increased with increasing throughout the entire range of values tested, which was similar to the arsenic adsorption behavior without NM. At the same value, the amounts of arsenic adsorbed on Ti 2 with the presence of various types of NM were generally less than that without the presence of NM. The observed effect on adsorption by Ti 2 in the absence of NM is in good agreement with the literature (Dutta et al., 24; Pena et al., ). In natural waters, both and Ti 2 can be viewed as weak acids. The proton dissociation reactions for and Ti 2 are presented in Table 2. Aqueous speciation distributions at various values can be calculated based on Reactions A1 A3 shown in Table 2. Within the range of , the primary aqueous species include the neutral H 3 As 3 and the negatively charged H 2 As 3. When is below 7., over 99% of exists as H 3 As 3. When ranges from 7. to 9.4, exists as a mixture of H 3 As 3 (99 35%) and H 2As 3 (1 65%). According to Reactions A4 A5 in Table 2, Ti 2 suspension is composed of the mixture of positively charged TiH þ 2 and neutral species of TiH when is below pzc (i.e., 6.7); when is above pzc, the primary species of the

5 WATER RESEARCH 42 (28) Table 1 Arsenic adsorption onto Ti 2 in simulated -contaminated raw waters with various initial DC and concentrations Initial concentration (mm) DC As-adsorbed DC As-adsorbed DC As-adsorbed (mg/l as C) (mmol=g-ti 2 ) (mg/l as C) (mmol=g-ti 2 ) (mg/l as C) (mmol=g-ti 2 ) 6:18 :7 26:6 1:4 5:6 7: :67 : :5 1: :6 2: :95 : :7 1: :9 2: 7.2 2:22 : :1 1: :7 2: :55 : :6 1: :1 :1 Arsenic adsorbed (µmol/g Ti 2 ) No NM With GFA With MEHA With MPHA With SRNM With MAHA an increased arsenic adsorption. The slow increasing of arsenic adsorption onto Ti 2 with the increase of at high values can be due to the formation of Ti and H 2 As 3, both of which are negatively charged and their repulsion to each other may result in a compromised adsorption. In studying the vibration spectroscopic properties of adsorbed at Ti 2 surface, Pena et al. (26) observed that the band position of the uncomplexed As shift from 795 to 78 cm 1 when moved from water to Ti 2 surface, which is indicative of the formation of As Ti bond. It should be noted that Reactions (1) (15) only represent the possible bonds that may form when is adsorbed onto Ti 2. They do not necessarily reflect the actual coordination number of Ti 2 interaction. That is, the Ti 2 association does not necessarily be monodentate mononuclear. 1. Adsorption of through an As Ti bond. Fig. 3 Effect of on arsenic adsorption by Ti 2. Ti 2 concentration,.5 g/l; initial concentration, 1 lm; initial NM concentration, 8 mg/l as C; ionic strength,.1 M NaCl; mixing time, 2 h; error bars show the range of data for duplicate samples. TiH þ 2 þ H 3As 3 () TiHAsðHÞ 3 þ H þ, (1) TiH þ H 3 As 3 () TiAsðHÞ 3 þ Hþ, (2) Ti þ H 3 As 3 () TiAsðHÞ 3, (3) TiH þ H 2 As 3 () TiAsðHÞ 2 2 þ H þ, (4) suspended Ti 2 include the neutral species of TiH and the negatively charged Ti. Therefore, the adsorption of on Ti 2 can be viewed as the interaction between H 3 As 3 and TiHþ 2 = TiH when is below 7., or the interaction between H 3 As 3 =H 2As 3 and TiH = Ti when is above 7.. Based on the above discussion, we propose the following reactions (Reactions (1) (15)) to explain the adsorption mechanisms of onto Ti 2 surface. Briefly, aqueous species adsorb onto Ti 2 surface through the formation of As Ti, As Ti, and hydrogen bonds. The fact that more was adsorbed onto Ti 2 at high values than at low values suggests that the formation of As Ti bond (Reactions (1) (5)) should be the major mechanism controlling the arsenic adsorption under the experimental conditions, as a proton that is originally associated with oxygen at Ti 2 surface should be released when the As Ti bond forms, and the increase of solution favors the association between and Ti 2, leading to Ti þ H 2 As 3 () TiAsðHÞ 2 2. (5) 2. Adsorption of through an As Ti. TiH þ 2 þ H 3As 3 () ðh 2ÞTiAsðHÞ þ 3, (6) TiH þ H 3 As 3 () ðhþtiasðhþ 3, (7) Ti þ H 3 As 3 () ðþtiasðhþ 3, (8) TiH þ H 2 As 3 () ðhþtiasðhþ 2, (9) Ti þ H 2 As 3 () ðþtiasðhþ 2 2. (1) 3. Adsorption of through a hydrogen bond. TiH þ 2 þ H 3As 3 () TiH 2HAsðHÞ þ 2, (11) TiH þ H 3 As 3 () TiHHAsðHÞ 2, (12) Ti þ H 3 As 3 () TiHAsðHÞ 2, (13)

6 2314 WATER RESEARCH 42 (28) Table 2 Deprotonation reactions for and Ti 2 in water No. Reaction pk a Ref. A1 H 3 As 3 () H 2 As 3 þ Hþ 9.23 Brown and Allison (1987) A2 H 2 As 3 () HAs2 3 þ H þ 12.1 Brown and Allison (1987) A3 HAs 2 3 () As 3 3 þ H þ Brown and Allison (1987) A4 TiH þ 2 () TiH þ Hþ o pzc Dutta et al. (24) A5 TiH () Ti þ H þ 4 pzc Dutta et al. (24) TiH þ H 2 As 3 () TiHHAsðHÞ, (14) Ti þ H 2 As 3 () TiHAsðHÞ2. (15) 5 4 GFA MEHA MPHA SRNM MAHA NM is negatively charged in natural waters. It can be associated with Ti 2 through electrostatic attraction. Theoretical and experimental studies of various organic compounds adsorption by Ti 2 indicate that the carboxyl and phenolic functional groups in NM can bind to Ti or atom on Ti 2 surface (Persson and Lunell, 2; Roddick-Lanzilotta and McQuillan, 2; Langel and Menken, 23). The following reactions ((16) (29)) are thus proposed as the possible mechanisms for NM adsorption onto Ti 2 surface. 1. Adsorption of NM through electrostatic attraction. TiH þ 2 þ NM () TiH 2 NM. (16) 2. Adsorption of NM through a hydrogen bond. UV 4 removal (%) TiH þ 2 þ NM 2C () TiH 2 2C2NM, (17) TiH þ NM 2C () TiH2C2NM, (18) TiH þ 2 þ NM 2C 6 H 4 2 () TiH 2 22H 4 C 6 2NM, (19) TiH þ NM 2C 6 H 4 2 () TiH22H 4 C 6 2NM, (2) Ti þ NM 2C 6 H 4 2 () Ti22H 4 C 6 2NM 2. (21) 3. Adsorption of NM through a Ti bond. TiH þ 2 þ NM 2C () ðh 2 ÞTiC2NM, (22) TiH þ NM 2C () ðhþtic2nm, (23) Ti þ NM 2C () ðþtic2nm 2, (24) TiH þ 2 þ NM 2C 6 H 4 2 () ðh 2 ÞTi2H 4 C 6 2NM, () TiH þ NM 2C 6 H 4 2 () ðhþti2h 4 C 6 2NM, (26) Ti þ NM 2C 6 H 4 2 () ðþti2h 4 C 6 2NM 2. (27) Fig. 4 UV 4 removal during arsenic adsorption onto Ti 2 under different values. Ti 2 concentration,.5 g/l; initial concentration, 1 lm; initial NM concentration, 8 mg/l as C; ionic strength,.1 M NaCl; mixing time, 2 h; error bars show the range of data for duplicate samples. 4. Adsorption of NM through condensation. TiH þ 2 þ NM 2CH () TiH2C2NM þ þ H 2, (28) TiH þ NM 2CH () Ti2C2NM þ H 2. (29) Reactions (1) (29) suggest that NM can compete with for available binding sites, such as Ti or atoms, on Ti 2 surface. Therefore, the presence of NM in solution can reduce the arsenic adsorption through competitive adsorption. The measurement of UV 4 remaining in solution confirms that a significant amount of NM was adsorbed on Ti 2 (Fig. 4). The adsorption of NM onto Ti 2 decreased with increasing, reflecting the anionic character of NM. If competitive adsorption is the only reason for the decreased arsenic adsorption onto Ti 2 surface, more arsenic should be adsorbed at higher values as more NM is adsorbed at lower values. Fig. 3, however, shows that with the presence of NM (i.e., SRNM, GFA, MEHA, and MPHA) the arsenic adsorption decreased with increasing within the range of , implying that mechanisms other than

7 WATER RESEARCH 42 (28) the competitive adsorption causing the decreased arsenic adsorption exist. Previous studies have shown that NM can affect adsorption onto metal (hydro)oxide through changing the speciation distribution of in water (Redman et al., 22; Ko et al., 24). In order to clarify this, we determined the arsenic speciation in solution before and after the addition of Ti 2 using an IC ICP-MS. The sum concentration of and obtained from the IC ICP-MS corresponded well with the total arsenic concentration measured directly using the ICP-MS, indicating that NM did not form any complexes with (Liu et al., 26). This ruled out the possibility of aqueous NM As complexation that may desorb from Ti 2 surface. The measurements of arsenic redox species in the filtered solution after adsorption demonstrate that part of was oxidized into under alkaline conditions, and the presence of NM facilitated the oxidation (Fig. 5). Without NM, was detected only when solution was Fig. 5 Arsenic redox speciation in the filtrates after arsenic adsorption onto Ti 2 surface in the simulated - contaminated raw waters with and without the presence of NM. Solution contains (a) no NM, (b) GFA, (c) MEHA, (d) MPHA, (e) SRNM, and (f) MAHA. Initial NM concentration, 8 mg/l as C; initial concentration, 1 lm; Ti 2 concentration,.5 g/l; ionic strength is.1 M NaCl; mixing time, 2 h; error bars show the range of data for duplicate samples.

8 2316 WATER RESEARCH 42 (28) above 8.5, and the detected concentration accounted for less than 1% of the total arsenic. With NM, however, the percentage of over total arsenic in solution was up to 88%. With GFA, MEHA, or MPHA, an appreciable amount of was detected even when the solution was as low as 5.. In the presence of NM and at high values, the arsenic adsorption onto Ti 2 actually represents for the adsorption behavior of an and mixture. It appears that the different trends of arsenic adsorption versus should also be ascribed to the NM-facilitated oxidation on Ti 2 surface besides the competitive adsorption. Because Ti 2 adsorbs less than at high values (Lee and Choi, 22; Dutta et al., 24), the increase of fraction in solution should lead to a decrease of the overall arsenic adsorption. The different adsorption behavior in the presence of different types of NM should be a mixed result of (i) different NM s capabilities in facilitating oxidation at Ti 2 surface and (ii) different NM s capabilities in competing with and for the adsorption sites on Ti 2. The slight oxidation of to by Ti 2 in the absence of air, light, and NM is probably due to the vacancies of bridging oxygen atoms on Ti 2 surface, which are caused by the adsorption through the formation of Ti As bonds (Reactions (1) (5)). The energy cost of breaking Ti bonds can be compensated by the adsorption of water molecules in the form of H 2 or H (Ménétry et al., 24). Since the adsorption of H is more energetically favored to the oxygen vacancy than the adsorption of H 2 (Schaub et al., 21; Tilocca and Selloni, 23), the oxidation of is relatively more remarkable at high than that at low. Moreover, since the standard reduction potential of the / couple decreases with increasing (Lee and Choi, 22), should be less difficult to oxidize under alkaline conditions. Despite the fact that NM may desorb from Ti 2, the association of NM with Ti atoms at the surface (Reactions (22) (27)) may weaken the bond strength of Ti in Ti As, boosting the bridging oxygen atom to escape from the surface. Therefore, NM can facilitate oxidation by Ti 2 in the absence of light and 2. Because the coordination of NM and H can inhibit the NM Ti association (Langel and Menken, 23), the effect of NM on oxidation is more significant at high than at low. The possible mechanism for oxidation to on Ti 2 surface in the presence of NM and in the absence of light and 2 are illustrated in Fig. 6. The oxidation on Ti 2 surface in the absence of 2 could also be due to the insufficient light exclusion when Ti Ti Ti H (Aqueous Ti 2 surface) Adsorption + H H As or H As - H H (Aqueous species) (Aqueous Ti 2 surface with adsorbed on it) Ti Ti Ti H As H H As H H + H + - xidation H - or RC - H As - H Ti Ti Ti H C R (Aqueous Ti 2 surface with adsorbed NM on it) + or or - As H - - As H + (Aqueous species) Fig. 6 Illustration of the possible mechanism for oxidation to on Ti 2 surface in the presence of NM and in the absence of 2 and light.

9 WATER RESEARCH 42 (28) performing the experiment (Foster et al., 1998), for the photogenerated valence-band holes in Ti 2 may oxidize the adsorbed. If light had been involved in the oxidation on Ti 2 surface, NM might have served as an electron scavenger or oxidant radical donor (e.g.,. 2 ) to facilitate the oxidation of (Yang et al., 1999; Lee and Choi, 22). The effect of NM on oxidation would then be more significant at low than that at high because more NM was adsorbed at low values (Fig. 4). This is opposite to what we have observed in our experiments (Fig. 5). Therefore, the insufficient light exclusion cannot be the major reason for the oxidation under our experimental conditions. A previous study by Lee and Choi (22) confirms that, with the presence of light and oxygen, NM facilitated the oxidation on Ti 2 surface more significantly at 3. than at 9.. Arsenic adsorbed (µmol/g Ti 2 ) No NM With GFA With MEHA With MPHA With SRNM With MAHA Arsenic equilibrium concentration (µm) Fig. 7 Arsenic adsorption isotherms with and without the presence of NM. Ti 2 concentration,.5 g/l; initial DC concentration, 8 mg/l as C;, 6: :1; ionic strength,.1 M NaCl; mixing time, 2 h; error bars show the range of data for duplicate samples. Solid and dashed lines represent the Freundlich and Langmuir modelling, respectively Arsenic adsorption isotherms The adsorption capacities of Ti 2 for arsenic in the simulated raw waters with and without the presence of various types of NM were examined (Fig. 7). Langmiur and Freundlich adsorption equations, which are the two most common adsorption models, are used to fit the experimental data. The parameters obtained by non-linear least-square regression for both Langmuir and Freundlich isotherms for arsenic adsorption with and without NM are presented in Table 3. The observed arsenic adsorption capacity of Degussa P Ti 2 in the absence of NM at 6. was 49:2 12:5 mmol=g- Ti 2, which is in reasonable agreement with the literature value of 32 mmol=g-ti 2 at 6.3 (Ferguson et al., ). The adsorption capacity of Ti 2 for arsenic decreased with the presence of 8 mg/l NM as C in solution. Both Langmuir and Freundlich isotherms fit the experimental data well. For the water free of NM, Freundlich isotherm fits the data better than Langmuir isotherm, which is indicative of a multisite adsorption of arsenic on Ti 2 surface (Dutta et al., 24). For the waters containing NM, Langmuir isotherm gives a better fit for the experimental data than Freundlich isotherm, implying that some available binding sites for arsenic were occupied by NM. The examination of adsorption dependence on the two different adsorption isotherms supports the fact that NM competes with for adsorption at the Ti 2 surface. 4. Conclusions Ti 2 is an ideal adsorbent for arsenic removal from raw water. Compared to other commonly used adsorbents such as granular activated carbon, Ti 2 has a relatively large specific surface area and has a high affinity to both and within the range of natural waters (Mohan and Pittman, 27). More importantly, light-irritated Ti 2 can accelerate the oxidation of to, which is less toxic and less difficult to remove through conventional water treatment methods (Yang et al., 1999; Bissen et al., 21; Lee and Choi, 22; Dutta et al., ; Ferguson et al., ; Xu et al., ). The oxidation efficiency and the overall arsenic Table 3 Isotherm parameters for arsenic adsorption in the simulated -contaminated raw waters with and without the presence of NM Type of NM Langmuir equation Freundlich equation (½AsŠ ad ¼ KL t½asš eq 1þK½AsŠ ) eq (½AsŠ ad ¼ K F ½AsŠ 1=n eq ) K ð1=mmþ L t (mmol=g-ti 2 ) r 2 K F n r 2 No NM : :14 49:2 12: :2 1:44 1:94 :27.98 GFA :49 :34 15:3 3: :11 1:45 2:41 :88.9 MEHA :37 :27 13:1 3: :79 1:14 2:24 :79.9 MPHA :57 :32 11:3 9: :13 1:1 2:61 :9.9 SRNM :7 :5 57:3 27:5.98 4:66 :98 1:4 :2.98 MAHA :32 :15 27:6 4: :8 1:65 1:96 :48.96

10 2318 WATER RESEARCH 42 (28) removal through the coupled TP-adsorption process can be affected by the adsorption of / onto Ti 2 surface (Ferguson et al., ). We examined the arsenic adsorption onto a commercially available Ti 2 in simulated -contaminated raw waters in the presence and absence of five types of NM that represent different environmental origins. The adsorption experiments were performed in the absence of light and oxygen. We found that NM decreased the arsenic adsorption within the tested range of The effects of NM on arsenic adsorption at low to neutral values are due to NM competition with for available binding sites on Ti 2.At high values, some was oxidized to due to the vacancies of bridging oxygen atoms on Ti 2 surface, and the presence of NM dramatically facilitated this oxidation process. Thus the observed effect of NM on the arsenic adsorption under alkaline conditions should be a mixed result of NM-facilitated oxidation and / adsorption. The investigation suggests that NM is an important element controlling arsenic speciation and adsorption on Ti 2 surface. Many previous studies have shown that Ti 2 is an effective photocatalyst to convert to with the assistance of light and 2 (Yang et al., 1999; Bissen et al., 21; Lee and Choi, 22; Dutta et al., ; Ferguson et al., ; Xu et al., ). The application of Ti 2 in practice can be limited due to the high energy cost for irritating the photocatalyst. ur study reveals that can be rapidly oxidized into by Ti 2 in NM-rich waters without light and 2, i.e., a light supplier may not be necessary for the oxidation treatment. Additionally, this finding indicates that Ti 2 may be applied as an in situ treatment technology to convert to the less toxic in groundwaters, as many -contaminated groundwaters are anaerobic in nature (Welch et al., 2) and contain pronounced amounts of NM (Mukherjee and Bhattacharya, 21). Acknowledgments This study was funded by the Notre Dame Environmental Molecule Scientific Institute through the National Science Foundation (NSF EAR ). The Ti 2 used in this study was provided as a gift by Degussa Company. Dr. Manfred Wolf (GSF Institute for Groundwater Ecology, Neuherberg) provided a ground water fulvic acid sample (Gohy-573 fulvic acid). We thank the two anonymous reviewers for their helpful suggestions. R E F E R E N C E S Bang, S., Patel, M., Lippincott, L., Meng, X.G.,. Removal of arsenic from groundwater by granular titanium dioxide adsorbent. Chemosphere 6, Bauer, M., Blodau, C., 26. Mobilization of arsenic by dissolved organic matter from iron oxides, soils and sediments. Sci. Total Environ. 354, Bednar, A., Garbarino, J., Ranville, J., Wildeman, T., 22. Preserving the distribution of inorganic arsenic species in groundwater and acid minedrainage samples. Environ. Sci. Technol. 36, Bissen, M., Frimmel, F.H., 23. Arsenic a review. Part I: ccurrence, toxicity, speciation mobility. Acta Hydrochim. Hydrobiol. 31, Bissen, M., Vieillard-Baron, M.-M., Schindelin, A.J., Frimmel, F.H., 21. Ti 2 -catalyzed photooxidation of arsenite to arsenate in aqueous samples. Chemosphere 44, 1 7. Brown, D.S., Allison, J.D., An equilibrium metal speciation model: Users Manual. Environmental Research Laboratory, ffice of Research and Development U.S. Environmental Protection Agency, Report No. EPA/6/3-87/12. Dutta, P.K., Ray, A.K., Sharma, V.K., Millero, F.J., 24. Adsorption of arsenate and arsenite on titanium dioxide suspensions. J. Colloid Interface Sci. 278, Dutta, P.K., Pehkonen, S.., Sharma, V.K., Ray, A.K.,. Photocatalytic oxidation of arsenic (III): evidence of hydroxyl radicals. Environ. Sci. Technol. 39, EC, European Commission Directive, 98/83/EC, related with drinking water quality intended for human consumption. EC, Brussel, Belgium. Ferguson, M.A., Hoffmann, M.R., Hering, J.G.,. Ti 2 -photocatalyzed oxidation in aqueous suspensions: reaction kinetics and effect of adsorption. Environ. Sci. Technol. 39, Foster, A.L., Brown, G.E., Parks, G.A., X-ray absorption pinestructure spectroscopy study of photocatalyzed, heterogeneous oxidation on kaolin and anatase. Environ. Sci. Technol. 32, Grafe, M., Eick, M.J., Grossl, P.R., 21. Adsorption of arsenate (V) and arsenite (III) on goethite in the presence and absence of dissolved organic carbon. Soil. Sci. Soc. Am. J. 65, Grafe, M., Eick, M.J., Grossl, P.R., Saunders, A.M., 22. Adsorption of arsenate and arsenite on ferrihydrite in the presence and absence of dissolved organic carbon. J. Environ. Qual. 31, Hoffmann, M.R., Martin, S.T., Choi, W.Y., Bahnemann, D.W., Environmental applications of semiconductor photocatalysis. Chem. Rev. 95, Katsoyiannis, I.A., Zouboulis, A.L., 26. Comparative evaluation of conventional and alternative methods for the removal of arsenic from contaminated groundwaters. Rev. Environ. Health 21 (1), 41. Katsoyiannis, I.A., Hug, S.J., Ammann, A., Zikoudi, A., Hatziliontos, C., 27. Arsenic speciation and uranium concentrations in drinking water supply wells in Northern Greece: correlations with redox indicative parameters and implications for groundwater treatment. Sci. Total Environ. 383, Kim, J.I., Buckau, G., Li, G.H., Duschner, H., Psarros, N., 199. Characterization of humic and fulvic acids from Gorleben groundwater. Fresenius J. Anal. Chem. 338, Ko, I., Kim, J.Y., Kim, K.W., 24. Arsenic speciation and sorption kinetics in the As-hematite-humic acid system. Colloids Surf. A 234, Langel, W., Menken, L., 23. Simulation of the interface between titanium oxide and amino acids in solution by first principles MD. Surf. Sci. 538, 1 9. Lee, H., Choi, W., 22. Photocatalytic oxidation of arsenite in Ti 2 suspension: kinetics and mechanisms. Environ. Sci. Technol. 36, Letterman, R., Amirtharajah, A., Melia, C., Coagulation and flocculation. In: Letterman, R. (Ed.), Water Quality & Treatment. McGraw-Hill Inc., New York, Liu, G.J., Zhang, X.R., Jain, J., Talley, J.W., Neal, C.R., 26. Stability of inorganic arsenic species in simulated raw waters with the presence of NM. Water Sci. Technol.: Water Supply 6,

11 WATER RESEARCH 42 (28) Mandal, B.K., Suzuki, K.T., 22. Arsenic round the world: a review. Talanta 58, Masscheleyn, P.H., Delaune, R.D., Patrick, W.H., Effect of redox potential and on arsenic speciation and solubility in a contaminated soil. Environ. Sci. Technol., Ménétry, M., Markovits, A., Minot, C., Pacchioni, G., 24. Formation of Schottky defects at the surface of Mg, Ti 2, and Sn 2 : a comparative density functional theoretical study. J. Phys. Chem. B 18, Mohan, D., Pittman Jr., C.U., 27. Arsenic removal from water/ wastewater using adsorbents a critical review. J. Hazard. Waste Mater. 142, Mukherjee, A.B., Bhattacharya, P., 21. Arsenic in groundwater in the Bengal delta plain: slow poisoning in Bangladesh. Environ. Rev. 9, Pena, M., Meng, X.G., Korfiatis, G.P., Jing, C.Y., 26. Adsorption mechanism of arsenic on nanocrystalline titanium dioxide. Environ. Sci. Technol. 4, Pena, M.E., Korfiatis, G.P., Patel, M., Lippincott, L., Ment, X.,. Adsorption of and by nanocrystalline titanium dioxide. Water Res. 39, Persson, P., Lunell, S., 2. Binding of bi-isonicotinic acid to anatase Ti 2 (1 1). Sol. Energy Mater. Sol. Cells 63, Redman, A.D., Macalady, D.L., Ahmann, D., 22. Natural organic matter affects arsenic speciation and sorption onto hematite. Environ. Sci. Technol. 36, Roddick-Lanzilotta, A.D., McQuillan, A.J., 2. An in situ infrared spectroscopic study of glutamic acid and of aspartic acid adsorbed on Ti 2 : implications for the biocompatibility of titanium. J. Colloid Interface Sci. 227, Schaub, R., Thostrup, P., Lopez, N., Lægsgaard, E., Stensgaard, I., Nørskov, J.K., Besenbacher, F., 21. xygen vacancies as active sites for water dissociation on rutile Ti 2 (1 1 ). Phys. Rev. Lett. 87, Smedley, P., Nicolli, H.B., Macdonald, D.M.J., Barros, A.J., Tullio, J.., 22. Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina. Appl. Geochem. 17, Smith, A., Lingas, E., Rahman, M., 2. Contamination of drinking-water by arsenic in Bangladesh: a public health emergency. Bull. WH 78, Tilocca, A., Selloni, A., 23. Reaction pathway and free energy barrier for defect-induced water dissociation on the (1 1) surface of Ti 2 -anatase. J. Chem. Phys. 119, USEPA, 21a. National primary drinking water regulations: arsenic and clarifications to compliance and new source contaminants monitoring. Fed. Regist. 66(14), USEPA, 21b. Fact sheet: drinking water standard for arsenic. U.S. Environmental Protection Agency, Report No. EPA 815-f--15. Welch, A.H., Westjohn, D.B., Helsel, D.R., Wanty, R.B., 2. Arsenic in ground water of the United States: occurrence and geochemistry. Ground Water 38, WH, Guidelines for Drinking-water Quality, second ed. WH, Geneva, Switzerland. Xu, T., Kamat, P.V., Shea, K.E.,. Mechanistic evaluation of arsenite oxidation in Ti 2 assisted photocatalysis. J. Phys. Chem. A 19, Yang, H., Lin, W.-Y., Rajeshwar, K., Homogeneous and heterogeneous photocatalytic reactions involving and species in aqueous media. J. Photochem. Photobiol. A 123,