The mechanism and applicability of in situ oxidation of trichloroethylene with Fenton's reagent

Fenton's reagent is the result of reaction between hydrogen peroxide (H(2)O(2)) and ferrous iron (Fe(2+)), producing the hydroxyl radical (-*OH). The hydroxyl radical is a strong oxidant capable of oxidizing various organic compounds. The mechanism of oxidizing trichloroethylene (TCE) in groundwater and soil slurries with Fenton's reagent and the feasibility of injecting Fenton's reagent into a sandy aquifer were examined with bench-scale soil column and batch experiment studies. Under batch experimental conditions and low pH values ( approximately 3), Fenton's reagent was able to oxidize 93-100% (by weight) of dissolved TCE in groundwater and 98-102% (by weight) of TCE in soil slurries. Hydrogen peroxide decomposed rapidly in the test soil medium in both batch and column experiments. Due to competition between H(2)O(2) and TCE for hydroxyl radicals in the aqueous solutions and soil slurries, the presence of TCE significantly decreased the degradation rate of H(2)O(2) and was preferentially degraded by hydroxyl radicals. In the batch experiments, Fenton's reagent was able to completely dechlorinate the aqueous-phase TCE with and without the presence of soil and no VOC intermediates or by-products were found in the oxidation process. In the soil column experiments, it was found that application of high concentrations of H(2)O(2) with addition of no Fe(2+) generated large quantities of gas in a short period of time, sparging about 70% of the dissolved TCE into the gaseous phase with little or no detectable oxidation taking place. Fenton's reagent completely oxidized the dissolved phase TCE in the soil column experiment when TCE and Fenton's regent were simultaneously fed into the column. The results of this study showed that the feasibility of injecting Fenton's reagent or H(2)O(2) as a Fenton-type oxidant into the subsurface is highly dependent on the soil oxidant demand (SOD), presence of sufficient quantities of ferrous iron in the application area, and the proximity of the injection area to the zone of high aqueous concentration of the target contaminant. Also, it was found that in situ application of H(2)O(2) could have a gas-sparging effect on the dissolved VOC in groundwater, requiring careful attention to the remedial system design.     

Two-dimensional DNAPL migration affected by groundwater flow in unconfined aquifer

The dense non-aqueous phase liquid (DNAPL) migration process was experimentally investigated in a laboratory-scale tank (150 cm width, 82.5 cm height, and 15 cm depth) to assess a site characterization on DNAPL contamination below a groundwater table. The heterogeneous ground of the tank model consisted of Toyoura sand (hydraulic conductivity, k = 1.5 x 10(-2) cm/s for void ratio, e = 0.62) and silica #7 sand (k = 2.3 x 10(-3) cm/s for e = 0.72). A series of experiments was carried out with or without lateral groundwater flow. Hydrofluoroether was used as a representative DNAPL. The main results obtained in this study are as follows: (1) the DNAPL plume does not invade into the less permeable soil layer with higher displacement pressure head; (2) the DNAPL plume migrates faster with lateral groundwater flow than without it; (3) lateral groundwater flow does not affect lateral DNAPL migration; rather, it promotes downward migration; and (4) pore DNAPL pressure without groundwater flow is higher than that with it. The above experimental results were compared with numerical analysis. The fundamental behaviors of DNAPL source migration observed experimentally are expected to be useful for assessing the characteristics of two-dimensional DNAPL migration in an aquifer.     

Force analysis and visualization of NAPL removal during surfactant-related floods in a porous medium

Governing mechanisms of dense non-aqueous phase liquid (DNAPL) removal during surfactant and surfactant-foam (SF) flooding were studied by porous-patterned glass model experiments. Physical forces, viscous forces and capillary forces, acting on trichloroethylene (TCE) blobs were quantified to understand DNAPL removal mechanisms during the floods, simultaneously visualizing the removal mechanisms. The viscous force of the remedial fluid was intimately related to TCE removal from the porous medium. The remedial fluid with a high viscous force displaced more TCE blobs. Displacement of residual TCE by the remedial fluid began as viscous pressure of flooding was closed to the capillary pressure of the porous medium. In the region of viscous pressure less than the capillary pressure, residual TCE was either retained or solubilized, not displaced, implying that TCE solubilization was the dominant TCE removal process. Glass porous model visualization validated a dominance of the capillary forces during a surfactant flush and a dominance of the viscous forces of the displacing fluid during a SF flood.     

Steel dust catalysis for Fenton-like oxidation of polychlorinated dibenzo-p-dioxins

An advanced oxidation process (AOP) for degrading toxic contaminants, specifically polychlorinated dibenzo-p-dioxins (PCDDs), was developed to utilize steel dust, a steel industry by-product, as the heterogenous catalyst for a Fenton-like oxidation. The steel dust was treated using a chemical acid etchant (HCl) and ultrasound to remove surface anchored groups, reduce aggregation, and thereby increase the specific surface areas, resulting in increased access to catalytic sites. The removal of PCDD was optimized through various reaction conditions. The removal percentage of 1,2,3,4-tetrachlorintated dibenzo-p-dioxins (1,2,3,4-TCDD, 3.1 microM) after 3 h of Fenton-like oxidation under the conditions of 3 g/L (88 mM) H(2)O(2) and pH 3 was approximately 97% with 10 g/L of steel dust, compared to approximately 99% when 5 g/L metallic iron was used as a control. When a PCDD mixture (0.5-0.7 nM) was treated, 10 g/L (92 mM) steel dust achieved approximately 88% removal, comparable to the removal with 5 g/L (89 mM) Fisher iron with 3 g/L (88 mM) H(2)O(2.) These results indicate that the steel dust is a potentially viable catalyst for removing PCDDs from contaminated water.     

Chemical oxidation of methylene blue using a Fenton-like reaction

Oxidation by Fenton-like reactions is proven and economically feasible process for destruction of a variety of hazardous pollutants in wastewater. We report herein the oxidation of methylene blue, a basic dye of thiazine series using a Fenton-like reaction at normal laboratory temperature and at atmospheric pressure. The effects of different parameters like the initial concentrations of dye, Fe2+, and H2O2, pH of the solution, reaction temperature, and added electrolytes on the oxidation of the dye present in dilute aqueous solution in the concentration range (3.13-9.39)x10(-5)mol dm(-3) (10-30 mg l(-1)) have been assessed. The results indicate that the dye can be most effectively oxidized in aqueous solution at dye:Fe(2+):H2O2 molar ratio of 1:1.15:14.1. More than 98% removal of the dye could be achieved in 1h in the pH range 2.2-2.6 at 299 K which corresponds to about 81% reduction of the initial COD. The results will be useful for designing the treatment systems of various dye-containing wastewaters.     

Photo-Fenton process as an efficient alternative to the treatment of landfill leachates

Advanced oxidation processes (AOPs), namely photo-Fenton, Fenton-like, Fenton and UV/H(2)O(2), have been investigated in the removal of organic matter and colour from landfill leachates. The leachate was characterised by high COD, low biodegradability and intense dark colour. Evaluation of COD removal as a function of the operation variables (H(2)O(2), Fe(2+), Cu(2+), UV) led to results that ranged between 30% and 77% and it was observed that the removal efficiencies decreased in the order: photo-Fenton>Fenton-like>Fenton>UV/H(2)O(2)>UV. Thus, a detailed experimental analysis was carried out to analyse the effect of the hydrogen peroxide and iron concentrations and the number of reagent additions in the photo-Fenton process, observing that: (i) the COD removal ranged from 49% to 78% depending on the H(2)O(2) dose, (ii) the total amount of organic matter removed was increased by adding the reagent in multiple steps (86%), (iii) iron concentration corresponding to a Fe(2+)/COD mass ratio=0.33 was found to be the most favourable and, (iv) after a neutralization step, the colour and residual concentrations of iron and H(2)O(2) were practically negligible in the final leachate solution.     

Application of H2O2 lifetime as an indicator of TCE Fenton-like oxidation in soils

Hydrogen peroxide decomposition and trichloroethylene (TCE) oxidation kinetics were studied through batch slurry experiments, performed on two TCE contaminated soils (a sandy soil and a clay soil), characterized by different texture and organic fraction; besides, experiments were also performed on sandy soil columns, in order to more closely reproduce the typical conditions of an in situ treatment. The results of the batch tests indicated that hydrogen peroxide lifetime was correlated to the oxidation efficiency; namely, complete TCE oxidation was achieved only for the conditions characterized by longer hydrogen peroxide lifetime, that was obtained by addition of a proper stabilizer (KH(2)PO(4)). The soil properties were also observed to influence both hydrogen peroxide decomposition and TCE oxidation kinetics, probably as a consequence of the different TOC content. The soil column experiments, performed on 10, 20, and 30 cm long columns, indicated that hydrogen peroxide decomposition, which was almost complete at 30 cm depth, was on the contrary negligible when the stabilizer was added. In agreement with this observation, the performance of TCE oxidation were greatly improved in the latter case. Based upon the collected results, it can be concluded that hydrogen peroxide experiments may be useful, at least in the first screening phase of the design activity, for selecting, among the different operating conditions, those that may be potentially more effective for the oxidation treatment.     

Waste green sands as reactive media for groundwater contaminated with trichloroethylene (TCE)

Waste green sands are byproducts of the gray iron foundry industry that consist of sand, binding agents, organic carbon, and residual iron particles. Because of their potential sorptive and reactive properties, tests were conducted to determine the feasibility of using waste green sands as a low cost reactive medium for groundwater treatment. Batch and column tests were conducted to determine the reactivity, sorptive characteristics, and transport parameters for trichloroethylene (TCE) solutions in contact with green sands. Normalized rate constants for TCE degradation in the presence of iron particles extracted from green sands were found to be comparable to those for Peerless iron, a common medium used to treat groundwater. Rate constants and partition coefficients obtained from the batch tests were found to be comparable to those from the column tests. Analytical modeling shows that reactive barriers containing green sand potentially can be used to treat contaminated groundwater containing TCE at typical concentrations observed in the field.     

Reaction efficiencies and rate constants for the goethite-catalyzed Fenton-like reaction of NAPL-form aromatic hydrocarbons and chloroethylenes

The contaminants present as nonaqueous phase liquids (NAPLs) in the subsurface are long-term sources for groundwater pollution. Fenton-like reaction catalyzed by natural iron oxides such as goethite in soils is one of the feasible in situ chemical reactions used to remediate contaminated sites. This research evaluated the Fenton-like reaction of five chlorinated ethylenes and three aromatic hydrocarbons using goethite as the catalyst. The reaction efficiencies and rate constants of these compounds in NAPL and dissolved forms were compared. The content of goethite used in batch experiments was in the range similar to those found in subsurfaces. Low H2O2 concentrations (0.05 and 0.1%) were tested in order to represent the low oxidant concentration in the outer region of treatment zone. The results showed that at the tested goethite and H2O2 ranges, the majority of contaminants were removed in the first 120 s. When aromatics and chloroethylenes were present as NAPLs, their removal efficiencies and reaction constants decreased. The removal efficiencies of 0.02 mmol NAPL contaminants were 26-70% less than those of the dissolved. The measured rate constants were in the order of 10(9) M(-1) s(-1) for dissolved chlorinated ethylenes and aromatic hydrocarbons, but were 25-60% less for their NAPL forms. The initial dosage of H2O2 and NAPL surface areas (18.4-38.2 mm2) did not significantly affect reaction efficiencies and rate constants of chlorinated NAPLs. Instead, they were related to the octanol-water partition coefficient of compounds. For both dissolved and NAPL forms, aromatic hydrocarbons were more reactive than chlorinated ethylenes in Fenton-like reaction. These results indicated that the decrease in reaction efficiencies and rate constants of NAPL-form contaminants would pose more negative impacts on the less reactive compounds such as benzene and cis 1,2-DCE during goethite-catalyzed Fenton-like reaction.     

Assessment of the generation of chlorinated byproducts upon Fenton-like oxidation of chlorophenols at different conditions

Homogeneous Fenton-like (H(2)O(2)/Fe(3+)) oxidation proved to be highly efficient in the degradation of monochlorophenols but some important issues need to be considered depending on the operating conditions. When using the stoichiometric amount of H(2)O(2) and a dose of Fe(3+) in the range of 10-20mg/L, complete breakdown of 4-CP up to CO(2) and short-chain acids was achieved. Nevertheless, when substoichiometric amounts of H(2)O(2) or lower concentrations of iron were used, significant differences between the TOC measured and the calculated from the identified species were found. These differences were attributed to condensation byproducts, including chlorinated species, formed by oxidative coupling reactions. PCBs, dioxins and dichlorodiphenyl ethers were identified. A solid residue was also formed consisting mainly in carbon, oxygen and chlorine including also Fe. The occurrence of these highly toxic species must be carefully considered in the application of Fenton oxidation to wastewaters containing chlorophenols. The possibility of reducing costs by lowering the H(2)O(2) dose below the stoichiometric one needs to take this into account.     

Petroleum oxidation using Fenton's reagent over beach sand following a spill

Removal and oxidation of petroleum adhered onto the beach sand after a spill over Guanabara Bay in Rio de Janeiro (Brazil) have been studied using Fenton's reagent (Fe2+ + H2O2). Jar tests were done on 5 and 20 g sand suspended in 200 ml aqueous solution containing iron(II) salt and hydrogen peroxide under constant stirring. The H2O2(g):Fe(g)2+ ratio varied from 0.5:1 to 50:1, pH was 2.0 and 6.0 and reaction time 1 and 3 h. Initially, the contaminated sand content of oil and grease (O&G) was 32 g/kg sand. The statistical analysis showed time and iron-sand and H2O2-iron-sand interactions to be the most significant variables, with an average O&G removal from the contaminated sand being just 30% after 3 h reaction. However, oil was removed from the sand (by up to 97%) and passed to the aqueous phase, making waste final disposal easier. The post-reaction analysis showed the supernatant to be biodegradable. Chromatographic analysis results were that the Fenton's reaction favored both the change and reduction of oil saturated and aromatic fractions.     

Application of surfactant enhanced permanganate oxidation and bidegradation of trichloroethylene in groundwater

The industrial solvent trichloroethylene (TCE) is among the most ubiquitous chlorinated solvents found in groundwater contamination. The main objectives of this study were to evaluate the feasibility of using non-ionic surfactant Simple Green™ (SG) to enhance the oxidative dechlorination of TCE by potassium permanganate (KMnO₄) employing a continuous stir batch reactor system (CSBR) and column experiments. The effect of using surfactant SG to enhance the biodegradation of TCE via aerobic cometabolism was also examined. Results from CSBR experiments revealed that combination of KMnO₄ with surfactant SG significantly enhanced contaminant removal, particularly when the surfactant SG concentrated at its CMC. TCE degradation rates ranged from 74.1% to 85.7% without addition of surfactant SG while TCE degradation rates increased to ranging from 83.8% to 96.3% with presence of 0.1 wt% SG. Furthermore, results from column experiments showed that TCE was degraded from 38.1 μM to 6.2 μM in equivalent to 83.7% of TCE oxidation during first 560 min reaction. This study has also demonstrated that the addition of surfactant SG is a feasible method to enhance bioremediation efficiency for TCE contaminated groundwater. The complete TCE degradation was detected after 75 days of incubation with both 0.01 and 0.1 wt% of surfactant SG addition. Results revealed that surfactant enhanced chemical oxidation and bioremediation technology is one of feasible approaches to clean up TCE contaminated groundwater.     

Remediation of groundwater contaminated with DNAPLs by biodegradable oil emulsion

Emulsion-based remediation with biodegradable vegetable oils was investigated as an alternative technology for the treatment of subsurface DNAPLs (dense non-aqueous phase liquids) such as TCE (trichloroethylene) and PCE (perchloroethylene). Corn and olive oil emulsions obtained by homogenization at 8000rpm for 15min were used. The emulsion droplets prepared with corn and olive oil gave a similar size distribution (1-10microm) and almost all of initially injected oil, >90%, remained in a dispersed state. In batch experiments, 2% (v/v) oil emulsion could adsorb up to 11,000ppm of TCE or 18,000ppm of PCE without creating a free phase. Results of one-dimensional column flushing studies indicated that contaminants with high aqueous solubility could be efficiently removed by flushing with vegetable oil emulsions. Removal efficiencies exceeded 98% for TCE and PCE with both corn and olive oil emulsions. The results of this study show that flushing with biodegradable oil emulsion can be used for the remediation of groundwater contaminated by DNAPLs.     

Two-stage reduction/subsequent oxidation treatment of 2,2',4,4'-tetrabromodiphenyl ether in aqueous solutions: kinetic, pathway and toxicity

The effectiveness of a two-stage reduction/subsequent oxidation (T-SRO) treatment of BDE-47, consisting of Fe-Ag reduction and Fenton-like oxidation, was investigated in this study. As an oxidation-resisting pollutant, BDE-47 (5 mg L(-1)) was difficult to be degraded by homogeneous Fe-Ag/H(2)O(2) system coupled with ultrasound (US) in 30 min. However, when this solution was firstly treated with Fe-Ag/US, the final debrominated product could be rapidly oxidized by the succeeding Fenton-like reactions, resulting in an efficient debromination of BDE-47 and a 100% mineralization of diphenyl ether (DPE). To scrutinize the degradation mechanism, several analytical techniques including HPLC, LC-MS/MS and GC/MS, were employed to monitor the major intermediates and final products. Moreover, luminescent bacteria test showed that the acute toxicity of the original solution (before reduction) was evidently lower than that of Fe-Ag/US reduction-treated solution; no toxicity was detected after Fenton-like oxidation. Evidence for the significance of a T-SRO treatment to decompose BDE-47 was presented.     

Treatment of petroleum-hydrocarbon contaminated soils using hydrogen peroxide oxidation catalyzed by waste basic oxygen furnace slag

The contamination of subsurface soils with petroleum hydrocarbons is a widespread environmental problem. The objective of this study was to evaluate the potential of applying waste basic oxygen furnace slag (BOF slag) as the catalyst to enhance the Fenton-like oxidation to remediate fuel oil or diesel contaminated soils. The studied controlling factors that affect the removal efficiency of petroleum hydrocarbons included concentrations of H₂O₂, BOF slag dosages, types of petroleum hydrocarbons (e.g., fuel oil and diesel), and types of iron mineral. Experimental results indicate that oxidation of petroleum hydrocarbon via the Fenton-like process can be enhanced with the addition of BOF slag. Results from the X-ray powder diffraction analysis reveal that the major iron type of BOF slag/sandy loam system was iron mineral (e.g., α-Fe₂O₃ and α-FeOOH). Approximately 76% and 96% of fuel oil and diesel removal were observed (initial total petroleum hydrocarbon (TPH) concentration = 10,000 mg kg⁻¹), respectively, with the addition of 15% of H₂O₂ and 100 g kg⁻¹ of BOF slag after 40 h of reaction. Because BOF slag contains extractable irons such as amorphous iron and soluble iron, it can act as an iron sink to supply iron continuously for Fenton-like oxidation. Results demonstrate that Fenton-like oxidation catalyzed by BOF slag is a potential method to be able to remediate petroleum-hydrocarbon contaminated soils efficiently and effectively.     

Application of potassium permanganate as an oxidant for in situ oxidation of trichloroethylene-contaminated groundwater: a laboratory and kinetics study

The objectives of this bench-scale study were to (1) determine the optimal operational parameters and kinetics when potassium permanganate (KMnO4) was applied to in situ oxidize and remediate trichloroethylene (TCE)-contaminated groundwater and (2) evaluate the effects of manganese dioxide (MnO2) on the efficiency of TCE oxidation. The major controlling factors in the TCE oxidation experiments included molar ratios of KMnO4 to TCE (P value) and molar ratios of Na2HPO4 to Mn2+ (D value). Results show that the second-order decay model can be used to describe the oxidation when P value was less than 20, and the observed TCE decay rate was 0.8M(-1)s(-1). Results also reveal that (1) higher P value corresponded with higher TCE oxidation rate under the same initial TCE concentration condition and (2) higher TCE concentration corresponded with higher TCE oxidation rate under the same P value condition. Results reveal that significant MnO2 production and inhibition of TCE oxidation were not observed under acidic (pH 2.1) or slightly acidic conditions (pH 6.3). However, significant reduction of KMnO(4) to MnO2 would occur under alkaline condition (pH 12.5), and this caused the decrease in TCE oxidation rate. Results from the MnO2 production experiments show that MnO2 was produced from three major routes: (1) oxidation of TCE by KMnO4, (2) further oxidation of Mn2+, which was produced during the oxidation of TCE by KMnO4, and (3) reduction of MnO4(-1) to MnO2 under alkaline conditions. Up to 81.5% of MnO2 production can be effectively inhibited with the addition of Na2HPO4. Moreover, the addition of Na2HPO4 would not decrease the TCE oxidation rate.     

Degradation of trichloroethylene by Fenton reaction in pyrite suspension

Degradation of trichloroethylene (TCE) by Fenton reaction in pyrite suspension was investigated in a closed batch system under various experimental conditions. TCE was oxidatively degraded by OH in the pyrite Fenton system and its degradation kinetics was significantly enhanced by the catalysis of pyrite to form OH by decomposing H(2)O(2). In contrast to an ordinary classic Fenton reaction showing a second-order kinetics, the oxidative degradation of TCE by the pyrite Fenton reaction was properly fitted by a pseudo-first-order rate law. Degradation kinetics of TCE in the pyrite Fenton reaction was significantly influenced by concentrations of pyrite and H(2)O(2) and initial suspension pH. Kinetic rate constant of TCE increased proportionally (0.0030 ± 0.0001-0.1910 ± 0.0078 min(-1)) as the pyrite concentration increased 0.21-12.82 g/L. TCE removal was more than 97%, once H(2)O(2) addition exceeded 125 mM at initial pH 3. The kinetic rate constant also increased (0.0160 ± 0.005-0.0516 ± 0.0029 min(-1)) as H(2)O(2) concentration increased 21-251 mM, however its increase showed a saturation pattern. The kinetic rate constant decreased (0.0516 ± 0.0029-0.0079 ± 0.0021 min(-1)) as initial suspension pH increased 3-11. We did not observe any significant effect of TCE concentration on the degradation kinetics of TCE in the pyrite Fenton reaction as TCE concentration increased.     

Persulfate regeneration of trichloroethylene spent activated carbon

The objective of this study was to demonstrate the regeneration of trichloroethylene (TCE) spent activated carbon using persulfate oxidation and iron activated persulfate (IAP) oxidation. Both processes resulted in decreases in the adsorbability of regenerated activated carbons. IAP was shown to rapidly degrade the aqueous TCE and causes a significant mineralization of the TCE. The release of chloride ions provided evidence of this. Persulfate oxidation mainly resulted in desorption of TCE from the activated carbon and only partial mineralization of the TCE through a carbon activated persulfate reaction mechanism. Concerning destruction of the TCE, in the regeneration test using persulfate, 30% of the original TCE was present in the solution and 9% remained on the activated carbon after the first regeneration cycle. In contrast, in the test that used IAP, it was observed that no TCE was present in the solution and only approximately 5% of the original TCE remained on the activated carbon after the first regeneration. Following the regeneration cycles, elemental analysis was carried out on the samples. BET surface area and EDS analysis showed some effects on the physico-chemical properties of the activated carbon such as a slight decrease in the surface area and the presence of iron precipitates on the carbon.     

Hydrogen peroxide lifetime as an indicator of the efficiency of 3-chlorophenol Fenton's and Fenton-like oxidation in soils

In this work the possibility of using the hydrogen peroxide lifetime as indicator of the oxidation efficiency of Fenton's and Fenton-like processes for soil treatment was explored. A reactivity scale, in terms of the oxidizing power in the different tested operating conditions (pH, iron sulfate concentration and stabilizer concentration) was built for each soil as a function of the hydrogen peroxide lifetime. Its validity was then confirmed through 3-chlorophenol Fenton's and Fenton-like slurry-phase oxidation experiments. The proposed reactivity scale proved to be effective for comparing the different operating conditions for the same soil, but failed when used to compare the oxidation performances for different soils, since the different adsorptive behavior of the tested soils may have influenced the contaminant removal rate.     

Removal of arsenic from water by zero-valent iron

Batch and column experiments were conducted to investigate the effect of dissolved oxygen (DO) and pH on arsenic removal with zero-valent iron [Fe(0)]. Arsenic removal was dramatically affected by the DO content and the pH of the solution. Under oxic conditions, arsenate [As(V)] removal by Fe(0) filings was faster than arsenite [As(III)]. Greater than 99.8% of the As(V) was removed whereas 82.6% of the As(III) was removed at pH 6 after 9h of mixing. When the solution was purged with nitrogen gas to remove DO, less than 10% of the As(III) and As(V) was removed. High DO content and low solution pH also increased the rate of iron corrosion. The removal of arsenic by Fe(0) was attributed to adsorption by iron hydroxides generated from the oxic corrosion of Fe(0). The column results indicated that a filtration system consisting of an iron column and a sand filter could be used for treatment of arsenic in drinking water.