In situ treatment of arsenic-contaminated groundwater by air sparging

Journal of Contaminant Hydrology 159 (2014) 20–35

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology journal homepage: www.elsevier.com/locate/jconhyd

In situ treatment of arsenic-contaminated groundwater by air sparging Joseph H. Brunsting a,⁎, Edward A. McBean b a b

School of Engineering, University of Guelph, 50 Stone Road East, Guelph, Ontario N1G 2W1, Canada School of Engineering, University of Guelph, Canada Research Chair of Water Supply Security

a r t i c l e

i n f o

Article history: Received 31 December 2012 Received in revised form 22 December 2013 Accepted 7 January 2014 Available online 7 February 2014 Keywords: Arsenic Groundwater In situ treatment Air sparging Bangladesh

a b s t r a c t Arsenic contamination of groundwater is a major problem in some areas of the world, particularly in West Bengal (India) and Bangladesh where it is caused by reducing conditions in the aquifer. In situ treatment, if it can be proven as operationally feasible, has the potential to capture some advantages over other treatment methods by being fairly simple, not using chemicals, and not necessitating disposal of arsenic-rich wastes. In this study, the potential for in situ treatment by injection of compressed air directly into the aquifer (i.e. air sparging) is assessed. An experimental apparatus was constructed to simulate conditions of arsenic-rich groundwater under anaerobic conditions, and in situ treatment by air sparging was employed. Arsenic (up to 200 μg/L) was removed to a maximum of 79% (at a local point in the apparatus) using a solution with dissolved iron and arsenic only. A static “jar” test revealed arsenic removal by co-precipitation with iron at a molar ratio of approximately 2 (iron/arsenic). This is encouraging since groundwater with relatively high amounts of dissolved iron (as compared to arsenic) therefore has a large theoretical treatment capacity for arsenic. Iron oxidation was significantly retarded at pH values below neutral. In terms of operation, analysis of experimental results shows that periodic air sparging may be feasible. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Arsenic, widely acknowledged as biologically harmful, is a contaminant in groundwater in many areas of the world, including Cambodia, Argentina, Chile, China, Hungary, Laos, Mexico, Mongolia, Nepal, Pakistan, Taiwan, Thailand, Vietnam, and the USA (Ahmed, 2003). However, the most widespread and serious groundwater arsenic levels are evident in West Bengal (India) and Bangladesh. The Bangladeshi arsenic problem has been described as “the largest poisoning of a population in history, with millions of people exposed” (Smith et al., 2000, pg. 1093).

Abbreviations: DI, deionized; DO, dissolved oxygen; HFO, hydrous ferric oxide; HRT, hydraulic retention time; ORP, oxidation–reduction potential; RmV, relative milli-volts. ⁎ Corresponding author. Tel.: +1 519 362 6429 (cell). E-mail address: [email protected] (J.H. Brunsting). 0169-7722/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jconhyd.2014.01.003

The current World Health Organization guideline for inorganic arsenic in drinking water is 10 μg/L (WHO, 2008). However, not all jurisdictions follow this guideline, including Bangladesh and India, where a guideline of 50 μg/L is used (Chakraborti et al., 2009; Flanagan et al., 2012; WQAA Government of India). Chronic arsenic exposure may result in severe health effects with skin lesions, hyperkeratosis, and increased risk of cancers. Although anthropogenic sources of arsenic exist (e.g. smelting operations), the most widespread problems are of natural geochemical origin. Groundwater arsenic concentrations reported in the literature range from b 0.5 μg/L to 5000 μg/L under natural conditions (Smedley and Kinniburgh, 2002). Oxides of iron, aluminum, and manganese are likely the most important sources and sinks for arsenic in aquifer sediments (Stollenwerk, 2003). Arsenic may be mobilized from soil as a result of reducing conditions in groundwater, as occurs in West Bengal and Bangladesh. The reducing conditions are the result of oxidation

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

of buried organic matter in sediments. This causes mobilization of arsenic by reduction of iron oxyhydroxides, reductive dissolution, and change in structure of iron oxide minerals (BGS and DPHE, 2001; Nickson et al., 1998; Smedley and Kinniburgh, 2002). Treatment of arsenic-impacted groundwater is feasible, and many options exist including oxidation and sedimentation, coagulation and filtration, sorption techniques, and membrane techniques. While being capable of effectively removing arsenic, many treatment technologies have drawbacks due to requirements for use of chemicals, disposal of arsenic-rich wastes, and/or technological complexity. As an alternative option, in situ treatment (by creation of oxidizing conditions) is a fairly simple procedure, does not require chemicals, and does not require disposal of arsenic-rich wastes. The basis for in situ treatment of arsenic is the same as that for in situ treatment of iron and manganese, which has been utilized in Europe for decades. Oxidation–reduction potential (ORP) conditions can be changed in the subsurface by introducing dissolved oxygen (DO), causing oxidation of ferrous iron and other metals (in solution and on the soil grains). This process creates ferric iron oxyhydroxides capable of adsorbing ferrous iron and other oxyanions such as arsenic (Appelo and de Vet, 2003; Rott, 1985; Rott and Lamberth, 1993; van Beek, 1985; van Halem et al., 2010a). These oxidation processes are enhanced by autocatalytic effects from oxidation products (Rott and Friedle, 2000; Sung and Morgan, 1980; Tamura et al., 1980). One method of in situ treatment for arsenic in groundwater involves injection of aerated water into the aquifer. This method has met with moderate success to date (Rott and Friedle, 1999, 2000; Sen Gupta et al., 2009; van Halem et al., 2010a, 2010b, 2010c). One possible concern regarding in situ treatment is the possibility that pore spaces in the aquifer may become clogged. However, this is not a significant problem in reality. Subsurface treatment for iron in groundwater (by method of injection of aerated water) has been used in Europe, and clogging has not been found to be an issue, even after more than a decade of operation (van Halem et al., 2011). Iron may initially precipitate as hydrous ferric oxide (HFO, of low crystallinity), but in the subsurface it ages and changes to thermodynamically more stable and less voluminous crystallized forms such as goethite (Mettler,

21

2002; Rott and Friedle, 2000; Smedley and Kinniburgh, 2002; Stollenwerk, 2003), and this prevents clogging. Another option for in situ treatment of arsenic in groundwater is direct air sparging in the saturated zone. However, besides literature regarding treatment on a deep well (Miller, 2006, 2008) as well as clean-up from a lead smelter site (Miller et al., 2002), there appears to have been limited research on this option. Grombach (1985) suggests that introduction of air directly into the aquifer is the easiest method but did not undertake experimental observations. The investigation described herein is a lab-scale study to investigate the potential of in situ treatment of arseniccontaminated water by air sparging. As described below, the experiments utilized a simple solution of dissolved inorganic arsenic and dissolved ferrous iron in order to demonstrate the concept, potential, and key factors of this type of treatment, as a precursor to possible field trials. 2. Materials and methods The experimental apparatus is illustrated in Fig. 1. Dimensions and sampling port labels are illustrated in Fig. 2. A large, sealable, food-grade polyethylene barrel (Fig. 1) was used for the inlet solution, which was subjected to nitrogen sparging to remove dissolved oxygen. The main apparatus was made of plexiglass. A small precision-flow peristaltic pump (Fisher model CON3386) was used to transport solution from the reservoir barrel to the inlet column. Rotameters of appropriate size were used for measuring gas flow. The apparatus had both an inlet and an outlet reservoir, fitted with a fine stainless steel mesh to retain the sand medium. A uniformly graded (rounded, nominal size 0.40 mm) sand was rinsed with deionized (DI) water and used as aquifer soil for purposes of the laboratory simulation. In chemical composition, the sand was as follows: SiO2 N99.5%; TiO2 ~0.10%; K2O ~0.10%; CaO ~0.03%; Fe2O3 ~0.03%; Al2O3 ~0.01%; Loss on ignition ~0.12% The outlet height was used to adjust the hydraulic gradient. Air sparging was accomplished using an aquarium aeration pump (Hagen, model Maxima-R) with tubing attached to a rotameter. The outlet from the rotameter ran to a small (2.5 cm) alumina diffuser stone (Fisherbrand model ME46944C) at the sparging point in the apparatus, as shown in Fig. 1.

air lock

Apparatus Concept peristaltic pump P.P.

Water le level el

220 L barrel

large pond diffuser

N2 gas

small diffuser stainless steel mesh

effluent

air diffuser air flow

Fig. 1. Apparatus conceptual design.

sampling ports

stainless steel mesh

22

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

200 vent

10

4 3.3 Water level

44

In

22

Expected design head loss head loss

A1

B1

C1

D1

E1

F1

G1

H1

I1

J1

A2

B2

C2

D2

E2

F2

G2

H2

I2

J2

A3

B3

C3

D3

E3

F3

G3

H3

I3

J3

50

Out

40.7 20.3

Width (i.e. into the drawing) = 5 cm

40 Fig. 2. Apparatus dimensions (cm) and sample port labels.

Experiments were completed as follows: – Experiment 1: to examine the dispersion of oxygen within the porous medium while sparging with air, and pumping plain DI water through the apparatus – Experiment 2: to examine the removal of arsenic with no dissolved iron present – Experiment 3: to examine the removal of arsenic with dissolved iron present – Experiment 4: a static jar test, with the same solution constituents as experiment 3 – Experiment 5: similar to experiment 3, but over a much longer duration and with a slightly higher air injection rate. Experiments were prepared by first flooding the drained apparatus with nitrogen while it was empty of water. Nitrogen was injected at a flow rate of 1 L/min, through the sparging point, over a period of 24 h. Solution was then introduced, by pumping a solution of 1000 μg/L of arsenic solution (50% As[III], 50% As[V] in experiment 2, and 100% As [V] in other experiments, explanation to follow in Results). This was either with or without 5.3 mg/L Fe(II), depending on the experiment. It was found that, while pumping a solution of 200 μg/L arsenic, breakthrough of arsenic took a prohibitive amount of time. Thus, the initial solution of 1000 μg/L was introduced to more quickly saturate available adsorption sites on the sand medium. Once arsenic was detected at above 200 μg/L at all sampling ports, the solution was switched to 200 μg/L and pumped until all ports showed 200 μg/L or less. There was no replacement with fresh sand between experiments (i.e. the same sand media was used throughout all experiments in this investigation). Two hundred μg/L arsenic and 5.3 mg/L iron were chosen as experimental concentrations. According to calculations by Roberts et al. (2004) using data from a groundwater survey, these are mean concentrations for Bangladeshi groundwater with an arsenic concentration of 50 μg/L or greater. Experimental solutions were prepared first by lowering DO to less than 0.05 mg/L using nitrogen gas sparging. Constituents were then added using granular ferrous chloride tetrahydrate, sodium arsenite solution, and sodium arsenate solution.

These specific chemicals were used to represent the chemical species present in solution in a reduced groundwater environment (i.e. ferrous iron, and a mix of both inorganic trivalent and inorganic pentavalent arsenic). Neither sodium nor chloride is cited as a major ion affecting arsenic adsorption (Stollenwerk, 2003), nor are they correlated with arsenic in Bangladeshi groundwater (Anawar et al., 2003). Hence, the sodium compounds of arsenic and the chloride compound of ferrous iron were selected as chemicals for use in this study, with the assumption that the effects of the associated sodium and chloride ions would be minimal. In Bangladeshi groundwater, the ratio of As(III)/[As(III) + As(V)] = As(III)/As(total) ranges from less than 0.1 to greater than 0.9, but averages around 50% to 60% (BGS and DPHE, 2001; Paul et al., 2008; Rasul et al., 2002). A ratio of 50% was used in initial experimentation, as explained in Section 3. Measurement of iron was accomplished using a Varian 220 SpectrAA flame atomic absorption spectrometer. Measurement of arsenic was done using either a Varian 880 SpectrAA graphite furnace atomic absorption spectrometer or a Shimadzu AA-6300 graphite furnace atomic absorption spectrometer. Samples for analysis of metals were taken using a new BD luer-lok 60 mL syringe for each sample. Metals samples were preserved using trace-metals grade nitric acid (adjusting to pH less than 2), and refrigerated at 4 °C. Analyses were done within one week of sampling. To ensure accuracy and precision, duplicate samples were taken with each sample set and assessed. A new calibration was performed with prepared standards for every sample run, and mid-point standard checks were also performed with each sample run. Duplicate samples and mid-point standard checks were assessed for repeatability within 10% for iron and 20% for arsenic. Dissolved oxygen measurements were completed using an Orion 083005MD membrane probe attached to an Orion 4 Star pH-DO portable meter. For dissolved oxygen samples from the apparatus, samples were taken using a 60 mL syringe with a 3-way stopcock. After withdrawal of the sample from the apparatus, the stopcock was set to seal the

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

23

Dissolved Oxygen (mg/L)

A & B Column DO 7 6 5

B1

4

B2

3

B3

2

A2

1 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Dissolved Oxygen (mg/L)

D Column DO 7 6 5 4

D1

3

D2

2

D3

1 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Dissolved Oxygen (mg/L)

G Column DO 7 6 5

G1

4

G2

3

G3

2 1 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Dissolved Oxygen (mg/L)

J Column DO 7 6 5

J1

4

J2

3

J3

2 1 0 0

5

10

15

20

25

30

Time (hours) Fig. 3. Results of experiment 1.

35

40

45

50

24

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

tip, and then the plunger was removed using gentle force against the vacuum. The probe was then used to gently stir the sample in the syringe while taking a reading. pH measurements were taken using an Orion 8102BNUWP probe attached to an Orion 4 Star pH-ISE Benchtop meter. For ORP readings, an Orion 9179BNMD epoxy body gel-filled ORP triode was used, attached to an Orion 4 Star pH-ISE Benchtop meter (since this pH meter is also capable of ORP readings). Speciation for As(V) and As(III) was done using a field method developed by Clifford et al. (2004). This relies on lowering of pH using acetic acid and complexing of iron with ethylenediaminetetraacetic acid (EDTA) to remove interference of iron, and capturing As(V) with a chloride-form resin in glass mini-columns, hence allowing As(III) to pass through. In terms of iron speciation, iron can be considered to be in Fe(II) (ferrous) dissolved form until oxidation and precipitation, at which point it converts to Fe(III) (ferric) form as iron(III) oxide-hydroxide. This appears as rusty brown precipitate.

As(V)

As(III)

3. Results 3.1. Experiment 1 — to examine the dispersion of DO In experiment 1, solution flow rate was 16 mL/min. Based on measured porosity of the sand of 0.35, hydraulic retention time (HRT) was calculated as 16 h (assuming uniform flow). The airflow sparging rate was 16 ± 5 mL/min. Results are shown in Fig. 3, organized by column of sampling ports. Port A2 is located up-gradient from sparging, and represents inlet conditions. As shown in Fig. 3, the DO concentration near the inlet remained below 0.6 mg/L throughout the experiment. Above the sparging point, the DO concentration rose to between 4 and 5 mg/L, with the lowest sampling port (B3) directly above the sparging point showing the quickest rise. The air traveled in bubbles, creating airflow channels. Over the extent of vertical rise, these airflow channels migrated approximately 20 cm both up-gradient and down-gradient of the sparging point, forming a “V” pattern. At the “D” column, DO concentrations appeared to rise slightly faster at D3, but effective transfer of oxygen to ports D1 and D2 was observed as the DO levels were 5 mg/L by the experiment's end. Concentrations of DO rose to approximately 5 mg/L along the bottom at sampling ports G3 and J3. However, there was not effective transfer of oxygen to migrated water sampled at ports G1, G2, J1, or J2. This indicates the existence of a region of lower velocity flow in the apparatus, as flow velocity was insufficient to transfer water from the sparging point to sampling ports G1, G2, J1, or J2 by the end of the experiment. Non-uniform grading of sand in the apparatus likely caused the zone of low-velocity flow referred to above, approaching ports G1, G2, J1, and J2. During experimental set-up, sand was first rinsed in a bucket, and then was deposited in the apparatus under submersion in DI water with a continuous flow of water from the inlet towards the outlet. Some residual dust was observed as cloudiness in the water during deposition of the sand. This dust and fine sand particles were likely deposited in the upper part of the

Fig. 4. Range of ORP and pH of experiments, plotted on an Eh–pH diagram for aqueous arsenic species in the system As–O2–H2O at 25 °C and 1 bar total pressure (Smedley and Kinniburgh, 2002, pg. 521).

apparatus near the outlet end because of the water flow maintained during application of sand in the apparatus. 3.2. Experiment 2 — to assess arsenic removal with no iron Characteristics and operating conditions of Experiment 2 were as follows: – Total water flow volume of 42.5 L over 48.8 h, so average flow of 14.3 mL/min – HRT calculated as 18 h – Initial pH 7.2 at the inlet and 5.7 at the outlet. This drop in pH through the apparatus was unexpected. It is possible that oxidation of As(III) under unstable conditions (see Fig. 4) caused release of H+, resulting in a mildly acidic environment in an unbuffered solution (i.e. DI water). – Air sparging flow rate at 16 ± 5 mL/min – Inlet solution containing 100 μg/L As(III) and 100 μg/L As(V) to examine behavior of arsenic species. Air sparging commenced immediately at the start of the experiment. In the inlet sampling (for the first two samplings at 0 h and 9 h), the As(III) was measured at 95 and 107 μg/L, respectively, and the As(V) was measured as 109 and 135 μg/L, respectively. This shows a balance between As(III) and As(V) Table 1 Relationship between DO and ORP in DI water with 5.3 mg/L iron and 200 μg/L arsenic. DO (mg/L)

ORP (RmV)a

0.41 7.67

266 340

a RmV refers to “relative milli-volts”, the milli-volt reading relative to the standard hydrogen electrode.

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

was apparently found to be present in the inlet but not at sample points within the sand matrix. The rest of the summary of experiment 2 deals only with As(total), taken as As(V). The DO levels at each port are shown in Fig. 5, showing concentrations at the beginning and end of air sparging. As shown, there was distinct oxygen transfer to ports A1, B1, B2, B3, E1, E2, E3, J3, and the outlet. It appears that the airflow channel (from sparging) migrated sufficiently to transfer some oxygen to port A1. In this experiment, DO acts as a tracer, indicating which areas of the apparatus received water flow from the sparging area during the experiment. As such, it is clear that J1 and J2 did not receive water from the sparging area by the end of the experiment, due to the low flow zone described previously. In addition, measurements taken prior to the start of sparging indicated that the concentrations of arsenic at J1 and J2 were lower than at other ports at the start of sparging (i.e. this area had a lower “saturation” of available adsorption sites), and as such may have exhibited biased results. J1 and J2 are therefore omitted in analyses to follow. Results for arsenic are summarized by “inlet and outlet” and by row of sampling ports (i.e. top row 1, middle row 2, and bottom row 3) in Fig. 6. Samples were taken at these points in order to examine spatial heterogeneity both upstream and downstream of air sparging. Statistical paired t-tests were completed, the results of which indicate greater than 90% certainty that arsenic concentrations decreased across the apparatus at all sampling ports. Although iron was not present in solution in this experiment, some metal oxides were nonetheless present in the sand. These could be responsible for some adsorption of arsenic, with subsequent regeneration of adsorption sites upon oxidation by DO. Summarized results for ports downstream of sparging are shown in Table 3. From examination of Table 3, it appears that the ports along the bottom flow path (i.e. E3 and J3) show lower removal values for arsenic than the ports at E1 and E2. Ports E1 and E2 were likely along a slower flow path (approaching the lower-flow zone) and as such the water samples at these ports had a higher contact time with the sand than corresponding samples along the bottom flow path. This suggests that removal of arsenic is

Table 2 Relationship between DO and ORP during air sparging trials in the apparatus (using only DI water). Sampling port

DO (mg/L)

ORP (RmV)

Barrel Inlet column Inlet Outlet

0.07 1.05 2.28 4.16

332 450 460 545

25

(approximately half each), with a total arsenic value of approximately 200 μg/L being achieved. However, speciation tests for samples taken from within the sand matrix and at the outlet (i.e. all sampling ports except the inlet) showed the large majority at zero As(III), with a few scattered tests showing up to 50 μg/L As(III). It was not expected that As(III) would be so readily oxidized, in particular because the DO was low in the system at the start of the experiment. It is apparent that As(III) was not stable under the conditions achieved in this experiment. During pre-experiment trials, and during experiment 1, ORP readings were taken under various conditions as per Tables 1 and 2. ORP is affected by DO levels, but is also affected by many additional parameters such as pH, temperature, and small reaction currents. Based on these results, one limitation of this experimental setup was that reducing conditions (encountered in Bangladeshi groundwater high in arsenic) could not be achieved by just deaerating DI water. As such, the approach was to control DO concentrations in the solution and apparatus as closely as possible. The approximate pH and ORP range encountered during all experiments in this investigation is plotted in Fig. 4. As per Fig. 4 under conditions of these experiments, As(III) was unstable. As(III) is typically very slow to oxidize in a simple solution of water with DO, on the order of months (Cherry et al., 1979), but the rate of transformation of As(III) to As(V) increases rapidly with manganese- and iron-oxides/-hydroxides (Rott and Friedle, 2000). Metal oxides are present in the sand matrix, and this catalyst effect may explain why As(III)

DO Levels in the Apparatus During Experiment 2 5

DO (mg/L)

4 3

Start 49 hours end

2 1 0 In

A1

A2

A3

B1

B2

B3

E1

E2

E3

J1

J2

J3

Sampling Port Fig. 5. DO levels in the apparatus during experiment 2.

Out

26

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

Inlet and Outlet - As(total) 250

As (µg/L)

200 150

In

100

Out

50 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Top Sample Ports - As(total) 250

As (µg/L)

200

A1

150

B1 100

E1

50 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Middle Sample Ports - As(total) 250

As (µg/L)

200

A2

150

B2 100

E2 50 0 0

5

10

15

20

25

30

35

40

45

50

Time (hours)

Bottom Sample Ports - As(total) 250 200

As (µg/L)

A3 150

B3 E3

100

J3

50 0 0

5

10

15

20

25

30

35

Time (hours) Fig. 6. Results of experiment 2 for arsenic.

40

45

50

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35 Table 3 Calculated percent removals of arsenic for ports downstream of air sparging that exhibited heightened DO by the end of experiment 2. Mean inlet arsenic conc. Port End arsenic (μg/L) (μg/L) 226

E1 E2 E3 J3 Out

155 165 187 190 152

conc.a Removalb (%) 31 27 17 16 34

a

End of air sparging. Calculated for the average inlet concentration compared to the concentrations at sampling ports at the end of air sparging. b

dependent on contact time under these conditions, with removal at E1 and E2 in the range of 30%. Measurements at the outlet show the highest removal; however, the result here was likely affected by solution flow from J1 and J2, which were omitted from analysis for reasons stated previously. 3.3. Experiment 3 — removal of arsenic with iron present Characteristics and operating conditions of experiment 3 were as follows: – Mean flow of 19.3 mL/min of water – HRT calculated as 13 h – Initial pH of 5.8 at the inlet and 5.1 at the outlet. The initial low pH was due to hydrolysis of the ferrous chloride upon dissolution in water, and the pH dropped further across the apparatus as in experiment 2. – Initial DO readings at the inlet and outlet of 0.32 mg/L and 0.53 mg/L respectively – Air sparging flow rate at 16 ± 5 mL/min. – Experiment run for 48 h with air sparging, followed by 24 h without air sparging – Inflow conditions with approximately 200 μg/L As(V) (used because As[III] was found to be quickly oxidized) and approximately 5 mg/L iron as Fe(II). Experimental results for iron are summarized in Fig. 7. A highlighted data point on each graph marks the end of air sparging. Fig. 7 shows that the concentration of iron varied somewhat at the inlet for the first 30 h of the experiment (between 4 and 5 mg/L) but the outlet concentration of iron was, with the exception of one measurement point, below the inlet concentration. It is unknown why one measurement period differs from the dominant trend; however, a paired t-test was done to consider all measurements as whole (i.e. all sample times at the inlet and outlet). This showed 97.5% certainty that there was removal of iron between the inlet and outlet a level greater than zero. Examination of ports A2, B2, and E2 demonstrate the effect of air sparging, since following the start of air sparging, the concentrations at B2 and E2 decreased slightly from levels at A2. The removal of iron is low, particularly since iron is typically easily oxidizable (Mettler, 2002), but the low pH (between 5.1 and 5.8 from the inlet to the outlet) shows that the oxidation rate of Fe(II) is highly dependent on pH. The oxidation rate of ferrous iron is very significantly slowed at pH values below 7 (Morgan and Lahav, 2007), and hence, the

27

iron removal in experiment 3 would be greatly increased if the pH had been closer to neutral. Buffering with sodium hydroxide was attempted, but this caused precipitation of iron and as such was not used in experimentation. Although the rate of iron oxidation was quite slow, this slow rate did allow observation of the relationship between removal of iron and arsenic in small increments. Examination of this is experimentally valuable in predicting removal of arsenic where greater precipitation of iron is expected. The concentrations of arsenic in the apparatus throughout experiment 3 are shown in Fig. 8. The highlighted data points mark the end of air sparging. As illustrated in Fig. 8, the arsenic concentration at the outlet was considerably lower than that of the inlet. The average difference in iron concentrations between the inlet and outlet over the first two readings (i.e. before the end of the first HRT, as an estimate of the background treatment effect) indicates a removal of approximately 0.4 mg/L. It is apparent that even a fairly small drop in iron concentration (i.e. 9%, 0.4 mg/L drop from an inlet concentration of 4.3 mg/L, averaged over the first two data points) leads to a relatively large removal effect on arsenic, in this case a drop of 76 μg/L (i.e. 42% drop from an inlet concentration of 180 μg/L, averaged over the first two data points). From Fig. 8, it follows that arsenic was in steady decline at the outlet throughout the entire duration of air sparging in Experiment 3, from an initial value of 109 μg/L down to the lowest value of 41 μg/L at the end of air sparging. This downward trend is also seen for ports B1, E1, E2, E3, and J3. The removal percentages for various sampling ports exposed to extra DO through air sparging, based on the end-of-sparging arsenic concentrations as compared to the average inlet arsenic concentration over the course of air sparging (i.e. 176 μg/L), are shown in Table 4. The higher removal values shown here, as compared to those from experiment 2, show the importance of the presence of dissolved iron for the removal of arsenic. At the outlet, estimated removal was improved from 34% to 77% from experiment 2 to experiment 3. Recovery of arsenic concentrations post-sparging is difficult to assess in this experiment, due to the apparent spike in arsenic at ports A1, A2, and A3 after air sparging ended. The reason for this spike is unknown, but it is apparent that limiting recovery to one day was not nearly sufficient to see arsenic recover to pre-sparging levels. This is illustrated in Fig. 8 as ports E3 and J3 only recovered to 67 μg/L and 25 μg/L respectively after a 24-hour recovery period, as compared to 114 μg/L and 98 μg/L respectively at the beginning of the experiment. This “lag effect” for recovery suggests that applying sparging intermittently may be feasible. Experiments were also carried out with sparging rates of 75 and 250 mL/min, and actually resulted in less effective transfer of oxygen to solution. The higher flow rate likely resulted in a lessened air/water contact time (as compared to a lower flow rate of air), resulting in a lower DO concentration in solution. 3.4. Experiment 4 — static test for conditions of experiment 3: removal of arsenic with iron present Experiment 4 was carried out to examine characteristics of iron and arsenic co-precipitation in a simplified environment,

28

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

Inlet and Outlet - Fe 6 5

Fe (mg/L)

4

In

3

Out 2 1 0 0

10

20

30

40

50

60

70

80

Time (hours)

Top Sample Ports - Fe 6 5

Fe (mg/L)

4

A1 3

B1

2

E1

1 0 0

10

20

30

40

50

60

70

80

Time (hours)

Middle Sample Ports - Fe 6

Fe (mg/L)

5 4

A2

3

B2 2

E2 1 0 0

10

20

30

40

50

60

70

80

Time (hours)

Bottom Sample Ports - Fe 6

Fe (mg/L)

5 4

A3

3

B3

2

E3 J3

1 0 0

10

20

30

40

50

Time (hours) Fig. 7. Experiment 3 results for iron.

60

70

80

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

in free solution in a static test. The purpose was to demonstrate reaction kinetics under the experimental conditions created in this investigation. A 750 mL anaerobic solution was prepared in a 1 L glass beaker (by nitrogen sparging followed by addition of iron and arsenic solutes), samples were taken, and then vigorous aeration was carried out for 10 min. Sampling was continued over successive days to examine the time needed for settling of iron flocs with arsenic adsorption. Samples were taken from 1 cm below the surface of the water to allow effects of flocculation and settling to be easily observed. At the end of the static test, the solution was stirred vigorously and sampled to test for re-suspension of metals. Experimental observations and results were as follows: – – – –

pH (initial, after nitrogen sparging): 7.3 DO (initial, after nitrogen sparging): 0.10 mg/L pH (after iron and arsenic added, before aeration): 5.5 DO (after aeration): 7.75 mg/L.

To test if adsorption to the side of the container made any difference to arsenic concentrations, the experiment was carried out again, but just with DI water and 200 μg/L As(V) in solution. All arsenic samples tested to within ±8%, with no clear positive or negative trend. As such, adsorption of arsenic to glass was considered to be negligible. As shown in Table 5, there was a small drop in iron concentration over the course of the experiment, with the ending (five-day) concentration of 5.05 mg/L being 0.24 mg/ L lower than the initial concentration of 5.29 mg/L (i.e. 5% removal). However, this drop in iron concentration made a very large difference to the arsenic concentration, with the ending (five-day) concentration of 47 μg/L being 157 μg/L lower than the initial concentration of 204 μg/L (i.e. 77% removal). This explains how a relatively small drop in iron concentrations (as in previous experiments) has a large effect on arsenic removal. The mass ratio of removal of iron to arsenic (as calculated from total removal over the course of the experiment) is: ð240 μg=L ironÞ=ð157 μg=L arsenicÞ ¼ 1:5 giron =garsenic : This translates to a molar ratio of: 1:5 ðgiron =garsenic Þð1 moliron =55:85 giron Þ  ð74:92 garsenic =1 molarsenic Þ ¼ 2 moliron =molarsenic : The 77% removal of arsenic in this experiment compares favorably with calculated removals at the outlet at the end of air sparging in experiment 3 (also 77%). Brownish precipitate was observed on the bottom of the beaker at the experiment's end. Due to this finding, and that vigorous stirring re-suspended iron and arsenic, co-precipitation is confirmed as the removal mechanism. The results of experiment 4 also verify that the oxidation and settling process for iron is slow under these experimental conditions, as a five-day reaction time only produced a 0.24 mg/L drop in iron concentration. According to Morgan and Lahav (2007), reaction around neutral pH would achieve a precipitous drop in iron in only 1 h, due to rapid oxidation. However, in this experiment, a high degree of removal of arsenic is shown to be possible

29

with sufficient time, when iron is over an order of magnitude higher in concentration (i.e. when the concentration of iron is much higher than that of arsenic, even a small amount of iron removal, relative to its total concentration, may in this case cause removal of a large portion of arsenic present at an iron/ arsenic molar removal ratio of 2.0). 3.5. Experiment 5 — examining removal of arsenic with iron present, but over an extended period This experiment was similar to experiment 3 but over an extended period (six days of air sparging followed by nine days of flow with no aeration). The purpose was to examine the effect of allowing longer times for air sparging and contact time of solution within the medium. The recovery of the system was also examined (i.e. the rise of arsenic concentration post-sparging following its drop in concentration during sparging). Characteristics and operating conditions for experiment 5 were as follows: – – – –

Average flow of 19.1 mL/min HRT of 13 h Initial pH of 6.0 at the inlet and 5.4 at the outlet Initial DO readings at the inlet and outlet of 0.27 mg/L and 0.57 mg/L, respectively – Air sparging flow rate of 30 ± 5 mL/min (since greater oxygen transfer was not achieved at higher flow rates tested, and a flow rate of 30 mL was more easily and consistently kept steady than the 16 mL/min flow rate used previously). For the beginning and end of the six-day air sparging period, DO levels in the apparatus are shown in Fig. 9. No terminal DO measurement for the outlet was taken and as such is not shown. As shown in Fig. 9, there was effective oxygen transfer to the middle and top sampling ports downstream of sparging, but not nearly as much along the bottom. It is considered likely that some disturbance of the sand above the sparging point occurred after trials with a higher airflow rate (250 mL/min in an experiment mentioned at the end of Section 3.3). With a disturbed sand medium above the sparging point, air bubbles may have had faster travel through the lower depth of the apparatus, and hence had less transfer of oxygen to solution in this section. In addition, there was little oxygen transfer to solution above the sparging point at column B. It is considered likely that there was greater lateral migration of airflow channels in the downstream direction during this experiment. It was noted during this experiment that there was an increasing amount of rusty brown iron precipitate accumulating on the peristaltic tubing and on the plexiglass in the inlet column as well. This illustrates the potential for iron precipitation at fairly low DO levels as experienced at the inlet (i.e. at 0.5 mg/L or lower), given sufficient time. Results for iron removal are shown in Fig. 10, with a highlighted data point on each graph marking the end of air sparging. It is apparent that the outlet concentration is generally below that of the inlet, with the exception of one point at 73 h at which the concentrations are equivalent. On average, the outlet is below the inlet by 0.25 mg/L over the course of the experiment. This pattern matches well with previous

30

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

Inlet and Outlet - As(V) 250

As (µg/L)

200 150

In 100

Out

50 0 0

10

20

30

40

50

60

70

80

Time (hours)

Top Sample Ports - As(V) 250

As (µg/L)

200

A1

150

B1 100

E1

50 0 0

10

20

30

40

50

60

70

80

Time (hours)

Middle Sample Ports - As(V) 250

As (µg/L)

200

A2

150

B2 100

E2

50 0 0

10

20

30

40

50

60

70

80

Time (hours)

Bottom Sample Ports - As(V) 250

As (µg/L)

200

A3

150

B3 100

E3 J3

50 0 0

10

20

30

40

50

Time (hours) Fig. 8. Experiment 3 results for arsenic.

60

70

80

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35 Table 4 Percent removal of arsenic at the end of air sparging for various ports in experiment 3. Mean inlet arsenic conc. Port End arsenic conc. (μg/ Removal (μg/L) L) (%) 176

B1 B2 B3 E1 E2 E3 J3 Out

49 84 104 36 27 41 20 41

72 52 41 80 85 77 89 77

experiments. However, a noted difference in this experiment compared to previous experiments is that there was sufficient time for flow to reach the sampling ports at J1 and J2 (as demonstrated by the elevated DO levels). Since J1 and J2 are in the low-flow area, it is likely that the water at these points had a contact time in the apparatus much greater than that of the HRT. The effect of this on iron concentration is shown at ports J1 and J2 in Fig. 10 (i.e. a visibly lower iron concentration than at E1 and E2, respectively, from approximately 200 to 250 h). Interpreting these findings in concert with those from experiment 4, the lower iron concentration is due to longer reaction time once water reaches ports J1 and J2. Results for arsenic are shown in Fig. 11. All three rows of sampling points (top, bottom, and middle) show a decline in arsenic concentrations at ports downstream of air sparging after the start of air sparging, but this trend is much more distinct at points E1, E2, J1, and J2. This higher removal of arsenic at ports E1, E2, J1, and J2 is likely from a combination of two factors. Firstly, DO is transferred very effectively to E1, E2, J1, and J2 as all had DO concentrations at 4.0 mg/L or greater at the end of air sparging, whereas E3 and J3 had concentrations at 1.3 mg/L or lower. This is likely responsible for the higher removal of arsenic at ports E1 and E2. Over the course of the experiment the iron removal at E1 and E2 is greater than that at E3 (with mean concentration differences of 90 and 70 μg/L comparing E3–E1 and E3–E2, respectively). A second reason for the higher removal of arsenic is that J1 and J2 lie in a lower-flow zone. As previously mentioned, iron is removed at J1 and J2 to a greater extent than at other ports, because these lie in a lower-flow region and reaction times were longer, sufficient to allow more iron to precipitate. The recovery of the system is also evident as the arsenic concentrations at downstream ports eventually exhibit an upward trend after the end of air sparging, and this is shown

Table 5 Results of experiment 4. Sample time (days)

Fe (mg/L)

As(V) (μg/L)

Start 1 2 3 4 5 End mix

5.29 5.23 5.21 5.19 5.08 5.05 5.28

204 177 164 121 73 47 122

31

much more quickly at ports E3 and J3, with ports J1 and J2 having a much more gradual recovery. Even after nine days of recovery time post-sparging, the arsenic concentrations at J1 and J2 have not returned to near original levels. The estimated percent removal of arsenic for each port downstream of sparging is shown in Table 6 (based on the concentration of arsenic at the end of air sparging). When compared to the removal values for arsenic in experiment 3 (shown in Table 4), the removal values in Table 6 tend to be smaller. It was noted that, when comparing initial conditions, there was less initial breakthrough of arsenic in experiment 3 than in subsequent experiments (i.e. at initial conditions at the start of air sparging, there was a larger difference between inlet and outlet concentrations in experiment 3 than in experiments 4 and 5). This contributed to a higher measured performance in terms of arsenic removal. However, comparing removal values for different ports within experiment 5 reveals that the removal of arsenic is distinctly higher at J1 and J2, from a greater amount of co-precipitation with iron. This further demonstrates the distinct removal of arsenic with a relatively small change in iron concentrations, when iron exists in solution at much higher concentrations. 4. Discussion It is clear that iron oxidation is slow (on the order of days) at pH values of 5.5 to 6.0. Judging by the rate kinetics shown in Morgan and Lahav (2007), a neutral pH would likely cause a further marked increase in iron removal. The pH-neutral conditions typical for Bangladeshi groundwater are therefore conducive to rapid iron oxidation as compared to these experiments. It is important to note that, while arsenic did not reach drinking water standards (i.e. 10 μg/L) in these experiments, performance may be improved under neutral conditions with more rapid iron oxidation. This requires further assessment. An important design aspect to consider is the effect of the concentration of DO on the removal of iron and subsequent removal of arsenic. While higher DO did not result in obviously higher iron removal (e.g. comparing ports E1 and E2 with port E3 in experiment 5), this did result in higher arsenic removal, suggesting that there was some slightly greater removal of iron resulting in more pronounced co-precipitation. However, given the sensitivity of iron oxidation kinetics to pH, this behavior (showing the benefit of greater DO concentrations) may change at concentrations closer to neutral pH. This requires further assessment in conditions more comparable to those in the field, since treatment zone volume may be more important than DO concentration according to van Halem et al. (2010a). In regard to the effect of contact time on arsenic removal, results from experiment 5 suggest that contact time (under those specific experimental conditions) was important, as allowing sufficient time in that experiment for lowered iron levels (i.e. at ports J1 and J2 in a low-flow zone) resulted in a dramatically lower level of arsenic. However, it is important to note that this behavior partly resulted from below-neutral pH causing a depressed reaction rate for the oxidation of iron, and may not be completely applicable to neutral conditions. In regard to the effect of time on the adsorption of arsenic to iron minerals, some experiments indicate a fairly rapid process (Banerjee et al., 2008; Pierce and Moore, 1982)

32

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

DO Levels in the Apparatus During Experiment 5 8 7

DO (mg/L)

6 5 4

Start

3

6 days end

2 1 0 In

A1

A2

A3

B1

B2

B3

E1

E2

E3

J1

J2

J3

Out

Sampling Port Fig. 9. DO levels in the apparatus during experiment 5.

with over 93% of adsorption of arsenate occurring in 24 h at pH 6 in one study (O'Reilly et al., 2001). Performance under these experimental conditions has shown maximum performance in the range of 79% removal of arsenic (considering experiment 5, not experiment 3 in which there was a smaller amount of initial breakthrough of arsenic). Measured between the inlet and outlet, removal of 59% was observed. These experiments also illustrate the high potential for effective treatment in solutions in which iron is much higher than arsenic. However, it is important to consider performance in a more complex field environment when assessing future potential of this method. A neutral pH will result in greater performance in regard to removal of iron (and subsequent co-precipitation of arsenic), but there is also likely to be significant hindrances because of competing ions, such as phosphate, sulfate, carbonate, and silica. Phosphate is a particular concern, since phosphate binds to HFO almost identically as arsenate (Waychunas et al., 1993), is a significant competitor for arsenic adsorption (Stollenwerk, 2003), and may hinder in situ treatment (Brennan and McBean, 2011; Brunsting, 2012; Brunsting and McBean, 2014; van Halem et al., 2010a). Although below-neutral pH did cause suppression of iron oxidation kinetics in these experiments, this did provide a benefit in allowing examination of the relationship of iron to arsenic in co-precipitation at small increments of iron removal. Experiment 4 showed iron and arsenic to be removed by co-precipitation at a molar ratio in the range of two (iron/ arsenic) and, as such, when the concentration of iron is many times that of arsenic, a relatively small drop in iron concentration may cause a large impact on arsenic removal. This is demonstrated in apparatus experiments (particularly in experiment 5, see Figs. 10 and 11 for ports J1 and J2). Groundwater in Bangladesh with high arsenic (N 50 μg/L) is typically high in iron as well, with an average of 5300 μg/L (Roberts et al., 2004), and therefore may be conducive to in situ treatment for arsenic. As shown in experiment 5, there is some lag time in concentrations of arsenic returning to pre-sparging levels after cessation of air sparging. This is caused by dropping levels of DO, subsequent cessation of oxidation of iron, and exhaustion of adsorption sites on iron oxyhydroxide (van Halem et al., 2010a). More laboratory research may be beneficial, in examining more complex solutions at conditions more closely simulating

those in the natural aquifer, with other constituents such as bicarbonate, calcium, magnesium, phosphate, and other ions. Examination of the effect of arsenic speciation on treatment effectiveness may also be beneficial, in examining treatment under conditions in which both As(III) and As(V) are present and fairly stable. An additional consideration for further research, especially in applying this technique in the field, is the temporary effect of treatment (i.e. oxidizing conditions diminishing once air injection ceases). Depending on the rate of recovery, sparging may not be needed continuously during treatment, and may be applied periodically. Examination of the effect of intermittent versus continuous injection of air may be beneficial in further laboratory and field study, to investigate how this may dictate proper operational conditions to ensure effectiveness. 5. Conclusions The main conclusions are as follows: – Maximum removal of arsenic was 79% under conditions of these experiments, at a local point in the apparatus (down to 33 μg/L from 158 μg/L, see experiment 5). Measured between the inlet and the outlet, removal of 59% was observed. – Iron and arsenic were removed by co-precipitation (in a simple solution where only these two solutes are present) at a molar ratio of approximately 2 (iron/arsenic). This shows that the theoretical natural treatment capacity for arsenic is very high in waters with iron at substantially higher amounts, such as in many Bangladeshi groundwaters. – There is a lag effect post-sparging for iron and arsenic concentrations returning to pre-treatment levels. Further experiments could exploit and study this by applying periodic air sparging and investigating effects. – Iron oxidation is significantly retarded at pH values below neutral. If neutral conditions were achieved in an experimental solution, and experiments of this investigation were repeated, it is likely that there would be even greater removal of arsenic because of increased oxidation and precipitation of dissolved iron.

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

33

Inlet and Outlet - Fe 6

Fe (mg/L)

5 4

In

3

Out 2 1 0 0

50

100

150

200

250

300

350

Time (hours)

Top Sample Ports - Fe 6

Fe (mg/L)

5 4

A1

3

B1 E1

2

J1

1 0 0

50

100

150

200

250

300

350

Time (hours)

Middle Sample Ports - Fe 6

Fe (mg/L)

5 4

A2

3

B2 E2

2

J2

1 0 0

50

100

150

200

250

300

350

Time (hours)

Bottom Sample Ports - Fe 6

Fe (mg/L)

5 4

A3

3

B3 E3

2

J3

1 0 0

50

100

150

200

250

Time (hours) Fig. 10. Experiment 5 results for iron.

300

350

34

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35

As (µg/L)

Inlet and Outlet - As(V) 200 180 160 140 120 100 80 60 40 20 0

In Out

0

50

100

150

200

250

300

350

Time (hours)

As (µg/L)

Top Sample Ports - As(V) 200 180 160 140 120 100 80 60 40 20 0

A1 B1 E1 J1

0

50

100

150

200

250

300

350

Time (hours)

As (µg/L)

Middle Sample Ports - As(V) 200 180 160 140 120 100 80 60 40 20 0

A2 B2 E2 J2

0

50

100

150

200

250

300

350

Time (hours)

As (µg/L)

Bottom Sample Ports - As(V) 200 180 160 140 120 100 80 60 40 20 0

A3 B3 E3 J3

0

50

100

150

200

250

Time (hours) Fig. 11. Experiment 5 results for arsenic.

300

350

J.H. Brunsting, E.A. McBean / Journal of Contaminant Hydrology 159 (2014) 20–35 Table 6 Percent removal of arsenic at various ports during experiment 5. Mean inlet arsenic conc. (μg/L) 158

Port

End arsenic conc. (μg/L)

Removal (%)

E1 E2 E3 J1 J2 J3 Out

57 50 96 37 33 89 64

64 68 39 77 79 44 59

Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) and the Ontario Graduate Scholarship program for scholarship monies provided during the course of this research. The Ontario Research Foundation and the Canada Research Chair program also provided funding for experimental work.

References Ahmed, M.F., 2003. Arsenic contamination: regional and global scenario. In: Ahmed, M.F. (Ed.), Arsenic Contamination: Bangladesh Perspective. ITN, Bangladesh Dhaka, pp. 1–20. Anawar, H.M., Akai, J., Komaki, K., Terao, H., Yoshioka, T., Ishizuka, T., Safiullah, S., Kato, K., 2003. Geochemical occurrence of arsenic in groundwater of Bangladesh: sources and mobilization processes. J. Geochem. Explor. 7, 109–131. Appelo, C.A.J., de Vet, W.W.J.M., 2003. Modeling in situ iron removal from groundwater with trace elements such as As. In: Welch, A.H., Stollenwerk, K.G. (Eds.), Arsenic in Groundwater. Kluwer Academic Publishers, Dordrecht, pp. 381–401. Banerjee, K., Amy, G.L., Prevost, M., Nour, S., Jekel, M., Gallagher, P.M., Blumenschein, C.D., 2008. Kinetic and thermodynamic aspects of adsorption of arsenic onto granular ferric hydroxide (GHG). Water Res. 42 (13), 3371–3378. BGS, DPHE, 2001. Arsenic contamination of groundwater in Bangladesh. In: Kinniburgh, D.G., Smedley, P.L. (Eds.), British Geological Survey (Technical Report, WC/00/19. 4 Volumes). British Geological Survey, Keyworth. Brennan, R., McBean, E., 2011. The role of orthophosphate and dissolved oxygen in the performance of arsenic–iron removal plants in Bangladesh. J. Environ. Sci. Health, Part A: Tox. Hazard. Subst. Environ. Eng. 46 (4), 426–435. Brunsting, J., 2012. A Laboratory-Scale Investigation of In Situ Treatment of Arsenic-Contaminated Groundwater by Air Sparging. Masters dissertation University of Guelph. Brunsting, J., McBean, E., 2014. Phosphate interference during in situ treatment for arsenic in groundwater. J. Environ. Sci. Health A 49 (6), 671–678. Chakraborti, D., Das, B., Rahman, M.M., Chowdhury, U.K., Biswas, B., Goswami, A.B., Nayak, B., Pal, A., Sengupta, M.K., Ahamed, S., Hossain, A., Basu, G., Roychowdhury, T., Das, D., 2009. Status of groundwater arsenic contamination in the state of West Bengal, India: a 20-year study report. Mol. Nutr. Food Res. 53 (5), 542–551. Cherry, J.A., Shaikh, A.U., Tallman, D.E., Nicholson, R.V., 1979. Arsenic species as an indicator of redox conditions in groundwater. J. Hydrol. 43 (1–4), 373–392. Clifford, D.A., Karori, S., Ghurye, G., Samanta, S., 2004. Field Speciation Method for Arsenic Inorganic Species. American Water Works Association Research Foundation, Denver. Flanagan, S.V., Johnston, R.B., Zheng, Y., 2012. Arsenic in tube well water in Bangladesh: health and economic impacts and implications for arsenic mitigation. Bull. World Health Organ. 90, 839–846. Grombach, P., 1985. Groundwater treatment in situ in the aquifer. Water Supply 3 (1), 13–18. Mettler, S., 2002. In Situ Removal of Iron from Ground Water: Fe(II) Oxygenation, and Precipitation Products in a Calcareous Aquifer. Doctoral dissertation Swiss Federal Institute of Technology, Zurich (Retrieved from ETH Bibliothek Knowledge Portal).

35

Miller, G.P., 2006. Subsurface Treatment for Arsenic Removal. WERC and AWWA Research Foundation, USA. Miller, G.P., 2008. Subsurface Treatment for Arsenic Removal Phase II Investigation. WERC and AWWA Research Foundation, USA. Miller, R.J., Thompson, W., Walker, M., 2002. Control of a groundwater arsenic plume by air sparging. In: Gavaskar, A.R., Chen, A.S.C. (Eds.), Remediation of Chlorinated and Recalcitrant Compounds, Proceedings of the Third International Conference on Remediation of Chlorinated and Recalcitrant Compounds. Monterey CA, May 2002. Battelle Press, Columbus, OH, pp. 2085–2094. Morgan, B., Lahav, O., 2007. The effect of pH on the kinetics of spontaneous Fe(II) oxidation by O2 in aqueous solution — basic principles and a simple heuristic description. Chemosphere 68 (11), 2080–2084. Nickson, R., McArthur, J.M., Burgess, W.G., Ahmed, K.M., Ravenscroft, P., Rahman, M., 1998. Arsenic poisoning of Bangladesh groundwater. Nature 395, 338. O'Reilly, S.E., Strawn, D.G., Sparks, D.L., 2001. Residence time effects on arsenate adsorption/desorption mechanisms on goethite. Soil Sci. Soc. Am. J. 65, 67–77. Paul, S.C., Rahman, M.A., Siddique, N., Shafiqul Alam, A.M., 2008. Voltammetric study of arsenic (III) and arsenic (V) in ground water of Hajigonj and Kalkini in Bangladesh. Pak. J. Anal. Chem. 9 (1), 49–55. Pierce, M.L., Moore, C.B., 1982. Adsorption of arsenite and arsenate on amorphous iron hydroxide. Water Res. 16 (7), 1247–1253. Rasul, S.B., Munir, A.K.M., Hossain, Z.A., Khan, A.H., Alauddin, M., Hussam, A., 2002. Electrochemical measurement and speciation of inorganic arsenic in groundwater of Bangladesh. Talanta 58, 33–43. Roberts, L.C., Hug, S.J., Reuttimann, T., Billah, M.D.M., Khan, A.W., Rahman, M.T., 2004. Arsenic removal with iron(II) and iron(III) in waters with high silicate and phosphate concentrations. Environ. Sci. Technol. 38 (1), 307–315. Rott, U., 1985. Physical, chemical and biological aspects of the removal of iron and manganese underground. Water Supply 3 (2), 143–150. Rott, U., Friedle, M., 1999. Subterranean removal of arsenic from groundwater. In: Chappell, W.R., Abernathy, C.O., Calderon, R.L. (Eds.), Arsenic Exposure and Health Effects. Elsevier Science Ltd., Oxford, pp. 389–396. Rott, U., Friedle, M., 2000. Eco-friendly and cost-efficient removal of arsenic, iron and manganese by means of subterranean ground-water treatment. Water Supply 18, 632–636. Rott, U., Lamberth, B., 1993. Groundwater clean up by in-situ treatment of nitrate, iron and manganese. Water Supply 11 (3–4), 143–156. Sen Gupta, B., Chatterjee, S., Rott, U., Kauffman, H., Bandopadhyay, A., DeGroot, W., Nag, N.K., Carbonell-Barrachina, A.A., Mukherjee, S., 2009. A simple chemical free arsenic removal method for community water supply — a case study from West Bengal, India. Environ. Pollut. 157 (12), 3351–3353. Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution of arsenic in natural waters. Appl. Geochem. 17 (5), 517–568. Smith, A.H., Lingas, E.O., Rahman, M., 2000. Contamination of drinking-water by arsenic in Bangladesh: a public health emergency. Bull. World Health Organ. 78 (9), 1093–1103. Stollenwerk, K.G., 2003. Geochemical processes controlling transport of arsenic in groundwater: a review of adsorption. In: Welch, A.H., Stollenwerk, K.G. (Eds.), Arsenic in Ground Water: Geochemistry and Occurrence. Kluwer Academic Publishers, Dordrecht, pp. 67–100. Sung, S., Morgan, J.J., 1980. Kinetics and product of ferrous iron oxygenation in aqueous systems. Environ. Sci. Technol. 14 (5), 561–568. Tamura, H., Kawamura, S., Hagayama, M., 1980. Acceleration of the oxidation of Fe2+ ions by Fe(III)-oxyhydroxides. Corros. Sci. 20 (8–9), 963–971. van Beek, C.G.E.M., 1985. Experiences with underground water treatment in the Netherlands. Water Supply 3 (2), 1–11. van Halem, D., Olivero, S., de Vet, W.W.J.M., Verberk, J.Q.J.C., Amy, G.L., van Dijk, J.C., 2010a. Subsurface iron and arsenic removal for shallow tube well drinking water supply in rural Bangladesh. Water Res. 44 (19), 5761–5769. van Halem, D., Heijman, S.G.J., Amy, G.L., van Dijk, J.C., 2010b. Subsurface arsenic removal for small-scale application in developing countries. Desalination 251, 241–248. van Halem, D., Heijman, S.G.J., Johnston, R., Huq, I.M., Ghosh, S.K., Verberk, J.Q.J.C., Amy, G.L., van Dijk, J.C., 2010c. Subsurface iron and arsenic removal: low-cost technology for community-based water supply in Bangladesh. Water Sci. Technol. 62 (11), 2702–2709. van Halem, D., de Vet, W., Verberk, J., Amy, G., van Dijk, H., 2011. Characterization of accumulated precipitates during subsurface iron removal. Appl. Geochem. 26, 116–124. Waychunas, G.A., Rea, B.A., Fuller, C.C., Davis, J.A., 1993. Surface chemistry of ferrihydrite: Part 1. EXAFS studies of the geometry of coprecipitated and adsorbed arsenate. Geochim. Cosmochim. Acta 57 (10), 2251–2269. WHO, 2008. Guidelines for Drinking-water Quality: Third Edition Incorporating the First and Second Addenda. World Health Organization, Geneva. WQAA, no date. Water Quality Standards for Various Uses of Water: Drinking Water Standards of BIS (IS: 10500: 1991). Water Quality Assessment Authority, Government of India. wqaa.gov.in/Content/wqs.aspx.