Adsorption and Heterogeneous Reduction of ... - ACS Publications

---

Chapter 4

Adsorption and Heterogeneous Reduction of Arsenic at the Phyllosilicate-Water Interface 1

1,2

3

1,4

L. Charlet , S. Chakraborty , S. Varma , C. Tournassat , M. Wolthers , D. Chatterjee , and G. Roman Ross 1

2

1

1

Environmental Geochemistry Group, LGIT-OSUG, University of Grenoble, BP 53, F-38041 Grenoble, Cedex 9, France Department of Chemistry, University of Kalyani, Kalyani 741 235, West Bengal, India Institute of Physics, Bhubaneswar 751005, India Current address: BRGM, BP 6009, F-45060 Orléans, Cedex 2, France 2

3

4

The Bengal Delta Plain (BDP) shallow aquifer is extensively used for public water supply but is severely polluted with naturally occurring arsenic, threatening the health of millions of people. Arsenic (III) is the most toxic and mobile aqueous arsenic species, but homogeneous reduction of As(V) is usually sluggish. Since in our study site field data indicate near equal amounts of As(III) and As(V), we investigated the heterogeneous reduction of arsenic at the muscovite-water interface, as muscovite is a mica frequently found in the B D P sediment. As(V) adsorption on muscovite was studied as a function of pH and compared to As(III) adsorption. The p H dependence for As(III) and As(V) adsorption are remarkably similar, and this similarity contrasts with the distinct reactivity of many minerals such as Fe and M n (oxyhydr)oxides and other phyllosilicates towards the two species. The reacted muscovite samples were investigated by PIXE (Particle Induced X-Ray Emission) and X P S (X-Ray Photoelectron Spectroscopy)spectroscopy, and it was revealed that As(V) is completely reduced upon adsorption. This was also observed for muscovite flakes sampled in the upper aquifer sediment, and thus only As(III) sorption on muscovite was described © 2005 American Chemical Society

41

42 using surface complexation modelling. In the presence of aqueous Fe(II), As(V) is also partly reduced to As(III) at the montmorillonite/water interface. The two investigated mineral surfaces have therefore been shown to catalyse, in different way, the reduction of As(V). Thus arsenic immobilization may be intimately linked to adsorption and electron transfer phenomena occurring at the phyllosilicate/water interface, and this may affect the transport of arsenic in anoxic groundwaters.

Introduction The occurrence of elevated levels of As in soil and groundwater can compromise soil and water quality. For instance, over the past three decades, millions of tube-wells have been installed in the Ganges-Brahmaputra Bengale Delta Plain to provide pathogen free water for cooking and drinking. This unexpected arsenic-contaminated water is now affecting at least 30-50 million people in Bangladesh and West Bengal, India. Another 10 million people are also affected in Vietnam, Argentina, and Central America, and arsenic has now been recognized as one of the most widespread and problematic water contaminants ( i , 2). In the environment, As(V) is present as ionic species, usually H A s 0 " or H A s 0 " , while the dominant form of As(III) species is the non-ionic H A s 0 ° complex, a rather poorly sorbing and lipophyllic species. Arsenic toxic and teratogenic effects depend strongly on this speciation. Elevated arsenic levels are likely to cause numerous cancers of the skin, liver, lungs and other internal organs (3, 4). The two usual ways of absorption of arsenic are by inhalation or ingestion, although there may be some degree of skin absorption of trivalent arsenic species since it is more lipid soluble than pentavalent species (5). Gastronintestinal uptake amounts to 60% for arsenate, 80% for arsenite and nearly 100% for methylated arsenic species (6). Once in the cell, As(III) may interact with specific functional groups such as protein thiol (RSH) and sufhydryl (-SH) groups, inhibiting a variety of enzymes, e.g. in the pyruvate oxidase system (7). In contrast, As(V) uptake and metabolic disturbance is due to its substitution for phosphate, in particular in the glycolase reaction chain (8, 9, 10). As a water contaminant, As(III) is more problematic than As(V). Previous studies seeking to understand the mechanism of arsenic mobilization have usually focused on sediment iron (oxyhydr)oxide coatings, due to the high affinity of these solids for arsenic (V) (1, 11). However, in the reductive conditions prevailing in the Holocene B D P shallow aquifer, little Fe 2

2

4

3

4

3

43 (oxyhydr)oxide is found. The grey, fine- to medium-sized sand particles of this aquifer are made of quartz, feldspar and mica often coated with mixed Fe(II)/Fe(III) (oxyhydr)oxides (12, 13, 14). The reductive conditions prevailing in the B D P aquifer are thought to arise from oxygen consumption coupled to mineralisation of organic matter either deposited over the past 10,000 years, as sea-level rose from the previous glacial low stand (13, 15) or introduced in the last 50 years through irrigation and latrines (16). Once dissolved oxygen and nitrate supplied by recharge are consumed, further mineralisation of organic matter is thought to be coupled to the reductive dissolution of As-rich Fe (oxyhydr)oxides, particularly in river beds. This microbially-driven reduction, coupled to the release of bicarbonate and As(V) aqueous species has frequently been invoked as an arsenic mobilization mechanism (17, 18, 16, 19), and bacterial species involved have only recently been identified (20, 21); further reduction of the released As(V) to As(III) may be driven by the presence of sulfide ions. A n alternative explanation explored in the present paper is that As may be significantly reduced upon adsorption on mica and clay particles. Up to now, only few studies have investigated the adsorption and redox transformation of arsenic on aluminosilicate minerals, despite the abundance of these materials in the terrestrial environment. Kaolinite, montmorillonite and illite were shown to have higher affinities for As(V) than for As(III) (22, 11), and heterogeneous reduction of As(V) on phlogopite, an Fe(II) free micas, was shown to occur in the presence of Fe(II) (23). By studying arsenic surface reactions on muscovite and montmorillonite, the objective of the present study was to determine i f a direct link could be established between groundwater As speciation and reactions occurring at the phyllosilicate/water interface. X P S and PIXE studies were conducted with large crystals of muscovite previously reacted with As(V) solution in order to determine the localisation of As sorption and the redox speciation of As ions sorbed on the mica surfaces. The very small crystallite size of montmorillonite particles prevents any fine characterisation of the As adsorption localisation with PIXE. Furthermore, montmorillonite is not able to sorb a sufficient amount of As to be detected by X P S and PIXE. Hence, only solution chemistry experiments were performed to assess the potential of specifically sorbed Fe(II) (24) towards the reduction of As(V) in solution. The results of these experiments are compared to field data collected in Chakdaha, West Bengal, India. X P S spectra of a natural Bengal muscovite and As speciation in groundwater as a function of pe, pH and Fe concentration equilibrium conditions are discussed.

44

Field Site, Material and Methods

Field Site and Sampling Field investigations were conducted on a deep, unconfmed silty sand aquifer in Chakdaha, 65 km North of Calcutta, West Bengal, India. The study area covers 19 km (Latitude 23° 04' 00" Ν - 23° 06Ό0"; longitude 88° 30' 00" Ε 88° 33' 00" Ε) and is situated in the heart of the BDP along the Hoogly River, a main tributary of the river Ganga. The B D P encompasses an integral part of the Ganga-Brahmaputra delta, the world's largest deltatic alluvium. The Ganges and Brahamaputra River sediment load is extremely high, reaching as much as 13 million tons per day during the monsoon season. The large alluvial plain, spreading southward down to the Bay of Bengal, is formed by successions of a fining upward sequence with occasional clay layers. The Ganga has shifted from time to time to the west during the Holocene. There is a gentle gradient with an elevation difference of 2 to 4 m generally sloping towards the southeast direction. At our site, a natural levee separates the paddy-field floodplain located to the West, along the Hoogly River, from the inhabited area located to the East, along the railway (Figure 1). Most tube wells sampled throughout the study area are 20 to 25 m deep and are located in the shallow Holocene aquifer, below a 15-20 m thick silt to silty clay layer (79). Water sampling was conducted in March-April 2002. Groundwater was siphoned during steady pumping using a hand-pump until pH and redox potential reading had stabilized (i.e. long enough to flush several times the 21-liters interval of a typically 20-meter deep well). Samples for chemical analysis of well waters were collected in previously acid-washed 60-ml high-density polyethylene bottles and subsamples to measure major and trace elements were immediately acidified to 1% HC1 (Fisher Optima). Whenever possible, a multiprobe (multilane YSI) fitted with calibrated platinum redox and dissolved-oxygen electrodes was introduced directly into the groundwater. Chemical analytical methods are described in detail elsewhere (19). Briefly, samples for measurement of Fe(II) and sulfide were immediately collected with a syringe and filtered through a 0.45 μπα cellulose ester filter membrane directly into pre-prepared colorimetric reagent solutions (25). Redox-sensitive constituents (Fe(II) and sulfide ions) were measured colorimetrically with a Hach D R 2010 spectrophotometer. Dissolved ferrous iron was measured using the O-phenantrolin method (26), and was found to be closed to saturation with respect to siderite, while Ca concentrations are at equilibrium with calcite (19). Calculations were made with Phreeqc2 and the Llnl.dat database. Sulfide was analysed using the method adapted from Fonselius (27). pH was measured on-site with a W T W 197 pH meter. To prevent any organic growth or decay, samples were refrigerated in an icebox within minutes after collection and 2

45 laboratory. Sampling, sample filtration, treatment and measurement of Fe(II) and H S took about 2-3 h. 2

Figure 1. Water table drawdown contour with well positions and groundwater total arsenic concentration at the study site, Chakdaha Block, Nadia district, 65 km north of Calcutta in West Bengal, India (data from (19)). The railway runs through the middle of the city. The black box outlines the 4 km χ 5 km study area, located on the Hoogly River East bank. (See page 1 of color insert.)

A n undisturbed sediment core was extracted from the 18-m-deep silty clay layer nearby the school (Figure 1). The 60-cm tube was sealed, put in a sealed bag filled with nitrogen and transported in an ice box to Grenoble, France, where the bag and the tube were opened in a glove box. Sediments were also sampled during manual drilling and collected at a depth of 12-15m. These samples were air-dried and muscovite micas were separated by hand plucking from dried sediment samples using a needle. Mineral sorbents The reference muscovite used in this study was obtained from Ward's Natural Science Establishment. It was ground with a pestle and mortar, to pass a