arsenic in groundwater: testing pollution mechanisms for … · 2010. 2. 12. · arsenic in...

Arsenic in groundwater: Testing pollution mechanismsfor sedimentary aquifers in Bangladesh

J. M. McArthur,1 P. Ravenscroft,2 S. Safiulla,3 and M. F. Thirlwall4

Abstract. In the deltaic plain of the Ganges-Meghna-Brahmaputra Rivers, arsenicconcentrations in groundwater commonly exceed regulatory limits (.50 mg L21) becauseFeOOH is microbially reduced and releases its sorbed load of arsenic to groundwater.Neither pyrite oxidation nor competitive exchange with fertilizer phosphate contribute toarsenic pollution. The most intense reduction and so severest pollution is driven bymicrobial degradation of buried deposits of peat. Concentrations of ammonium up to 23mg L21 come from microbial fermentation of buried peat and organic waste in latrines.Concentrations of phosphorus of up to 5 mg L21 come from the release of sorbedphosphorus when FeOOH is reductively dissolved and from degradation of peat andorganic waste from latrines. Calcium and barium in groundwater come from dissolution ofdetrital (and possibly pedogenic) carbonate, while magnesium is supplied by bothcarbonate dissolution and weathering of mica. The 87Sr/86Sr values of dissolved strontiumdefine a two-component mixing trend between monsoonal rainfall (0.711 6 0.001) anddetrital carbonate (,0.735).

1. Introduction

Aquifers ,300 m deep (most ,100 m) provide Bangladeshand West Bengal with .90% of its drinking water. Thegroundwater contains .50 mg L21 of arsenic in up to 1,000,000water wells and adversely affects health, putting up to 20 mil-lion people at risk [Dhar et al., 1997; Ullah, 1998; Mandal et al.,1998; Department of Public Health Engineering (DPHE), 1999](see also Bangladesh international community news homepage at http://www.bicn.com/acic/). We use new data for Ban-gladesh well waters and literature data to test three mecha-nisms invoked to explain arsenic release to this groundwater,i.e., reductive dissolution of FeOOH and release of sorbedarsenic to groundwater, oxidation of arsenical pyrite, and an-ion (competitive) exchange of sorbed arsenic with phosphatefrom fertilizer. We show that neither fertilizer phosphate norpyrite oxidation cause arsenic pollution (a term meaning theaddition to the environment of a species in amounts sufficientto cause environmental harm). We postulate that the severityand distribution of arsenic pollution is controlled by the dis-tribution of buried peat deposits rather than the distribution ofarsenic in aquifer sediments, as the former drives reduction ofFeOOH. This postulate has wide applicability because theprocess of FeOOH reduction is generic and not limited bygeography or by time.

2. Ganges-Meghna-Brahmaputra Delta Plain2.1. Arsenic Pollution

The area to the west of the Meghna and north of the Gangesis occupied by slightly elevated alluvial terraces of the Barindand Madhupur Tracts (Figure 1), which are underlain by de-posits of lower Pleistocene age [Alam et al., 1990]. Aquifersbeneath these areas are assigned to the Dupi Tila Formation.There are sharp lateral contrasts in age between the terracesand the Holocene floodplains [Ravenscroft, 2000] owing to theeffect of river incision during the Pleistocene sea level low.Maximum incision occurred 18,000 years ago when world sealevel was ;120 m below the present level. The main rivers mayhave cut down .100 m along the axial courses [Umitsu, 1993]and formed a broad plain ;50 m below the present surface ofthe modern coastal plains [Goodbred and Kuehl, 1999, 2000].Rapid sedimentary infilling resulted in regional fining upwardsequences. The alluvial infill ranges from coarse sand andgravel at the base and passes upward through sand deposits,laid down by braided rivers, into more heterogeneous sand andsilts, laid down by meandering streams. Extensive peat depositsaccumulated during the mid-Holocene climatic optimum[Reimann, 1993; Umitsu, 1993].

Aquifers beneath the elevated alluvial terraces (Dupi TilaFormation) are almost free of arsenic pollution. In aquifersbeneath the Holocene floodplains, within the alluvial and del-taic plains of the Ganges, Meghna, and Brahmaputra (in Ban-gladesh, Jamuna) Rivers, concentrations of arsenic (Figure 1)commonly exceed the Bangladesh drinking water standard (50mg L21). The distribution of pollution is very patchy, beingcommonest in the southeast and northeast of Bangladesh.Limited data show that highest arsenic concentrations occur atdepths of ;30 m [Frisbie et al., 1999; Karim et al., 1997; RoyChowdhury et al., 1999; Acharyya et al., 1999; Asian ArsenicNetwork (AAN), 1999]. Using 2024 new data pairs of well depthand arsenic concentration [DPHE, 1999], we have graphed, asa function of depth, the percentage of wells that exceed regu-latory limits (Figure 2) and so confirm that the highest per-centage of contaminated wells occurs at depths between 28 and

1Geological Sciences, University College London, London, England,United Kingdom.

2Mott MacDonald International, Dhaka, Bangladesh.3Department of Environmental Sciences, Jahangirnagar University,

Dhaka, Bangladesh.4Department of Geology, Royal Holloway University of London,

Egham, England, United Kingdom.

Copyright 2001 by the American Geophysical Union.

Paper number 2000WR900270.0043-1397/01/2000WR900270$09.00

WATER RESOURCES RESEARCH, VOL. 37, NO. 1, PAGES 109–117, JANUARY 2001

109

Figure 1. Map of Bangladesh with circled areas showing study areas of DPHE [1999, 2000]. CN, ChapaiNawabganj; F, Faridpur; L, Lakshmipur. Shading shows the percentage of wells that exceed an arsenicconcentration of 0.05 mg L21, as estimated from union averages of 18,471 data and based on the center ofeach union, calculated using a fixed radius of 7.5 km, a 1.5 km grid, and 3125 union centers. Unions areadministrative areas. A color version is available from J. M. McArthur.

McARTHUR ET AL.: ARSENIC IN GROUNDWATER110

45 m. Hand-dug wells are mostly ,5 m deep and are usuallyunpolluted by arsenic. Below 45 m a reduction occurs in thepercentage of wells that are contaminated, but risk remainssignificant until well depth exceeds 150 m.

2.2. Water Composition

We use data from Nickson et al. [2000], new data in Table 1,and published data from DPHE [1999, 2000] for two areas ofBangladesh, namely, Faridpur and Lakshmipur (Figure 1). An-alytical methods used to obtain DPHE data are given by DPHE[1999]. Our 87Sr/86Sr data (Table 1) were obtained on unfil-tered, acidified water samples using the method given byMcArthur et al. [1991]. We do not use DPHE data for Nawab-ganj because those EC and bicarbonate data are suspect (J. M.McArthur et al., unpublished data, 2000). We use d13C datafrom DPHE [1999] rather than the modified data by DPHE[2000] as we believe the former more accurately reflect aquifervalues. When discussing chemical mechanisms rather than ar-senic distributions, we use data only for wells ,100 m depth asthe most severe arsenic pollution occurs at these depths (Fig-ure 2). Full data for wells are available in downloadable formatfrom http://www.bgs.ac.uk/arsenic/Bangladesh/home.htm.

The waters in the Ganges-Meghna-Brahmaputra delta plain(GMBD) are anoxic, calcium-magnesium bicarbonate waters[DPHE, 2000]. Typically, they contain neither dissolved oxygennor nitrate, which have been removed by reduction. Localizedpollution adds nitrate and/or sulfate to a few wells and, in a fewothers, especially where sodium and chloride are high, sulfatemay be remnant from marine connate water. Waters com-monly contain concentrations of ammonium and phosphorusin the milligram per liter range, and hundreds of microgramsper liter of arsenic. Values of pH range from 6.4 to 7.6 (min-imum 5.9 at Nawabganj [DPHE, 2000]). Concentrations ofsilica (as H4SiO4) reach 131 mg L21. Free methane occurs inthe aquifer [Ahmed et al., 1998].

Saturation indices, calculated with WATEQF embedded inNETPATH [Plummer et al., 1994], show that most waters areat close to equilibrium with calcite and dolomite, with satura-tion indices (SI) for both ranging from 10.6 to 20.4 in Farid-pur and from 11.2 to 21.2 in Lakshmipur. Manganese ismostly undersaturated with respect to rhodochrosite (SI from21.4 to 10.6 at Faridpur and 20.6 to 10.2 at Lakshmipur).Waters are mostly oversaturated with vivianite (SI mostly 12

to 13.5 at Faridpur and 20.4 to 14.2 at Lakshmipur) andsiderite (SI 10.5 to 11.4 at Faridpur and 10.1 to 11.5 atLakshmipur). Such oversaturation may reflect slow precipita-tion kinetics or the stabilization of iron in solution by organiccomplexing.

3. Arsenic Pollution MechanismsThree mechanisms have been invoked to explain arsenic

pollution of groundwater in the GMBD: (1) arsenic is releasedby oxidation of arsenical pyrite in the alluvial sediments asaquifer drawdown permits atmospheric oxygen to invade theaquifer [Mallick and Rajagopal, 1996; Mandal et al., 1998; RoyChowdhury et al., 1999]; (2) arsenic anions sorbed to aquiferminerals are displaced into solution by competitive exchangeof phosphate anions derived from overapplication of fertilizerto surface soils [Acharrya, 1999]; (3) anoxic conditions permitreduction of iron oxyhydroxides (FeOOH) and release ofsorbed arsenic to solution [Bhattacharya et al., 1997; Nickson etal., 1998, 2000].

We discount pyrite oxidation as a mechanism for arsenicpollution, even though trace pyrite is present in the aquifersediments [Public Health Engineering Department (PHED),1991; AAN, 1999; Nickson et al., 1998, 2000]. Measured sulfurconcentrations in aquifer sediments represent both pyritic andorganic sulfur but allow upper limits to be placed on pyriteabundance of 0.3% [Nickson et al., 2000], 0.02% [AAN, 1999],0.1% (J. M. McArthur, unpublished data, 2000) and 0.06%[DPHE, 1999]. The presence of pyrite shows that it has notbeen oxidized and that it is a sink for, not a source of, arsenicin Bangladesh groundwater. Were pyrite to be oxidized, itsarsenic would be sorbed to the resulting FeOOH [Mok andWai, 1994; Savage et al., 2000], rather than be released togroundwater. Furthermore, Bangladesh groundwaters, whichare anoxic, would contain iron and sulfate in the molar ratio of0.5 were pyrite oxidation releasing arsenic; in reality, theseconstituents are mutually exclusive in solution [DPHE, 2000],as are arsenic and sulfate, i.e., arsenic concentrations above 50mg L21, are found only where sulfate concentrations are ,30mg L21 [DPHE, 1999, 2000]. Finally, arsenic pollution is un-common in hand-dug wells [DPHE, 1999] which, being shal-lowest, are the most exposed to atmospheric oxygen and theones that would be most polluted were arsenic derived frompyrite by oxidation.

Figure 2. Percentage of wells in Bangladesh exceeding spec-ified arsenic concentrations, shown as a function of depth (datafrom regional survey [DPHE, 1999]).

Table 1. Strontium Isotopic Composition of Well WatersFrom Faridpur

Number DPHE Numbera Sr, mg L21 87Sr/86Sr

S1 BTS208 0.480 0.72107S2 0.400 0.72103S3 BTS243 0.330 0.71536S4 BTS260 0.370 0.71603S6 BTS258 0.470 0.72228S7 BTS206 0.360 0.71798S8 BTS241 0.420 0.72065S9 BTS242 0.400 0.71903S10 BTS214 0.300 0.71349S11 0.554 0.72343S12 0.486 0.72333S14 0.497 0.72400S15 0.591 0.72510

a Department of Public Health Engineering (DPHE) well numbersare those of DPHE [1999, 2000].

111McARTHUR ET AL.: ARSENIC IN GROUNDWATER

Arsenic pollution may be caused by the displacement ofarsenic from sorption sites on aquifer minerals as a result ofcompetitive (anion) exchange by fertilizer phosphate, whichmay leach from soils after excessive use of fertilizer [e.g.,Acharyya et al., 1999]. We reject this idea because the watersattain a bicarbonate concentration of at least 200 mg L21

before phosphorus, arsenic, or iron, are found in significantamounts (Figure 3). Waters lowest in bicarbonate are theyoungest and least evolved, but they would contain most phos-phorus (and so arsenic) were phosphorus supplied from sur-face application of fertilizer. Furthermore, phosphorus con-centrations increase with depth in both Faridpur andLakshmipur (J. M. McArthur (unpublished data, 2000) basedon DPHE [2000]). Finally, the areal distribution of phosphorusin aquifer waters [Davies and Exley, 1992; Frisbie et al., 1999]show that areas high in phosphorus are also arsenical; thiscoincidence implies that if fertilizer phosphate promotes ar-senic release, the process operates only in some areas of Ban-gladesh, which seems unlikely. The arguments above suggestthat competitive exchange with fertilizer phosphate neither

worsens nor causes arsenic pollution. Nevertheless, concentra-tions of phosphorus in the milligram per liter range are re-leased to groundwater from latrines and from the fermentationof buried peat deposits (see section 4). Concentrations of ar-senic covary with those of phosphorus for waters from Laksh-mipur (Figure 4b) but not for waters from Faridpur (Figure4b), suggesting that competitive exchange with phosphate gen-erated in situ may contribute to arsenic pollution. For reasonsgiven later, we believe this contribution to be small.

Reduction of FeOOH is common in nature and has beeninvoked previously to explain the presence of arsenic in anoxicsurface waters [Aggett and O’Brien, 1985; Cullen and Reimer,1989; Belzile and Tessier, 1990; Ahmann et al., 1997] and anoxicground waters [Matisoff et al., 1982; Cullen and Reimer, 1989;Korte, 1991; Korte and Fernando, 1991; Bhattacharya et al.,1997; Nickson et al., 1998, 2000, and references therein]. Re-duction of FeOOH,

8FeOOH 1 CH3COO2 1 15H2CO33

8Fe21 1 17HCO32 1 12H2O, (1)

is driven by microbial metabolism of organic matter [Chapelleand Lovley, 1992; Nealson, 1997; Lovley, 1997; Banfield et al.,1998; Chapelle, 2000]. That FeOOH reduction is common andintense in GBMD aquifers is shown by several observations.First, high concentrations of dissolved iron have been reportedby DPHE [2000] (#24.8 mg L21), by Nickson et al. [1998, 2000](#29.2 mg L21), and by Safiullah [1998] (#80 mg L21). Sec-ond, at concentrations above ;200 mg L21 of bicarbonate,iron shows a weak correlation with bicarbonate (line A ofFigure 3a). The relation is not stoichiometric for reduction ofFeOOH (equation (1)), but data fall on or to the right of lineA, the slope of which (molar HCO3/Fe of 13) is within a factor

Figure 3. Relation of bicarbonate to (a) Fe21, (b) P, and (c)As in Bangladesh groundwater. For the significance of line A,see text. Data from Nickson et al. [2000] are open circles, anddata from DPHE [2000] are from Faridpur (triangles) andLakshmipur (squares).

Figure 4. Relation of (a) As to Fe and (b) As to P, in Ban-gladesh groundwater for arsenic concentrations below 500 mgL21. Data are from DPHE [2000] and Nickson et al. [2000].Symbols are as in Figure 3.


of 2 of that (;30) given for FeOOH reduction by Chapelle andLovley [1992]. Samples enriched in bicarbonate relative to lineA have possibly derived additional bicarbonate from otherredox reactions, calcite dissolution, and weathering of micaand feldspar or have lost iron into precipitated phases (seebelow).

The data of Nickson et al. [2000] show a relation betweenarsenic and bicarbonate that was interpreted as evidence thatarsenic was derived from reduction of FeOOH; arsenic andbicarbonate data of DPHE [2000] do not show such a covari-ance (Figure 3c). Concentrations of iron and arsenic covary inaquifer sediments, with molar ratios of Fe/As (oxalate-extractable) of between 1500 and 6000 [DPHE, 1999] andFe/As (diagenetically available) ratios of 1800 [Nickson et al.,1998, 2000]. Nevertheless, concentrations of arsenic and irondo not covary in solution (Figure 4a). This may be because,first, arsenic and iron may be sequestered differentially intodiagenetic pyrite [Moore et al., 1988; Rittle et al., 1995] and sodo not behave conservatively in solution. Second, dissolvediron may also be derived from weathering of biotite. Third, theiron/arsenic ratio in dissolving FeOOH is variable. Finally, ironmay be removed from solution into vivianite, siderite, or mixedvalency oxides or hydroxycarbonates (R. Loeppert, personalcommunication, 2000).

4. Redox DriverThe lateral and vertical differences in arsenic concentration

in well water (Figures 1 and 2) cannot arise from variations inthe abundance of arsenic in aquifer sediments: these are mi-caceous quartzo-feldspathic sands and are not unusual in theirconcentrations of arsenic, which are commonly in the rangebetween 1 and 30 mg kg21 [Nickson et al., 1998, 2000; AAN,1999; DPHE, 1999]. Arsenic at these concentrations is presentas a dispersed element sorbed to dispersed FeOOH. Higherconcentrations of arsenic, e.g., 196 ppm of Roy Chowdhury etal., [1999], are uncommon and occur where (rare) localizedpyrite has formed during burial diagenesis and scavenged ar-senic from solution [Moore et al., 1988; Rittle et al., 1995; AAN,1999].

Arsenic in Bangladesh sediments will not be released fromFeOOH unless organic matter is present to drive microbialreduction (or release phosphate for competitive exchange), sowe postulate that it is the distribution of organic matter, par-ticularly peat, in the aquifer sediments that is the primarycontrol on arsenic pollution. Peat beds are common beneaththe Old Meghna Estuarine Floodplain in Greater Comilla[Ahmed et al., 1998], in Sylhet, and in the Gopalganj-KhulnaPeat Basins [Reimann, 1993]. Many wells in the area aroundFaridpur may be screened in waterlogged peat [Safiullah,1998], and the aquifer in Lakshmipur contains peat [DPHE,1999]. Peat is often found in geotechnical borings (piston sam-ples), although it is rarely recorded during rotary drilling forwater wells because such drilling masks its presence unless thepeat is very thick. One indicator of peat is the total organiccarbon (TOC) content of some aquifer sediment; a samplefrom a depth of 2.1 m at Gopalganj (100 km SW of Dhaka)contained 6% TOC [Nickson et al., 1998], and sediment from adepth of 23 m at Tepakhola (Faridpur municipality) contained7.8% TOC [Safiullah, 1998].

A further indicator of buried peat is the covariance (Figure5) of the concentrations of iron, phosphorus, ammonium, andd13C of dissolved inorganic carbon (DIC), which suggests all

are controlled by a master process, which we take to be themicrobial metabolization of buried peat. Complexing moitiesderived from fermentation of peat (e.g., short-chain carboxylicacids and methylated amines) will drive redox reactions andammonium production [Bergman et al., 1999]. Furthermore,methane is common in groundwater [Ahmed et al., 1998; Hoqueet al., 2000], in places in amounts sufficient to impede pumpingof groundwater and to provide domestic fuel. Where metha-nogenesis is not seen directly, the chemical signature of metha-nogenic CO2 is visible as low pH (.5.9 [DPHE, 2000]) andhigh pCO2 of 1020.7 to 1021.5 atm. We postulate also thatdecreasing pH with increasing bicarbonate (in Faridpur) anddissolved H4SiO4 (Figure 6) reflects the reaction of methano-genic CO2 with carbonates, micas, and feldspars in the aquifer.Concentrations of H4SiO4 (#131 mg L21) approach saturationvalues for amorphous silica (195 mg L21 [Parkhurst, 1995]) andsuggest active weathering is occurring. Values of d13C (DIC)

Figure 5. Relation between d13C (dissolved inorganic carbon(DIC)), P, Fe, and NH4 in Bangladesh groundwater. Data arefrom DPHE [1999] for d13C (DIC); otherwise, they are fromDPHE [2000]. Symbols are as in Figure 3. In Figure 5b thelarge arrow represents N/P ratio of 16 for degrading organicmatter; the small arrow shows departure from this N/P ratio asreductive dissolution of FeOOH adds additional phosphorusto groundwater.


range up to 110‰ [DPHE, 1999], a common upper limit formethanogenic CO2 [Whiticar, 1999].

In Faridpur wells, values of d13C (DIC) decrease as thecalcium concentration increases (Figure 7) because methano-genic CO2 (d13C of 15 to 110‰) dissolves (and equilibrateswith) detrital calcite (d13C of 0 to 26‰ [Quade et al., 1997;Singh et al., 1998]) and, possibly, pedogenic calcite, whichwould be more 13C-depleted (compare d13C values to 212‰in pedogenic carbonates of the Siwalik Group [Quade et al.,1997]). Isotopic lightening of ground water may result fromoxidation in situ of 13C-depleted methane, but the importanceof this mechanism cannot be established with current data. Thed13C (DIC) values of Lakshmipur ground waters scatter con-siderably and show no trend.

That calcite dissolution and subordinate mica weatheringare important controls on the calcium and magnesium concen-tration in Bangladesh well water is shown by the good corre-lation between Ca and Mg for many waters (Figure 8a), thegood correlation between Ca and 87Sr/86Sr (Figure 8b), and anisotopic mixing trend for strontium that defines two end-members with 87Sr/86Sr values of ;0.711 and 0.735 (Figure 8c).These values are close to those of monsoonal rain in Bang-ladesh (0.710 to 0.712 [Galy et al., 1999]) and modern detritalcarbonate (#0.735 [Quade et al., 1997; Singh et al., 1998]).

The coliform count of Bangladesh wells [Hoque, 1998] co-varies with ammonium concentrations (Figure 9). Latrineshave been recorded within 2 m of wells, possibly allowingpollution access to wells via insecure casing. This source willalso supply phosphorus to groundwater. That another sourceof ammonium, and so phosphorus, exists is shown by the factthat wells with a faecal coliform counts of zero have ammo-nium concentrations up to 6.6 mg L21 (Figure 9 [Hoque, 1998])and the fact that latrines are found throughout the country butphosphorus enrichment parallels the distribution of arsenic

Figure 8. Relation of Ca to (a) Mg and (b) 87Sr/86Sr and (c)87Sr/86Sr to 1/Sr in Bangladesh groundwater. Symbols are as inFigure 3. In Figure 8a, arrow A shows trend for mica weath-ering, and arrow B shows trend for carbonate dissolution.

Figure 6. Relation of pH to (a) H4SiO4 and (b) bicarbonatein Bangladesh groundwater. Data are from DPHE [2000]. Sym-bols are as in Figure 3.

Figure 7. Relation of Ca to d13C (DIC) in Bangladeshgroundwater. Symbols are as in Figure 3. Arrow shows trendfor carbonate dissolution from methanogenic CO2, M, to de-trital/pedogenic carbonate, P.


enrichment and is concentrated mostly in the northeast andsoutheast of Bangladesh (Figure 1). This other source of am-monium and phosphorus must be buried peat. A few wellscontain amounts of ammonium and phosphorus that reflect themaximum ratio likely to be found in common wetland vegeta-tion (Figure 5; molar N/P of 16 [Redfield et al., 1963; Bedford etal., 1999]), suggesting that both come from this source. Atlower concentrations, molar N/P values ,,16 indicate a sourceof additional phosphorus, which we take to be phosphorussorbed to FeOOH and released during its reductive dissolu-tion; likely vegetative sources have N/P ratios .16 [Bedford etal., 1999; Richardson et al., 1999]. From Figure 5 we estimatethat .70% of phosphorus comes from reduction of FeOOH atphosphorus concentrations below 3 mg L21. It therefore seemslikely that most arsenic also is derived this way rather than bycompetitive exchange with phosphate derived from organicmatter.

In Hungary, arsenic-polluted wells contain methane, ammo-nium concentrations between 1 and 5 mg L21, and often highconcentrations of iron (M. Csanady, personal communication,2000). The similarity with Bangladesh groundwaters may indi-cate a common pollution driver: burial and degradation of peatdeposits. Haskoning [1981] noted that ammonium was a minornuisance in deep (.200 m) production wells at Khulna, south-western Bangladesh, so some deep wells in Bangladesh may besusceptible to arsenic pollution not because of leakage of pol-luted water from overlying aquifers but because in situ degra-dation of organic matter may drive reduction of FeOOH andrelease of arsenic.

The arguments presented above suggest that the areal dis-tribution of arsenic pollution corresponds closely to the arealdistribution of buried peat. The geographic distribution ofarsenic pollution shows some concordance with the distribu-tion of paludal basins recorded by Goodbred and Kuehl [2000].Peat deposits are, and were, formed in waterlogged areasrather than active river channel deposits, a fact that helps todefine today’s areal pattern of pollution. Umitsu [1987, 1993,personal communication, 1998] proposed that much peatlanddevelopment occurred in the GMBD during a climatic/sealevel optimum some 5000 years B.P. The high number of pol-luted wells with depths of 28–45 m may result from their beingscreened near the depth of this major peat horizon. As peatmust have formed at other times, other peat layers, at otherdepths and of differing ages, might explain why arsenic pollu-tion also peaks at depths of 55, 75, 100, and 130 m (Figure 2).

5. ImplicationsArsenic pollution by oxidation of arsenical pyrite is a mech-

anism that is valid for oxic environments, typically surfacewaters. It may apply to the subsurface where high permeabilityallows polluted surface water access to the subsurface, as inZimapan, Mexico [Armienta et al., 1997]. It may apply whereoxic conditions invade a previously anoxic environment hostingsulfide ore, for example, in northeastern Wisconsin [Schreiberet al., 2000], where a commercially prospective sulfide ore bodyup to 3 m thick is exposed to oxic conditions by water leveldrawdown and in domestic boreholes. Oxidation of the oreresults in pollution of groundwater by high concentrations ofarsenic (#15 000 mg L21), sulfate (,618 mg L21), iron (,160mg L21), and acidity (pH $2.1) [Schreiber et al., 2000; A.Weissbach, personal communication, 2000].

Where arsenic pollution occurs in most subsurface and most

anoxic environments, the pyrite oxidation model is inappropri-ate, and a different model is needed. Reduction of FeOOH(invoked before for groundwater [e.g., Matisoff et al., 1982;Cullen and Reimer, 1989; Korte, 1991; Bhattacharya et al., 1997;Nickson et al., 1998, 2000, and references therein]) will serve inmost instances. As the process is generic and not site-specific,it should be examined (not necessarily accepted) wherevernaturally occurring arsenic pollution occurs in groundwater,such as in Argentina [Nicolli et al., 1989], Taiwan [Chen et al.,1996], China [Wang and Huang, 1994; Sun et al., 2000], Hun-gary, and the United States [Welch et al., 2000]. It is likely thatany fluvial or deltaic basin that has hosted marshland andswamp will be prone to severe arsenic contamination of bore-hole water. In many areas of the world, agriculture and urban-ization occur on lowland coastal plains in a setting similar intype, although not always in scale, to that seen in Bangladesh.It might be expected that such areas would be afflicted byarsenic contamination, if not pollution, and it should be lookedfor. Vulnerable regions include the deltas of the Mekong, Red,Irrawaddy, and Chao Phraya Rivers.

6. ConclusionsNeither pyrite oxidation nor competitive exchange of fertil-

izer phosphate for sorbed arsenic cause arsenic pollution ofgroundwater in the Ganges-Meghna-Brahmaputra deltaicplain. Indeed, pyrite in Bangladesh aquifers is a sink for, not asource of, arsenic. Pollution by arsenic occurs because FeOOHis microbially reduced and releases its sorbed load of arsenic togroundwater. The reduction is driven by microbial metabolismof buried peat deposits. Dissolved phosphorus comes mainlyfrom FeOOH, as it is reductively dissolved, with subordinateamounts being contributed by degradation of human organicwaste in latrines and fermentation of buried peat deposits.Dissolved ammonium in the aquifer derives predominantlyfrom microbial fermentation of buried peat deposits, but sig-nificant amounts are contributed by unsewered sanitation. Am-monium ion is not therefore an infallible indicator of faecalcontamination of groundwater. Reduction of FeOOH and re-lease of sorbed arsenic serve as a generic model for arseniccontamination of aquifers where waters are anoxic, particularlywhere organic matter is abundant, e.g., in deltaic or fluvialareas that supported peatland during climatic optimums.

Figure 9. Relation between NH41 and faecal coliform count

in Bangladesh wells. Data are from national survey by Hoque[1998]. Zero colliform wells contain up to 6.6 mg L21 of am-monium.


Acknowledgments. J.M.M. thanks the Friends of UCL for travelfunds to visit Bangladesh and West Bengal during the preparation ofthis paper. We thank Bill Cullen, W. Berry Lyons, R. Loeppert, and ananonymous reviewer for constructive reviews that helped us improvethe manuscript. The 87Sr/86Sr measurements were made by J.M.M. inthe Radiogenic Isotope Laboratory at RHUL, which is supported, inpart, by the University of London as an intercollegiate facility. P.R.thanks the Department of Public Health Engineering, Government ofBangladesh, for permission to use its data.

ReferencesAcharyya, S. K., Comment on Nickson et al. Arsenic poisoning of

Bangladesh groundwater, Nature, 401, 545, 1999.Aggett, J., and G. A. O’Brien, Detailed model for the mobility of

arsenic in lacustrine sediments based on measurements in LakeOhakuri, Environ. Sci. Technol., 19, 231–238, 1985.

Ahmann, D., L. R. Krumholz, H. H. Hemond, D. R. Lovley, andF. M. M. Morel, Microbial mobilization of arsenic from sediments ofthe Aberjona watershed, Environ. Sci. Technol., 31, 2923–2930, 1997.

Ahmed, K. M., M. Hoque, M. K. Hasan, P. Ravenscroft, and L. R.Chowdhury, Occurrence and origin of water well methane gas inBangladesh, J. Geol. Soc. India, 51, 697–708, 1998.

Alam, M. K., A. K. M. S. Hassan, M. R. Khan, and J. W. Whitney,Geological map of Bangladesh, Geol. Surv. of Bangladesh, Dhaka,1990.

Armienta, M. A., R. Rodriguez, A. Aguayo, N. Ceniceros, G. Villase-nor, and O. Cruz, Arsenic contamination of groundwater at Zima-pan, Mexico, Hydrogeol. J., 5, 39–46, 1997.

Asian Arsenic Network (AAN), Arsenic contamination of groundwa-ter in Bangladesh: Interim report of the research at Samta village, 90pp., Tokyo, Japan, April 1999.

Banfield, J. F., K. H. Nealson, and D. R. Lovley, Geomicrobiology:Interactions between microbes and minerals, Science, 280, 54–55,1998.

Bedford, B. L., M. R. Walbridge, and A. Aldous, Patterns in nutrientavailability and plant diversity of temperate North American wet-lands, Ecology, 80, 2151–2169, 1999.

Belzile, N., and A. Tessier, Interactions between arsenic and ironoxyhydroxides in lacustrine sediments, Geochim. Cosmochim. Acta,54, 103–109, 1990.

Bergman, I., P. Lundberg, and M. Nilsson, Microbial carbon mineral-isation in an acid surface peat: Effects of environmental factors inlaboratory incubations, Soil Biol. Biochem., 31, 1867–1877, 1999.

Bhattacharya, P., D. Chatterjee, and G. Jacks, Occurrence of arsenic-contaminated groundwater in alluvial aquifers from the Delta Plain,eastern India: Options for a safe drinking water supply, Water Res.Dev., 13, 79–92, 1997.

Chapelle, F. H., The significance of microbial processes in hydrogeol-ogy and geochemistry, Hydrogeol. J., 8, 41–46, 2000.

Chapelle, F. H., and D. R. Lovley, Competitive exclusion of sulfatereduction by Fe(III)-reducing bacteria: A mechanism for producingdiscrete zones of high-iron groundwater, Ground Water, 30, 29–36,1992.

Chen, S.-L., S. R. Dzeng, M.-H. Yang, K.-H. Chiu, G.-M. Shieh, andC. M. Wai, Arsenic species in groundwaters of the Blackfoot diseasearea, Taiwan, Environ. Sci. Technol., 28, 877–881, 1996.

Cullen, W. R., and K. J. Reimer, Arsenic speciation in the environ-ment, Chem. Rev., 89, 713–764, 1989.

Davies, J., and C. Exley, Hydrochemical character of the main aquiferunits of central and northeastern Bangladesh and possible toxicity ofgroundwater to fish and humans, Tech. Rep. WD/92/43R, Suppl.WD/92/44R, Br. Geol. Surv., Keyworth, England, 1992.

Department of Public Health Engineering (DPHE), Main report andvolumes S1–S5, report on Phase I, Groundwater studies for arseniccontamination in Bangladesh, Dhaka, Bangladesh, 1999.

Department of Public Health Engineering (DPHE), Report on PhaseI (supplementary data), groundwater studies for arsenic contamina-tion in Bangladesh, Dhaka, Bangladesh, 2000.

Dhar, R. K., et al., Groundwater arsenic calamity in Bangladesh, Curr.Sci., 73, 48–59, 1997.

Frisbie, S. H., D. M. Maynard, and B. A. Hoque, The nature and extentof arsenic-affected drinking water in Bangladesh, in Metals andGenetics, edited by B. Sarkar, pp. 67–85, Kluwer Acad., Norwell,Mass., 1999.

Galy, A., C. France-Lanord, and L. A. Derry, The strontium isotopic

budget of Himalayan rivers in Nepal and Bangladesh, Geochim.Cosmochim. Acta, 63, 1905–1925, 1999.

Goodbred, S. L., and S. A. Kuehl, Holocene and modern sedimentbudgets for the Ganges-Brahmaputra river system: Evidence forhighstand dispersal to flood-plain, shelf and deep-sea depocenters,Geology, 27, 559–562, 1999.

Goodbred, S. L., Jr., and S. A. Kuehl, The significance of large sedi-ment supply, active tectonism, and eustasy on margin sequencedevelopment: Late Quaternary stratigraphy and evolution of theGanges-Brahmaputra delta, Sediment. Geol., 133, 227–248, 2000.

Haskoning, A., Khulna Water Supply Project: Feasibility Study, FinalReport, vol. 3, Water Resources, vol. 3.1, Water Resources: Annexes,Dep. of Public Health Eng., Dhaka, Bangladesh, 1981.

Hoque, B. A., Biological contamination of tubewell water, Environ.Health Programme, Int. Cent. for Diarrhoeal Disease Res., Dhaka,Bangladesh, 1998.

Hoque, M., P. Ravenscroft, and K. Hassan, Investigation of ground-water salinity and gas problems in southeast Bangladesh, Bang-ladesh Cent. for Adv. Stud., Dhaka, in press, 2000.

Karim, M., Y. Komori, and M. Alam, Arsenic occurrence and depth ofcontamination in Bangladesh, J. Environ. Chem., 7, 783–792, 1997.

Korte, N., Naturally-occurring arsenic in groundwaters of the midwest-ern United States, Environ. Geol. Water Sci., 18, 137–141, 1991.

Korte, N. E., and Q. Fernando, A review of arsenic(III) in groundwa-ter, Crit. Rev. Environ. Control, 21, 1–39, 1991.

Lovley, D. R., Microbial Fe(III) reduction in subsurface environments,FEMS Microbiol. Rev., 30, 305–313, 1997.

Mallick, S., and N. R. Rajagopal, Groundwater development in thearsenic-affected alluvial belt of West Bengal—Some questions, Curr.Sci., 70, 956–958, 1996.

Mandal, B. K., T. R. Chowdhury, G. Samanta, D. Mukherjee, C. R.Chanda, K. C. Saha, and D. Chakraborti, Impact of safe water fordrinking on five families for 2 years in West Bengal, India, Sci. TotalEnviron., 218, 185–201, 1998.

Matisoff, G., C. J. Khourey, J. F. Hall, A. W. Varnes, and W. Strain,The nature and source of arsenic in northeastern Ohio groundwater, Ground Water, 20, 446–455, 1982.

McArthur, J. M., J. Turner, W. B. Lyons, M. F. Thirlwall, andA. Osborn, Hydrochemistry on the Yilgarn Block: Ferrolysis inacidic brines, Geochim. Cosmochim. Acta, 55, 1273–1288, 1991.

Mok, W. M., and C. M. Wai, Mobilization of arsenic in contaminatedriver sediment, in Arsenic in the Environment, part 1, Cycling andCharacterization, edited by J. Nriagu, pp. 99–118, John Wiley, NewYork, 1994.

Moore, J. N., W. H. Ficklin, and C. Johns, Partitioning of arsenic andmetals in reducing sulfidic sediments, Environ. Sci. Technol., 22,432–437, 1988.

Nealson, K. H., Sediment bacteria: Who’s there, what are they doing,and what’s new?, Annu. Rev. Earth Planet. Sci., 25, 403–434, 1997.

Nickson, R. T., J. M. McArthur, W. G. Burgess, P. Ravenscroft, K. Z.Ahmed, and M. Rahman, Arsenic poisoning of Bangladesh ground-water, Nature, 395, 338, 1998.

Nickson, R. T., J. M. McArthur, P. Ravenscroft, W. B. Burgess, andK. Z. Ahmed, Mechanism of arsenic poisoning of groundwater inBangladesh and West Bengal, Appl. Geochem., 15, 403–413, 2000.

Nicolli, H. B., J. M. Suriano, M. A. Gomez Peral, L. H. Ferpozzi, andO. M. Baleani, Groundwater contamination with arsenic and othertrace elements in an area of the Pampa, Province of Cordoba,Argentina, Environ. Geol. Water Sci., 14, 3–16, 1989.

Parkhurst, D. L., User’s guide to PHREEQC-A computer program forspeciation, reaction-path, advective-transport, and inverse geo-chemical calculations, U.S. Geol. Surv. Water Res. Invest. Rep., 95-4227, 1995.

Plummer, L. N., E. C. Prestemon, and D. L. Parkhurst, NETPATH:An interactive code (NETPATH) for modeling NET geochemicalreactions along a flow PATH, version 2.0, U.S. Geol. Surv. Water Res.Invest. Rep., 94-4169, 1994.

Public Health Engineering Department (PHED), National drinkingwater mission: Submission project, arsenic pollution in groundwaterin West Bengal, final report, Steering Comm., Arsenic Invest. Proj.,Dhaka, Bangladesh, 1991.

Quade, J., L. Roe, P. G. DeCelles, and T. P. Ojha, The late Neogene87Sr/86Sr record of lowland Himalayan Rivers, Science, 276, 1828–1831, 1997.

Ravenscroft, P., An overview of the hydrogeology of Bangladesh, inGroundwater Resources of Bangladesh, edited by A. A. Rahman,


P. Ravenscroft, and S. Hug, Bangladesh Cent. for Adv. Stud.,Dhaka, in press, 2000.

Redfield, A. C., B. H. Ketchum, and F. A. Richards, The influence oforganisms on the composition of sea-water, in The Sea, vol. 2, editedby N. M. Hill, pp. 26–77, John Wiley, New York, 1963.

Reimann, K.-U., Peat deposits of Bangladesh, in The Geology of Ban-gladesh, edited by K.-U. Reiman, Gebruder Borntraeger, Stuttgart,Germany, 1993.

Richardson, C. J., G. M. Ferrell, and P. Vaithiyanathan, Nutrienteffects on stand structure, resorption efficiency, and secondary com-pounds in Everglades sawgrass, Ecology, 80, 2182–2192, 1999.

Rittle, K. A., J. I. Drever, and P. J. S. Colberg, Precipitation of arsenicduring bacterial sulfate reduction, Geomicrobiol. J., 13, 1–11, 1995.

Roy Chowdhury, T., et al., Comment on Nickson et al. 1998, Arsenicpoisoning of Bangladesh groundwater, Nature, 401, 545–546, 1999.

Safiullah, S., CIDA Arsenic project report: Monitoring and mitigationof arsenic in the ground water of Faridpur Municipality, 96 pp.,Jahangirnagar Univ., Dhaka, Bangladesh, 1998.

Savage, K. S., N. T. Tracey, P. A. O’Day, G. A. Waychunas, and D. K.Bird, Arsenic speciation in pyrite and secondary weathering phases,Mother Lode Gold District, Tuolumne County, California, Appl.Geochem., 15, 1219–1244, 2000.

Schreiber, M. E., J. A. Simo, and P. G. Freiberg, Stratigraphic andgeochemical controls on naturally occurring arsenic in groundwater,eastern Wisconsin, USA, Hydrogeol. J., 8, 161–176, 2000.

Singh, S. K., K. P. Trivedi, R. Ramesh, and S. Krishnaswami, Chemicaland strontium, oxygen and carbon isotopic compositions of carbon-ates from the lesser Himalaya: Implications to the strontium isotopiccomposition of the source waters of the Ganga, Ghaghara, andIndus Rivers, Geochim. Cosmochim. Acta, 62, 743–755, 1998.

Sun, G. F., J. B. Pi, B. Li, X. Y. Guo, H. Yamavchi, and T. Yoshida,Introduction of present arsenic research in China, paper presented

at 4th International Conference on Arsenic Exposure and HealthEffects, Soc. for Geochem. and Health, San Diego, Calif., June 2000.

Ullah, S. S., Geochemical mapping and speciation of arsenic ingroundwater of Faridpur municipality, Bangladesh, J. BangladeshAcad. Sci., 22, 143–147, 1998.

Umitsu, M., Late Quaternary environment and landform evolution inBengal, Geogr. Rev. Jpn., Ser. B, 60, 164–178, 1987.

Umitsu, M., Late Quaternary environment and landforms in theGMBD, Sediment. Geol., 83, 177–186, 1993.

Wang, L., and J. Huang, Chronic arsenism from drinking water insome areas of Xinjiang, China, in Arsenic in the Environment, part II,Human Health and Ecosystem Effects, edited by J. O. Nriagu, pp.159–172, John Wiley, New York, 1994.

Welch, A. H., D. B. Westjohn, D. R. Helsel, and R. B. Wanty, Arsenicin ground water of the United States: Occurrence and geochemistry,Ground Water, 38, 589–604, 2000.

Whiticar, M. J., Carbon and hydrogen isotope systematics of bacterialformation and oxidation of methane, Chem. Geol., 161, 291–314,1999.

J. M. McArthur, Geological Sciences, University College London,Gower Street, London WC1E 6BT, England, UK. ([email protected])

P. Ravenscroft, Mott MacDonald International, 122 Gulshan Ave-nue, Dhaka 1212, Bangladesh.

S. Safiulla, Department of Environmental Sciences, JahangirnagarUniversity, Savar, Dhaka 1212, Bangladesh.

M. F. Thirlwall, Department of Geology, Royal Holloway Universityof London, Egham, Surrey TW20 0EX, England, U.K.

(Received March 30, 2000; revised August 29, 2000;accepted August 31, 2000.)


arsenic in groundwater: testing pollution mechanisms for … · 2010. 2. 12. · arsenic in...

Documents