Content uploaded by C. B. Dowling
Author content
All content in this area was uploaded by C. B. Dowling on Apr 15, 2014
Content may be subject to copyright.
Content uploaded by C. B. Dowling
Author content
All content in this area was uploaded by C. B. Dowling on Feb 07, 2014
Content may be subject to copyright.
Geochemical study of arsenic release mechanisms in the Bengal
Basin groundwater
Carolyn B. Dowling,
1
Robert J. Poreda, Asish R. Basu, and Scott L. Peters
Department of Earth and Environmental Sciences, University of Rochester, Rochester, New York, USA
Pradeep K. Aggarwal
International Atomic Energy Agency, Vienna, Austria
Received 26 September 2001; revised 28 February 2002; accepted 18 March 2002; published XX Month 2002.
[1]To investigate arsenic mobility in the Bengal Basin groundwater, we sampled water
wells and sediments throughout the region. There are strong correlations among high
levels of dissolved arsenic and iron, ammonia, and methane, especially in samples from a
single site (Laxmipur). No linkage is seen between As and agricultural tracers such as
phosphate. The association of As and Fe occurs because arsenic strongly adsorbs
onto FeOOH particles in river water. They flocculate with other fine-grained particles at
the freshwater/saltwater transition zone. Subsequent bacterially mediated reduction of
FeOOH in the clay releases the adsorbed arsenic. Weathering of As-bearing mica plays a
significant role in the As budget. The ‘‘correlated’’ presence of As, CH
4
, and NH
4
in water
supply wells is the result of diffusion out of organic-rich clay into the more permeable
zones. Arsenic is mainly released from recent sediments at <50 m depth deposited in the
GBR floodplain as sea level rose throughout the Holocene. INDEX TERMS:1065
Geochemistry: Trace elements (3670); 1829 Hydrology: Groundwater hydrology; 1831 Hydrology:
Groundwater quality; KEYWORDS:arsenic, groundwater, Bengal Basin, hydrogeochemistry, iron oxy-
hydroxides, microbes
1. Introduction
[2] Since the 1970s, the World Health Organization has
drilled more than two million tube wells in Bangladesh for
the population to have a safe, bacteria-free drinking water
supply. Unfortunately, under certain conditions, the ground-
water frequently contains elevated levels of natural arsenic,
greater than the World Health Organization’s maximum
contaminant level (MCL) of 0.01 ppm (0.13 mM) and often
times greater than the Bangladesh’s MCL of 0.05 ppm (0.66
mM). Arsenic, a notorious bio-accumulating poison, has
been adversely affecting the health of millions of people
in the Bengal Basin (Bangladesh and the West Bengal State,
India) for the last twenty years [Das et al., 1995, 1996;
Mandal et al., 1996; Dhar et al. 1997;Biswas et al., 1998;
Karim, 2000; Smith et al., 2000].
[3] The Bengal Basin lies within in the floodplain of the
Ganges and Brahmaputra Rivers (GBR). As one of the
world’s largest river systems, the GBR transports the single
largest sediment flux and the fourth highest water discharge
to the oceans [Holeman, 1968; Coleman, 1969; Milliman
and Meade, 1983]. Based on the average As of 0.07 mmol/L,
the GBR delivers an estimated 7.0 10
7
mol/yr of dissolved
As to the Bay of Bengal annually. The majority of this
arsenic is most likely removed during the flocculation of
fine-grained particles at the saltwater/freshwater interface in
the estuaries of the Bengal Basin or, under certain extreme
reducing conditions, bacterially reduced and precipitated as
diagenetic pyrite in the swampy, anoxic wetlands of the
GBR delta. From the combined effects of sea level change,
migration of river channels and variable patterns of sedi-
mentation, the boundary between fresh and seawater has
varied greatly over the past million years. These sedimentary
processes have led to a complicated subsurface distribution
of arsenic, which is presently being liberated into the
groundwater of the Bengal Basin.
[4] There have been several recent studies of the Bengal
Basin groundwater in an attempt to understand the mecha-
nisms that mobilize arsenic into the groundwater. Several
existing and competing theories have been suggested to
explain the arsenic release into groundwater, such as the
oxidation of pyrite [Mallick and Rajagopal, 1996; Mandal et
al., 1998; Chowdhury et al., 1999], competitive exchange
with fertilizer phosphate [Acharyya et al., 1999, 2000], and
dissolution of iron oxy-hydroxides [Nickson et al., 1998,
2000; McArthur et al., 2001]. The pyrite oxidation process
would require either dissolved O
2
or NO
3
as an electron
acceptor, acidified water, and measurable quantities of
sulfate. In general, none of these conditions are met for a
significant number of groundwater samples [Nickson et al.,
2000; McArthur et al., 2001; this study]. Phosphate, an
important component of fertilizer, may replace sorbed
arsenic on the sediment surfaces through competitive
exchange thereby releasing As to the groundwater. We
would expect to find higher As in shallow wells near recently
cultivated fields.
[5] Some studies have focused on the arsenic cycle with
an emphasis on adsorption and desorption reactions on iron
oxides. The results of these studies show that iron oxy-
1
Now at Byrd Polar Research Center, Ohio State University, Columbus,
Ohio, USA.
Copyright 2002 by the American Geophysical Union.
0043-1397/02/2001WR000968$09.00
X-1
WATER RESOURCES RESEARCH, VOL. 38, NO. 0, 10.1029/2001WR000968, 2002
hydroxides (FeOOH) strongly adsorb arsenic, as As
2
O
4
2
or
AsO
4
3
, in the river water [Moi and Wai, 1994], and the As-
laden FeOOH fine-grained sediments are deposited in the
organic-rich estuaries and wetlands of the Bengal Basin.
Many researchers have identified a correlation between Fe
and As in the groundwater and hypothesized that the
dissolution of iron oxy-hydroxides releases As into the
groundwater [Nickson et al., 1998; Cummings et al., 1999;
Nickson et al., 2000; McArthur et al., 2001]. Cummings et al.
[1999] and McArthur et al. [2001] believe that, in the
moderately reducing groundwater of Bangladesh, the micro-
bial mediated reductive dissolution of FeOOH is liberating
arsenic from the sediment into the water. The redox boun-
dary for the ferric/ferrous transition is close to the arsenate/
arsenite boundary for neutral pH and suggests that a sig-
nificant amount of arsenic in groundwater may occur as
arsenite [As(III)] which is more mobile at lower pH values
and more toxic than arsenate [As(V)].
[6] The purpose of this study is to describe the ground-
water geochemistry and determine the mechanisms of
arsenic release from the sediment into the Bengal Basin
groundwater. An important component of our research is to
evaluate the groundwater residence time for both the As and
non-As bearing wells since earlier studies have suggested a
depth control to the As distribution and possible link
between recent agricultural activities (phosphatic fertilizer
and excessive extraction of groundwater for irrigation) and
elevated arsenic [Acharyya et al., 1999, 2000]. We collected
sixty-eight groundwater samples, specifically for determin-
ing major and trace elements, dissolved gas, helium iso-
topes, and tritium, according to standard methods. The
samples were taken from a variety of localities and meas-
ured for major and trace elements to establish the regional
geochemistry of the groundwater and the extent of the As
problem. Across the Bengal Basin, we sampled wells at
multiple depths (9– 335 m) to establish the vertical ground-
water velocity and, hence, the recharge rate to the aquifer by
using tritium,
3
He, and
4
He concentrations. Additionally,
sediment extraction and digestion experiments were per-
formed on a sediment drill core from Laxmipur, Bangladesh
(a location with high dissolved As in groundwater) to
determine the role of adsorption/desorption reactions on
sedimentary grain surfaces in controlling the trace metal
concentrations in the groundwater and to establish the
source of arsenic being released to the groundwater.
2. Geology of the Bengal Basin
[7] The drainage area of the Ganges-Brahmaputra Rivers
is approximately 2 10
6
km
2
[Holeman, 1968; Coleman,
1969], and its average annual water discharge is 1.0 ± 10
10
12
m
3
/yr [Subramanian, 1979]. Despite the high sediment
load to the Bay of Bengal (1.5 10
12
kg/yr), the shoreline
of the GBR delta is relatively stable. The delta is only
growing at 7 km
2
/yr while parts of the sediment-starved
western delta are retreating at 1.9 km
2
/yr [Allison, 1998].
Most of the sediment in the GBR is channeled into deeper
water by canyons to create one of the largest global
submarine deltas, the Bengal fan. The heavy sediment load,
strong oceanic currents, tectonic subsidence, major seismic
events, and the Swatch of No Ground canyon system
heavily influence the sedimentation rates of the subaerial
delta, shelf, and submarine delta. Conditions over the last
20,000 years have varied greatly as sea level rose 120 m.
The large GBR sediment discharge during this time was
sufficient to accommodate aggradation and subsidence and
maintain shoreline stability. Based on radiocarbon ages, the
overall accumulation rate for the Bengal Basin over the last
10,000 years ranges between 3.2 ± 0.8 mm/yr and 5.0 ± 1
mm/yr [Umitsu, 1993; Goodbred and Kuehl, 2000].
According to field measurements on the Brahmaputra River,
the present sedimentation rate varies from 7.5 to 11.5 mm/yr
leading to a 15 m/yr seaward advance of the Bengal Fan
[Michels et al., 1998].
[8] Within the Ganges-Brahmaputra floodplain, the Ben-
gal Basin consists of mostly quaternary deltaic sediments of
the Ganges and Brahmaputra rivers and the alluvial deposits
from the weathering of the Himalayas. The Bengal Basin is
1.4 10
5
km
2
in area and covers most of Bangladesh and
part of the West Bengal State in India (Figure 1). The
subsurface geology of the Bengal Basin has complex
interfingerings of coarse and fine-grained sediments from
numerous regressions and transgressions throughout geo-
logic time. Umitsu [1993] and Goodbred and Kuehl [2000]
classified the Late Quaternary sediments of the coastal
region. The Oxidized Facies (lowermost unit) are weathered
paleosols of the Late Pleistocene lowstand. The Sand Facies
(lower unit) is made of medium-fine sand, represents
channel fill and early transgressive alluvial valleys, and is
older than 12,000 years. Above the Oxidized and Sand
Facies, a sharp transition occurs to the fine-grained organic-
rich Lower Delta Mud (middle unit) whose thickness varies
between 5 and 30 m. The Lower Delta Mud, consisting of
clay and sandy clay layers, results from a change in the
depositional environment during the early Holocene caused
by the rapid rise in sea level and flooding on the lower delta.
The Lower Delta Muds range from 7857 and 9917 radio-
carbon years while the ages of the overlying sequence,
Muddy Sand (upper unit), vary between 4770 and 7382
years. The Muddy Sand unit (5– 35 m thick) is divided into
clay (lower), peat (middle), and peaty silt (upper) layer and
suggests a gradual environment transition from decelerating
sea-level rise with lower flow conditions, increased channel
mobility, and widespread distribution of fluvial sands across
the coastal plain.
[9] From 5000 to 7000 years ago, the Ganges River
sediment was still being transported to the western margin
of the basin. However, during this time, the Brahmaputra
river sediment was being trapped in the Sylhet Basin
causing sediment starvation and transgression along the
eastern coast. Around 5000 years ago, the Ganges river
began migrating easterly towards the Brahmaputra river.
The shoreline in the eastern delta has been prograding into
the Bay of Bengal for the past several thousand years. The
remaining unit, Thin Mud (uppermost; 5 – 20 m thick), has a
bottom layer of peaty clay and a top layer of silt that
characterizes the formation of marshy lands and floodplain
deposits for the last 5000 years.
[10] The Bengal Basin soils are a mixture of clays, quartz,
calcium carbonate, dolomite, and mica [Food and Agricul-
tural Organization (FAO), 1971]. In general, the subsurface
mineralogy of the Bengal Basin is dominated by quartz with
some plagioclase and potassium feldspar and volcanic,
metamorphic, and sedimentary fragments [Uddin and Lund-
X-2 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
berg, 1998]. Foster et al. [2000] divided the top 50 m in
Ramrail, Brahmanbaria (near Dhaka) into two provisional
sections where the upper section (6 – 25 m depth) is com-
posed of gray to black micaceous fine-grained sand and the
lower section (26 –48 m depth) consists of medium to
coarse-grained FeOOH coated quartz and weathered mica.
Our Laxmipur drill core and river sediment samples are
dominated by iron-stained quartz with some plagioclase and
micas and a minor amount of carbonates as confirmed by our
XRD and XRF analyses and optical microscopy. We esti-
mate that larger (150 – 250 mm) biotite and muscovite flakes
compose 10– 15% of Lax-3 (3 m depth), Lax-13 (13 m
depth), and Lax-48 (48 m depth) and 3 – 5% of the bulk
samples from Lax-23 (23 m depth) and Lax-39 (39 m depth).
Our sediment grain-size analysis loosely corresponds to the
units described by Umitsu [1993] and Goodbred and Kuehl
[2000] and demonstrates the influence of localized trans-
gressions and regressions on the stratigraphy. Silt (31– 32%)
and very fine-grained sands (31 – 36%) dominate the Lax-3,
Lax-13, and Lax-48 samples at Laxmipur. Whereas the
Figure 1. Bengal Basin groundwater and river sediment collection sites. This map shows the location of
groundwater (variety of symbols) and river sediment (thick open cirlces) samples in Bangladesh. The
sediment samples (thick open cirlces) are from Laxmipur and the northern and southern Ganges and
Brahmaputra Rivers (collected by W. S. Moore and A. R. Basu). The crosses designate the larger towns.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-3
sediment at 23 m (Lax-23) consists of almost equal amounts
of fine-grained sand (28%), very fine sand (26%) and silt
(26%), and Lax-39 is composed of fine-grained sand (59%)
and medium-fine sand (20%).
3. Sampling and Methodology
[11] Sixty-eight groundwater samples were collected
from monitoring and domestic wells in the Bengal Basin
throughout Bangladesh and parts of West Bengal, India in
May 1999 and January 2000 (Figure 1), according to
standard methods [Long and Martin, 1991; Greenberg et
al., 1992]. The sediment drill core was taken adjacent to the
groundwater well cluster at Laxmipur (Figure 1). W. S.
Moore (University of South Carolina) provided the river
sediment samples upstream from the confluence of the
Ganges (RW-54) and Brahmaputra (RW-53) Rivers [Sarin
et al., 1989]. Asish R. Basu collected the northern Ganges
and Brahmaputra river sediment samples at Rishikesh, India
(Ganges) and Guwahati, India (Brahmaptura). Please refer
to Figure 1 for the river sediment locations.
[12] We recorded field parameters (pH, temperature,
conductivity, bicarbonate, and latitude and longitude) at
each well site. The samples for major element (cations
and anions) and trace metal analyses were collected through
0.25 mm polycarbonate filters. For the trace metal and cation
samples, the filtered water was acidified to pH 2 with
ultrapure nitric acid. We collected the samples for dissolved
gas content and helium isotope analysis using 3/8-inch
(I.D.) copper tubes and sealed with refrigeration clamps
according to our established methods [Poreda et al., 1988].
Water samples for tritium analysis were stored in amber
glass bottles with polyethylene caps to minimize water-
vapor exchange.
[13] We measured major cations and anions in the
groundwater samples on a Dionex ion chromatograph
(IC), using CS12A and AS4A Ion Pac columns, according
to our established procedures at the University of Rochester
[Fehn et al., 2000]. The analytical errors for these IC
analyses were usually less than 6% for the cations and
anions. Gas concentrations and isotopic ratio measurements
of the groundwater samples were carried out at the Rare Gas
Facility at the University of Rochester. The dissolved gas
was extracted and processed on a high vacuum line [Poreda
et al., 1988]. We measured the helium isotope ratio meas-
urements with a VG 5400 noble gas mass spectrometer by
peak height comparison to a calibrated air standard with
errors of 2% [Poreda and Farley, 1992]. The tritium
values were determined using the
3
He ‘‘in-growth’’ techni-
que [Clarke et al., 1976]. The errors depend on the amount
of
3
H and are ±0.1 T. U. (Tritium Units) or ±5% at 5 T. U.,
whichever is greater.
[14] We sent the Laxmipur core samples to ALS Chemex,
Mississauga, Ontario, Canada for XRF (X-ray Fluores-
cence) and organic carbon analyses. The mineralogical
classification of the core was determined through X-ray
diffraction (XRD) analysis using a Scintag XDS 2000. The
bulk drill core was sieved for grain-sized analysis. We
separated the clay-sized particles from the bulk sediment
using floatation and the micas by hand using a Nikon SMZ-
U microscope.
[15] For the bulk and fine-grained sediments and mica
separates, we used an ammonium oxalate acid reagent to
extract the exchangeable and adsorbed trace metals
[McKeague, 1978]. Ten mL of 0.2 M oxalate solution was
added to 0.25 g of sediment (or any proportion thereof) and
agitated horizontally for four hours in the dark. We then
rinsed the bulk samples with 18 Mwater, added 30%
ultrapure H
2
O
2
and 0.02 M ultrapure HNO
3
, and heated
them (85C) for five hours to breakdown the organic
fraction (modified from Li et al. [1995]). Following the
oxalate extraction procedure, the clay-sized particles and
micas were then rinsed with 18 Mwater, dried, and then
digested using a Milestone MLS 1200 microwave digestion
system. The digested samples were evaporated to dryness
and re-acidified them using ultrapure nitric acid.
[16] Iron concentrations for groundwater and sediment
extracts were determined using a Shimandzu UV-Visible
(UV-Vis) Spectrophotometer. We prepared the samples by
adding hydroxylamine hydrochloride (10% w/v), FerroZine
iron reagent, and ammonium acetate buffer solution (ammo-
nium hydroxide and ultrapure glacial acetic acid) to them
[To et al., 1999]. Errors in the UV-Vis calibration curves
were generally less than 3%. To prepare the groundwater,
sediment extracts, organic fraction, and sediment digestions
for trace metal analysis on the ICP-MS, we diluted the
samples with 18 Mwater and acidified them with ultra-
pure nitric acid. The analyses were done, according to US
EPA Method 200.8 [Long and Martin, 1991], using the VG
ICP-MS Plasma Quad II+ at the University of Rochester. To
correct for drift, gallium, indium, and bismuth were added
as internal standards for the water and sediment extraction
samples. For the sediment digestions, we only used indium
and bismuth. A five-point calibration curve was created
using standards that are traceable to NIST. The standards
were analyzed before and after a set of samples. We
analyzed two NIST water standards (NIST 1640 and NIST
1643d) as unknowns. Based on these results, the errors in
the trace metal concentration analyses were less than 5% for
the samples measured on the VG ICP-MS Plasma Quad II+.
4. Results
4.1. Groundwater Chemistry
[17] The calcium carbonate rich groundwater in the
Bengal Basin contain some sodium chloride, no sulfate,
and background levels of most trace metals, except for Sr,
Ba, Fe, Mn, and As (C. Dowling et al., The groundwaters of
the Bengal Basin: Their geochemistry and trace metal flux
to the oceans, submitted to Geochimica Cosmochimica
Acta, 2002). Our results, listed in Table 1 and displayed
in Figure 2, show that the elevated arsenic levels are
concentrated in the shallow groundwater wells (<60 m).
Forty-seven out of 68 samples or 69% of the wells are
above the WHO standard (0.01 ppm; 0.13 mM). This plot
(Figure 2) of the As distribution with depth confirms the
findings of the British Geological Society and others that
show high As concentrations at shallow depths [Acharyya et
al., 1999; British Geological Survey and Mott MacDonald
Ltd., 1999; Chowdhury et al., 1999; Nickson et al., 2000;
McArthur et al., 2001]. Some of the deeper wells (e.g., BGD
8, 13) with high arsenic levels are multiple depth comple-
tion wells and are most likely tapping both shallow As-rich
water and deeper As-free groundwater. Figure 2 also dem-
onstrates that higher levels of NH
4
, Fe and Mo (as molyb-
date which is a redox sensitive oxy-anion, like arsenate) are
X-4 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
Table 1. Bengal Basin Groundwater Sample Locations, Well Depths, Selected Major Ions, Gas, and Trace Metal Data
a
Sample
Name Sample Date Sample Location Depth, m NH
4
CH
4
Mn Fe As Mo Ba
BGD 1 May 1999 Laximpur 244 104.44 * 3.23 37.17 0.02 bdl 1.76
BGD 2 May 1999 Sonapur 11 222.22 * 8.90 14.38 1.57 0.05 0.08
BGD 3 May 1999 Faridpur 20 273.33 1052.23 22.32 35.97 4.28 0.03 1.50
BGD 4 May 1999 Faridpur 30 313.33 917.76 8.32 33.53 3.76 0.03 1.05
BGD 5 May 1999 Faridpur 40 331.11 531.43 8.89 30.79 2.94 0.03 0.91
BGD 6 May 1999 Faridpur 150 nd 14.63 15.17 10.90 0.04 0.01 0.80
BGD 7 May 1999 Faridpur 50 346.67 301.66 4.62 38.28 2.57 0.02 1.19
BGD 8 May 1999 Faridpur 244 286.67 1265.87 4.26 64.60 2.53 0.02 2.09
BGD 9 May 1999 Faridpur 98 0.00 8.84 6.43 12.68 0.04 0.00 0.99
BGD 10 May 1999 Ahladipur 30 17.78 8.01 17.71 16.07 0.49 0.01 1.29
BGD 11 May 1999 Rajbari 48 nd * 9.01 3.94 0.06 0.00 0.77
BGD 12 May 1999 Rajbari 33 nd * 19.02 8.32 0.39 0.01 0.57
BGD 13 May 1999 Rajbari 127 46.67 8.68 13.37 46.91 0.72 0.01 1.10
BGD 14 May 1999 Talma 44 385.56 734.04 2.55 107.82 0.79 0.01 1.28
BGD 15 May 1999 Nagarkanda 29 224.44 268.32 8.26 121.48 2.92 0.02 1.84
BGD 16 May 1999 Samta 9 165.56 174.44 2.49 92.22 7.82 bdl 5.65
BGD 17 May 1999 Samta 49 164.44 294.84 1.58 62.12 2.74 0.01 2.26
BGD 18 May 1999 Samta 61 173.33 * 1.64 65.93 2.99 0.01 2.47
BGD 20 May 1999 Jhenidah 100 nd 3.08 8.15 10.12 0.47 0.01 0.84
BGD 21 May 1999 Burir Char 61 nd 7.24 2.77 7.24 0.04 0.04 0.07
BGD 22 May 1999 Burir Char 152 nd * 0.34 1.89 0.32 bdl 2.56
BGD 23 May 1999 Burir Char 280 nd 7733.29 1.46 0.00 0.23 bdl 6.59
BGD 24 May 1999 Barishal 335 nd 553.14 bdl 1.30 0.01 bdl 0.18
BGD 25 May 1999 Barishal 290 nd 160.57 0.38 1.25 0.01 bdl 0.17
BGD 26 May 1999 Madaripur 30 34.44 25.97 6.15 84.78 0.32 0.01 0.72
BGD 27 May 1999 Choto Arjundi 27 376.67 204.62 16.48 159.10 3.74 0.02 1.05
BGD 28 May 1999 Mugraoara * nd * 17.18 0.21 0.02 0.01 0.08
BGD 29 May 1999 Patalpara * 160.00 645.68 38.16 39.05 12.12 0.12 0.73
BGD 30 May 1999 Adampur 91 nd 1.04 26.30 4.14 0.03 0.00 0.06
BGD 31 May 1999 Laxmipur 50 1155.56 1549.67 8.33 93.34 8.94 0.04 2.60
BGD 32 May 1999 Laxmipur 40 1066.67 2057.20 11.04 110.93 7.76 0.05 4.11
BGD 33 May 1999 Laxmipur 30 977.78 1902.18 11.56 140.07 9.50 0.04 3.21
BGD 34 May 1999 Laxmipur 20 622.22 1734.50 18.95 81.68 5.79 0.07 2.46
BGD 35 May 1999 Laxmipur 10 nd * 10.19 14.31 2.07 0.16 0.46
BGD 36 May 1999 Laxmipur 150 nd 2.29 22.02 56.84 0.25 bdl 1.33
BGD 37 January 2000 Shibganj 42 0.00 * 12.87 0.00 0.00 0.01 0.84
BGD 38 January 2000 Nawabganj 9.1 37.78 * 24.75 72.05 0.26 0.01 1.79
BGD 39 January 2000 Nawabganj 19.2 374.44 * 20.04 74.81 5.77 0.04 2.12
BGD 40 January 2000 Nawabganj 34 52.22 * 15.85 58.43 5.56 0.04 1.86
BGD 41 January 2000 Iswardi 36 nd * 11.46 0.00 0.00 0.01 0.25
BGD 42 January 2000 Iswardi 30 nd * 35.06 16.45 0.01 0.01 0.37
BGD 43 January 2000 Padma River 33 nd * 17.86 23.49 0.02 0.00 0.58
BGD 44 January 2000 Baghail Dotala Sako 35 6.67 2.47 15.80 0.57 1.09 0.04 0.64
BGD 45 January 2000 Baghail Dotala Sako 35 7.78 * 21.73 14.18 2.64 0.03 0.83
BGD 46 January 2000 Ruppur 18.3 11.11 * 18.66 6.44 0.35 0.02 1.16
BGD 47 January 2000 Meherpur 32.3 nd * 4.91 0.49 0.29 0.01 1.01
BGD 48 January 2000 Refugee Para 27 25.56 3.94 15.99 6.75 12.75 0.03 0.96
BGD 49 January 2000 Ghosh Para 36 65.56 * 9.11 65.29 0.91 0.02 2.25
BGD 50 January 2000 Ujjalpur 40 50.00 * 12.07 75.42 5.93 0.04 2.00
BGD 51 January 2000 Ghosh and Refugee 36 nd * 10.62 66.90 1.83 0.04 1.91
BGD 52 January 2000 Kustia 33 nd * 16.40 12.53 0.02 0.01 0.60
BGD 53 January 2000 Kustia 37 68.89 191.13 3.10 71.77 1.48 0.02 1.87
BGD 54 January 2000 Kustia 35 nd * 3.00 0.77 1.57 0.02 0.37
BGD 55 January 2000 Paurosabha 82.3 10.00 * 18.07 2.75 0.06 0.02 0.62
BGD 56 January 2000 Kustia Town 88.4 67.78 * 9.37 18.22 0.71 0.01 0.71
IND 1 January 2000 Moyna 50.3 43.33 67.06 1.44 44.63 12.53 0.03 0.65
IND 2 January 2000 Moyna 114 8.89 * 1.71 0.54 0.03 0.00 1.23
IND 3 January 2000 Moyna 47.2 nd 2.13 22.34 0.00 0.45 0.02 0.44
IND 4 January 2000 Moyna 18.3 36.67 564.96 4.22 56.53 2.42 0.02 1.67
IND 5 January 2000 Moyna 46 64.44 * 1.33 37.56 1.89 0.03 2.01
IND 6 January 2000 Birohi 18.3 nd 221.83 4.00 53.20 1.74 0.02 2.80
IND 7 January 2000 Birohi 126 nd * 2.88 11.44 1.19 0.02 2.63
IND 8 January 2000 Birohi 207 6.67 * 1.83 1.28 0.02 0.01 0.65
IND 9 January 2000 Baruipur 293 nd 7.64 0.66 4.07 0.04 0.02 0.96
DAC 1 January 2000 Dhaka * nd * 0.39 0.00 0.00 bdl 0.12
DAC 2 January 2000 Dhaka * nd * 0.75 0.00 0.11 bdl 0.52
DAC 3 January 2000 Dhaka * nd * 0.49 10.72 0.01 bdl 0.29
a
All data are in mmol/L. An asterisk means data were not available; nd is not detected, and bdl is below detection limit.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-5
also found at depths of less than 60 m in a crude correlation
with As-bearing localities.
[18] Several studies have linked the release of As to the
dissolution of FeOOH [Nickson et al., 1998; Cummings et
al., 1999; Nickson et al., 2000; McArthur et al., 2001]. A
positive correlation between dissolved arsenic and iron
implies that the arsenic is probably coming from the
dissolution of FeOOH, and we confirm a weak positive
correlation between Fe and As in Figure 3. This plot also
examines the associations between As and NH
4
,CH
4
, and
Mo because ammonia and methanogenic methane are good
indicators of microbial activity in the anoxic groundwater
and molybdate, as a redox sensitive oxy-anion, would
demonstrate the importance of adsorption and desorption
reactions on the sediment surfaces. For all the Bengal Basin
wells, arsenic versus methane, iron, molybdenum and
ammonia have weak to modest positive correlations (r
2
ranging from 0.37 to 0.55). Note the poor correlations
between As and Fe when all the samples are included.
Some groundwaters have high Fe but low As concentrations
suggesting that the simple breakdown of FeOOH may not
be sufficient to explain high levels of dissolved As.
[19] However, when the data from only arsenic laden
areas (Laxmipur and Faridpur) are plotted in Figure 4, much
stronger correlations occur between As and CH
4
, Fe, and
NH
4
(r
2
ranging from 0.8 to 0.9). In the Laxmipur water
samples, the molar ratio of As/Fe ranges from 0.070 to
0.144 with an average of 0.090 ± 0.032. The groundwater at
Faridpur (20– 50 m) has a comparable As/Fe ratio (0.099 ±
0.02) that varies between 0.067 and 0.119. The As-Mo
relation indicates that adsorption-desorption reactions may
also play a significant role in controlling the final dissolved
concentrations of As, Fe, and Mo. At Laxmipur and
Faridpur, there are strong positive correlations (r
2
= 0.9)
between arsenic and methane and ammonia with the highest
levels occurring at less than 60 m deep at the two multilevel
sites (Figure 4). The As-CH
4
and As-NH
4
correlations
indicate that both methanogens and other bacteria are active
in the anoxic groundwater and suggest that a source of
organic carbon is essential to the overall release of arsenic.
When combined with the iron data, the associations
between arsenic and methane, ammonia, and iron support
the microbially mediated reductive dissolution of FeOOH as
the dominant reaction for arsenic release into groundwater.
There is a strong correlation (r
2
= 0.8) between dissolved As
and Fe in the Faridpur and Laxmipur groundwater samples
with greater than 30 mmol/L CH
4
. It seems that methane is a
key indicator of microbial activity and signifies the impor-
tance of organic carbon decomposition to the release of As.
In the shallow wells, water samples have been stripped of
their atmospheric gases from the high levels of biogenic
methane. Since there is no
4
He associated with these
shallow samples (Table 2) and thermogenic methane has
measurable levels of helium (>10 ppm), methanogenic
microbes are most likely producing the methane in this
reducing environment.
[20] Groundwater residence times (
3
He/
3
H age dating)
can provide important information about the arsenic
buildup and the possible relation to subsurface stratigraphy.
By using
3
He/
3
H groundwater ages, we can estimate the
recharge rate to the aquifers and subsurface travel time
[Solomon et al., 1992, 1993]. The calculated recharge rate
to the Bengal Basin groundwater is 60 ± 20 cm/yr. The
groundwater vertical velocity ranges from 3 m/yr in the
Figure 2. Depth profiles of dissolved As, Fe, Mo, and NH
4
. The groundwater arsenic concentrations
above the WHO standard (0.01 ppm; 0.13 mM) along with elevated levels of iron, molybdenum, and
ammonia are concentrated in the shallow wells (<60 m). The higher levels of As in the deeper wells (>60 m)
are most likely from the wells screening both deep, As-free and shallow, As-rich groundwater.
X-6 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
active flow system to <0.5 m/yr in the low conductivity
sediments of Laxmipur and Faridpur. This sizeable contrast
in vertical velocities suggests that a complex stratigraphy
of high and low conductivity layers exist within the
subsurface of the Ganges-Brahmaputra floodplain, which
is consistent with its geologic history [Lindsay et al., 1991;
Umitsu, 1993; Uddin and Lundberg, 1998; Goodbred and
Kuehl, 2000; Dowling et al., submitted manuscript, 2002].
A schematic cross section from A-A
0
(of Figure 1) is
shown in Figure 5. For a more complete discussion about
groundwater ages and vertical velocities of the Bengal
Basin, please refer to Dowling et al. (submitted manuscript,
2002).
[21] The samples from the Laxmipur and Faridpur sites
contain tritium only in the most shallow 10 – 20 m interval
(Table 2). The other groundwater samples at Laxmipur and
Faridpur are tritium-free (<0.15 T. U.) below 20 m, indica-
tive of low permeability sediments and groundwater resi-
dence times of more than 60 years. This older (>60 years)
shallow groundwater contains elevated concentrations of
arsenic, methane, and ammonia but little phosphate and
challenges any proposed link between arsenic and recent
applications of phosphate-rich fertilizer [Acharyya et al.,
1999, 2000]. Fine-grained sediments, mostly silt and very
fine sand, dominate the depositional environment in the
Laxmipur and Faridpur regions and are suggested as the
dominant source of the As and CH
4
.
[22] There are several deeper wells within the upper
aquifer (Table 2; 100 m to 250 m; BGD-6, 9, 14, 15, 29,
36) that have residence times of greater than 60 years (shaded
triangles in Figure 5). They have no tritium and a moderate
concentration of radiogenic helium (10 – 50 mcc/kg). Model
ages, using the measured radiogenic
4
He accumulation rate
from the aquifer solids to the groundwater (0.02 mcc/kg
water
/
yr), suggest groundwater residence times of 100 – 500 years
(Dowling et al., submitted manuscript, 2002). In three cases
(BGD 6, 9, and 36), the wells lie beneath the arsenic,
methane-rich clusters at Faridpur and Laxmipur and yet the
wells do not contain elevated levels of either methane or
arsenic. In other wells with residence times of greater than 60
years (such as the shallow samples at Laxmipur and Faridpur
along with BGD 14, 15 and 29), As has a clear association
with CH
4
but not with age.
[23] In the very deepest parts of the groundwater systems
(>200 m), there is evidence of long groundwater residence
times (>1000 yrs) with elevated levels of
4
He and CH
4
and
background concentrations of As (<10 nM). Although the
d
13
C of this methane was not analyzed, the most likely
source of this methane is from thermogenic breakdown of
deep organic matter in the Bengal Fan. The association of
4
He and CH
4
is more typical of thermogenic rather than
biogenic natural gas. Groundwater in this deeper zone (see
Figure 5) is partially separated from the shallow (<200 m)
system by an aquitard of marine clay (estimated to be
>40,000 years based on the sedimentation rate of 5 mm/
yr). Based on the accumulation of helium and methane in
this lower aquifer, the recharge to this lower unit is
estimated at less than 1 cm/yr. With the exception of
BGD-8 (As = 2.53 mM) and BGD-23 (As = 0.23 mM), all
deep wells (>200 m) contain background levels of As.
Based on the chemistry of BGD-8, there is reason to believe
that the wells screen both shallow, As-rich, CH
4
-rich water
(30– 50m) and deep He-rich, As and CH
4
-free waters. There
is an adjacent well, BGD-9, that taps only As-free, deep
Figure 3. Dissolved As versus CH
4
, Fe, Mo, and NH
4
. Dissolved arsenic levels from all the wells in the
Bengal Basin are plotted against methane, iron, molybdenum, and ammonia. There are only weak to
modest correlations between arsenic and methane, iron, molybdenum, and ammonia.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-7
groundwater and has the same helium concentration and
isotope ratio as BGD-6 and BGD-8.
4.2. Sediment Data
[24] Sediments influence aqueous chemistry of the
ground and river water via adsorption and desorption
reactions on sediment surfaces (e.g. grain-size) and through
weathering of the sediment (e.g. mineralogy). To fully
examine the processes that control the dissolved levels of
As, Ba, Fe, Mo, and Mn in the groundwater, we measured
the adsorbed trace metals on river sediments (northern and
southern Ganges, northern and southern Brahmaputra), and
drill core samples from Laxmipur, Bangladesh (Figure 1) by
the standard oxalate extraction method. The oxalate proce-
dure [McKeague, 1978] removes adsorbed cations and
breaks down Mn and Fe oxy-hydroxides. Table 3a displays
the As, Ba, Fe, and Mn (Mo is not included because the
adsorbed concentrations are below the detection limit)
concentrations adsorbed on bulk river sediments (RW-53
and RW-54), river mica separates (Rishikesh and Guwa-
hati), and three fractions of the Laxmipur core (bulk, fine-
grained, and mica). Table 3b presents the As, Ba, Fe, and
Mn data from the organic fraction of the river and Laxmipur
core bulk samples and the digestion of the river fine-grained
fraction, river mica separates, and the Laxmipur core fine-
grained and mica fractions. In some of the drill core (23 m
and 39 m) and river (RW-53 and RW-54) samples, there was
insufficient mica for the ammonium oxalate extraction. The
exchangeable and adsorbed trace elements extraction pro-
cedure did not dissolve the carbonates because there was no
measurable calcium (<0.05 ppm) in the oxalate supernatant.
In the Laxmipur oxalate extract, magnesium was the dom-
inant cation present with a lesser amount of potassium and
sodium. The river sediment supernatant had magnesium as
the only measurable cation.
[25] Figure 6 displays the results of select trace elements
(As, Ba, Fe, and Mn) adsorbed on the Laxmipur sediment
core and dissolved in the Laxmipur groundwater versus
depth. The organic fraction is not included because the
average As in the organic fraction of the Laxmipur core is
0.66 ± 0.16 mmol/L (n = 3) and less than 5% of the amount
found in the oxalate extraction. Two C
organic
samples (Lax-
23 and Lax-48) have As levels below the detection limit,
suggesting that the oxidation of organic carbon itself
(<0.4% for all depths), although essential to the As release
mechanisms, is not a contributor itself in the overall As
budget in the Bengal Basin (Table 3b). The groundwater As,
Ba, and Fe increase with depth, until 30 m where fairly
constant values are reached. Manganese does not show a
similar trend suggesting that Mn quickly re-adsorbs onto the
aquifer protolith and the breakdown of Mn oxides is
probably not an important source of As or, alternatively,
arsenic is released by the breakdown of manganese oxides
but remains in solution while Mn re-adsorbs. The adsorbed
Figure 4. Dissolved As versus CH
4
, Fe, Mo, and NH
4
in Faridpur and Laxmipur. Dissolved arsenic
levels from wells located in Faridpur and Laxmipur are plotted against methane, iron, molybdenum, and
ammonia. There are reasonable correlations (r
2
= 0.8– 0.9) between arsenic and methane, iron, and
ammonia. The As-CH
4
and As-NH
4
correlations indicate that there are active microbes in the anoxic
groundwater. The As, CH
4
, and NH
4
associations combined with the Fe data support bacterial reduction
of FeOOH as the main release mechanism of As into the groundwater. The modest relation with Mo (r
2
=
0.6) indicates that adsorption reactions on sediment surfaces influence the trace metal concentrations in
the groundwater. The samples experiencing low levels of microbial activity (based on CH
4
levels) are
excluded from the As-Fe correlation.
X-8 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
As, Ba, Fe, and Mn have comparable patterns with depth
with the exception of Ba at the surface. The river sediments
with their high values of adsorbed Ba can easily supply the
barium to the groundwater through desorption (Dowling et
al., submitted manuscript, 2002). However, only active
weathering of the aquifer protolith can provide the As, Fe,
and Mn to the groundwater.
[26] Figure 6 clearly demonstrates the role that grain size
plays in controlling the solid phase distribution of many trace
elements. In the oxalate extraction of the sediment, the bulk
samples of the Laxmipur core have an average As concen-
tration of 35 ± 14 mmol/kg and a range from 12.5 to 51.4
mmol/kg whereas the river sediment averages 26.5 ± 8 mmol/
kg of As and varies between 20.6 and 32.4 mmol/kg. The
average arsenic on the fine-grained sediment samples (<5
mm) is about five times higher (165 ± 116 mmol/kg), and the
concentrations vary between 50.9 and 337.3 mmol/kg. The
pattern for As with depth is a direct function of grain size and
surface area in the bulk and fine-grained fractions. The major
drop in As, Ba, Fe, and Mn seen at 39 m corresponds to a
sandier interval that contains mostly quartz, smaller amount
of mica, and fewer clay-sized particles (<5 mm). The amount
of fine-grained material in the sediment influences the overall
concentration of the adsorbed trace metal in the bulk sedi-
ment.
[27] We use As-Fe molar ratios to examine the correla-
tions between As and Fe in the solid phase as well as in
solution. In most environments (pH < 8.5), iron oxy-hydrox-
ides have a positive surface charge and will preferentially
adsorb anions (like AsO
4
3
). The adsorbed and dissolved
iron-arsenic ratios should be similar if the sources and
processes that affect the solubility and adsorption of As
and Fe are the same. Typically, manganese-arsenic ratios are
not applicable since manganese oxides are often negatively
charged and will adsorb cations. However, Mn oxides have
been known to adsorb arsenic since the surface charge can be
altered by cation adsorption eventually leading to a positive
charge [Moi and Wai, 1994] and could be a minor source of
As. Figure 7 demonstrates the general correlation (r
2
= 0.72)
between adsorbed iron and adsorbed arsenic in the river and
Table 2. Dissolved Gas Data and Ages of the Bengal Basin Groundwaters
a
Sample
Name Sample Location Depth, m
Total
4
He,
mcc/kg
Radiogenic
4
He, mcc/kg
Ne,
mcc/kg
N
2
,
cc/kg
Ar,
cc/kg
CH
4
,
cc/kg R/Ra
3
H, TU
3
He/
3
H
Age, years
BGD 2 Sonapur 11 * * * * * * * 2.4 <20
BGD 3 Faridpur 20 18.6 9.1 62.2 5.4 0.2 23.6 0.90 <0.15 >80
BGD 4 Faridpur 30 18.9 0.0 77.1 5.8 0.2 20.6 0.94 <0.15 >80
BGD 5 Faridpur 40 25.1 3.6 103.2 6.6 0.2 11.9 0.92 <0.15 >80
BGD 6 Faridpur 150 99.1 72.9 106.5 8.1 0.2 0.3 0.53 <0.15 >80
BGD 7 Faridpur 50 36.9 2.9 140.8 8.9 0.2 6.8 0.96 <0.15 >80
BGD 8 Faridpur 244 53.2 29.9 109.8 7.6 0.2 28.4 0.55 <0.15 >80
BGD 9 Faridpur 98 53.6 28.9 119.4 7.6 0.2 0.2 0.56 <0.15 >80
BGD 10 Ahladipur 30 39.4 1.9 171.0 16.1 0.3 0.2 2.51 8.4 28.5
BGD 11 Rajbari 48 * * * * * * * 4.2 <20
BGD 12 Rajbari 33 * * * * * * * 6.2 30
BGD 13 Rajbari 127 66.2 1.0 260.9 14.6 0.3 0.2 1.03 0.5 14.7
BGD 14 Talma 44 55.9 28.7 101.0 5.8 0.2 16.4 0.49 * *
BGD 15 Nagarkanda 29 52.1 5.0 184.2 10.9 0.3 6.0 0.85 0.3 >80
BGD 16 Samta 9 26.5 3.2 109.6 9.7 0.2 3.9 1.84 8.4 17.9
BGD 17 Samta 49 44.6 5.6 191.4 11.2 0.3 6.6 1.03 0.2 40.1
BGD 18 Samta 61 * * * * * * * <0.15 >80
BGD 20 Jhenidah 100 65.6 6.5 271.9 14.5 0.3 0.1 1.78 5.8 30.3
BGD 21 Burir Char 61 51.7 2.7 210.8 11.5 0.3 0.2 2.39 4.8 40.0
BGD 22 Burir Char 152 * * * * * * * <0.15 *
BGD 23 Burir Char 280 5302.1 5276.0 116.8 10.7 0.3 173.2 0.09 <0.15 >100
BGD 24 Barishal 335 3277.8 3109.4 610.7 36.7 0.6 12.4 0.30 <0.15 >100
BGD 25 Barishal 290 895.2 845.1 200.0 14.6 0.2 3.6 0.29 <0.15 >100
BGD 26 Madaripur 30 53.8 5.1 233.1 14.0 0.3 0.6 1.05 3.4 8.4
BGD 27 Choto Arjundi 27 40.1 6.7 169.2 8.9 0.2 4.6 1.17 0.8 36.6
BGD 28 Mugraoara * * * * * * * * <0.15 *
BGD 29 Patalpara * 51.0 12.4 144.1 9.0 0.2 14.5 0.76 0.3 8
BGD 30 Adampur 91 311.5 265.8 184.9 11.6 0.3 0.0 0.45 <0.15 >80
BGD 31 Laxmipur 50 3.5 8.6
b
8.6 1.1 0.0 34.7 0.76 0.4 >80
BGD 32 Laxmipur 40 3.9 8.9
b
8.9 1.4 0.1 46.1 0.66 0.2 >80
BGD 33 Laxmipur 30 3.8 8.9
b
8.4 1.5 0.1 42.6 0.73 <0.15 >80
BGD 34 Laxmipur 20 8.8 1.3 23.0 2.4 0.1 38.9 0.79 0.2 >80
BGD 35 Laxmipur 10 * * * * * * * 2.8 <20
BGD 36 Laxmipur 150 73.2 26.3 180.6 9.8 0.2 0.1 0.59 0.4 51.9
BGD 40 Nawabganj 34 * * * * * * * <0.15 *
BGD 44 Baghail Dotala Sako 35 67.3 1.5 260.7 16.4 0.4 0.1 1.45 5.3 27.1
BGD 48 Refugee Para 27 53.2 4.5 223.8 14.1 0.3 0.1 1.32 2.7 23.9
BGD 53 Kustia 37 82.0 25.6 231.8 18.5 0.4 4.3 1.00 <0.15 >80
IND 1 Moyna 50 57.5 1.9 228.2 13.5 0.3 1.5 1.41 3.3 28.5
IND 3 Moyna 47.2 60.3 2.0 227.0 14.0 0.3 0.0 1.18 0.6 46.3
IND 4 Moyna 18.3 78.3 4.1 297.4 15.1 0.3 12.7 1.15 1.8 23.0
IND 6 Birohi 18.3 40.4 2.2 167.2 10.4 0.2 5.0 1.85 3.7 32.1
IND 9 Baruipur 293 254.6 188.5 261.4 15.4 0.4 0.2 0.37 <0.15 >100
a
The sample locations are shown in Figure 1. An asterisk means data were not available.
b
See explanation in Dowling et al. (submitted manuscript, 2002).
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-9
core sediments signifying that the sources of As and Fe in all
solid phases (bulk, fine-grained, and mica) at a given depth
may be the same—the dissolution of FeOOH. Overall, the
bulk sediment has the lowest As and Fe levels (e.g. the sandy
Lax-39-B) and the fine-grained material has the highest As-
Fe concentrations (e.g. Lax-49-F), which is indicative of the
influential role that grain size plays in trace metal concen-
trations.
[28] Figure 8 displays the adsorbed As/Fe molar ratios of
the river and core sediments plotted against depth. The
average ratio of the river bulk (0.0017 ± 0.0006) is similar
to that of the river fine-grained (0.0016 ± 0.0001) fraction.
The As/Fe molar ratios of the Laxmipur core fractions (bulk
and fine-grained) vary between 0.0005 and 0.0023 with a
pronounced decrease with depth resulting from the older,
deeper sediments having more fresh water flowing through
and removing some of the adsorbed As. In comparison, the
British Geological Survey and Mott MacDonald Ltd. [1999]
found that the sediments (from 3 m to 150 m deep) in
Nawabganj, an arsenic hotspot in northwest Bangladesh,
have generally lower and more variable As/Fe molar ratios
between 0.0001 and 0.0019. The variability may relate to
the age and hydraulic conductivity of the sediments.
[29] Even though the Laxmipur sediment is dominated by
iron-stained quartz, there is considerable mica (5 – 15% of
the samples), partially altered to vermiculite, found in the
sediment. The major and trace element chemistry data
imply that the weathering of the mica can have a significant
impact on the groundwater chemistry. The average
exchangeable arsenic in the Laxmipur mica separates (grain
size: 150– 250 mm) is 53.6 ± 4 mmol/kg which is slightly
higher than the bulk sediment but much lower than the fine-
grained sediments (Tables 3a and 3b) and is most likely
related to the FeOOH coatings on the mica grains. How-
ever, the complete digestion of Laxmipur mica after the
oxalate extraction shows elevated levels of arsenic (aver-
age: 239.8 ± 73 mmol/kg; 17.99 ppm) or about five times
the fine-grained sediment. Thus weathering can provide an
additional source of As to the groundwater via silicate
weathering. Pristine mica from both northern river sedi-
ments (northern Ganges River (Rishikesh, India) and north-
ern Brahmaputra (Guwahati, India)) contains a sizeable
reservoir of both extractable As and total As. The Rishikesh
mica has 14 times more extractable arsenic than the
Guwahati mica (294.8 versus 21.5 mmol/kg) but the differ-
ence in ‘‘structural’’ As (digestion after oxalate extraction)
is considerably less (239 versus 120 mmol/kg). Based on the
As concentrations and the percentage of mica in the sedi-
ment, the weathering of this arsenic-laden mica throughout
the GBR floodplain may supply a significant portion of the
Figure 5. Schematic cross section of groundwater flow in the Bengal Basin along the NW-SE (A-A
0
)
line. All wells are projected to the A-A
0
line in Figure 1. There are many municipal wells (up to 150 m)
that contain tritium and have a
3
He/
3
H groundwater age. Shallow wells at Laxmipur, Faridpur, and Kustia
are tritium dead with high biogenic methane. Deeper wells in the upper aquifer, with no tritium and
moderate amounts of radiogenic
4
He, are more than 100 years old. In the lower aquifer, the groundwater
has no tritium, elevated methane and helium concentrations, and residence times of greater than 1000
years.
X-10 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
Table 3a. Results of the Sediment Oxalate Extractions
a
Samples Oxalate Extractions K
d
Values, mL/g
Name
Depth,
m
Bulk As,
mmol/kg
Bulk Ba,
mmol/kg
Bulk Fe,
mmol/kg
Bulk Mn,
mmol/kg
Fine-
Grained
As,
mmol/kg
Fine-
Grained
Ba,
mmol/kg
Fine-
Grained
Fe,
mmol/kg
Fine-
Grained
Mn,
mmol/kg
Mica As,
mmol/kg
Mica Ba,
mmol/kg
Mica Fe,
mmol/kg
Mica Mn,
mmol/kg
Bulk
Sediment
b
As
Bulk
Sediment
b
Ba
Bulk
Sediment
b
Fe
Bulk
Sediment
b
Mn
River
RW-53 0 20.60 107.09 15.86 2.06 84.53 352.9 54.81 10.04 ** ** ** ** 103.0 716.2 *** ***
RW-54 0 32.44 127.87 14.99 3.39 63.32 183.1 37.45 10.46 ** ** ** ** 162.2 855.1 *** ***
Rishikesh0++++++++ 294.8 5.54 29.17 NM ++++
Guatahi 0 ++++++++ 21.5380.5615.99NM++++
Core
LAX 03 3 37.23 28.51 23.30 1.90 214.56 118.39 93.84 12.18 57.82 46.54 53.71 NM *** *** *** ***
LAX 14 14 35.73 37.18 22.70 2.39 75.16 149.63 68.25 13.46 52.84 34.80 63.57 NM 17.3 80.5 1586.2 234.3
LAX 23 23 38.22 55.32 43.15 3.77 148.08 153.43 143.07 13.09 ** ** ** ** 6.6 22.5 528.3 198.9
LAX 39 39 12.51 27.73 17.82 0.77 50.91 155.88 98.37 12.35 ** ** ** ** 1.6 6.8 160.7 69.9
LAX 48 48 51.38 45.99 103.46 2.20 337.28 224.73 472.70 28.18 50.19 35.17 95.27 NM 5.7 17.8 1108.5 264.0
a
NM, not measured; +, did not perform extractions on the bulk or fine-grained fractions; **, not enough mica for extractions; ***, no water data available.
b
Used average GBR surface and corresponding Laxmipur groundwater concentrations to calculate distribution coefficients.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-11
Table 3b. Results of the Sediment Digestions and the Organic Fraction Extraction
c
Samples Organic Fraction Microwave Digestion
Name
Depth,
m Bulk C
organic
,%
Bulk As,
mmol/kg
Bulk Ba,
mmol/kg
Bulk Fe,
mmol/kg
Bulk Mn,
mmol/kg
Fine-
Grained
As,
mmol/kg
Fine-
Grained
Ba,
mmol/kg
Fine-
Grained
Fe,
mmol/kg
Fine-
Grained
Mn,
mmol/kg
Mica As,
mmol/kg
Mica Ba,
mmol/kg
Mica Fe,
mmol/kg
Mica Mn,
mmol/kg
River
RW-53 0 NM bdl 28.92 NM 0.40 15.72 276.10 NM 0.82 ** ** ** **
RW-54 0 NM 1.10 11.62 NM 0.01 23.22 424.62 NM 0.90 ** ** ** **
Rishikesh0 NM+ + ++++++ 239.02 2095.42 NM NM
Guatahi 0 NM + + ++++++ 120.03 1450.96 NM NM
Core
LAX 03 3 0.2 0.50 22.33 NM 0.52 42.82 2296.22 NM 10.38 291.05 1102.85 NM NM
LAX 14 14 0.19 0.81 23.64 NM 0.46 52.69 559.73 NM 5.47 271.93 1551.88 NM NM
LAX 23 23 0.36 bdl 46.07 NM 0.75 143.82 620.91 NM 7.43 ** ** ** **
LAX 39 39 0.06 0.67 10.99 NM 0.15 bdl 1275.79 NM 3.80 ** ** ** **
LAX 48 48 0.14 bdl 34.43 NM 0.78 92.67 1576.08 NM 11.06 156.43 2135.41 NM NM
c
NM, not measured; **, not enough mica for digestions; +, did not perform digestions or organic fraction extractions on the bulk or fine-grained fractions.
X-12 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
arsenic that ultimately adsorbs on the FeOOH, thus making
it the definitive source of arsenic released to the ground-
water. However, we cannot easily distinguish between the
two possible sources of arsenic: (1) the arsenic adsorbed
onto FeOOH after weathering out of high-arsenic mica in
the aquifer, (2) the arsenic from As-bearing iron oxy-
hydroxides that flocculated at the seawater/freshwater inter-
face.
5. Discussion
5.1. Microbial Activity
[30] From the analysis of the correlation plots (Figures 3
and 4), we can begin to interpret the variable As concen-
trations in the Bangladeshi groundwaters. Several studies,
including this one, have observed that the highest As
concentrations are located in the shallowest sediments
(<60 m) [Acharyya et al., 1999; British Geological Survey
and Mott MacDonald Ltd., 1999; Chowdhury et al., 1999;
Nickson et al., 2000; McArthur et al., 2001]. These sedi-
ments began to be deposited about 11,000 years ago and
have had the smallest amount of groundwater flushing
through them compared to the rest of the stratigraphic
column. As Figure 8 demonstrates, there is a general
decrease in the As/Fe molar ratios with depth, probably
resulting from more water flowing through the lower sedi-
ments thus removing some of the adsorbed As. Assuming
no further addition of arsenic by weathering of mica and
using the Laxmipur recharge rate and dissolved As levels, it
would take 300– 500 years and 15 L of As-free ground-
water flushing through 1 L of aquifer to lower the dissolved
As concentration below 0.5 mmol in the Laxmipur wells. In
the deeper samples, there appears to be a limit on dissolved
As and a finite supply of As even with high levels of
dissolved Fe released to the groundwater.
[31] When we examined individual sites (e.g., Laxmipur,
Faridpur), the positive correlations between arsenic and
methane, iron, and ammonia became readily apparent. The
samples in Laxmipur and Faridpur (Figure 4) have methane
levels greater than 30 mmol/L and a strong correlation
between Fe and As (r
2
= 0.8). The average Laxmipur
groundwater As/Fe molar ratio is 0.090 ± 0.030 or hundred
times greater than the adsorbed As/Fe molar ratios, which
indicates that the As release is not a simple stoichiometric
Figure 6. Groundwater and sediment oxalate extraction data plotted against depth at Laxmipur,
Bangladesh. Select dissolved and adsorbed data from the water and sediment oxalate extraction analyses
(arsenic, barium, iron, and manganese) are plotted against depth at Laxmipur. The concentrations of
dissolved As, Fe, Mn, and Ba are 5 times less than the adsorbed bulk fraction. Adsorbed As, Fe, and Mn
display a similar trend with one another while barium does not. Grain size plays an important role in
controlling the concentrations of trace metals dissolved in the groundwater and adsorbed onto sediment.
The fine-grained sediment shows the same pattern as the bulk and mica; however, the patterns are more
pronounced and the concentrations are higher. The drop in the adsorbed metals seen at 39 m corresponds
to a sandier interval in the drill core.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-13
FeOOH breakdown. Adsorption of dissolved Fe and/or the
precipitation of siderite may occur. Once siderite (FeCO
3
)
solubility is reached in the groundwater, dissolved iron
levels should no longer increase because of mineral precip-
itation. Curiously, the dissolved As/Fe molar ratios in the
groundwater are relatively consistent and independent of the
arsenic and iron concentrations. Even though arsenic is
more soluble in groundwater than iron, the arsenate and
arsenite oxy-anions are still subject to adsorption/desorption
reactions on mineral surfaces. The modest correlation of
total As with the molybdate oxy-anion, MoO
4
2
(r
2
= 0.6)
may reflect the influence of surface adsorption. Other redox
Figure 7. Adsorbed Fe versus adsorbed As in the sediment fractions. There is an overall correlation (r
2
=
0.72) between adsorbed Fe and As indicating that Fe and As come from a similar source. It also points out
the importance of grain size on trace metals because the bulk sediment has the lowest concentrations of As
and Fe while the fine-grained fraction has the highest levels.
X-14 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
sensitive species, such as methane and ammonia, that are
generated by bacterial reduction of organic matter, display
strong correlation coefficients with As (r
2
= 0.9). Neither
methane nor ammonia are adsorbed on mineral surfaces
(unlike MoO
4
2
) or released by the breakdown of FeOOH.
We agree with the hypothesis put forth by McArthur et al.
[2001] who propose that the biogenic decomposition of
organic matter is the primary driver behind the dissimilatory
reduction of iron and the presence of arsenic, phosphate,
methane, and ammonia in the groundwater.
[32]Cummings et al. [1999] have shown that As can be
released into solution by bacterially reductive dissolution of
iron. Shewanella alga BrY, a dissimilatory iron-reducing
bacterium, facilitates As mobilization from crystalline ferric
arsenate as well as from sediment sorption sites [Cummings
et al., 1999]. Even though Shewanella alga BrY may cause
As to be mobilized in both synthetic substrates and natural
sediment, the oxidization state of arsenic [As(V)] remains
unaltered. It may be that As(III), the more mobile and toxic
arsenic, in groundwater is a twofold process. The iron-
reducing bacteria introduce arsenic into the solution through
the dissolution of FeOOH. If the arsenic is dominantly
As(V), then arsenic reduction happens either directly from
arsenic reducing bacteria or indirectly by chemical reduc-
Figure 8. Adsorbed As-Fe molar ratios plotted against depth. There is an overall decrease in the ratios
with depth, resulting from more groundwater flushing the lower sediments and removing some of the
adsorbed As from the system.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-15
tion [Cummings et al., 1999]. Ahmann et al. [1994] reports
that a microorganism, MIT-13, can reduce arsenate [As(V)]
to arsenite [As(III)]. Those organisms that are capable of
both Fe(III) and As(V) reduction, such as Geospirillum
barnesii SES-3, would promote rapid As mobilization
[Laverman et al., 1995]. These microbes could release As
into solution and reduce the available iron oxides to adsorb
arsenic, thereby forcing As to stay in solution.
5.2. Modeling
[33] We modeled the groundwater system using our dis-
solved data from the Laxmipur well cluster by constructing a
1-D flow/chemical reactor model to evaluate the rate of iron
and arsenic release (Figure 9) [Peters,2001].Weused
simplifying assumptions based on CaCO
3
equilibrium for
a given P
CO2
and measured bicarbonate values. The initial
condition of P
CO2
, and the rates at which FeOOH and O
2
are
consumed are important variables that influence the micro-
bial mediated breakdown of FeOOH, and the subsequent
release of arsenic in the groundwater. We used literature
values for the constant parameters (equilibrium constants
and activity coefficients). The dynamic variables are the
rates of CaCO
3
dissolution, FeCO
3
precipitation, the expo-
nential decay of Fe production (l), consumption of dis-
solved O
2
in the aerobic zone (k
bug 1
), and biological release
of iron via the breakdown of FeOOH in the anaerobic zone
(k
bug 2
). We ran the model with several combinations and
ranges of variables to establish the best fit to the field data.
[34] Rainwater percolates into the vadose zone, and the
water equilibrates with the elevated soil CO
2
from plant
respiration and the CaCO
3
in the vadose zone yielding high
dissolved HCO
3
and P
CO2
as initial conditions. The water
must enter the phreatic zone saturated with CaCO
3
and at
high P
CO2
or the model [H
+
] and total CO
2
at depth will be
too low. Once the water further flows into the phreatic zone
(6 m), it becomes isolated from exchange with the soil
gas. Aerobic organisms (bug zone 1) consume the dissolved
oxygen (O
2
= 0.3 mM) in the groundwater to metabolize
organic matter and release CO
2
. The addition of carbon
dioxide in equilibrium with calcium carbonate increases the
concentration of bicarbonate in the groundwater. After
aerobic microbes consume all the dissolved oxygen, the
Figure 9. Schematic diagram of the Bengal Basin groundwater system. The rainwater, in equilibrium
with atmospheric gas, percolates into the vadose zone where the water dissolves carbonates and other
soluble minerals from the soil. The water further flows into the phreatic zone (6 m) where it becomes
isolated from exchange with the atmosphere. The aerobic organisms consume the oxygen to metabolize
organic matter and introduce carbon dioxide into the groundwater. Once all the oxygen is used, anaerobic
microbes break down iron oxy-hydroxides for energy which introduces CO
2
, Fe, and As into the
groundwater. The concentration of dissolved arsenic is controlled by biological activity and adsorption
reactions. Not drawn to scale.
X-16 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
anaerobic microbes (bug zone 2) become the dominant
carbon dioxide producers by breaking down iron oxy-
hydroxides to oxidize organic matter [4FeOOH +
3CH
2
O (bug 2) !4Fe
2+
+5H
2
O + 3CO
2
]. The bacterial
dissimilatory reduction of FeOOH releases CO
2
, Fe, and
any adsorbed trace metals (e.g. As) into the groundwater.
Even though methanogens are active in the groundwater
and generate both CO
2
and CH
4
(2C
organic
+2H
2
O=CO
2
+
CH
4
), we do not consider them in our model since they do
not produce Fe or As during the decomposition of organic
carbon. However, the dissolved CH
4
,NH
4
, Fe, and As
correlations strongly relate microbial activity and the respi-
ration of organic carbon to the dissimilatory reduction of
FeOOH and the subsequently high concentrations of dis-
solved As.
[35] The dissolved data and the model results are dis-
played in Figure 10. The system of equations are written as
follows [Peters, 2001]:
CO2ðgÞþH2OKCO2
! H2CO3
PCO2CG
KCO2 ¼101:47
H2CO3
K1
! HþþHCO
3
CGCHCB
K1¼106:35
HCO
3
K2
! HþþCO2
3
CBCHCC
K2¼1010:33
CaCO3ðsÞKCal
! Ca2þþCO2
3
CCa CC
KCal ¼108:48
Fe2þþCO2
3
KFe
! FeCO3ðsÞ
CFe CC
KFe ¼1010:89
CH2OþO2
KBUG 1
! CO2þH2O
Bug 1 COCG
3CH2Oþ4FeOOHðAsÞKBUG 2
! 3CO2þ4Fe2þþ5H2O
Bug 2 CFO CGCF
We assume that the first equation occurs instantaneously;
therefore it has been omitted when creating the mass
balances and replaced with C
G
=P
CO2
*K
CO2
. The mass
balances on each species are as follows [Peters, 2001]:
Figure 10. Comparison of model results and measured data at the Laxmipur well cluster. (a) The natural
iron production slows with depth due to an unknown biological factor (l) such as As poisoning, a
limiting reagent that inhibits the microbial breakdown of FeOOH, organic carbon availability, or another
unknown factor. The model iron concentration is most similar to the field data when we use a lvalue of
0.049 yr
1
. (b) The rate that aerobic microbes consume dissolved O
2
affects the depth at which aerobes
are replaced by anaerobes. Based on our modeling, the change from aerobic to anaerobic activity is made
at 9.0 ± 0.5 m. The bacterial breakdown of FeOOH produces 20 mM/yr of dissolved iron and represents
our best model fit based on the data.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-17
dCG
dt ¼K1k1mCGþk1mgHCO
3CBCHþkbug1þkbug 2
dCB
dt ¼K1k1mCGk1mgHCO
3CBCHK2k2mgHCO
3
CBþk2mgCO2
3CCCH
dCH
dt ¼K1k1mCGk1mgHCO
3CBCH
þK2k2mgHCO
3CBk2mgCO2
3CCCH
dCC
dt ¼K2k2mgHCO
3CBk2mgCO2
3CCCH
þkCal KCal
gCO2
3CC
gCa2þCCa
!
þkFe KFe
gCO2
3CC
gFe2þCFe
!
dCCa
dt ¼kdiss K3
gCO2
3CC
gCa2þCCa
!
dCFe
dt ¼4
3kbug 2elt
þkFe KFe
gCO2
3CC
gFe2þCFe
!
dCO
dt ¼kbug 1:
We choose an initial P
CO2
of 10
1.5
atm to match our field
pH and bicarbonate values at calcium carbonate saturation,
and we use our measured groundwater velocity of 0.4 m/yr.
Bicarbonate and calcium concentrations have the greatest
discrepancies between the model results and the measured
data since we have not accounted for all the sources of
solid-phase carbonates (e.g. dolomite) or the high levels of
calcium from seawater flooding. The best fit to the data
requires that the natural iron production decreases exponen-
tially (l), caused by either As poisoning, a limiting reagent
such as Mo that inhibits the bacterial breakdown of FeOOH,
organic carbon availability, or another unknown factor.
[36] The rate that aerobic microbes (k
bug 1
) consume O
2
affects the depth at which aerobes are replaced by anaerobes.
Figure 10b shows that the best fit for the change from aerobic
to anaerobic activity occurs at 9.0 ± 0.5 m (3 m below the
water table). The 20 mM/yr generation of dissolved iron via
bacterial breakdown represents our best model fit based on
our measured data. If the average measured dissolved As/Fe
is 0.09 ± 0.03, then the rate at which As is liberated into the
groundwater is 1.8 ± 0.6 mM/yr. Based on the rapidity of the
microbial mediated breakdown of the FeOOH, its subse-
quent release of arsenic, the elevated levels of adsorbed As,
and the rate of groundwater flow, the high levels of dissolved
arsenic in the groundwater will be a long-term problem in the
shallow wells of the Bengal Basin.
5.3. Dual Permeability
[37] Another important factor to consider is the influential
role of sediment permeability in the release and distribution
of arsenic to the groundwater. The highest levels of dis-
solved As are observed in the shallow fine-grained organic-
rich sediments of Faridpur and Laxmipur. However, many
of the wells that supply the villages in the Bengal Basin
have mildly anaerobic waters (field Eh of 0 to +150 mv and
arsenate/arsenite ratios of 1) that are not conducive to CH
4
and NH
4
production but still have elevated As which
correlates with methane and ammonia. These water supply
wells are screened in high permeability, sandy zones (e.g.,
BGD 14, 15, 29) in contrast to the low permeable high-As
horizons of Faridpur and Laxmipur. We would propose that
the levels of As, CH
4
, and NH
4
dissolved in the ground-
water are controlled by the relative proportions of low and
high permeability sediments. The dissolved As, CH
4
, and
NH
4
, are diffusing from clay-rich areas into the sandy units
with low levels or no As, CH
4
, and NH
4
. The Fe does not
spread into the sandy layers as readily since it re-precipitates
either as FeCO
3
or FeOOH.
[38] Understanding the spatial distribution of arsenic on
the sediments from As-bearing localities will also have a
strong bearing on our conceptual model of As release. A
preliminary examination of five samples from a single core
(3 to 49 m depth) at Laxmipur reveals an average value of
35 ± 14 mM of adsorbed As on the bulk sediments while the
associated waters display an average As concentration of
6.8 ± 3 mM. In addition, the sandy modern day sediments,
collected near the confluence of the Ganges and Brahmapu-
tra, contain a comparable amount of adsorbed (oxalate-
extractable) arsenic (average 26.5 ± 8 mM). Thus, in the
case of these bulk sediment samples from Laxmipur drill
core and the sediments from the Ganges and Brahmaputra
riverbeds, the analyzed bulk sediments do appear capable of
supplying all of the dissolved arsenic to the water. It is most
likely coming from the finer-grained sediments (e.g. clay-
sized) that contain five times more sorbed As and Fe than
the bulk sediments or the mica separates. In the British
Geological Survey and Mott MacDonald Ltd. [1999] study,
the highest levels of adsorbed As (released by oxalate
extraction) also occur in the clay-rich layers as opposed to
the more sandy lenses within the Nawabganj core.
[39] The fine-grained As-bearing FeOOH colloids are
typically deposited in low-energy environments at the sea-
water/freshwater interface (i.e. an estuary or tidal channel).
The flocculation hypothesis also accounts for the geograph-
ical concentration of As-bearing wells that exist in a broad
band trending WNW-ESE, subparallel to the course of the
Ganges (Figure 1) that presumably coincides with the
river– seawater paleo-transition zone. Another theory com-
patible with our results is that the distribution of peat and
organic matter in the subsurface controls the As release and
distribution in the groundwater [McArthur et al., 2001]. Peat
and organic matter are found at depth in the Bengal Basin
[Umitsu, 1993; Goodbred and Kuehl, 2000] and forms in
the low energy environments as organic matter collects in
estuaries or wetlands rather than active river channels.
Ultimately, they form low permeability zones within the
subsurface and a source of energy for microbes.
[40] An additional reason for the present distribution of
As-bearing groundwater is the Recent age of the shallow
sediments. Sediments older than 10,000 years old were
deposited when sea level was lower and large quantities of
fresh water flowed through the sediments removing nearly
all adsorbed As. If the average present day recharge rate of
60 ± 20 cm/yr existed for the past 100 kyr, the sediments
from 50 to 250 m depth have been flushed by more than 100
X-18 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
pore volumes of water. Conversely, most of the shallow
sediments (<60 m), deposited since the last significant drop
in sea level, have seen fewer than 20 pore volumes. The
volume of water flushing through fine-grained, low perme-
ability sediments would be considerably less.
[41] The ultimate source of arsenic in the groundwater
may be from the weathering of micas and the result of a
multistage weathering process. Mica, with its high intrinsic
arsenic, weathers into clay and releases arsenic and other
trace elements into the water. The As is scavenged by
FeOOH and flocculates as fine-grained particles in estuaries
and other low energy environments. The XAFS (X-ray
absorption fine structure) data of Foster et al. [2000] show
that the more toxic As(III) species dominate the upper gray
to black micaceous fine-grained sediment (6 –25 m) in
Ramrail, Brahmanbaria (near Dhaka) while the less mobile
As(V) is found sorbed onto FeOOH in the oxidized quartz
and weathered mica sediment (26 – 48 m depth). Through
sediment desorption reactions, bacterial dissolution of
FeOOH, and the microbial mediated transformation of
As(V) to As(III), As(III) is being introduced into the
groundwater. However, we are unable to distinguish
between As coming from the As-FeOOH colloids that were
flocculated at the freshwater/saltwater interface and the
arsenic adsorbed on FeOOH after the As was weathered
from the high-arsenic mica in the aquifer.
[42] We expect the arsenic concentrations to remain
above the WHO standard in post-glacial sediments for a
long period of time in As-bearing localities because of the
enhanced microbial activity in the groundwater, the distri-
bution of organic matter and peats, dual permeability sedi-
ments in the subsurface, the diffusion of As into the sandy
layers, and the young age of the shallow sediments.
6. Conclusions
[43] Dissolved arsenic in the groundwater of the Bengal
Basin represents a considerable health risk to a population
that relies on tube wells for their water supply. The majority
of high As groundwater wells in the Bengal Basin are
moderately reducing, have high dissolved Fe, CH
4
, and
NH
4
, and contain no dissolved O
2
or SO
4
. The As-laden
wells of Laxmipur and Faridpur have groundwater resi-
dence times of greater than 60 yrs, which eliminates the
causal link between the recent application of phosphatic
fertilizers and excessive extraction of groundwater for
irrigation in recent years and As release into the ground-
water. We believe that the arsenic is being liberated from the
sediment into the moderately reducing groundwater by the
microbial mediated reductive dissolution of iron oxy-
hydroxides. The strong correlations between high levels of
dissolved arsenic with iron, methane, and ammonia support
the bacterial breakdown of FeOOH.
[44] Physical adsorption and desorption reactions have
significant effects on the concentrations of dissolved trace
elements in groundwater. The analyzed bulk sediments do
appear capable of supplying all of the dissolved arsenic,
mostly through desorption and FeOOH dissolution, to the
groundwater. The fine-grained particles in Laxmipur core
have elevated levels of extractable arsenic compared to the
bulk sediments and mica separates, implying that the sur-
face area has a significant control on groundwater arsenic
levels. The adsorbed As-Fe molar ratios decrease with depth
throughout the stratigraphic column suggesting that some of
the arsenic has been flushed from the system since the lower
sediments have had more water flowing through them than
the upper layers. The data from the digestions of the mica
separates indicate that the weathering of mica at least
contributes and may be the ultimate source of arsenic.
Arsenic released from mica is subsequently adsorbed onto
iron oxy-hydroxides and now is being re-introduced into the
groundwater through the microbial dissolution of FeOOH.
However, we are unable to differentiate between the arsenic
coming from As-bearing iron oxy-hydroxides being floccu-
lated and the arsenic adsorbed on FeOOH after the As was
weathered from the aquifer mica. Our model illustrates that
the bacterial breakdown of iron oxy-hydroxides can release
As into the groundwater at a rate of 1.8 ± 0.6 mM/yr.
[45] The hydraulic conductivity of the sediments also has
a considerable impact on groundwater chemistry. Ground-
water residence times and subsurface stratigraphy suggest
that the presence of the As in water supply wells is the result
of diffusion out of the organic-rich clay into the more
permeable sandy zones. The data also suggests that the As
is mainly released from Holocene sediments since large
quantities of water have not flushed the dissolved arsenic
from the aquifers.
[46] The dissolved arsenic in the groundwater will remain
a long-term problem for the people of Bengal Basin
because of microbial dissolution of iron oxy-hydroxides,
the subsurface distribution of organic matter and peats, dual
sediment permeability, the diffusion limited release of As
into the sandy layers, and the young age of the shallow
sediments.
[47]Acknowledgments. We would like to thank the International
Atomic Energy Agency (IAEA), Vienna, Austria, for sponsoring the field
excursions in the Bengal Basin, and the Atomic Energy Agency and the
Water Development Board of Bangladesh (BWDB) and the Central
Groundwater Board of West Bengal, India for field support. In particular,
we are appreciative of Reazuddin Ahmed and Mizanur Rahman of the
BWDB, and K. M. Kulkarni, now of the IAEA in Vienna, for their
assistance and guidance in the field procurement of the groundwater
samples in Bangladesh and West Bengal State, India, respectively. We
thank Billy Moore from the University of South Carolina for the Ganges-
Brahmaputra River sediment samples, and Andrew G. Hunt, Magdalyn J.
Renz, and Gregory L. Wortman for lab assistance. We also would like to
express our thanks to the Associate Editor and two anonymous reviewers
for their constructive comments and suggestions. This research was
partially funded by NSF 9730743.
References
Acharyya, S. K., P. Chakraborty, S. Lahiri, B. C. Raymahashay, S. Guha,
and A. Bhowmilk, Arensic poisoning in the Ganges Delta, Nature,401,
545, 1999.
Acharyya, S. K., S. Lahiri, B. C. Raymahashay, and A. Bhowmilk, Arsenic
toxicity of groundwater in parts of the Bengal Basin in India and Ban-
gladesh: The role of Quaternary stratigraphy and Holocene sea-level
fluctuation, Environ. Geol.,39, 1127 – 1137, 2000.
Ahmann, D., A. L. Roberts, L. R. Krumholz, and F. M. M. Morel, Microbe
grows by reducing arsenic, Nature,371, 750, 1994.
Allison, M. A., Historical changes in the Ganges-Brahmaputra Delta front,
J. Coastal Res.,14, 1269– 1275, 1998.
Biswas, B. K., R. K. Dhar, G. Samanta, B. K. Mandal, D. Chakraborti,
I. Faruk, K. S. Islam, M. M. Chowdhury, A. Islam, and S. Roy, Detailed
study report of Samta, one of the arsenic-affected villages of Jessore
District, Bangladesh, Curr. Sci.,74, 134 – 145, 1998.
British Geological Survey and Mott MacDonald Ltd., Main report: Ground-
water studies for arsenic contamination in Bangladesh, Keyworth, UK,
1999.
DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER X-19
Chowdhury, T. R., et al., Arsenic poisoning in the Ganges Delta, Nature,
401, 545– 546, 1999.
Clarke, W. B., W. J. Jenkins, and Z. Top, Determination of tritium by mass
spectrometric measurement of
3
He, Int. J. Appl. Radiat. Isot.,27, 515–
522, 1976.
Coleman, J. M., Brahmaputra River: Channel processes and sedimentation,
Sediment. Geol.,3, 129– 239, 1969.
Cummings, D. E., F. Caccavo Jr., S. Fendorf, and R. F. Rosenzweig, Ar-
senic mobilization by the dissimilatory Fe(III)-reducing bacterium She-
wanella alga BrY, Environ. Sci. Technol.,33, 723 – 729, 1999.
Das, D., A. Chatterjee, B. K. Mandal, G. Samanta, D. Chakraborti, and B.
Chanda, Arsenic in ground water in six districts of West Bengal, India;
the biggest arsenic calamity in the world; part 2, Arsenic concentration in
drinking water, hair, nails, urine, skin-scale and liver tissue (biopsy) of
the affected people, in Speciation of Elements: Trace Elements in Human
Health, edited by Y. Thomassen and E. Nieboer, pp. 917 – 924, Chem.
Soc., Cambridge, England, 1995.
Das, D., G. Samanta, B. K. Mandal, T. R. Chowdhury, C. R. Chanda, P. P.
Chowdhury, G. K. Basu, and K. Chakraborti, Arsenic in groundwater in
six districts of West Bengal, India, Environ. Geochem. Health,18, 5 – 15,
1996.
Dhar, R. K., et al., Groundwater arsenic calamity in Bangladesh, Curr. Sci.,
73, 48– 59, 1997.
Fehn, U., G. Synder, and P. K. Egeberg, Dating of pore waters with
129
I;
relevance for the origin of marine gas hydrates, Science,289, 2332 –
2335, 2000.
Food and Agricultural Organization (FAO), Soil survey project, Bangla-
desh, AGL:SF/PAK 6, Tech. Rep. 3, Rome, 1971.
Foster, A. L., G. N. Breit, A. H. Welch, J. W. Whitney, J. C. Yount, M. S.
Islam, M. M. Alam, M. K. Islam, and M. N. Islam, In-situ identification
of arsenic species in soil and aquifer sediment from Ramrail, Brahman-
baria, Bangladesh, Eos Trans. AGU,81(48), Fall Meet. Suppl., abstract
H21D-01, 2000.
Goodbred, S. L., and S. A. Kuehl, The significance of large sediment
supply, active tectonism, and eustasy on margin sequence development:
Late Quaternary stratigraphy and evolution of the Ganges-Brahmaputra
Delta, Sediment. Geol.,133, 227– 248, 2000.
Greenberg, A. E., L. S. Clesceri, and A. D. Eaton, Standard Methods for the
Examination of Water and Wastewater, 18th ed., Am. Public Health
Assoc., Washington, D.C., 1992.
Holeman, J. N., The sediment yield of major rivers in the world, Water
Resour. Res.,4, 737 – 747, 1968.
Karim, M. M., Arsenic in groundwater and health problems in Bangladesh,
Water Res.,34, 304 – 310, 2000.
Laverman, A. M., J. S. Blum, J. K. Schaefer, E. J. P. Phillips, D. R. Lovley,
and R. S. Oremland, Growth of strain SES-3 with arsenate and other
diverse electron acceptors, Appl. Environ. Microbiol.,61, 3556– 3561,
1995.
Li, X., B. J. Coles, M. H. Ramsey, and I. Thornton, Sequential extraction of
soils for multielement analysis by ICP-AES, Chem. Geol.,124, 109 –
123, 1995.
Lindsay, J. F., D. W. Holliday, and A. G. Hulbert, Sequence stratigraphy
and the evolution of the Ganges-Brahmaputra Delta complex, AAPG
Bull.,75, 1233– 1254, 1991.
Long, S. E., and T. D. Martin, Method 200.8, Determination of trace ele-
ments in waters and wastes by inductively coupled plasma-mass spectro-
metry, in Methods for the Determination of Metals in Environmental
Samples, pp. 83 – 122, U.S. Environ. Prot. Agency, Washington, D.C.,
1991.
Mallick, S., and N. R. Rajagopal, Groundwater development in the arsenic
affected alluvial belt of West Bengal—Some questions, Curr. Sci.,70,
956 – 958, 1996.
Mandal, B. K., et al., Arsenic in groundwater in seven districts of West
Bengal, India—The biggest arsenic calamity in the world, Curr. Sci.,70,
976 – 986, 1996.
Mandal, B. K., G. T. R. Chowdhury, G. Samanta, D. Mukherjee, C. R.
Chanda, K. C. Saha, and D. Chakraborti, Impact of safe water for drink-
ing on five families for two years in West Bengal, India, Sci. Total
Environ,218, 185 – 201, 1998.
McArthur, J. M., P. Ravenscroft, S. Safiullah, and M. F. Thirlwall, Arsenic
in groundwater: Testing pollution mechanisms for sedimentary aquifers
in Bangladesh, Water Resour. Res.,37, 109– 117, 2001.
McKeague, J. A., Manual on soil sampling and methods of analysis, 2nd
ed., Can. Soc. of Soil Sci., Ottawa, Ontario, 1978.
Michels, K. H., H. R. Kudrass, C. Hubscher, A. Suckow, and M. Wiedicke,
The submarine delta of the Ganges-Brahmaputra: Cyclone-dominated
sedimentation patterns, Mar. Geol.,149, 133 – 154, 1998.
Milliman, J. D., and R. H. Meade, World-wide delivery of river sediment to
the oceans, J. Geol.,91, 1– 21, 1983.
Moi, W. M., and C. M. Wai, Mobilization of arsenic in contaminated river
sediment, in Arsenic in the Environment, part I, Cycling and Character-
ization, edited by J. Nriagu, pp. 99 – 118, John Wiley, New York, 1994.
Nickson, R., J. McArthur, W. Burgess, K. M. Ahmed, P. Ravenscroft, and
M. Rahman, Arsenic poisoning of Bangladesh groundwater, Nature,395,
338, 1998.
Nickson, R. T., J. M. McArthur, P. Ravenscroft, W. G. Burgess, and K. M.
Ahmed, Mechanism of arsenic release to groundwater, Bangladesh and
West Bengal, Appl. Geochem.,15, 403 – 413, 2000.
Peters, S. L., Modeling the mechanism for arsenic release in the ground-
water of Bangladesh, M.Sc., Univ. of Rochester, Rochester, New York,
2001.
Poreda, R. J., and K. A. Farley, Rare gases in Samoan xenoliths, Earth
Planet. Sci. Lett.,113 , 129 – 144, 1992.
Poreda, R. J., T. E. Cerling, and D. K. Solomon, Tritium and helium iso-
topes as hydrologic tracers in a shallow unconfined aquifer, J. Hydrol.,
103, 1– 9, 1988.
Sarin, M. M., S. Krishnaswami, K. Dilli, B. L. K. Somayajulu, and W. S.
Moore, Major ion chemistry of the Ganges-Brahmaputra river system:
Weathering processes and fluxes to the Bay of Bengal, Geochim. Cos-
mochim. Acta,53, 997– 1009, 1989.
Smith, A. H., E. O. Lingas, and M. Rahman, Contamination of drinking
water by arsenic in Bangladesh: A public health emergency, Bull. World
Health Org.,78, 1093 – 1103, 2000.
Solomon, D. K., R. J. Poreda, S. L. Schiff, and J. A. Cherry, Tritium and
helium 3 as groundwater age tracers in the Borden aquifer, Water Resour.
Res.,28, 741– 755, 1992.
Solomon, D. K., S. L. Schiff, R. J. Poreda, and W. B. Clarke, A validation
of the
3
H/
3
He method for determining groundwater recharge, Water Re-
sour. Res.,29, 2951 – 2962, 1993.
Subramanian, V., Chemical and suspended sediment characteristics of rivers
of India, J. Hydrol.,44, 37 – 55, 1979.
To, T. B., D. K. Nordstrom, K. M. Cunningham, J. W. Ball, and R. B.
McCleskey, New method for the direct determination of dissolved Fe(III)
concentration in acid mine waters, Environ. Sci. Technol.,33, 807 – 813,
1999.
Uddin, A., and N. Lundberg, Cenozoic history of the Himalayan-Bengal
system: Sand composition in the Bengal Basin, Banglasdesh, Geol. Soc.
Am. Bull.,110 , 497 – 511, 1998.
Umitsu, M., Late Quaternary sedimentary environments and landforms in
the Ganges Delta, Sediment. Geol.,83, 177– 186, 1993.
P. K. Aggarwal, International Atomic Energy Agency, A-1400, Vienna,
Austria.
A. R. Basu, S. L. Peters, and R. J. Poreda, Department of Earth and
Environmental Sciences, University of Rochester, Rochester, NY 14627,
USA. (carolyn@earth.Rochester.edu)
C. B. Dowling, Byrd Polar Research Center, Ohio State University, 108
Scott Hall, 1090 Carmack Road, Columbus, OH 43210-1002, USA.
(dowling.37@osu.edu)
X-20 DOWLING ET AL.: ARSENIC RELEASE IN BENGAL BASIN GROUNDWATER
