NATURAL ARSENIC IN GROUNDWATER: OCCURRENCE, REMEDIATION AND MANAGEMENT - CHAPTER 16 pptx

Natural Arsenic in Groundwater: Occurrence, Remediation and Management – Bundschuh, Bhattacharya and Chandrasekharam eds © 2005, Taylor & Francis Group, London, ISBN 04 1536 700 X Natura

Section 3: Arsenic biogeochemistry in

groundwater

Natural Arsenic in Groundwater: Occurrence, Remediation and Management –

Bundschuh, Bhattacharya and Chandrasekharam (eds)

© 2005, Taylor & Francis Group, London, ISBN 04 1536 700 X

Natural enrichment of arsenic in groundwaters of Brahmanbaria

district, Bangladesh: geochemistry, speciation modeling and

multivariate statistics

Ondra Sracek

Institute of Geological Sciences, Faculty of Science, Masaryk University, Brno,

Czech Republic

Prosun Bhattacharya, Mattias von Brömssen, Gunnar Jacks

Groundwater Arsenic Research Group, Department of Land and Water Resources Engineering,

Royal Institute of Technology (KTH), Stockholm, Sweden

Kazi Matin Ahmed

Department of Geology, University of Dhaka, Dhaka,Bangladesh

Holocene sedimentary aquifers of the Bengal Delta Plain (BDP) The present study was carried out in Brahmanbaria district, covering an area of 18 km2in northeastern Bangladesh The Chandina Formation is the main hydrostratigraphic unit of the area, which comprises silt and clay with high content of organic matter Dissolved arsenic concentrations in groundwater are high, reaching

⬎400 ␮g/L in some wells Groundwater is reducing with general lack of detectable dissolved oxy-gen (DO) and contains low concentrations of nitrate and sulfate Concentrations of dissolved Fe are high, which is in general in agreement with the reductive dissolution of ferric oxide and hydroxide hypothesis Results of speciation modeling indicated the possibility of precipitation of siderite, and to less extent, vivianite for many samples The log PCO2values were extremely high (⬎⫺1.0 atm), suggesting production of CO2in redox reactions involving the organic matter in the sediments Redox potential values calculated on the basis of different redox couples and field Eh measurement indicated redox disequilibrium Hierarchical cluster analysis (HCA) performed in paired groups mode using the program PAST indicated highest degree of similarity among redox-sensitive elements NO3, Mn, Fe, PO4, SO4, As, and pH Na and Cl form a distinct group, which indicate the influence of sea water Bicarbonate generated in several redox reactions and carbonate dissolution was linked to almost all parameters and this holds even more for the electrical conductiv-ity (EC) Principal components analysis (PCA) yielded Principal Component 1 (PC1) correspon-ding to sea water, and Principal Component 2 (PC2) corresponcorrespon-ding to redox reactions with generally high arsenic concentrations In summary, combination of speciation modeling and multivariate statistics proved to be useful in testing of conceptual model of geochemical evolution of arsenic-rich groundwater

Natural arsenic in concentrations above the safe drinking water limits of World Health Organization (10␮g/L; WHO 2001), and above the national drinking water standard (50 ␮g/L) is present in groundwater of the Bengal Delta Plain (BDP) in many districts of Bangladesh (Mukherjee & Bhattacharya 2001, Smedley & Kinniburgh 2002, Ahmed et al 2004) The source of arsenic is geogenic and is related to the sediments deposited by the rivers Ganges (Padma), Brahmaputra

(referred as Jamuna in Bangladesh) and Meghna (Nickson et al 2000, Bhattacharya et al 2002a,b) Arsenic contaminated groundwater is common in the aquifers of alluvial lowlands, comprising the floodplains of Padma and Brahmaputra (Jamuna) rivers, and also the Ganges Delta In this paper, we present applications of geochemical modeling and multivariate statistics

in development and support of conceptual model of arsenic behavior

2 GEOLOGICAL SETTING

The BDP is a large sedimentary basin drained by the Ganges, Brahmaputra (Jamuna) and Meghna (GBM) rivers (Fig 1) Huge amounts of sediments have been transported and converged at the lower reaches, forming the pro-grading delta at the head of the Bay of Bengal In general, two broad physiographic units characterize the BDP – elevated Pleistocene Terraces such as the Barind and Madhupur tracts, floored with thick surficial oxidized clay and silty clay deposits, and the Holocene lowlands

The Holocene lowlands include piedmont plains, flood plains, delta plains and coastal plains (Umitsu 1987 and 1993, Brammer 1996, Ahmed et al 2004) The area of present investigation is located in the Meghna Deltaic Plain comprising coarse-grained channel-fill deposits and fine grained overbank deposits During the late Holocene period, in several parts of the BDP, sediments were deposited in marshy environments, as evidenced by occurrence of continuous layers of peat (Umitsu 1993, Ravenscroft et al 2001, Ahmed et al 2004)

The present study was carried out in an area of 18 km2covering the Ashuganj and Brahmanbaria Sadar Upazilas (sub-districts) in Brahmanbaria district in eastern Bangladesh (Fig 2) The Chandina Deltaic Plain (CDP), the major physiographic units covering the study area (Bakr 1977),

is generally flat and occurring at relatively higher levels than the surrounding floodplains The

Figure 1 Map of Bangladesh showing the network of the rivers Ganges (Padma), Jamuna (Brahmaputra) and Meghna rivers and the location of Brahmanbaria area The major geomorphic domains, Barind and Madhupur tracts of Pleistocene age (lighter tone) are seen distinctly within the vast tract of Holocene alluv-ium (Resolution: 625 meters; MODIS Data Type: MODIS-PFM; MODIS Band Combination: 1, 4, 3)

sediments of the CDP are composed of silt, silty loam, silty clay belonging to the Chandina Formation The Chandina Formation is overlain by the Meghna alluvium and underlain by the Pleistocene Madhupur Clay and Pliocene Dupi Tila Formation

Neotectonic uplifts and course shifting of the Meghna and Old Brahmaputra rivers have influenced sedimentation in this area, particularly during the late Holocene time The source ter-rains of most of these sediments were located in the areas around the Shillong Plateau in the north and Tripura Hills on the east A thick sequence of fine to very fine Holocene sediments overlies the Pleistocene Madhupur Clay and Pliocene Dupi Tila sediments The Holocene sediments are gen-erally gray in color and contain high amount of organic matter while underlying older sediments are characterized by reddish-brown, light brown and yellowish brown color and contain only low amounts of organic matter (Table 1)

Groundwater from the Holocene sandy sedimentary aquifer is extracted by shallow hand tube wells Water levels in the Holocene aquifers fluctuate with annual recharge/discharge conditions, with a maximum depth of 5–7 m bgs in pre-monsoon months During the monsoon most of the area if flooded and the groundwater level reaches the ground surface The multiple aquifer system

Figure 2 Geological map of a part of the Chandina Delta Plain (CDP) in the Brahmanbaria district, Bangladesh showing the location of the groundwater sampling.

Table 1 Hydrostratigraphy of the study area.

arsenic rich Late Pleistocene Chandina Formation Grey silt, silty loam, silty clay Aquitard

fine sand

in this area is characterized by variable hydraulic conductivity and water quality Water quality is often good except for occurrences of pockets of brackish water, remnants of paleo-seawater Occurrences of biogenic methane gas have also been reported from a number of places with the BDP and particularly in the vicinity of the study area (Ahmed et al 1998, Ravenscroft et al 2001) Arsenic and iron concentrations are frequently high in the Holocene aquifers However, their conc-entrations are significantly lower in the underlying Dupi Tila aquifers (BGS & DPHE 2001, van Geen et al 2003) This deeper aquifer probably receives recharge at their outcrops in the Tippera Hills region, outside the political boundary of Bangladesh

3 MATERIALS AND METHODS

Samples of groundwater were collected during late November, 2000 from 30 domestic and governmental tube wells placed at varying depths of 18–150 m (Fig 2) Parameters like pH, redox potential (Eh), temperature, and electrical conductivity (EC) were taken in the field The pH was measured using a Radiometer Copenhagen PHM 80 instrument using a combination electrode (pH C2401-7) The Eh was measured in a flow-through cell using a combined platinum electrode (MC408Pt) equipped with a calomel reference cell Samples collected for analyses included: (a) filtered (using Sartorius 0.45
m online filters) for major anions; (b) filtered and acidified with suprapure HNO3(14 M) for the cations and other trace elements including arsenic (Bhattacharya

et al 2002b) Arsenic speciation was performed with Disposable Cartridges(r) (MetalSoft Center, PA) in the field, Meng et al (2001) The cartridges adsorb As(V), but allows As(III) to pass through Sulfide was precipitated in the field by addition of Zn acetate Major anions, Cl, and

SO4were analyzed in filtered water samples, with a Dionex DX-120 ion chromatograph with an IonPac As14 column NO3-N and PO4-P was analyzed spectrophotometrically with a Tecator Aquatec 5400 Ammonium (NH4) was analyzed spectrophotometrically with a Tecator Aquatec

5400 at 540 nm wavelength The major cations (Ca, Mg, Na and K) and minor and trace elements (Fe, Mn, As) were analyzed by inductively coupled plasma (ICP) emission spectrometry (Varian Vista-PRO Simultaneous ICP-OES) at Stockholm University Dissolved organic carbon (DOC) in the water samples were determined on a Shimadzu 5000 TOC analyzer (0.5 mg/L detection limit with a precision of10% at the detection limit Speciation modeling was performed by program PHREEQC (Parkhurst 1995) Thermodynamic data for arsenic were taken from data base of pro-gram MINTEQA2 (Allison et al 1991) Multivariate statistics analysis was performed to verify the hydrogeochemical similarities among geochemical parameters The data were analyzed using multivariate statistics implemented in the program PAST (Hammer et al 2001)

4 GENERAL HYDROGEOCHEMISTRY

Selected results of the groundwater chemical analyses are presented in Table 2 In addition, Bhattacharya et al 2004 (in press) provide more detailed discussion on the trends of spatial variability of water chemistry Shallow groundwater (50 m) in Brahmanbaria region was of Ca-Mg-HCO3and Ca-Na-HCO3types (Fig 3a) Groundwater samples had very variable HCO3

(74–562 mg/L) and SO4(bdl-32.9 mg/L) concentrations In the intermediate and deeper aquifers groundwater of Na-Cl-HCO3type was also found (Fig 3b) Groundwater pH values were between 6.2 and 7.6 Field Eh values corrected with respect to hydrogen electrode were from0.180

to0.29 V, indicating moderately reducing conditions However, these results do not represent redox status of groundwater, possibly because of aeration of groundwater in hand-pump wells and during pumping as discussed later

Concentrations of total arsenic (Astot) in shallow wells varied from 10 to 335
g/L and in interm-ediate wells reached up to 439
g/L Concentrations of dissolved Fe were highly variable, from 0.28 mg/L to 10.3 mg/L However, no correlation between dissolved iron (Fetot) and Astotin shal-low samples and only shal-low correlation in intermediate depth samples were observed (Bhattacharya

et al 2004, in press) Most of dissolved arsenic (up to 99.5%) was present as As(III) Concentrations

Copyright © 2005 Taylor & Francis Group plc, London, UK

of NH 4 were high in some shallow wells, reaching 12.2 mg/L High concentrations of dissolved organic matter (DOC, up to 21.8 mg/L) were consistent with reducing character of groundwater Dissolved sulfide with concentrations up to 2.1 mg/L was found in several wells, indicating the presence of sulfate reduction Spatial variability of the distribution of arsenic in the shallow ground-waters and its relationship with other chemical parameters is discussed in detail in Bhattacharya et al (2004, in press) Three domains were defined: Domain 1 with high concentrations Astot, and

PO4 and low Fetot, and anomalous SO4 concentrations; Domain 2 with high concentrations of

Astot, and PO4, and low concentrations of Fetot, and sulfate; Domain 3 with low concentrations

of Astot, and PO4, and with high Fetot, and sulfate concentrations (Fig 4)

5 GEOCHEMICAL MODELLING

Results of calculations of saturation indices (SI) together with calculated log PCO2values are pres-ented in Table 3 There was no significant complexation of Fe with other inorganic anions and the

Figure 3 Major ion characteristics of Brahmanbaria groundwaters plotted on a piper diagram (a) Shallow wells ( 50 ) (b) Intermediate wells (50–150 m, black circles) and deep wells (150 m, data not included in discussion).

Figure 4 Spatial variability of total As (Astot) in the shallow wells of Brahmanbaria in eastern Bangladesh.

principal aqueous species of Fe was Fe2 and minor species was FeHCO3 Concentrations of Fe(III) were low, and the principal species were Fe(OH)3and, at lower pH, Fe(OH) 2 Low Mn conc-entrations in groundwater within the reduced domains could possibly be due to precipitation of rhodochrosite (MnCO3) (Sracek et al 2000, McArthur et al 2001, Ahmed et al 2004) Calculated log PCO2values were very high, reaching in some cases values higher than – 1.0 This is related to the generation of CO2in redox reactions like dissolution of ferric oxide and hydroxides in reaction with organic matter This is consistent with high calculated values of DIC (up to 1.27 2mol/L) Many samples (60%) were at equilibrium or supersaturated with respect to siderite (FeCO3) sugg-esting that this mineral phase might have acted as a sink for dissolved iron Some samples (30%) were also supersaturated with respect to vivianite Fe3(PO4)2.8H2O, but the degree of saturation is generally lower than in the case of siderite Few samples (13.3%) are also supersaturated with respect to rhodochrosite Some samples are close to equilibrium with calcite and dolomite (not shown), but saturation is reached only in very limited number of samples

The speciation program was also used to calculate Eh values based on As(III)/As(V) couple and S(VI)/S(-II) couple determined analytically Results of these calculations are presented in Table 4 Typical feature is the strong disequilibrium between measured field Eh values adjusted with respect to hydrogen electrode and values of Eh calculated on the basis of arsenic couple Values of redox potential based on arsenate to arsenite ratios are lower than the field Eh values This holds even more for sulfur redox couple, suggesting strong redox disequilibrium However, the field Eh values truncated at0.180 V are possibly unreliable and must be interpreted with caution There

Table 3 Results of speciation calculations.

also is a possibility of precipitation of secondary sulfide minerals like mackinawite This mineral

is a precursor of pyrite and may incorporate some arsenic, acting as a sink for arsenic in ground-water where sulfate reduction takes place

6 MULTIVARIATE STATISTICS

Results of Hierarchical Cluster Analysis (HCA) performed in Ward’s mode using the program PAST are given in Figure 5 They indicate highest degree of similarity among NO3, Mn, Fetot,

PO4 , SO4 , Astot, and pH Most of them are redox sensitive species, except for PO4 which is linked to Fe due to its release during reductive dissolution of ferric oxide and hydroxides The effect of pH is however not clear, although this parameter plays a role in precipitation of minerals like siderite and vivianite Naand Clform a distinct group, which most likely indicates relict sea water entrapped in the aquifers Ca2and Mg2are separated from Naand Clbecause they are related not only to relict sea water in the sediments, but also to processes like dissolution of carbonates, and weathering of silicates enhanced by production of CO2in redox reactions HCO3 

Table 4 Comparison of redox data.

- 3

Mn Fe

pH SO

0

-100

-200

-300

-400

-500

-600

-700

-800

Figure 5 Results of Hierarchical Cluster Analysis (HCA).

generated in several redox reactions and carbonate dissolution is linked to almost all parameters and this holds even more for EC

Results of Principal Components Analysis (PCA) presented in Figure 6 indicate two principal components and identified as: PC1 with high loadings for EC, Na, Cl, and HCO3 , and PC2 with high loading for HCO3and with relatively low, but significant loading for arsenic The PC1 corresponds to the influence of relict seawater entrapped in the sediments to the groundwater, and PC2, which corresponds to the impact of redox reactions These principal components explain 92.01% and 6.71%, respectively, of total variance in sample set The samples at the bottom right (27 and 28, Fig 6) indicate strong influence of saline water, most likely reflect to be relict seawa-ter entrapped in the sediments On the other hand, samples at the top of the graph (4, 11, 21, 22,

25 etc., Fig 6) are strongly influenced by redox reactions and they generally have high arsenic concentrations Samples at the bottom left (1, 2, 3, 6, 17 etc., Fig 6) are relatively less influenced

by both processes and have low arsenic concentrations It seems that redox processes relatively less influence the samples showing signature of relict saline water from the marine sources However, the impact of palaeo-seawater relicts in the aquifers seems to be a local phenomenon in the Brahmanbaria area, which is also seen in many other areas of Bangladesh

7 DISCUSSION AND CONCLUSIONS

The conceptual model of arsenic and iron behavior in groundwater in Bangladesh can be summar-ized as follows:

(a) reductive dissolution of ferric oxide and hydroxides in reaction with organic matter like peat after consumption of more favored electron acceptors in a reaction like:

1 2 3

4

5

6

7

8

9

10

11

12

13 14

15

16

17

18

19

22

25

26

27

28

29 30

400

300

200

100

1000

Component 1

0

-100

-200

Figure 6 Results of Principal Components Analysis (PCA).