Journal of Environmental Science and Health Part A (2007) 42, 1869–1878Copyright C© Taylor & Francis Group, LLCISSN: 1093-4529 (Print); 1532-4117 (Online)DOI: 10.1080/10934520701567122

A simple and effective arsenic filter based on compositeiron matrix: Development and deployment studiesfor groundwater of Bangladesh

ABUL HUSSAM1 and ABUL K. M. MUNIR2

1Department of Chemistry and Biochemistry, MSN 3E2, George Mason University, Fairfax, Virginia 22030, USA2Manob Sakti Unnayan Kendro (MSUK), Kushtia, Bangladesh

Drinking groundwater contaminated with naturally occurring arsenic is a worldwide public health issue. This work describes theresearch, development and distribution of a filter used by thousands of people in Bangladesh to obtain arsenic-free safe water. Thefilter removes arsenic species primarily by surface complexation reactions: =FeOH + H2AsO−

4 → =FeHAsO−4 + H2O (K = 1024)

and =FeOH + HAsO2−4 → =FeAsO2−

4 + H2O (K = 1029) on a specially manufactured composite iron matrix (CIM). The filterwater meets WHO and Bangladesh standards, has no breakthrough, works without any chemical treatment (pre- or post-), withoutregeneration, and without producing toxic wastes. It costs about $40/5 years and produce 20–30 L/hour for daily drinking andcooking need of 1–2 families. The spent material is completely non toxic-solid self contained iron-arsenate cement that does not leachin rainwater. Approved by the Bangladesh Government, about 30,000 SONO filters were deployed all over Bangladesh and continueto provide more than a billion liters of safe drinking water. This innovative filter was also recognized by the National Academy ofEngineering – Grainger Challenge Prize for sustainability with the highest award for its affordability, reliability, ease of maintenance,social acceptability, and environmental friendliness, which met or exceeded the local government’s guidelines for arsenic removal.

Keywords: Arsenic filter, composite iron matrix, Grainger challenge prize, groundwater, Sono filter

Introduction

Arsenic poisoning in drinking water is now identified asone of the worst natural disasters on Earth. It is estimatedthat of the 140 million people of Bangladesh, between 77–95 million are drinking groundwater containing more than50 µg/L (50 ppb or 0.05 mg/L) maximum contaminationlevel (MCL) from 10 millions tubewells.[1,2] The prolongeddrinking of this water has caused serious illnesses in theform of hyperkeratosis on the palms and feet, fatigue symp-toms of arsenicosis, and cancer of the bladder, skin andother organs.[3] The only way to solve this crisis is to drinkclean potable water free from arsenic and other toxic impu-rities. In this endeavor, we have developed an arsenic wa-ter filter and deployed it on a large scale. The filter hasbeen thoroughly studied and passed through several en-vironmental technologies verification programs for arsenic

Address correspondence to Abul Hussam, Department of Chem-istry and Biochemistry, MSN 3E2, George Mason University,Fairfax, VA 22030, USA. E-mail: ahussam@gmu.edu

mitigation (ETVAM) projects and approved by the Govern-ment of Bangladesh (GOB) for household use.[4] Recently,the filtration technology has been given the highest awardfrom the National Academy of Engineering-Grainger Chal-lenge Prize for Sustainability[5] after testing 15 other com-petitor technologies. NAE has recognized this innovativetechnology for its affordability, reliability, ease of mainte-nance, social acceptability, and environmental friendliness,which met or exceeded the local government’s guidelinesfor arsenic removal.

The arsenic measurement and mitigation research bythe group started in 1997. Since then, we have publishedkey papers on both measurement and mitigation. Notably,we have developed a computer controlled electrochemi-cal analyzer for arsenic measurement which passed twointer-laboratory method validation studies conducted byInternational Atomic Energy Agencies.[6,7] The ability tomeasure ppb level As(III) and As(V) allowed us to test thefiltration technology with real groundwater in the field. Thefirst mitigation technology paper was published in 2000.[8]

Since then, several other papers were published on im-provement of the technology. The technology was patentedin 2002. About 30,000 SONO filters were deployed in 16

1870 Hussam and Munir

Fig. 1. (a) Filter schematic (left) and (b) filter in use in a village hut.

districts all over Bangladesh; about 825 SONO filters wereinstalled for primary school children. Many of these filtershave been in continuous use for 5 years without a break-through. We estimate that more than a billion liters of cleandrinking water was consumed from these filters and theycontinue to provide high-quality water for drinking andcooking. The following is a description of the SONO fil-ter and its performance based on the data obtained fromour research, development, and extensive participation inETVAM.

Materials and methods

Analytical methods

Measurements of metal concentrations in groundwater andfiltered water were carried out by Inductively CoupledPlasma Atomic Emission Spectroscopic (ICP-AES, As de-tection limit 16 µg/L), Zeeman-effect Atomic AbsorptionSpectrometer with a Graphite Furnace (AASGF-Z, Asdetection limit 2 µg/L), and Anodic Stripping Voltam-metry (ASV, As detection limit 2 µg/L). The specia-tion procedure and analytical method validation were de-scribed elsewhere.[6−8] Anion analysis were performed byLachat QuickChem 8000 Ion Chromatography (IC) fol-lowing standard protocol.[9] Standard laboratory instru-ments and practices were followed for other water qualityparameters.

Filter design

The unit- SONO filter is shown in Figures 1a and 1b.[10]

A list of basic materials and their characteristic functionsis described in Table 1.

Results and discussion

Chemistry and general considerations

In groundwater (pH = 6.5–7.5) arsenic is present in twooxidation states (As(III) in H3AsO3 and As(V) in H2AsO−

4and HAsO2−

4 ). It is known that in most groundwater inBangladesh more than 50% of total arsenic is presentas the neutral H3AsO3 at groundwater pH. The other50% is divided equally in two As(V) species H2AsO−

4 andHAsO2−

4 . An ideal filter must remove all three species with-out chemical pretreatment, without regeneration, and with-out producing toxic wastes. The unit SONO filter and itspredecessor (3-Kolshi filter. Kolshi is a round containershown in Figure 1) had satisfied these requirements. Theoriginal SONO 3-Kolshi filtration systems also passed theETVAM tests and a similarly made system was tested inNepal by a MIT group with an arsenic removal capacity of20 µg/L.[11,12]

The primary active material in the SONO filter is thecomposite iron matrix (CIM), a mass made of cast ironturnings through a proprietary process to maintain active

Developing an arsenic filter for Bangladesh 1871

Table 1. Materials used in SONO filter–general specification, function and manufacturing method. Please use Figure 1 as a guide

Material (Refer Figure 1) Function and characteristics Brief manufacturing method and availability

General filter specifications: Top Bucket (Red): 45L, dia/height: 46/44 cm, 1.5 kg; Bottom Bucket (Green): 23 L, 38/36, 1 kg.Dimension: h/w/l 1.22/0.42/0.45 meter, Wt 56 kg. Flow rate: 20-30 L/hour continuous.

Top bucket (32 kg)Coarse river sand (CRS)—10 kg

wet (Fm = 1.5–2, 95% SiO2, 5%other metal oxides)

CRS is an inactive material used as coarseparticulate filter, disperser, flow stabilizer,and to provide mechanical stability.Groundwater with high soluble iron isoxidized and preicpitate as Fe(OH)3(s) inthis media.

CRS is obtained from local river andthoroughly washed before use.

Composite iron matrix(CIM)—5–10 kg, Fe 92–94%, C4–5%, SiO2 1–2%, Mn 1–2%, S, P1–2%

CIM is the active surface for complexationand immobilization of inorganic arsenicand many toxic metals cations. The finalproduct is porous, lighter than originalturnings and produce less fines for filterstability.

CIM is manufactured from various ironturnings obtained from local foundry ormachine shops. Turnings are throroughlywashed, dried and treated with food gradeacids to enhance HFO formation in aproprietary process to a composite ironmatrix (CIM).

Coarse river sand and Brick chips.CRS-10 kg wet, BC-2.5 kg (b)

CRS and BC are inactive material and hassimilar functions. In combination these areused as a protection barrier for thefree-flow junction outlet.

Same as top. BS are from local brickmanufacturer, thoroghly washed anddisinfected with bleach.

Bottom bucket (25 kg)Coarse river sand (CRS)–10 kg wet Similar as above. This stage retains residual

iron leached from the first stage CIM asHFO. Filter life span can be estimatedfrom the residual iron from top bucket.

Same as above.

Wood charcoal (WC) Wood charcoal is known to adsorb organics(odor causing compounds, pesticideresidues etc). WC is passive to arsenic butimparts a better tasting water.

WC is obtained from firewood used forcooking. Large quantities are collectedfrom local hotels and villagers.

Fine river sand and (FRS)–9 kg wetBrick chips (BC)–3.5 kg

FRS is the fine filtration media to catch anyresidual particulates. BC are used as flowstabilizing media.

Both obtained from local manufacturers,thoroghly washed.

Other materialsPlastic buckets–40 L Container. Only food-grade high density

polypropylene (HDPP) buckets are usedLocal plastic moulding industries. Buckets

were retrofitted with top cover and outletsfor flow controller taps.

Flow controllers Control flow to maintain optimum residencetime for best arsenic removal. This is fixedin the factory.

Moulded plastic or metal taps are avilable inlocal hardware stores.

Metallic filter stand Support for the buckets Made by local welders.(b) Brick chips: Silica 55%, Alumina 30%, iron oxide 8%, magnesia 5%, lime 1% and others 1%.

CIM integrity for years. The CIM removes the inorganicarsenic species quantitatively through possible reactionsshown in Table 2. Infrared spectroscopy (IRS)[13] and ex-tended X-ray absorption fine structure (EXAFS)[14] showthat arsenate and arsenite form bidentate, binuclear surfacecomplexes with =FeOH (or =FeOOH or hydrous ferricoxide, HFO) as the predominant species tightly immobi-lized on the iron surface. Also, inorganic As(III) speciesare oxidized to As(V) species by the active O−

2 , which isproduced by the oxidation of soluble Fe(II) with dissolvedoxygen. Manganese (1–2% by wt) in CIM catalyzes ox-idation of As(III) to As(V). Therefore, the process doesnot require pretreatment of water with external oxidizing

agents such as hypochlorite or potassium permanganate.Dynamic electrochemical study (Tafel plot) shows that wa-ter alone can also act as the primary oxidant for castiron.[15] As(V) species (H2AsO−

4 and HAsO2−4 ) are then

removed by surface-complexation reactions on the sur-face of hydrated iron (=FeOH ). New =FeOH is gener-ated insitu as more water is filtered. In addition to arsenicspecies, =FeOH is also known to remove many othertoxic species.[16−19] The primary reactions are: =FeOH +H2AsO−

4 →=FeHAsO−4 + H2O (K = 1024) and =FeOH

+ HAsO2−4 →=FeAsO2−

4 + H2O (K = 1029). These intrin-sic equilibrium constants indicate very strong complexationand immobilzation of inorganic arsenic species. It is known

1872 Hussam and Munir

Table 2. Possible physicochemical reactions in different parts of the filtration process.[20−25] All surface species are indicated by =Xare solids. CIM- Composite Iron Matrix

Reaction Location Reactions

Top layer: Oxidation of As(III)(Equations are balanced for reactive species only)

As(III) + O.−2 → As(IV) + H2O2

As(III) + CO.−3 → As(IV) + HCO−

3As(III)OH.− → As(IV)As(IV) + O·−

2 → As(V) + O.−2

Top bucket: Oxidation of soluble ironOxidation of ferrous to ferric through active oxygen species

Fe(II) + O2 → O.−2 + Fe(III)OH+

2Fe(II) + O.−

2 → Fe(III)+ H2O2

Fe(II) + CO.−3 → Fe(III)+ HCO−

3CIM hydrous ferric oxide (HFO)

Fe(III) complexation and precipitation.=FeOH + Fe(III) + 3 H2O → Fe(OH)3 (s, HFO) +=FeOH + 3H+(=FeOH is surface of hydrated iron )

CIM – HFO surfaceSurface complexation and precipitation of As(V) species

=FeOH + AsO3−4 + 3 H+ →=FeH2AsO4+ H2O =FeOH +

AsO3−4 + 2 H+ →=FeHAsO−

4 + H2O =FeOH + AsO3−4 +

H+ →=FeAsO2−4 + H2O =FeOH +

AsO3−4 →=FeOHAsO3−

4Top two buckets: Precipiation of other metals

Bulk precipitation of arsenate with soluble metal ionsM(III) + HAsO2−

4 → M2 (HAsO4)3 (s), M=Fe, Al, M(II) +HAsO2−

4 → M(HAsO4) (s) and other arsenatesM= Ba, Ca, Cd, Pb, Cu, Zn and other trace metals

CIM and Sand interfaceReactions with iron surface and sand can produce a poroussolid structure with extremely good mechanical stability for thefilter known as solid CIM

=FeOH+ Si(OH)4 →=FeSiO(OH)3 (s) +H2O=FeOH+ Si2O2(OH)−5 + H+ →=FeSi2O2(OH)5 (s) +H2O=FeOH+ Si2O2(OH)−5 →=FeSi2O3(OH)−4 (s) +H2O=FeOHAsO3−

4 + Al(III) →=FeOHAsO4Al (s)=FeOHAsO3−

4 + Fe(III) →=FeOHAsO4Fe(s)=FeOH.HAsO2−

4 + Ca(II) →=FeOH.HAsO4Ca (s)

that excess Fe2+, Fe3+, and Ca2+ in groundwater enhancepositive charge density of the inner Helmholtz plane of theelectrical double layer and specifically binds anionic arse-nates to form surface complexes. We found that As(III)and As(V) removal process was independent of the inputarsenic concentration i.e., a zero-order reaction. Detailedthermodynamics and kinetics of these reactions are still un-der investigation.

Tests and performance

The SONO filters were tested only with real groundwatercontaminated with arsenic and other species. From the in-ception of our research, we realized that the fastest wayto test for filter efficacy was to use real groundwater con-taining varied concentrations of arsenic, iron, other inor-ganic species and water quality parameters in Bangladesh.Thus we have selected 6 tubewells in 6 different householdswhere SONO filters were installed. Table 3 shows that allthe filters remove arsenic to less than10 µg/L from an in-put range of 32–2423 µg/L As(Total). All filters removedsoluble Fe below 0.26 mg/L, even from the highest inputFe of 21 mg/L. It is important to mention that we haveidentified arsenicosis patients in the last 3 locations wherearsenic concentrations are above the suggested 300 µg/L.These experimental filters continue to provide clean potablewater for the households and we are monitoring them con-tinuously.

In combination with the CIM, the sand, the charcoal, andthe optimum arrangement of the materials, the SONO filterremoves arsenic, iron, manganese and many other inorganicspecies to a potable water.

Figure 2a shows typical test results in which 25,000 L oftubewell water containing 1139–1600 µg/L of As (Total)was filtered to produce potable water with less than 14 µg/LAs (Total) until reaching the detection limit (2 µg/L). Thisprogressive decrease in effluent arsenic is unique to theSONO CIM filter. This is also confirmed in recent ETVAMtests in comparison to activated alumina, cerium hydroxideion exchange resin, and microfine iron oxide based filters.[26]

We attribute this unique property of the SONO filter tothe generation of new complexation sites on CIM throughinsitu iron oxidation and surface chemical reactions asshown in Table 2. It also indicates that the arsenic com-plexation reaction has zero-order kinetics with respect tothe influent As(total) concentration. This means high ar-senic removal capacity independent of input As(total) (upto 2400 µg/L), and also implies no breakthrough of arsenicfor the life of the filter. The filter life span can be estimatedby assuming a 1As:1Fe surface complexation reaction inwhich all available CIM Fe is used and where the filter loses500 µg/L of iron (from CIM) at 200 L/day use. It wouldtake 274 years to loose 1000 g of iron when 20,000,000 L(20 Million L!) water is filtered.

The filter life span can also be estimated from our data onFreundlich isotherm with CIM (log (X/M) = log K + (1/n)

Developing an arsenic filter for Bangladesh 1873

Table 3. Results from six SONO filters monitored for 2.3–4.5 years in active use by householders. Location: Kushtia district,Bangladesh. Test period: 2000–2005

Filter 1 Filter 2 Filter 3 Filter 4 Filter 5 Filter 6Parametersa Fatic Courtpara Zia Alampur Kaliskhnpur Juniadah

Years in use 2.32 4.5 2.66 3.6 4.4 2.5Water yield (L) 67,760 125,000 77,840 104,960 128,480 72,960Number of Measurements 10 110 12 14 56 8As(Total)- Input, µg/L 32 ± 7 155 ± 7 243 ± 9 410 ± 15 1139–1600 2423 ± 87As(Total), Filter, µg/L <2 7 ± 1 7 ± 1 8 ± 2 7 ± 2 8 ± 4Fe(Total), Input, mg/L 20.7 ± 0.6 4.85 ± 0.25 7.35 ± 0.3 10.86 ± 0.56 1.53 ± 0.08 0.6 ± 0.03Fe(Total) Filter, mg/L 0.22 ± 0.02 0.228 ± 0.04 0.25 ± 0.03 0.242 ± 0.03 0.25 ± 0.05 0.26 ± 0.03Cost per liter (Taka), (1 Taka = 0.016$) 0.031 0.016 0.026 0.02 0.016 0.028

aFlow rate 20–30 L/hour, Other water chemistry parameters are similar to that in Table 4. Consumption: 60–180 L/day. As (total) was measured byASV on a thin film gold electrode validated by IAEA interlaboratory comparison studies at SDC/MSUK, Kushtia Bangladesh and with GraphiteFurnace AA at GMU Chemistry Department. Iron was measured spectrophotometrically at SDC/MSUK. Cost per liter decreases as more water isfiltered.

log Cf, where X is the µg/L of As adsorbed, M is the gof CIM used, and Cf is the free arsenic in µg/L) with anadsorption capacity, K = 139.3 and adsorption intensity,n = −1.9. It was found to take about 14 years to reachthe 50 µg/L MCL breakthrough from influent water con-taining 300 µg/L As(total) at 80 L/day usage rate and afilter with 10,000 g of CIM. Although these calculationsare disparate, it shows, along with our observations thatthe filter will work for many years before breakthrough oc-curs. The MCL breakthrough is further extended by theco-precipitation of arsenate by HFO produced from Fe(II)present in groundwater even at low concentrations.

Our oldest filter has been working for 5 years nowwithout any changes and without a breakthrough. Figure2b shows that the filter can even work at 60 L/hourflow rate without breakthrough. However, due to theunknown water chemistry and varied As(total) in ground-

water, we fixed the flow rate to 20–30 L/hour to ensurelong-term use and effluent As(total) below 30 µg/L.In contrast, blank filters breakthrough occurs at 88 Lgroundwater as shown in Figure 3. This experiment alsodemonstrate that plain sand filter broke through theMCL almost instantaneously. A low iron (1.0 mg/L)containing groundwater was selected for this study toensure HFO precipiate from this iron would not bias theresults by coprecipiation and complexation of arsenicspecies.

Several NGOs have installed SONO filters in many ar-senic affected areas of Bangladesh. One such most af-fected area is Hajigong, where 165 SONO filters were in-stalled to supply water for 300 arsenicosis patients and3,000 family members. The results published show As (to-tal) in filtered water was <2 µg/L (70% samples), <10 µg/L(20% samples), <30 µg/L (10% samples), and none above

(a) (b)

y = -0.0003x + 8.9937R2 = 0.4001

0

2

4

6

8

10

12

14

0 5000 10000 15000 20000 25000

Water Yield, L

As(

To

tal)

, ug

/L

Series -02 Exp.(Twin)Flow rate vs As(total)

y = 0.4969x - 17.388R2 = 0.721

0

2

4

6

8

10

12

14

30 35 40 45 50 55 60Flow Rate, L/hour

As

(To

tal)

, u

g/L

Fig. 2. SONO filter performance at two different tubewells showing effluent As (total) vs. water yield. (a) Filter 5 (Table 3), Kushtia.Influent water As(total) 1139–1600 µg/L. Flow rate: 20 L/ hour. (b) Effect of flow rate on effluent As (total). influent As(total) 410µg/L (Filter 4 at Alampur). Effluent As(total) was measured only at flow rates above 40 L/hour. These tests were performed with ahigh flow SONO twin filter for accelerated testing.

1874 Hussam and Munir

0

100

200

300

400

500

600

700

500 1000 1500 2000

Total volume, L

As(

III)

or

As(

To

t), u

g/L

As(III), Sand Filter As(Tot), Sand Filter

As(III), Brick+Sand+Charcoal As(Tot), Brick+Sand+Charcoal

Fig. 3. Behavior of two filters made without the active CIM. Figureshows blank filters maximum contaminant limit (MCL) break-through occurs the first day even at 88 L. Groundwater composi-tion: As(III): 300 µg/L, As(total) 996 µg/L, and Fe(II) 1.0 mg/L.The filtered water properties: Sand filter: Temperature 27.3◦C,p 7.6 ± 0.1, TDS 210 ± 6 (µs/cm), Eh 158 ± 6 mV vs. NHE.Data for Brick + Sand + Charcoal filter: Temperature 25.7◦C,pH 7.9 ± 0.1, TDS 208 ± 12 (µs/cm), Eh 148 ± 6 mV vs. NHE.Filtered water had no detectable total Fe.

30 µg/L from the influent As(total) 600–700 µg/L with atleast 50% in the form of more toxic As(III).[27] The studyalso found no change in flow rate and no maintenance re-quired for 12 months.

Table 4. Comparisons of water quality from SONO filter, USEPA, World Health Organization (WHO) and Bangladesh Standards(1 mg/L = 1000 µg/L)

USEPA WHO, Bangladesh Influent SONO FilterConstituent (MCL) Guideline Standarda Groundwater Waterb

Arsenic (total)- µg/L 10 10 50 5 – 4000c 3 - 30Arsenic (III)- µg/L 5 – 2000d < 5Iron (total) - mg/L 0.3 0.3 0.3 (1.0) 0.2 – 20.7 0.19 ± 0.10pH 6.5–8.5 6.5–8.5 6.5–8.5 6.5–7.5 7.6 ± 0.1Sodium - mg/L 200 < 20.0 19–25Calcium - mg/L 75 (200) 120 ± 16 5–87Manganese - mg/L 0.5 0.1 - 0.5 0.1 (0.5) 0.04–2.00 0.22 ± 0.12Aluminum -mg/L 0.05–0.2 0.2 0.1(0.2) 0.015–0.15 0.11 ± 0.02Barium, mg/L 2.0 0.7 1.0 <0.30 <0.082Chloride, mg/L 250 250 200 (600) 3–12 4.0–20.0Phosphate, mg/L 6 < 12.0 0.9 ± 0.12Sulfate, mg/L 100 0.3–12.0 12 ± 2Silicate, mg/L — 10–26 18 ± 6

aBangladesh standard values are given as maximum desirable concentration with maximum permissible concentration in parentheses.bSONO filters. ICP multielement measurements of Cu, Zn, Pb, Cd, Se, Ag, Sb, Cr, Mo, and Ni show concentrations below the USEPA, and WHOlimits at all times. All other measurements show average of semi-continuous measurement of more than 394,000 L of groundwater filtered by us andETVAM in at least eight different water chemistries in different regions of Bangladesh. Water chemistry parameters were recoded for 23 metals, 9anions, Eh, pH, Temp, dissolved oxygen, conductivity, and turbidity for hundreds of samples. All prescribed parameters passed the drinking waterstandards of WHO and Bangladesh.cOne tubewell at Bheramara was found to contain As (total) 4000 µg/L. The filtered water had 7 µg/L. This well was later capped by the Government.dIn some wells As(III) concentrations exceeded 90% of the As(total).

Arsenic removal efficacy and system validation

The SONO filtration system was extensively tested by usand several technology verification projects (ETVAM) runby the Government of Bangladesh (Bangladesh ArsenicMitigation Water Supply Projects, BAMWSP).[4] Table 4summarizes the results of more than 590,000 L of ground-water filtered in ten experimental filters located through-out Bangladesh. Of these, 577,000 L were tested by the au-thors at 6 locations as described in Table 3, and 17,334 Lwere tested by ETVAM at four locations throughoutBangladesh (Bera-Pabna, Hajigong-Comilla, Manikgang-Dhaka, Faridpur, and Nawabgang- Rajshahi) under variedgroundwater chemistry. Clearly, the filter water parametersmet and exceeded USEPA, WHO, and Bangladesh stan-dards. It is noted that the SONO filter removes the mosttoxic As(III) species from groundwater without a chem-ical pretreatment below the detection limit (2 µg/L). Italso removes manganese which is now implicated as a toxictrace metal in Bangladesh groundwater.[28] The SONO fil-ter water, low in Ca and Fe, is a soft-water, and is there-fore lighter in taste and pleasant to drink. It is also in-teresting to note that many groundwater sources do notmeet the potable water quality standard even in the ab-sence of arsenic; in these locations the SONO filter can beused.

Naturally occurring soluble iron and phosphate can af-fect the performance of most filtration technologies. Figure4a shows that soluble iron can enhance the retention ofarsenic in the filter through co-precipitation and complexformation as expected. These results were obtained from

Developing an arsenic filter for Bangladesh 1875

(a)

020406080

100120140160180

0 20 40 60

Influent [PO4/As(T)]

Effl

uent

As(

T),

ppb

(b)

y = -1.8753x + 27.857

R2 = 0.6086

-10

0

10

20

30

40

0 5 10 15 20

Influent Fe, ppm

Effl

uent

As,

ppb

Fig. 4. (a) Figure shows the effect of influent iron on the effluentarsenic concentration for filters tested throughout Bangladesh.The influent arsenic concentration range was from 150–878 µg/Lfrom ETVAM test data. The average soluble iron concentration inthe effluent water was 0.07 mg/L showing more than 99% removalof iron. Total volume of groundwater filtered 39,500 L. (b) Effectof influent phosphate on effluent As(total). Influent phosphateconcentration was normalized by influent As(total) ca. 700 µg/L.Actual phosphate range was 0–50 mg/L. The 3 outlier data wereobtained with 1 filter in Hajigong.

fields where soluble iron in influent water was above thepotable limit. In all cases (including data in Table 3) theSONO filter not only removed the arsenic, it also removedthe soluble iron by 99% and made the water potable. Phos-phate is often considered the competing ion for arsenateand has the potential to negatively affect the performanceof the filter. The effect of phosphate on the effluent As(total)is shown in Figure 4b. We find no clear effect of phosphateon the removal capacity of arsenic. The outlying data foundfor one specific filter in Hajigong was found to be anoma-lous, because in the same place 165 SONO filters functionedproperly as described earlier. Our observations indicate thatphosphate does not affect the performance of SONO filtereven at 40–50 mg/L concentration.

Operation and maintenance

The SONO filter does not require any special maintenanceother than the replacement of the upper sand layers whenthe apparent flow rate decreases. Experiments show thatflow rate may decrease 20–30% per year if groundwater hashigh iron (>5 mg/L) due to formation and deposition ofnatural HFO in sand layers. The sand layers (about an inchthick) can be removed, washed and reused or replaced withnew sand. The presence of soluble iron and formation ofHFO precipitate is also a common problem with other fil-tration technologies. Water flow disturbance could also oc-cur through accumulation of sand/ HFO deposits into thetap nozzle, which can be removed by detaching and cleaningthe tap in a flowing water stream.

The use of tubewells to extract groundwater was done toavoid drinking surface water contaminated with pathogenicbacteria. However, pathogenic bacteria can still be found indrinking water due to unhygienic handling practice and inmany shallow tubewells, possibly located near unsanitarylatrines and ponds.[29] To investigate the issue of bacterialgrowth in SONO filters, a Bangladesh NGO named Vil-lage Education Resource Center (VERC) recently tested193 SONO filters at 61 locations in one of the remotestfields, Sitakundu, Bangladesh.[30] The report shows of the264 tests, 248 were found to have zero count of ttc/100 mL(ttc- thermo tolerance coliform) and 16 with 2 ttc/100 mL.Pouring 5 L hot water in each bucket every month hasshown to kill pathogenic bacteria and eliminate coliformcount. This recommended protocol can be followed once aweek where coliform counts are high. We have no recordsof diarrhea or other water borne diseases from drinkingSONO-filtered water. It appears that the SONO filtrationsystem does not foster pathogenic bacteria on its own.

Except for basic training in hygiene, no special skill isrequired to maintain the filter. The maintenance processrequires about 20–30 mins. Because the SONO filter hasno breakthrough, the active media does not require anyprocessing (backwashing, regeneration etc). The filter willproduce potable water for at least 5 years (time span ofour continuing test results). The actual filter life span isdetermined by the life span of the 6 experimental filters(in Table 3) running in the field. Except for manufactur-ing defects, mechanical damage due to mishandling, trans-portation, and natural disasters (flooding), none of thefilters showed MCL breakthrough to this date.

Manufacturing and dissemination

The filter is now manufactured by an NGO (Manob SaktiUnnyan Kendro -MSUK, Kushtia, Bangladesh) from in-digenous materials at 200–500 units per lot (Fig. 5, pho-tos). The filter materials are available almost anywhere inthe world except the CIM, which can be produced with anappropriate licensing agreement. Large-scale transporta-tion of the filters takes place with flat bed trucks, and filter

1876 Hussam and Munir

Fig. 5. (a) SONO filter production center at Kushtia, Bangladesh. (b) Filters are loaded onto a flat bed truck for distribution.

distribution in villages occurs with flat bed rickshaws. Fol-lowing simple instructions and without setup costs, the usercan set up the system in 20 minutes. At the time of writingthis paper, about 30,000 filters were distributed in 16 dis-tricts throughout Bangladesh. The large-scale deploymentwas accomplished through participation of a dozen localNGOs and international institutions. For example, MSUKhas distributed 825 filters to 325 primary schools serving67,000 students and teachers. The SONO filter is also usedby many local restaurants, tea/candy shops, and villagersin remote places. We estimate that about half-a-million peo-ple are the direct beneficiaries of the filtration system. It isworthy to note that many people, the authors included, arediscovering the joy of using soft-water in drinking and cook-ing free from arsenic and other toxic species. The filtrationsystem has been scaled up by connecting units in parallel.

This can easily enhance the flow rate for small communityuse.

Residue management

Table 5 shows the measurements on used sand and CIM-Fe by total available leaching protocol (TALP). TALP issimilar to EPA’s toxicity characteristic leaching procedure(TCLP) except the samples were ground to fine powder be-fore leaching at two different pH values. The procedureis also followed with Bangladesh rainwater (adjusted topH 7), where the primary mode of transport of water sol-uble species takes place during the rainy season. These re-sults show that the spent material is completely nontoxicwith less than 16 µg/L As(total), which is 300 times lessthan the EPA limit. Similar results were also reported byETVAM using EPA’s TCLP methods.

Table 5. Trace element distribution of leachates from filter spent material by total available leaching protocols (TALP). TALP wasimplemented with deionized water and Bangladesh rainwater at indicated pH values.

at pH = 7Element at pH = 7 at pH = 4 at pH = 7 at pH = 4 (Sand + iron)Conc. (mg/L) (Sand) (Sand) (Sand + Iron) (Sand + Iron) Rainwater

Al 0.028 0.02 0.049 0.014 0.02As (EPA = 5) <0.016 <0.016 <0.016 0.025 <0.016Ba 0.002 0.007 0.006 0.009 0.008Ca 2.65 8.39 4.89 8.13 4.00Cu 0.005 0.018 0.007 0.008 0.002Fe 0.02 0.031 0.041 0.051 0.016Mg 0.2 1.03 0.519 0.831 0.52Mn 0.013 0.081 0.005 0.047 0.002Ni 0.002 0.009 0.003 0.005 0.004K 0.82 1.26 0.608 0.527 0.62Na 2.64 0.750 3.35 1.85 7.2Sr 0.006 0.017 0.014 0.02 0.012Sn 0.007 0.011 0.003 0.006 0.003Zn 0.011 0.038 0.013 0.021 0.024

Note: The “<” sign indicates the instrumental detection limit for ICP-AES and specified as the concentration below the detection limit (bdl). For Asthe bdl is 0.016 mg/L or 16 ppb. Other metals- Pb, Cd, Se, Ag, Sb, Cr, and Mo were either below the method detection limit or the drinking waterlimit. Sand: Iron = 50:50. Iron is CIM-Iron.

Developing an arsenic filter for Bangladesh 1877

Further tests on backwash of filter waste showed SONOproduced the lowest concentrations of As(total), 93 mg/kg,in comparison to commercial filters based on micro fine ironoxide-2339 mg/kg, cerium hydroxide based ion exchangeresin-105 mg/kg, and activated alumina-377 mg/kg in solidwaste. These values are all below EPA limits of 5,000 mg/kg.Arsenic species in the filter’s used sand and CIM are inthe oxidized form and firmly bound with insoluble solidCIM. This is similar to a self-contained naturally occurringcompound in the earth’s crust. It is almost like disposingsoil on soil. Most importantly, the NAE- tests of the usedCIM of SONO filter was characterized as “non-detectableand non-hazardous (limit 0.50 mg/L)” by the TCLP.[31]

Social acceptability

Presently, at $35–40 per 5 years (equivalent to one-monthincome of a village labor in Bangladesh), SONO is one ofthe most affordable water filters in Bangladesh. Affordabil-ity can be enhanced by monthly payment schedules throughNGOs distributing the filter. The SONO filter does not re-quire any chemicals or consumables. The estimated oper-ating cost is no more than $10.00 over 5 years, only if theflow controller needs replacement, which is extremely rare.One unit can serve 2 families need for drinking and cookingwater for at least 5 years. The SONO filter setup and main-tenance do not require any special skill. The potable wateris collected within 2–3 hours after discarding the first twobatches. After overcoming the initial skepticism of drink-ing different water from a filter, people liked the taste ofsoft-water. Our experience shows that mostly women par-ticipate in the water collection and maintenance of the filter.The filter has been accepted extremely well because womendo not have to go far to find arsenic free wells (paintedgreen)—they can still use their contaminated wells. Theusers drinking this water for 2 years show some disappear-ance of arsenical melanosis with a general sense of well be-ing and improvement in health. These data are now beingcollected and will be reported. Besides drinking and cook-ing, at 20 L/hour flow rate the filters produce enough waterto be used for other purposes, such as cleaning and washingcooking utensils. Except for a reluctance to share filter waterwith neighbors, we find no critical social and cultural issueshindering the dissemination and use of the filter. We foundschool children drinking SONO water carrying bottles ofwater home at the end of the school day. Many NGOs havealso implemented intensive training and cultural programsto motivate people to drink arsenic-free water.

Future outlook

The present filtration system can be modified and improved.Scaling up for large volume water filtration is being studiedby connecting the units in parallel. Scaling down to a table-top unit while maintaining the same efficiency is also un-der investigation. The filter’s inherent capability to remove

thermo tolerant coliform is now proven. Further experi-ments to prove its efficacy in removing other pathogenicbacteria and virus are planned. It is now clear that inBangladesh and many other countries, while the surfacewater is not potable without treatment and filtration, a ma-jor portion of the groundwater is also not potable due to thepresence of many toxic species. It appears that the develop-ment of low-cost filters is the only way to solve the presentdrinking water crisis for many countries of the world.

Acknowledgment

The authors deeply appreciate the workers of SDC/MSUK, Bangladesh for their unwavering support. We alsoacknowledge assistance from Prof. A. H. Khan and Prof. A.Barkat, Dhaka University, Bangladesh; Prof. M. Alauddin,Wagner College, NY, USA; and Dr. S. S. Newaz, Polyor-ganix Inc., USA. The authors acknowledge Dr. Shahamat.U. Khan for his critical constructive suggestions during thedevelopment of this filter.

References

[1] Smith, A.H.; Lingas, E.O.; Rahman, M. Contamination of drinking-water by arsenic in Bangladesh: a public health emergency. Bull.World Health Org. 2000, 78(9), 1093–1103.

[2] Chatterjee, A.; Das, D.; Mandal, B.K.; Chowdhury, T.R.; Samanta,G.; Chakraborti, D. Arsenic in groundwater in six districts of WestBengal, India: the biggest arsenic calamity in the World. Part 1—arsenic species in drinking water and urine of the affected people.Analyst 1995, 120, 643–650.

[3] IARC (International Agency for Research on Cancer), IARC Mono-graphs on the Evaluation of the Carcinogenic Risk of Chemicals toHuman, IARC, Lyons. Arsenic in Drinking Water 2001 update. Na-tional Academy Press, Washington DC, 2001, 1980; p 20.

[4] Performance Evaluation and Verification of Five Arsenic RemovalTechnologies: ETVAM Field Testing and Technology VerificationProgram. Bangladesh Council of Scientific and Industrial Research(BCSIR), Dhaka, Bangladesh, September 2003.

[5] Grainger Challenge. http/www.nae.edu/nae/grainger.nsf.[6] Rasul, S.B.; Hossain, Z.; Munir, A.K.M.; Alauddin, M.; Khan,

A.H.; Hussam, A. Electrochemical measurement and speciationof inorganic arsenic in groundwater of Bangladesh. Talanta. 2002,58(1), 33–43.

[7] Report on the Arsenic PT TC Project BGD/08/018, IAEA, Analyti-cal Quality Control Services, Agency’s Laboratories, A-2444 Seibers-dorf, Austria. 2005.

[8] Khan, A.H.; Rasul, S.B.; Munir, A.K.M.; Habibuddowla, M.;Alauddin, M.; Newaz, S.S.; Hussam, A. Appraisal of a simple ar-senic removal method for groundwater of Bangladesh. J. Environ.Sci. Health 2000, A35(7), 1021–1041.

[9] Pfaff, J.D.; Hautman, D.P. Method 300.1: Determination of Inor-ganic Anions in Drinking Water by Ion Chromatography, Revision 1.National Exposure Research Laboratory, Office of Research andDevelopment, U.S. Environmental Protection Agency (USEPA),Cincinnati, Ohio 45268. 1997.

[10] Patent No. 1003935, 2002, Arsenic Removal Filter, Department ofPatents, Design and Trade Marks, Patents and Design Wings, Gov-ernment of Bangladesh, Dhaka, Bangladesh and U.S. Patent Pend-ing, 2007.

1878 Hussam and Munir

[11] Rapid Assessment of Arsenic Removal Technologies, Phase 1, FinalDraft Report, BAMWSP/DFID/Water Aid, 2001.

[12] Hurd, J.J. Evaluation of three arsenic removal technologies in Nepal.MS Thesis, Civil and Environmental Engineering, MassachusettsInstitute of Technology, USA, June 2001; pp. 47–50.

[13] Manning, B.A.; Fendorf, S.E.; Goldberg, S. Surface structures andstability of arsenic(III) on goethite: Spectroscopic evidence for innersphere complexes. Environ. Sci. Tech. No. 1 1998, 32, 2383–2388.

[14] Waychunas, W.A.; Rea, B.A.; Fuller, C.C.; Davis, J.A. Surface chem-istry of ferrihydrite, Part I: EXAFS studies of geometry of co pre-cipitated and adsorbed arsenate. Geochim. Cosmochim. Acta 1993,57, 2251–2270.

[15] Farrell, J.; Wang, J.; O’Day, P.; Conklin, M. G. Electrochemical andspectroscopic studies of arsenate removal from water using zero-valent iron media. Environ. Sci. Technol. ASAP Web Edition Feb26, 2001, 10.1021/es0016710.

[16] Hussam, A.; Habibuddowla, M.; Alauddin, M.; Hossain, Z.A.;Munir, A.K.M.; Khan, A.H. Chemical fate of arsenic and othertrace metals in groundwater of Bangladesh: Experimental measure-ment and chemical equilibrium model. J. Environ. Sci. Health. Pt.A 2003, A38(1), 71–86.

[17] Munir, A.K.M.; Rasul, S. B.; Habibuddowla, M.; Alauddin, M.;Hussam, A.; Khan, A. H. Evaluation of the Performance of theSONO3-Kolshi Filter for Arsenic Removal from Groundwater UsingZero Valent Iron Through Laboratory and Field Studies. Proceed-ings International Workshop on Technology for Arsenic Removalfrom Drinking Water, Bangladesh University of Engineering andTechnology and United Nations University, Japan, May 5, 2001;pp. 171–189.

[18] Alauddin, M.; Hussam, A.; Khan, A.H.; Habibuddowla, M.; Ra-sul, S.B.; Munir, A.K.M. Critical Evaluation of a Simple ArsenicRemoval Method for Groundwater of Bangladesh. In: Arsenic Ex-posure and Health Effects, Chappell, W.R., Abernathy, C.O., andCalderon, R.L. (Editors), Elsevier Science, B.V. 2001; pp. 439–449.

[19] Hussam, A.; Munir, A. K. M. Development and Deployment ofArsenic Filters for Groundwater of Bangladesh, in Green Chemistryand Water Purity, American Chemical Society 37th Middle AtlanticRegional Meeting, Rutgers University, New Jersey, USA, May 23rd,2005. Abstract 252.

[20] Dzombak, D.A.; Morel, F.M.M. Surface Complexation Modeling:Hydrous Ferric Oxide, Wiley-Interscience, New York, 1990.

[21] Wilkie, A.J.; Herring, J. Adsorption of arsenic onto hydrous fer-ric oxide: effects of adsorbate/adsorbent ratios and co-occurringsolutes. Coll. Surf. A: Physicochem. Eng. Aspects 1986, 107, 97–110.

[22] Schecher, W.D., McAvoy, D.C. MINEQL+: A Chemical Equilib-rium Program for Personal Computers, User’s Manual, Version 4.0,Environmental Research Software, Hallowell, ME 04347, 1998.

[23] MINTEQA2 Model System, Center for Exposure Assessment Mod-eling, Environmental Protection Agency, 960 College Station Road,Athens, GA 30605-2720, 2001.

[24] Davis, C.C.; Chen, H.-W.; Edwards, M. Modeling silica sorption toiron hydroxide. Environ. Sci. Technol. 2002, 36(4), 582–587.

[25] Stephen, J.H.; Canonica, L.; Wegelin, M.; Gechter, D.; Gunten, V.U.Solar oxidation and removal of arsenic at circumneutral pH in ironcontaining water. Environ. Sci. Technol. ASAP Web Edition, 2001,es001551s.

[26] Performance Evaluation and Verification of Five Arsenic RemovalTechnologies: ETVAM Field Testing and Technology VerificationProgram. Bangladesh Council of Scientific and Industrial Research(BCSIR), Dhaka, Bangladesh, September 2003.

[27] Rabbani, G.H. Double-Blind, Randomized. Placebo-Controlled Trialof Antioxidant Vitamins and Minerals in the Treatment of ChronicArsenic Poisoning in Bangladesh, Bangladesh Arsenic Control So-ciety Research Group Final Report Sponsored by UNICEF andGovernment of Bangladesh, 2003.

[28] Graziano, J. Poison in the well: The case of arsenic in drinking waterof Bangladesh. American Chemical Society, 37th Middle AtlanticRegional Meeting. Session: Assuring Water Purity, May 23, 2005.Abstract # 250

[29] Islam, M.S.; Siddika, A.; Khan, M.N.H.; Goldar, M.M.; Sadique,M.A.; Kabir, A.N.M.H.; Huq, A.; Colwell, R.R. MicrobiologicalAnalysis of Tube-Well Water in a Rural Area of Bangladesh. Appl.Environ. Microbiol. 2001, 67(7), 3328–3330.

[30] Arsenic Mitigation Pilot Project: Bacteriological Field Test Report,Dhalipara & Doazipara of Muradpur Union in Sitakunda, April2005–March 2006. Village Education Resource Center (VERC),B-30, Ekhlas Uddin Khan Road, Anandapur, Savar, Dhaka-1340,Bangladesh, 2007.

[31] Final Report: Evaluation of Grainger Challenge Arsenic TreatmentSystems-SONO Filter #29. Prepared by Shaw Environmental Inc.,under EPA Contract No. EP-C-05-056 and National Academy ofEngineering- Shaw PN 118205-03, December 2006.