Subsurface Iron and Arsenic Removal

Doris van Halem 

Subsurface Iron and 

Arsenic Removal 

for drinking water treatment in Bangladesh

Subsurface Iron and Arsenic Removal 

for drinking water treatment in Bangladesh 

Proefschrift 

ter verkrijging van de graad van doctor 

aan de Technische Universiteit Delft, 

op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben, 

voorzitter van het College van Promoties 

in het openbaar te verdedigen op maandag 3 oktober 2011 om 15:00 uur 

door 

Doris VAN HALEM 

civiel ingenieur 

geboren te Eindhoven

Dit proefschrift is goedgekeurd door de promotoren: 

Prof. ir. J.C. van Dijk 

Prof. dr. G.L. Amy 

Samenstelling promotiecommissie: 

Rector Magnificus 

voorzitter 

Prof. ir. J.C. van Dijk 

Technische Universiteit Delft, promotor 

Prof. dr. G.L. Amy 

UNESCO-IHE/Technische Universiteit Delft, promotor 

Dr. ir. J.Q.J.C. Verberk 

Technische Universiteit Delft 

Prof. dr. K.M.U. Ahmed University of Dhaka 

Prof. dr. ing- U. Rott 

Technische Universität Stuttgart 

Prof. dr. ir. H.H.G. Savenije Technische Universiteit Delft 

Prof. dr. P.J. Stuyfzand 

Vrije Universiteit Amsterdam 

Prof. dr. ir. W.G.J. van der Meer Technische Universiteit Delft, reservelid 

ISBN: 978-90-8957-022-2 

Published by Water Management Academic Press 

Copyright © 2011 by D. van Halem 

All rights reserved. No part of the material protected by the copyright may be reproduced or utilized in any form or by any means, 

electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without written 

permission from the copywright owner. 

Cover design: S. Willems (www. sjorswillems.nl) 

Printed in the Netherlands

Acknowledgements 

Acknowledgements 

The desire to make a difference has been the driving 

force to start my PhD research, but along the road I 

have also increasingly enjoyed the scientific aspects of 

it. This dissertation could not have been written without 

the people around me, so thank you all for making my 

PhD period anything but lonely. 

First of all I thank my two promotors, Prof. Hans 

van Dijk and Prof. Gary Amy, who have given me the 

opportunity to take this new research direction. Hans, 

your enthusiasm and confidence have encouraged me 

to approach this research in my own way. I enjoyed 

working with you, especially because you know how to 

transfer your knowledge (and gezond boerenverstand) 

without dominating the discussion. Gary, thank you 

for sharing your broad expertise with me, and I am 

glad that I also got to know you as a kind and caring 

person. With two promotors operating from a distance 

in the last year(s), either retired or working in Jeddah, 

it was essential to have a daily supervisor. Jasper, thank 

you for all the guidance and laughs, you are a valuable 

colleague. 

My time in Bangladesh was a success thanks 

to the great support of Rick Johnston (Unicef), his 

know-how and generosity helped me get started in this 

beautiful country. At the Department of Public Health 

Engineering I want to thank Ihtishamul Huq, Sudhir 

Ghosh and Shah Alam for tackling loads of political and 

practical hinders, and for assisting in the monitoring 

program. Dhonnobad. 

The field research in the Netherlands would not 

have been possible without the support and hospitality 

of Oasen Drinking Water Company and Vitens 

Drinking Water Supply at WTPs Lekkerkerk, De Put 

and Loosdrecht. I am also thankful for the funding of 

iii

Subsurface iron and arsenic removal for drinking water treatment in Bangladesh 

AgentschapNL in their InnoWator program. 

During the past four years I was lucky to 

supervise many (international) students, whose results 

have contributed to several of my published papers. I 

especially want to thank Samuele, David and Harmen 

who have battled the O 2 

and challenged me with 

surprising questions. 

At TU Delft, I want to thank my colleagues 

for accompanying me on this PhD journey, which 

took us from coffee breaks, past tropical beaches and 

inspiring discussions (on water bugs and hydrophobia), 

to running the 10km. With special thanks to Mieke, 

Weren, David, Bas, Luuk, Arne, Sheng, Anke, Petra, 

Tonny, Martine and Stefan. I wish Sandra, Moshiur, 

Zahid and Debasish all the best in pursuing their PhD 

within the newly started NWO WOTRO project on 

Subsurface Arsenic Removal. I also want to thank my 

colleagues at UNESCO-IHE, who have made the start of 

my PhD more enjoyable. Addionally, I want to mention 

the TU Delft and UNESCO-IHE laboratory staffs who 

have provided great assistance from analysis with GF- 

AAS to designing (oxygen-free) column installations. 

Apart from my direct colleagues, it has been 

very inspiring to discuss my findings with so many 

(international) scientists at conferences and meetings, 

and during peer-review processes. A PhD path is not 

something you walk by yourself in solitude, because 

it is the interaction with the world that makes you 

reconsider your route. 

Naturally it is impossible to enjoy four years of 

research without the occasional distraction of friends 

and family. I want to thank you all for asking me about 

my research, and maybe even more for not asking. 

In the final months of my PhD period it was great 

to have the support of my two paranimphs, Geerte and 

Harmen, who share my enthusiasm for research and 

helped me with the last hurdles. 

I realize that it is because of my loving parents 

that I approach life with an open and positive mind. 

Pap en mam, thanks for supporting me, your believe in 

me gives me wings. 

Most of all I want to thank Wieger, for always 

keeping me close whether I am in Delft or on the other 

side of the world. I do not know why we are so lucky, 

but life is amazing with you. And Ferre, you are my 

little explorer, curious and brave, the best example a 

scientist can get. 

Een rivier dankt haar naam 

aan de oevers, want het water 

stroomt anoniem voorbij. 

T. van Lieshout,1987 

iv

Table of contents 

Table of contents 

Acknowledgements iii 

1 Introduction 1 

2 Subsurface iron removal in the Netherlands 23 

3 Small-scale subsurface iron and arsenic removal in Bangladesh 37 

4 Simulation of adsorptive-catalytic oxidation in sand columns 51 

5 Cation exchange during subsurface iron removal 67 

6 Characterization of accumulated deposits 81 

7 Catalysis by accumulated deposits 99 

8 Influence of groundwater composition 111 

9 Concluding remarks 125 

Summary 135 

Samenvatting 139 

List of publications 145 

Biography 148 

v


vi

1 

Introduction 

Parts of this chapter are based on: 

van Halem et al. (2009) Drinking Water Engineering and Science 2: 29-34 

van Halem et al. (2010) Desalination 248: 241-248


1 An introduction to the arsenic 

problem 

A worldwide problem 

The World Health Organization estimated in 2001 that 

about 130 million people worldwide are exposed to 

arsenic concentrations above 50 μg.L -1 (WHO, 2001). 

Affected countries include Bangladesh (>30 million 

exposed people), India (40 million), China (1.5 million) 

and the United States (2.5 million). The problem of 

arsenic-contaminated source waters is, however, not 

confined to these countries, as illustrated in Figure 

1.1. According to the United Nations Synthesis report, 

arsenic poisoning is the second most important health 

hazard related to drinking water (Johnston et al., 2001). 

Only contamination by pathogenic microorganisms has 

a bigger impact worldwide. Arsenic contamination of 

groundwater (Box 1.1) has been found to occur due to 

(Smedley and Kinniburgh, 2002): 

• geothermally- influenced groundwater; 

• mineral dissolution (e.g., pyrite oxidation); 

• desorption in the oxidising environment, and; 

• reductive desorption and dissolution. 

Reductive dissolution of young arsenic-bearing 

sediments is the cause of the large-scale arsenic 

contamination of the strongly reducing aquifers in the 

Bengal Delta. Also in China the reducing conditions 

in the subsurface are the cause of arsenic mobilization. 

Figure 1.1 Arsenic-affected countries of the world (Smedley and Kinniburgh, 2002; Appleyard et al., 2006; 

Petrusevski et al., 2007; Smedley et al., 2007; Gunduz et al., 2009; Jovanovic et al., 2011) 

2

1 Introduction 

Box 1.1 

Arsenic speciation, including an Eh-pH diagram for aqueous arsenic species in the system As-O 2 

-H 2 

O at 25°C and 1 bar total 

pressure (Smedley and Kinniburgh, 2002) 

1 

Arsenic is stable in four oxidation states (+5, +3, 0, -3) 

under redox conditions occurring in aquatic systems 

(Furguson and Gavis, 1972). However, in the aqueous 

environment, arsenic occurs mainly as two species; trivalent 

arsenic, arsenite or As(III) and pentavalent arsenic, arsenate 

or As(V). Under oxidizing and aerated conditions, the 

predominant form of arsenic in water is As(V) (as HAsO 4 

2- 

or H 2 

AsO 4 

- 

around neutral pH), whereas if reducing 

conditions prevail, As(III) (around neutral pH as H 3 

AsO 3 

) is 

the predominant arsenic compound (WHO, 2001). As(III) 

is thermodynamically unstable in oxic environments and 

oxidizes to As(V); however studies have shown that this 

reaction proceeds slowly when oxygen is the only oxidant 

(AWWARF, 2000; Kim et al., 2000; Stollenwerk, 2003). 

Therefore both As species can be found in natural systems. 

The speciation of arsenic is controlled by pH within a 

particular oxidation state, while the redox potential controls 

the distribution of arsenic species between the two oxidation 

states (Masscheleyn et al., 1999; Bose and Sharma, 2002). 

Concentrations up to 1,800 μg.L -1 have been measured in 

Inner Mongolia, a northern province of China (Smedley 

et al., 2003). In Vietnam and Cambodia, arsenic 

concentrations were also observed to be high (up to 1,340 

μg.L -1 ) due to reductive dissolution of young sediments 

(Buschmann et al., 2007; Buschmann et al., 2008). 

Arsenic mobilization caused by mineral dissolution 

has been found in active volcanic areas of Italy (Aiuppa 

et al., 2003) and inactive volcanic regions in Mexico 

(Armienta and Segovia, 2008). Volcanism in the Andes 

has lead to arsenic contamination of groundwater in 

Chile and Argentina (Smedley and Kinniburgh, 2002). 

Also mining activities have been found to contribute to 

arsenic contamination in Latin American groundwater 

(Smedley and Kinniburgh, 2002). Mining activities may 

cause the oxidation of arsenic-bearing sulphide minerals 

resulting in the release of arsenic into groundwater. 

Smedley and Kinniburgh (2002) listed cases of arsenic 

contamination caused by mining activities in Canada, 

Germany, Ghana, Greece, Mexico, South Africa, 

Thailand, UK, USA and Zimbabwe. In the past years, 

more and more countries have found their waters to 

3


1 be affected by arsenic contamination due to mining 

wastes, e.g., Poland, Korea and Brazil (Marszałek and 

Wasik, 2000; Woo and Choi, 2001; Borba et al., 2003). 

More recently, groundwater in Burkino Faso was found 

to be contaminated by arsenic with concentrations up 

to 1,630 μg.L -1 , caused by mining activities (Smedley et 

al., 2007). Furthermore, Gunduz et al. (2009) reported 

elevated arsenic levels ( As(V) > monomethylarsonate 

(MMA) > dimethylarsinate (DMA) (Jain and Ali, 

2000). In areas with elevated arsenic concentrations in 

the environment, the exposure is not solely confined 

to drinking water. Arsenic (organic and inorganic) is 

also found in a wide range of food products, like fish, 

4


meat and rice (WHO, 2001b; Williams et al., 2006; 

Sambu and Wilson, 2008). Intake through air may also 

be significant, especially close to industrial sources 

(WHO, 2001b). 

The cancer risk at low-to-moderate exposure 

concentrations in drinking water is still under debate 

(Smith et al., 2002; Celik et al., 2008). Most risk 

estimations use data from Taiwanese studies (Tseng et al., 

1968; Chen et al., 1992), since limited epidemiological 

information is available from elsewhere in the world. 

In Bangladesh, results indicated at least a doubling of 

lifetime mortality risk from liver, bladder, and lung 

cancers (229.6 vs 103.5 per 100,000 population) owing 

to arsenic in drinking water (Chen and Ahsan, 2004). 

This has an enormous impact since it is estimated that 

of the 140 million inhabitants of Bangladesh, over 37 

million are at risk of arsenic poisoning (WHO, 2001; 

Chowdhury et al., 2006). The cancer risks from arsenic 

in drinking water were assessed by Smith and coworkers 

(1992) for the USA situation. At that time the 

national guideline in the USA was 50 μg.L -1 and they 

concluded that the lifetime risks of dying from cancer 

due to arsenic in drinking water was 21 in 1,000 adults. 

For a concentration of 2.5 μg.L -1 the risk would still be 

1 in 1,000 adults, which they found comparable to the 

lifetime cancer risk of passive smoking. More recently, 

based on male bladder cancer with an excess risk of 

1 in 10,000 for 75-year lifetime exposure, the arsenic 

guideline is recommended to be 3.4 μg.L -1 (Liao et al., 

2009). 

The World Health Organization has published an 

overview document on the toxicology and legislation 

for arsenic in drinking water (WHO, 2003). They 

conclude that the maximum likelihood “for bladder 

and lung cancer for US populations exposed to 10 μg 

of arsenic per litre in drinking water are, respectively, 

12 and 18 per 10,000 population for females and 23 

and 14 per 10,000 population for males”. The WHO 

has a general rule that no substance may have a higher 

lifetime risk of more than 1 in 100,000. Purely based on 

health effects, the WHO guideline of 10 μg.L -1 would, in 

that respect, not suffice. The main reason to maintain 

this guideline is, therefore, merely practical from 

economic and engineering perspective and not health 

related. The US Environmental Agency (EPA, 1998) and 

the US Natural Resources Defense Council (NRDC, 

2000) even recommend arsenic guidelines below 1 

μg.L -1 to attain an acceptable lifetime cancer risk. It is 

noteworthy that EPA considered life-time skin cancer 

risk only and did not include arsenic intake through 

food due to a lack of reliable data. The consumption 

of arsenic through food could overestimate the 

current risk calculations and EPA indicates a possible 

uncertainty of one order of magnitude. The WHO and 

EPA do not provide information on whether inorganic 

arsenic is genotoxic or non-genotoxic, because current 

epidemiologic studies are inadequate for that. In their 

background documents, these organizations describe 

the cancer risks based on both approaches. Although 

the uncertainties concerning the health risks due to 

5 

1


1 arsenic in drinking water are undeniable, it is clear that 

extremely low concentrations, or its complete absence, 

are desirable to avoid these potential risks. 

Household water treatment and 

safe storage 

Household water treatment and storage (HWTS) has 

been recognized to be an effective measure in reducing 

diarrheal diseases (Sobsey, 2002; Fewtrell et al., 2005; 

Clasen, 2008). HWTS treats the water in homes to 

remove microbial and/or chemical contaminants. The 

targeted microbial contaminants generally include 

helminthes, protozoa, bacteria and viruses, in order to 

reduce diarrhoea cases and diseases such as bilharzias 

and cholera. HWTS is a so-called point-of-use water 

treatment, greatly reduceing the risk of recontamination 

of the treated water during transport (Sobsey, 2002). 

HWTS systems to remove microorganisms include 

technologies based on chlorination, coagulationfiltration, 

solar disinfection, ceramic and (bio)sand 

filtration (Sommer et al., 1997; Stauber et al., 2006; 

van Halem et al., 2007). Although HWTS is widely 

promoted as an appropriate intervention (UNICEF, 

2009; WHO, 2011), studies have shown that not 

all systems are as effective in reducing pathogenic 

microorganisms. Hunter (2009) predicted that over 12 

months, ceramic filters were likely to be still effective 

at reducing disease, whereas solar disinfection, 

chlorination, and coagulation-chlorination had little if 

any long-term public health benefit. Additionally, the 

Biosand filter, alongside the ceramic filter, was found to 

have the greatest potential to become widely used and 

to be most effective in reducing waterborne disease and 

death (Sobsey et al., 2008). Such findings point out that 

not all HWTS interventions are sustainable, illustrating 

the need for knowledge on technology development, 

adoptation and socio-economic dynamics. 

In the past decade, HWTS solutions have 

also been developed for the removal of arsenic; most 

of these technologies have translated conventional 

treatment methods to a household solution. When 

considering arsenic contamination in (anoxic or 

anaerobic) groundwater, the dominant arsenic species is 

generally As(III). Around neutral pH As(III) (H 3 

AsO 3 

) 

is uncharged and therefore difficult to remove with 

processes that rely on surface charge (ion exchange, iron 

hydroxide adsorption). As(V) (HAsO 4 

2- 

or H 2 

AsO 4- 

) 

can be more easily removed because it is negatively 

charged and is, therefore, relatively easily incorporated 

into the iron oxide matrix during iron removal. In 

conventional groundwater treatment usually preoxidation 

to As(V) is required to remove As(III) from 

the water. Hypochlorite and permanganate are common 

oxidants to catalyze the reaction of As(III) oxidation. A 

lot of attention has been given to the development of 

technologies based on the co-precipitation of arsenic 

in flocs during coagulation (e.g., by dosing of ferric 

6


sulfate or ferric chloride) and arsenic adsorption 

to media, like activated alumina and granular iron 

oxide/oxide (GFH; Banerjee et al., 2008). In all cases 

the arsenic binds to the positively-charged surface 

of the (iron hydroxide) matrix. Especially at low to 

moderate arsenic concentrations, the technology 

of arsenic adsorption is relatively effective. Other 

arsenic removal technologies include ion exchange, 

coagulation/filtration and reverse osmosis (AWWARF, 

2000; EPA, 2007; Johnston et al., 2001). Most processes 

are, however, not designed for household application 

in a developing country context. The sustainability 

of a HWTS can be addressed based on five criteria 

(van Halem et al., 2009): accessibility, water quality, 

water quantity, functionality, and environmental 

footprint. Accessibility entails both the affordability 

and availability of the technology and should be 

expressed in money, time or distance. A product can be 

accessible through a local shop or organization, but in a 

globalizing world, also through an ordering system on 

the internet. The criteria of water quality and quantity 

are formulated by the World Health Organization 

(WHO, 2006; WHO, 2007), being 2L per person per day. Functionality depends on 

the operation and maintenance activities for the users, 

which strongly determines the social acceptance of a 

technology. The environmental footprint is generally 

not the main concern of most users, but long-term 

1 

brick 

chips 

coarse sand 

cast iron 


coagulation and 

flocculation bucket 

cloth filter 

scooping net 


charcoal 

fine sand 

brick 

chips 

media 

media 

perforated plate 

covered with 

cotton filter 

sand 

cloth filter 

sand 

EVOH resin 

geo textile 

synthetic cloth 

scooping net 

scooping net 

charcoal 

Figure 1.2 Arsenic household water treatment and storage (from left to right): Sono Filter, Alcan Filter, Shawdesh Filter and Read-F 

(BETV-SAM, 2011) 

7


1 issues regarding chemical/electricity consumption and 

production of a waste stream may seriously affect a 

rural community. 

The majority of the adsorption-based HWTS 

technologies targeting arsenic removal, rely on iron 

oxides, such as cast iron (Sono Filter, Figure 1.2a), 

iron coated bricks/sand (Shapla Filter/IHE Filter) 

and iron nails (Kanchan Filter), or activated alumina 

(Alcan Filter, Figure 1.2b). In order to remove iron 

from the groundwater these filters use a bucket-style 

sand filter. Also frequently combined with a sand 

filter are oxidation and coagulation-flocculation 

processes by dosing chemicals such as permanganate, 

sodium hypochlorite, iron or aluminum sulphate 

(DPHE-Danida Bucket Treatment Unit; 2-Kolshi 

Filter; Shapla Filter; Shawdesh Filter, Figure 1.2c) or 

atmospheric oxygen (Asia Arsenic Network Filter). Ion 

exchange resins have also been utilized for household 

arsenic removal (Read-F, Figure 1.2d; Tetrahedron), 

in combination with a cloth and/or sand filter. Most 

arsenic HWTS solutions consist of buckets on a tripod 

and appear to be similar, but reported efficacies vary 

widely between 90% (Sutherland, 2002; 

IGRAC, 2007; BETV-SAM, 2011). 

Many factors influence removal efficacies (e.g., 

water quality, consumer operation and maintenance), 

but before the implementation of HWTS one must 

have solid and statistically valid results showing 

that it is a safe technology. The post-deployment 

monitoring is generally poor or absent, even further 

demanding for a safe and robust verification of the 

newly developed technologies. The Bangladesh 

Environmental Technology Verification – Support to 

Arsenic Mitigation (BETV-SAM) program has tested 

15 arsenic removal technologies and has approved only 

6 of them for sale. The majority of arsenic removal units 

failed to perform according to the set guidelines (www. 

verification-unit.org). Nevertheless, there are arsenic 

HWTS solutions that show potential, although further 

research and documentation is needed to determine 

the long-term effectiveness. Additionally, the treatment 

of water in the homes requires capacity building in 

these rural communities. The users need to be aware 

of the operation and maintenance obligations. because 

all filters will at one stage show breakthrough of arsenic 

and many leave a waste stream that should be disposed 

of as hazardous waste. 

Subsurface Iron and Arsenic 

Removal 

A new HWTS approach to remove arsenic from 

groundwater is by retention in the subsurface, 

preventing the generation of a waste stream (Sarkar 

and Rahman, 2001; Rott et al., 2002). Immobilizing the 

arsenic in the aquifer provides the additional benefit 

that existing infrastructure, namely the (shallow) tube 

well hand pump, can be utilized for treatment. In that 

8


case there is no longer the need for above-ground water 

treatment. This novel technology, Subsurface Arsenic 

Removal (SAR), is based on the existing technology of 

Subsurface Iron Removal (SIR), or in-situ iron removal. 

Subsurface Iron Removal (SIR) 

With a patent originating from 1900 (von Oesten, 

1900), subsurface iron removal was first introduced 

in Sweden in 1971 (Mettler, 2002), but soon other 

European countries followed with subsurface iron 

removal plants, including Switzerland, Germany, 

Denmark, France and the Netherlands (van Beek, 

1985; Mettler, 2002). The principle of subsurface iron 

removal is that aerated water is periodically injected 

into an anoxic or anaerobic aquifer through a tube 

well, partially displacing the original Fe 2+ -containing 

groundwater. The O 2 

-rich injection water oxidizes the 

Fe 2+ in the subsurface environment around the tube 

well. When the flow is reversed, groundwater with 

low Fe concentrations is abstracted. More water with 

reduced iron concentrations can be abstracted (volume 

V) than was injected (volume Vi), i.e., this volumetric 

ratio (V/Vi) determines the efficiency of the system. 

When O 2 

-rich water is injected into the aquifer, the 

adsorbed Fe 2+ will oxidize heterogeneously: 

Equation 1.1 

S − OFe( II) + 0.25O + 1.5 H O →S − OFe( III)( OH ) + H 

+ 0 + 

2 2 2 

With S being the solid iron oxide surface. The freshly 

formed Fe 3+ iron hydroxides provide new adsorption 

sites, for soluble Fe 2+ during the abstraction phase. The 

adsorption of Fe 2+ occurs in the absence of oxygen and 

can be schematically formulated as: 

Equation 1.2 

o 2+ + + 

S − OH + Fe →S − OFe( II ) + H 

It should be noted that the use of S-OH 0 in the equation 

is simplified, as in reality adsorbed Fe 2+ may transfer 

an electron to the solid (Hiemstra and van Riemsdijk, 

2007). The adsorptive capacity of the soil depends on 

the type of iron oxides present in the soil, as adsorption 

capacities for the amorphous ferrihydrite are larger 

than for crystalline mineral structures with lower 

surface areas (goethite, lepidocrocite). The presence 

of Fe 3+ oxides are known to catalyze the Fe 2+ oxidation 

reaction (Tamura et al., 1980; Stumm and Morgan, 

1996), making this mode of iron immobilization very 

attractive. Besides this theory of adsorptive-catalytic 

oxidation (Rott, 1985; van Beek, 1985), it has also been 

proposed that the injection of O 2 

-rich water onsets the 

exchange of adsorbed Fe 2+ with other cations, such as 

calcium (Appelo et al., 1999; Appelo and Postma, 2005), 

that Fe 2+ is effectively incorporated into the mineral 

structure through interfacial electron transfer (Mettler, 

2002), or that iron oxidizing bacteria (IOB) enhance 

Fe 2+ removal (Hallberg and Martinell, 1976). 

It has been widely reported that a major benefit 

of subsurface iron removal is that the technology 

increases in efficacy with every injection-abstraction 

cycle (Boochs and Barovic, 1981; Braester and Martinell, 

9 

1


1 

Figure 1.3 Light microscope image (top left) of a region 

containing roots with large As/Fe ratio, along with 

the distribution of As, Fe and Mn (μ-XRF). Obtained 

with permission from Frommer et al. (2011) 

1988; Grombach, 1985; Hallberg and Martinell, 1976; 

Mettler, 2002; Rott, 1985; van Beek, 1985), providing 

it with a great advantage over the above-ground 

alternative, which is sensitive to clogging. Clogging of 

the aquifer by iron sludge has not been reported as a 

limitation of the subsurface iron removal technology. 

The deposited iron precipitates have been characterized 

as compact, crystalline mineral structures (Mettler at 

al, 2001). Additionally it has been proposed that iron 

precipitates at various distances from the well, resulting 

in an even spreading of the precipitates around the tube 

well (Boochs and Barovic, 1981; Appelo et al., 1999). 

In the past decades several well designs have 

been implemented, including the single-well pushpull 

design, the dual well design and the Vyredox TM 

method (Hallberg and Martinell, 1976). The Vyredox TM 

method consists of a circle of injection wells around 

one abstraction well, providing a subsurface oxidation 

screen for Fe 2+ adsorption. In the dual well design, one 

well is used for production while the other is being 

used for injection and vice versa. The single-well design 

can be found most attractive for the shallow tube well 

HWTS, as a normal production well can be relatively 

easily transitioned into a subsurface iron removal well. 

Apart from iron removal, the SIR technology has also 

been applied for the removal of manganese and arsenic, 

with varying results (van Beek, 1985; Rott et al., 2002). 

Manganese precipitates are generally found closer to 

the well, as oxidation kinetics are slower than for iron 

(Hallberg and Martinell, 1976; van Beek, 1985), whereas 

arsenic is mostly bound to the iron oxides (Dixit and 

Hering, 2003). The spatial separation of Fe, Mn and As 

deposits is beautifully visualized by a study of the O 2 

and groundwater interaction around a rice plant root 

in a Bangladeshi paddy field (Figure 1.3; Frommer et 

al., 2011). In the Netherlands subsurface iron removal 

is even operated at some locations to enhance the 

nitrification process in subsequent biosand filters (de 

Vet et al., 2009), illustrating the beneficiary side effects 

of SIR. 

Subsurface Arsenic Removal (SAR) 

Although the removal of arsenic during subsurface 

iron removal can be considered a side effect, it may 

10


also be approached as the main objective at sites highly 

contaminated with arsenic. In that case we consider 

iron removal to be the beneficiary side effect and 

Subsurface Arsenic Removal (SAR) the technology. 

Either way, arsenic removal depends greatly on the iron 

removal processes, as precipitated iron oxides form the 

adsorptive surfaces for the arsenic (Dixit and Hering, 

2003): 

Equation 1.3 

Equations 1.4a and 1.4b 

As(III) adsorption onto iron oxides is relatively 

insensitive to pH as found in the aquatic environment, 

though slightly favouring the pH range of 5 to 8 (Dixit 

and Hering, 2003). In natural waters, the adsorption 

(a) injection phase 

S − OH + H3AsO3 →S − H 

2AsO3 + H 

2O 

− 

S − OH + H 

2 

AsO4 

→ S − HAsO4 

+ H 

2O 

2− 

S − OH + HAsO4 → S − AsO4 

+ H 

2O 

− 

2− 

of the negatively charged As(V) is better at higher 

pH, because the point of zero charge of iron oxides is 

reportedly between 7 and 9 (Sposito, 1989; van Geen 

et al., 1994; Peacock and Sherman 2004). The type 

of iron oxide mineral structure also determines the 

arsenic adsorption, mainly due to the large difference 

in specific surface area and accompanying number of 

adsorptive surface sites. Apart from arsenic adsorption, 

there may also be arsenic immobilization through coprecipitation 

with Fe 2+ to ferrihydrite, or, depending on 

pH and Fe:As ratio, precipitation of other complexes 

(such as simplesite; Johnston, 2008). 

In Bangladesh, elevated iron and arsenic 

concentrations have been found to often co-occur in 

groundwater (Nickson et al., 2000). Presence of elevated 

iron levels is considered a key criterion for the technical 

feasibility of SAR. In the small-scale setting, subsurface 

arsenic removal relies on the existing infrastructure 

of a hand-pump/shallow tube-well and retains iron 

(b) abstraction phase 

1 

Ground water level 

containing Fe 2+ and As(III) 



O2 front 

iron oxide 

adsorbed iron oxideFe 2+ 

and As(III) 

with adsorbed 

Fe 2+ and As(III) 

Injected water front 

Figure 1.4 Principle of subsurface iron removal (a) injection and (b) abstraction 

11


1 and arsenic in the subsurface (Figure 1.4). As such, 

it has potential advantages over other household and 

community arsenic removal systems, such as SONO 

and Alcan (Sutherland et al., 2002; IGRAC 2007; 

BETV-SAM, 2011): 

• no costly filter media and maintenance is needed; 

• the tube well is the 1st preferred option for drinking 

water in rural Bangladesh (WSP/Worldbank, 2003); 

and available to a majority of the rural poor in their 

household; 

• (minimal) additional hardware beyond the existing hand 

pump is affordable and locally available/repairable; 

• iron is also removed which improves colour and taste 

of the water; greatly enhancing potential for social 

acceptance; 

• iron could be a visible indicator for arsenic presence (and 

aid in post-deployment monitoring of water quality); 

• groundwater-irrigation leading to arsenic accumulation 

in crops (rice) and harvest reduction may also be 

mitigated. 

Clearly, the SIR/SAR technology scores high for the 

sustainability criterion of accessibility. And the absence 

of a waste stream and electricity/chemicals shows 

the great potential of this technology regarding the 

environmental footprint. The social acceptance and 

proper operation and maintenance (functionality) is 

yet to be determined, but the co-removal of iron may 

well aid in that. The criteria of sufficient water quantity 

and water quality includes, for an arsenic removal 

system, the production of >2L with


to be investigated. In order to do so, knowledge needs 

to be gained regarding the subsurface processes that 

determine the site-specific efficacy of SIR/SAR. 

This thesis 

Arsenic contamination of shallow tube well drinking 

water in Bangladesh is an urgent developmental and 

health problem (British Geological Survey/DPHE, 

2001; World Health Organization, 2001; Smith et al., 

2002), disproportionately affecting the rural poor, 

i.e., those most reliant on this source of drinking 

water. Current arsenic mitigation solutions, including 

(household) arsenic removal options, do not always 

provide a sustainable alternative for safe drinking 

water in rural Bangladesh. A novel HWTS combines 

the removal of iron and arsenic by utilizing the existing 

hand pump infrastructure. This new technology shows 

great potential, but for safe implementation there is the 

need to address urgent research questions regarding 

the production of sufficient water quality and water 

quantity. The problem description of this thesis has 

been formulated as follows: 

Subsurface Iron and Arsenic Removal is a 

promising safe water solution, however, there is the 

need for better understanding of the (subsurface) 

processes determining the sustainable operation 

in diverse geochemical settings. 

This problem has been addressed by identifying research 

questions within three main knowledge gaps: (1) Fe 2+ / 

As(III) immobilization processes, (2) site-specific 

effectiveness, and (3) safe and sustainable application. 

The knowledge gaps include research topics that are 

within and outside the scope of this thesis, as has been 

summarized below. 

Fe 2+ /As(III) immobilization processes 

Although much work has already been done on the 

oxidation and adsorption kinetics of iron and arsenic, 

the rapidly changing redox conditions during injectionabstraction 

cycles make SIR/SAR a unique application. 

A combination of Fe 2+ oxidation, precipitation, 

adsorption and exchange may occur during injectionabstraction 

cycles, resulting in the immobilization 

of iron. Adsorptive-catalytic oxidation and cation 

exchange have been proposed to dominate the 

subsurface iron removal process (van Beek, 1985; Appelo 

et al., 1999), with arsenic removal being a beneficiary 

side effect. However, to target effective arsenic removal 

it is desirable to understand the behaviour of this 

constituent during the adsorptive-catalytic oxidation 

mechanism, including the speciation of arsenic. In this 

thesis the competition of iron oxidizing bacteria for the 

same dissolved oxygen as the chemical Fe 2+ oxidation 

reaction will not be addressed. This does, however, not 

mean that it is assumed that IOB communities play an 

insignificant role during SIR, as the observed presence 

of Gallionella spp. in subsurface treated groundwater 

13 

1


1 could point towards vivid microbial activity (van 

Halem et al., 2008). The focus of this thesis lies on the 

supply of oxygen to adsorbed and exchangeable Fe 2+ 

and to unravel the complex combination of adsorption, 

(heterogeneous) oxidation and cation exchange 

reactions that determine the removal efficacy of iron 

and arsenic. 

Site-specific effectiveness in diverse (geochemical) 

conditions 

The decentralized character of a household water 

treatment and storage solution requires that a smallscale 

SIR/SAR performs in a diversity of geochemical 

conditions. Fe 2+ and As(III) oxidation and adsorption 

reactions are influenced by groundwater quality 

parameters such as pH, temperature, Fe:As ratio, 

competing cations or anions, and more. Apart from 

the water composition, it is also reported in the 

literature that certain mineral structures may catalyze 

the oxidation and/or adsorption process (e.g., Fe 3+ 

oxides, calcite). On the other hand, the presence of 

other mineral structures may seriously threaten safe 

SIR/SAR operation, such as (arseno)pyrite. Both 

groundwater and soil composition are given parameters 

at a certain site, and may vary significantly from well 

to well. It is therefore evident that more knowledge 

needs to be gained on the performance of SIR/SAR in 

diverse geochemical settings. Clearly, local variations 

in socio-economic and geohydrological conditions 

could also greatly influence the effective small-scale 

Fe 2+ /As immoblization 

site-specific efficacy 

II: experience in the Netherlands 

safe and sustainable 

III: small-scale application 

IV: adsorptive-catalytic oxidation 

V: Fe 2+ exchange 

VI: clogging 

by deposits 

VII: catalysis by deposits 

VIII: influence groundwater composition 

Figure 1.5 Synthesis between chapters and the formulated knowledge gaps 

14


implementation of SIR/SAR. Although indisputable 

important, considerations regarding social acceptance, 

seasonal fluctuations and spatial solute transport were 

beyond the scope of this thesis. 

Safe and sustainable application 

Hand pump tube wells are the prime source of drinking 

water in rural Bangladesh and may therefore never be 

threatened in their ability to produce sufficient and safe 

water. The application of SIR/SAR as a HWTS requires a 

thorough investigation of the long-term environmental 

and health effects of this technology. The subsurface 

surroundings of the shallow tube well will be affected 

by the accumulation of Fe/As deposits once SIR/SAR 

is in operation. Knowledge needs to be gained on 

whether these deposits could clog the aquifer and/or 

well and whether these deposits may release arsenic 

during irregular injection-abstraction modes. Clogging 

of the aquifer has not been reported in large-scale SIR 

wells in Europe, but injection volumes at family owned 

hand pumps are one-thousandth of the 1,000-2,000 m 3 

in full scale plants. SIR/SAR at household level requires 

new boundary conditions for sustainable application, 

including operational modes and post-deployment 

monitoring. 

The defined knowledge gaps form the basis for the 

main research questions of the individual chapters 

of this thesis. The research questions overlap one or 

more knowledge gaps, as illustrated by the schematic 

synthesis overview in Figure 1.5. The research questions 

per chapter are: 

I. What lessons can be drawn from the past experiences 

with SIR in the Netherlands (Chapter 2)? 

II. Can SIR/SAR be applied at a hand pump/tube-well level 

in Bangladesh (Chapter 3)? 

III. What is the contribution of the adsorptive-catalytic oxidation 

mechanism on the retention of Fe 2+ and As(III) 

during SIR/SAR (Chapter 4)? 

IV. Does Fe 2+ exchange play a role during an injection-abstraction 

cycle (Chapter 5)? 

V. What kinds of deposits accumulate around a SIR well 

and do they clog the aquifer (Chapter 6)? 

VI. Does the presence of accumulated deposits catalyze the 

subsurface removal of Fe and As (Chapter 7)? 

VII. What is the influence of groundwater composition on 

the efficacy of SIR/SAR (Chapter 8)? 

In Chapter 2 an overview is given of three decades 

of experience with subsurface iron removal in the 

Netherlands. This overview focuses on current/past 

practices, site-specific efficacy and long-term operation 

and is the starting point of the research presented in 

this thesis. 

Chapter 3 focuses primarily on the technical feasibility 

of small-scale application of SIR/SAR at family level in 

Manikganj (Bangladesh), with injection volumes below 

1m 3 . 

Chapter 4 translates the field observations from 

Bangladesh and the Netherlands, to injection- 

15 

1


1 abstraction sand columns in order to investigate the 

adsorptive-catalytic oxidation mechanism during 

successive cycles. 

Additional sand column experiments were conducted 

to investigate the proposed occurrence of cation 

exchange during injection and abstraction, as presented 

in Chapter 5. 

Clogging of the aquifer during long-term operation of 

SIR is a concern that is addressed in Chapter 6, where 

results from characterized deposits near a 12-year-old 

subsurface iron removal well are presented. 

The same sediments were used for column studies to 

investigate the effect of the accumulated deposits on 

adsorptive-catalytic Fe 2+ oxidation, Fe 2+ exchange and 

As(III) removal during SIR/SAR (Chapter 7). 

Chapter 8 elaborates on the influence of groundwater 

composition on the efficacy of SIR/SAR. The limiting 

and/or catalyzing effect on SIR/SAR of commonly cooccurring 

ions, such as phosphate, nitrate and calcium, 

are investigated with sand columns in laboratory and 

natural groundwater. 

The final chapter, Chapter 9, provides the reader with 

the concluding remarks of this thesis. These conclusions 

provide a broader perspective than the individual 

chapters and recommendations are given for future 

research. 

References 

Aiuppa A., W. D’Alessandro, C. Federico, B. Palumbo and 

M. Valenza (2003) The aquatic geochemistry of arsenic in 

volcanic groundwaters from southern Italy, Appl Geochem 

18(9): 1283-1296. 

Appelo C. A. J., B. Drijver, R. Hekkenberg and M. de Jonge 

(1999) Modeling in situ iron removal from ground water, 

Ground Water 37(6): 811-817. 

Appelo C. A. J. and W. W. J. M. de Vet (2003) Modeling in situ 

iron removal from groundwater with trace elements such 

as As. In Arsenic in groundwater. A. H. Welch and K.G. 

Stollenwerk. Kluwer Academic, Boston. 

Appelo C.A.J. and D. Postma (2005) Geochemistry, groundwater 

and pollution. Balkema, Rotterdam, 2nd edition. 

Appleyard S. J., J. Angeloni and R. Watkins (2006) Arsenicrich 

groundwater in an urban area experiencing drought 

and increasing population density, Perth, Australia, Applied 

Geochemistry 21(1): 83-97. 

Armienta M. A. and N. Segovia (2008) Arsenic and fluoride 

in the groundwater of Mexico, Environ Geochem and Hlth 

30(4): 345-353. 

AWWA Research Foundation (2000) Arsenic treatability options 

and evaluation of residuals management issues, American 

Water Works Association, Denver. 

Banerjee K., G.L. Amy, M. Prevost, S. Nourc, M. Jekel, P.M. 

Gallagher and C.D. Blumenscheine (2008) Kinetic and 

thermodynamic aspects of adsorption of arsenic onto 

granular ferric hydroxide (GFH), Water Research 42: 3371 – 

3378. 

16


BETV-SAM: Bangladesh Environmental Technology Verification 

– Support to Arsenic Mitigation, www.verification-unit.org. 

Boochs P. W. and G. Barovic (1981) Numerical-Model Describing 

Groundwater Treatment by Recharge of Oxygenated Water, 

Water Resources Research 17(1): 49-56. 

Borba R. P., B. R. Figueiredo and J. Matschullat (20030 

Geochemical distribution of arsenic in waters, sediments and 

weathered gold mineralized rocks from Iron Quadrangle, 

Brazil, Environ Geol 44(1): 39-52. 

Bose P., A. Sharma (2002) Role of iron in controlling speciation 

and mobilization of arsenic in subsurface environment, 

Water Research 36: 4916-4926. 

Braester C. and R. Martinell (1988) The Vyredox and Nitredox 

methods of in situ treatment of groundwater, Water Science 

and Technology 20(3): 149-163. 

British Geological Survey/DPHE: Arsenic contamination of 

groundwater in Bangladesh (2001) Volume 2: Final report, 

BGS Technical Report WC/00/19. 

Buschmann J., M. Berg, C. Stengel and M.L. Sampson (2007) 

Arsenic and manganese contamination of drinking water 

resources in Cambodia: Coincidence of risk areas with low 

relief topography, Environ Sci and Technol 41(7): 2146-2152. 

Buschmann J., M. Berg, C. Stengel, L. Winkel, M.L. Sampson, 

P.T.K. Trang and P.H. Viet (2008) Contamination of drinking 

water resources in the Mekong delta floodplains: Arsenic and 

other trace metals pose serious health risks to population, 

Environ Int 34(6): 756-764. 

Celik I., L. Gallicchio, K. Boyd, T.K. Lam, G. Matanoski, X. 

Tao, M. Shiels, E. Hammond, L. Chen, K. A. Robinson, L.E. 

Caulfield, J.G. Herman, E. Guallar and A.J. Alberg (2008) 

Arsenic in drinking water and lung cancer: a systematic 

review, Environ Res 108: 48-55. 

Chen C.J., C.W. Chen, M.M. Wu and T.L. Kuo (1992) Cancer 

potential in liver, lunch bladder and kidney due to injected 

inorganic arsenic in drinking water, Br J Cancer 66: 888-892. 

Chen Y. and H. Ahsan (2004) Cancer burden from arsenic in 

drinking water in Bangladesh, Am J Public Health 94(5): 741- 

744. 

Chowdhury M.A.I., M.T. Uddin, M.F. Ahmed, M.A. Ali and 

S.M. Uddin (2006) How does arsenic contamination of 

groundwater cause severity and health hazard in Bangladesh, 

J Appl Sci 6(6): 1275-1286. 

Clasen T. (2008) Scaling Up Household Water Treatment: 

Looking Back, Seeing Forward. Geneva: World Health 

Organization. 

de Vet W.W.J.M., I.J.T. Dinkla, G. Muyzer, L.C. Rietveld and 

M.C.M. van Loosdrecht (2009) Molecular characterization 

of microbial populations in groundwater sources and sand 

filters for drinking water production, Water Research 43(1): 

182-194. 

Dixit S. and J. G. Hering (2003) Comparison of arsenic(V) and 

arsenic(III) sorption onto iron oxide minerals: implications 

for arsenic mobility. Environ. Sci. Technol. 37: 4182-4189. 

Dixit S. and J. G. Hering (2006) Sorption of Fe(II) and As(III) 

on goethite in single- and dual-sorbate systems, Chemical 

Geology 228(1-3): 6-15. 

Environmental Protection Agency: Arsenic Treatment 

Technologies, Environmental Technology Verification 

Program, EPA/600/S-07/007 (2007) www.epa.gov/etv/vtdws.html#traic. 

17 

1


1 

Environmental Protection Agency: Integrated Risk Information 

System, Arsenic, inorganic; CASRN 7440-38-2 (1998) www. 

epa.gov/NCEA/iris/subst/0278.htm. 

Ferguson J.F. and J. Gavis (1972) A review of the arsenic cycle in 

natural waters, Water Research 6: 1259-1274. 

Fewtrell L., R. B. Kaufman, D. Kay, W. Enanoria, L. Haller 

and J. M. Colford Jr (2005) Water, sanitation, and hygiene 

interventions to reduce diarrhoea in less developed countries: 

a systematic review and meta-analysis, Lancet Infect Dis. 5: 

42–52. 

Frommer J., A. Voegelin, J. Dittmar, M.A. Marcus and R. 

Kretzschmar (2011) Biogeochemical processes and arsenic 

enrichment around rice roots in paddy soil: results from 

micro-focused X-ray spectroscopy, European Journal of Soil 

Science 62: 305-317. 

Grombach P. (1985) Groundwater treatment in situ in the 

aquifer, Water Supply 3(1): 13-18. 

Gunduz O., C. Simsek and A. Hasozbek (2009) Arsenic pollution 

in the groundwater of Simav Plain, Turkey: its impact on 

water quality and human health, Water Air Soil Pollut 205 

(1-4): 43-62. 

Hallberg R. O. and R. Martinell (1976) Vyredox - in situ 

purification of groundwater, Ground Water, 14(2): 88-93. 

Hiemstra T. and W.H. van Riemsdijk (2007) Adsorption and 

surface oxidation of Fe(II) on metal (hydr)oxides, Geochim. 

Cosmochim. Acta 71: 5913-5933. 

Hunter P.R. (2009) Household Water Treatment in Developing 

Countries: Comparing Different Intervention Types Using 

Meta-Regression. Environmental Science and Technology 43 

(23): 8991–8997. 

IGRAC (2009) Arsenic in groundwater: Overview and evaluation 

of removal methods. International Groundwater Resources 

Assessment Centre, Report nr. SP 2007-2. 

International Agency for Research on Cancer: Some drinkingwater 

disinfectants and contaminants, including arsenic. 

(2004) IARC Monographs on the evaluation of carcinogenic 

risks to humans, Volume 84. 

Jain C.K. and I. Ali (2000) Arsenic: occurrence, toxicity and 

speciation techniques, Water Research 43(17):4304-4313. 

Johnston R., H. Heijnen and P. Wurzel (2001) Chapter 6. Safe 

Water Technology, in United Nations Synthesis Report on 

Arsenic in Drinking Water, World Health Organisation. 

Johnston R. (2008) Chemical interactions between iron and 

arsenic in water, PhD dissertation, University of North 

Carolina at Chapel Hill. 

Jovanovic, D., B. Jakovljevic, Z. Rasic-Milutinovic, K. Paunovic, 

G.Pekovic and T. Knezevic (2011) Arsenic occurrence in 

drinking water supply systems in ten municipalities in 

Vojvodina Region, Serbia, Environmental Research 111(2): 

315-318. 

Kim M.J. and J. Nriagu (2000) Oxidation of arsenite in 

groundwater using ozone and oxygen, The Science of the 

Total Environment 247: 71-79. 

Liao, C.M., H.H. Shen, C.L. Chen, L.I.Hsu, T.L. Lin, S.C. Chen 

and C.J. Chen (2009) Risk assessment of arsenic-induced 

internal cancer at long-term low dose exposure, Journal of 

Hazardous Materials 165: 652-663. 

Marszalek, H. and M. Wasik (2000) Influence of arsenic-bearing 

gold deposits on water quality in Zloty Stok mining area (SW 

Poland), Environ Geol 39(8): 888-892. 

18


Masscheleyn P.H., R.D. Delaune, W.H. Patrick (1999) Effect of 

redox potential and pH on arsenic speciation and solubility 

in a contaminated soil, Environmental Science & Technology 

25: 1414-1419. 

Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P. 

Weidler and U. von Gunten (2001) Characterization of iron 

and manganese precipitates from an in situ groundwater 

treatment plant, Ground Water 6: 921-930. 

Mettler S. (2002) In-situ removal of iron from groundwater: 

Fe(II) oxygenation, and precipitation products in a 

calcareous aquifer, PhD dissertation, Swiss Federal Institute 

of Technology, Zurich. 

Natural Recourses Defense Council: Arsenic and old laws: A 

scientific and public health analysis of arsenic occurrence 

in drinking water, its health effects, and EPA’s outdated 

arsenic tap water standard, 2000: http://www.nrdc.org/water/ 

drinking/arsenic/aolinx.asp. 

Nickson R., J. McArthur, P. Ravencroft, W.G. Burgess and K.M. 

Ahmed (2000) Mechanism of arsenic release to groundwater, 

Bangladesh and West Bengal, Applied Geochemistry 15(4): 

403-413. 

Petrusevski B., W. van der Meer, J. Baker, F. Kruis, S.K. Sharma 

and J.C. Schippers (2007) Innovative approach for treatment 

of arsenic contaminated groundwater in Central Europe, 

Water Sci Technol: Water Supply 7(3): 131–138. 

Peacock C. L. and D. M. Sherman (2004) Copper(II) sorption 

onto goethite, hematite and lepidocrocite: A surface 

complexation model based on ab initio molecular geometries 

and EXAFS spectroscopy, Geochimica Et Cosmochimica 

Acta 68(12): 2623-2637. 

Rott U. (1985) Physical, chemical and biological aspects of the 

removal of iron and manganese underground. Water Supply 

3(2), 143-150. 

Rott U., C. Meyer and M. Friedle (2002) Residue-free removal 

of arsenic, iron, mangenese and ammonia from groundwater. 

Water Science and Technology: Water Supply 2(1), 17-24. 

Sambu S. and R. Wilson (2008) Arsenic in food and water – a 

brief history, Toxicol Ind Health 24: 217-226. 

SAR Technology (2011) In-situ arsenic treatment. www. 

insituarsenic.org. 

Sarkar A. R. and O.T. Rahman (2001) In-situ removal of arsenic 

- experiences of DPHE-Danida pilot project. In Technologies 

for arsenic removal from drinking water, Bangladesh 

University of Engineering and Technology and The United 

Nations University, Bangladesh. 

Sen Gupta B., S. Chatterjee, U. Rott, H. Kauffman, A. 

Bandopadhyay, W. de Groot, N.K. Nag, A.A. Borbonell- 

Barrachina and S. Mukherjee (2009) A simple chemical free 

arsenic removal method for community, Environmental 

Pollution 157: 3351–3353. 

Smedley P.L and D.G. Kinniburgh (2002) A review of the source, 

behaviour and distribution of arsenic in natural waters, Appl 

Geochem 17: 517–568. 

Smedley P. L., M. Zhan, G. Zhang and Z. Luo (2003) Mobilisation 

of arsenic and other trace elements in fluviolacustrine aquifers 

of the Huhhot Basin, Inner Mongolia, Appl Geochem 18(9): 

1453-1477. 

Smedley P. L., J. Knudsen and D. Maiga (2007) Arsenic in 

groundwater from mineralised Proterozoic basement rocks 

of Burkina Faso, Appl Geochem 22(5): 1074-1092. 

19 

1


1 

Smith A.H., C. Hopenhayn-Rich, M.N. Bates, H.M. Goeden, I. 

Hertz-Picciotto, H.M. Duggan, R. Wood, M.J. Kosnett and 

M.T. Smith (1992) cancer risks from arsenic in drinking 

water, Environmental Health Perspectives 97: 259-267. 

Smith A.H., E.O. Lingas and M. Rahman (2000) Contamination 

of drinking-water by arsenic in Bangladesh: a public health 

emergency, Bulletin of the World Health Organization, 78 

(9): 1093-1103. 

Smith A.H., P.A. Lopipero, M.N. Bates and C.M. Steinmaus 

(2002) Arsenic epidemiology and drinking water standards, 

Science 296: 2145-2146. 

Sobsey M.D. (2002) Managing water in the home: accelerated 

health gains from improved water supply. Geneva: The World 

Health Organization (WHO/SDE/WSH/02.07) 

Sobsey, M.D., C.E. Stauber, L.M. Casanova, J.M. Brwon and 

M.A. Elliott (2008) Point of Use Household Drinking Water 

Filtration: A Practical, Effective Solution for Providing 

Sustained Access to Safe Drinking Water in the Developing 

World, Environ, Sci. Technol. 42: 4261–4267. 

Sommer B., A. Mariño, Y. Solarte, M.L. Salas, C. Dierolf, C. 

Valiente, D. Mora, R. Rechsteiner, P. Setter, W. Wirojanagud, 

H. Ajarmeh, A. Al-Hassan and M. Wegelin (1997) SODIS 

- An emerging water treatment process. Journal of Water 

Supply: Research and Technology - AQUA 46(3): 127-137. 

Sposito G. (1989) The Chemistry of Soils, New York, Oxford 

University Press. 

Stauber C.E., M.A. Elliot, F. Koksal, G.M. Ortiz, F.A. DiGiano 

and M.D. Sobsey (2006) Characterisation of the biosand filter 

for E. coli reductions from household drinking water under 

controlled laboratory and field use conditions, Water Science 

and Technology 54 (3): 1-7 

Stollenwerk K.G. Geochemical processes controlling transport 

of arsenic in groundwater: a review of adsorption (2003) In 

Arsenic in groundwater. A. H. Welch and K.G. Stollenwerk. 

Kluwer Academic, Boston. 

Stumm W. and J.J. Morgan (1996) Aquatic chemistry: chemical 

equilibrium and rates in natural waters. 3rd edition, Wiley 

Interscience, New York. 

Stuyfzand, P.J., P. van Rossem and I. Mendizabal (2006) Does 

arsenic, in groundwaters of the compound Rhine-Meuse- 

Scheldt-Ems delta, menace drinking water supply in the 

Netherlands? Proceedings of IAH meeting, Utrecht. 

Sutherland D., P.M. Swash, A.C. MacQueen, L.E. McWilliam, 

D.J. Ross and S.C. Wood (2002) A field based evaluation of 

household arsenic removal technologies for the treatment 

of drinking water, Environmental Technology 23(12): 1385- 

1404. 

Tamura H., S. Kawamura and M. Hagayama (1980) Acceleration 

of the oxidation of Fe 2+ ions by Fe(III)-oxyhhydroxides, 

Corrosion Science 20, 963-971. 

Tseng W.P., H.M. Chu, S.W. How, J.M. Fong, C.S. Lin and S. 

Yeh (1968) Prevalence of skin cancer in an endemic area of 

chronic arsenicism in Taiwan, J Nat Cancer Inst 40: 453-463. 

UNICEF (2009) Promotion of household water treatment 

and safe storage in UNICEF WASH programmes, Water, 

Environment and Sanitation Section, New York. 

van Beek C.G.E.M. (1985) Experiences with underground water 

treatment in the Netherlands, Water Supply 3: 1-11. 

van Geen A., A. P. Robertson and J.O. Leckie (1994) Complexation 

of Carbonate Species at the Goethite Surface - Implications 

20


for Adsorption of Metal-Ions in Natural-Waters, Geochimica 

Et Cosmochimica Acta 58(9): 2073-2086. 

van Halem, D., S.G.J. Heijman, A.I.A. Soppe, J.C. van Dijk and 

G.L. Amy (2007) Ceramic silver-impregnated pot filters for 

household drinking water treatment in developing countries: 

material characterization and performance study. Water 

Science and Technology, Water Supply 7(5-6), 9-17. 

van Halem, D., W. W. J. M. de Vet, G.L. Amy and J.C. van 

Dijk (2008) Subsurface iron removal for drinking water 

production: understanding the process and exploiting 

beneficial side effects, Water Quality Technology Conference, 

Cincinnati, American Water Works Association. 

van Halem, D., H. van der Laan, S.G.J. Heijman, J.C. van Dijk 

and G.L. Amy (2009) Assessing the sustainability of the 

silver-impregnated ceramic pot filter for low-cost household 

drinking water treatment, Physics and Chemistry of the 

Earth 34: 36-42. 

von Oesten, G. (1900) Verfahren zur Enteisenung von 

Grundwasser im Untergrund selbst, Patentschrift Nr. 114709, 

Kaiserliches Patentamt Berlin. 

Williams P.N., M.R. Islam, E.E. Adomako, A. Raab, S.A. Hossain, 

Y.G. Zhu, J. Feldmann and A.A. Meharg (2006) Increase in 

rice grain arsenic for regions of Bangladesh irrigating paddies 

with elevated arsenic in groundwater, Environ. Sci. Technol., 

40(16): 4903-4908. 

Woo, N. C. and M. C. Choi (2001) Arsenic and metal 

contamination of water resources from mining wastes in 

Korea, Environ Geol 40(3): 305-311. 

World Health Organization (2001) United Nations synthesis 

report on arsenic in drinking water, Geneva. 

World Health Organization (2001b) Arsenic and arsenic 

compounds, Environmental Health Criteria 224, Second 

edition, Geneva. 

World Health Organization (2003) Arsenic in drinking-water, 

Background document for development of WHO guidelines 

for drinking-water quality, Geneva. 

World Health Organization (2006) Guidelines for drinkingwater 

quality, First Addendum to Third Edition, Volume 1 

Recommendations, Geneva. 

World Health Organization (2007) Combating waterborne 

diseases at the household level, Geneva. 

World Health Organization (2011) Household water treatment 

and safe storage, www.who.int/household_water/en/ 

WSP/Worldbank (2003) Fighting Arsenic: Listening to Rural 

Communities, Willingness to Pay for Arsenic-Free Safe 

Drinking Water in Bangladesh. Geneva. 

21 

1

2 

Subsurface iron removal in the 

Netherlands 

This chapter is based on: 

van Halem et al. (2011) Journal of American Water Works Association: to be submitted


2 

An introduction to subsurface 

iron removal 

Iron removal in the Netherlands 

In the Netherlands there has been a long history of 

groundwater treatment, either above-ground or below 

the surface. Since the start of drinking water supply in 

the Netherlands, the preferred source has always been 

microbiologically safe groundwater (in the Netherlands 

this source can be found in confined sandy aquifers 

in most parts of the country; Smeets et al., 2009). 

Conventional above-ground treatment consists in 

general of the combination of aeration, e.g., cascade 

or plate aeration, and subsequent sand filtration. Prechlorination 

of the groundwater is not applied, as 

iron removal can also be achieved with the (slower) 

oxidation kinetics as obtained by aeration. Iron removal 

processes can be approached from a chemical and 

biological perspective, with the latter being considered 

only since several decades (Mouchet, 1992; Czekalla et 

al., 1985). Chemical iron removal can be subdivided 

into oxidation-flocculation and adsorptive-oxidation. 

The formation of flocs results in faster clogging of the 

filter bed, whereas adsorptive iron removal produces a 

neatly ordered coating onto the sand grains (Sharma 

et al., 2001). Biological iron removal relies on the 

presence of iron oxidizing bacteria, such as Gallionella 

spp., to immobilize iron in the filter bed. An additional 

advantage of the biosand filters is that nitrification of 

ammonium-containing anaerobic groundwater can 

be obtained as well. Apart from aeration-filtration, 

also nanofiltration, reverse osmosis, activated carbon 

filtration and pellet softening reactors are sometimes 

practised to treat groundwater in the Netherlands. 

After treatment, the groundwater is usually biologically 

stable and, therefore, post-chlorination of the water to 

prevent bacterial growth is not needed (Smeets et al., 

2009). 

The principle of SIR 

For groundwater treatment a new iron removal 

technology was introduced in the 1970s, based on 

similar oxidation-adsorption principles as sand 

filtration, but utilizing the natural treatment capacity of 

the aquifer. The principle of Subsurface Iron Removal 

(SIR), or in-situ iron removal, is that aerated water is 

A 

B 

1.0 

C 

C 0 

Injection Oxygen front 

C 

C 0 

0.5 

0 

1.0 

0.5 

0 

Injection 

front 

distance from the well 

Fe 2+ front Fe 2+ front 

Abstraction 

t 2 

t 1 


Figure 2.1 Behaviour of injection water, oxygen and iron front 

at a distance from the well (obtained from van Beek, 

1983) 

24

2 Subsurface iron removal in the Netherlands 

periodically injected into an anoxic or anaerobic aquifer 

through a tube well (Figure 2.1A), partially displacing 

the original Fe 2+ -containing groundwater. The O 2 

- 

rich injection water oxidizes Fe 2+ in the subsurface 

environment around the tube well. When the flow is 

reversed, groundwater with low Fe concentrations is 

abstracted (Figure 2.1B). More water with reduced iron 

concentrations can be abstracted (volume V) than was 

injected (volume Vi), i.e., this volumetric ratio (V/Vi) 

determines the efficiency of the system. 

When O 2 

-rich water is injected into the Fe 2+ 

saturated subsurface environment the adsorbed Fe 2+ 

will oxidize: 

Equation 2.1 


+ 0 + 

2 2 2 

for the amorphous ferrihydrite is larger than crystalline 

mineral structures with lower surface areas (goethite, 

lepidocrocite). Besides the adsorptive-catalytic 

oxidation theory (Rott, 1985; van Beek, 1985), it has 

also been proposed that the injection of O 2 

-rich water 

promotes the exchange of adsorbed Fe 2+ with other 

cations, such as calcium (Appelo et al., 1999; Appelo 

and Postma, 2005), that Fe 2+ is effectively incorporated 

into the mineral structure through interfacial electron 

transfer (Mettler, 2002) or that iron oxidizing bacteria 

(IOB) enhance Fe 2+ removal (Hallberg and Martinell, 

1976). 

2 

With S being the solid iron oxide surface. The freshly 

formed Fe 3+ iron hydroxides provide new adsorption 

sites, for soluble Fe 2+ during the abstraction phase. The 

adsorption of Fe 2+ occurs in the absence of oxygen and 

can be schematically formulated as: 

Equation 2.2 

o 2+ + + 

S − OH + Fe →S − OFe( II ) + H 

It should be noted that the use of S-OH 0 in the equation 

is simplified, as in reality adsorbed Fe 2+ may transfer an 

electron to the solid, creating an equivalent to Fe(OH) 2 

(Hiemstra and van Riemsdijk, 2007). The adsorptive 

capacity of the aquifer sediments depends on the type of 

iron oxides present in the soil, as adsorption capacities 

Figure 2.2 Subsurface iron removal patent from 1900 (obtained 

from Olthoff, 1986) 

25


2 

26 

SIR history 

With a patent originating from 1900 (von Oesten; 

Figure 2.2) subsurface iron removal by injection of 

oxygen-rich water into a tube well was first proposed in 

Berlin. In 1902 Piesch had a similar idea and patented 

a subsurface iron filter, where air was allowed to pass 

through the ground just before abstraction (Olthoff, 

1986; Figure 2.3). With only little experimental work 

in the following decades, it was not until 1971 that 

the Finns applied subsurface iron removal (Olthoff, 

1986), with a complete new multiple-well design 

(Vyredox TM ). Soon other European countries followed 

Figure 2.3 Subsurface iron removal patent from 1902 (obtained 

from Olthoff, 1986) 

with subsurface iron removal plants, including Sweden, 

Switzerland, Germany, Denmark, France and the 

Netherlands (van Beek, 1985; Mettler, 2002). 

It has been widely reported that a major 

benefit of subsurface iron removal is that the technology 

increases in efficacy with every injection-abstraction 

cycle (Boochs and Barovic, 1981; Braester and Martinell, 

1988; Grombach, 1985; Hallberg and Martinell, 1976; 

Mettler, 2002; Rott, 1985; van Beek, 1985), providing 

it with a great advantage over the above-ground 

alternative, which is sensitive to clogging. Clogging of 

the aquifer by iron sludge has not been reported as a 

limitation of the subsurface iron removal technology, 

and deposited iron precipitates have been characterized 

as compact, crystalline mineral structures (Mettler at al, 

2001; van Halem et al., 2011). Additionally it has been 

proposed that iron precipitates at various distances 

from the well, resulting in an even spreading of the 

precipitates around the tube well (Boochs and Barovic, 

1981; Appelo et al., 1999). The microbial water quality 

of the abstracted groundwater has not been found to 

be negatively affected by the injection of aerated water 

(van Beek, 1983). 

In the past decades several well designs have 

been implemented, including the single-well pushpull 

design, the dual well design and the Vyredox TM 

method (Hallberg and Martinell, 1976). The Vyredox TM 

method consists of a circle of injection wells around 

one abstraction well (also referred to as “star” or 

“satellite”), providing a subsurface oxidation screen for


Fe 2+ adsorption. In the dual well design, two wells are 

intermittently used for injection or abstraction. The 

single-well design can be found most in Dutch drinking 

water production practices, as a normal production well 

can be relatively easily transitioned into a subsurface 

iron removal well. Apart from iron removal, the SIR 

technology has also been applied for the removal of 

manganese and arsenic, with varying results (van Beek, 

1985; Rott et al., 2002; van Halem et al., 2010). In the 

Netherlands it is even operated at some locations to 

enhance the nitrification process in subsequent biosand 

filters (de Vet et al., 2009), illustrating the beneficial 

side effects of SIR. 

Although subsurface iron removal has been 

applied for several decades in Europe it has not found 

widespread attention elsewhere. The objective of this 

study was to present an overview of the experience at 

two water treatment plants (WTPs) in the Netherlands 

that have operated subsurface iron removal for several 

decades: WTP Corle and WTP Lekkerkerk. The 

overview includes information on the site-specific 

conditions and Fe removal efficacies, resulting in an 

analysis of the current knowledge on the influence of 

groundwater matrix and operational mode on the longterm 

and sustainable application of SIR. 

Subsurface iron removal sites 

Vitens Water Supply Company 

The majority of the subsurface iron removal 

treatment plants are located in the eastern provinces 

of the Netherlands (Figure 2.4). Vitens Water Supply 

Company is currently responsible for the operation of 

10 SIR plants. SIR wells have been operated since 1977 

at WTP Vorden and WTP ‘t Klooster, whereas WTP 

Corle, WTP Eerbeek and WTP de Pol started subsurface 

treatment in 1980. A decade later, WTP Aalten was 

constructed in 1990 and has been applying SIR ever 

since. The number of SIR wells varies per location with 

4 wells at WTP Aalten up to 14 wells at WTP De Pol; in 

total 77 SIR wells are currently in operation by Vitens. 

In general, Vitens operates their SIR wells based on the 

volumetric V/Vi ratio, where iron concentration may 

never exceed 1.8 µmol.L -1 . The V/Vi ratio is determined 

for every well individually and then operated for several 

years before this ratio is retested, unless of course, 

iron concentrations exceed the limit of 1.8 µmol.L -1 . 

At a number of the water treatment plants Vitens has 

started to enrich the injection water with pure oxygen 

to increase the oxygen content from 0.28 mmol.L -1 to 

0.55 mmol.L -1 in order to operate at higher V/Vi ratios. 

Currently, the treatment plants of Vitens operate their 

SIR wells with injection volumes of 2,000 or 3,000 m 3 

and V/Vi ratios varying between 6 and 20. 

2 

27


2 

Oasen Drinking Water Company 

The three locations in the west of the Netherlands, 

WTP Lekkerkerk (since 1998), WTP De Put (1981- 

1988, and 1996-current) and WTP ‘t Kromme Gat 

(1983-2010), are operated by Oasen Drinking Water 

Company. Oasen operates subsurface aeration, a less 

intensive form of subsurface iron removal, since 1980. 

The operation of the injection-abstraction wells are not 

optimized to target iron removal, but the purpose of 

1. Aalten 

2. Corle 

3. De Pol 

4. Dinxperlo 

5. Gorssel 

6. Noordijk 

7. ‘Klooster 

8. Vorden 

9. Zutphen 

10. Eerbeek 

11. Lekkerkerk 

12. ’t Kromme Gat 

13. De Put 

subsurface aeration is the enhancement of nitrification 

in the dry biofilters (de Vet, 2011). After subsurface 

aeration the water is mixed with other sources before 

treatment. The total amount of injection water is only 

1% of the total abstracted raw water. The retention of 

iron, arsenic and phosphate is, in this case, a side effect 

observed during subsurface treatment for enhanced 

nitrification. As iron and other constituents are allowed 

to arrive at elevated concentrations in the well, the 

measurements at these locations provide information 

on the constituent behaviour during an injectionabstraction 

cycle. 

Influence of pH 

The pH of the groundwater influences the efficiency 

of subsurface iron removal both during injection 

(oxidation) and abstraction (adsorption). During 

injection the heterogeneous oxidation reaction is 

catalyzed at higher pH ranges (Tamura et al., 1980; 

Sung and Morgan, 1980): 

d 

[ Fe ] 

dt 

Equation 2.3 

[ S − OH ][ Fe ][ O ] 

[ H ] 

0 2+ 

2+ 

2 

hetero 

= − k 

2 

+ 

Figure 2.4 Geographical locations of subsurface iron removal 

WTPs in the Netherlands (Gorssel has been shut 

down) 

The value of k 2 

depends on the type of mineral 

structure, being lowest for the more crystalline iron 

(oxy)hydroxides (goethite, lepidocrocite). When 

28


pH 

8.0 

7.5 

7.0 

6.5 

6.0 

0 5 10 15 

V/Vi 

WTP Corle 

van Beek, 1983 

Figure 2.5 Relation between pH and V/Vi ratio at the 12 SIR 

wells at WTP Corle and data from van Beek (1983); 

[O 2 

] of injection water was 0.28mM. 

pH 

7.8 

7.6 

7.4 

7.2 

7.0 

6.8 

5 10 15 20 25 

V/Vi 

Figure 2.6 Relation between pH and V/Vi ratio at 12 SIR wells 

at WTP Corle with [O 2 

] of injection water of 0.55mM 

the abstraction phase starts, the Fe 2+ -containing 

groundwater travels passed oxidized Fe 3+ oxide surfaces 

for adsorption. The adsorptive capacity of the soil 

depends on the type of Fe 3+ oxides present in the soil, 

as adsorption capacities for amorphous Fe 3+ oxides are 

larger than more crystalline mineral structures with 

lower surface areas. The adsorption of cations onto a 

surface is a result of two major forces: the chemical 

and the coulombic forces (Dzombak and Morel, 

1990; Stumm and Morgan, 1996). In case of an Fe 3+ 

oxide surface the chemical forces are most dominant, 

resulting in higher adsorption capacities at higher pH 

(Sharma, 2001; Dixit and Hering 2006). 

Figure 2.5 shows the relation between the V/Vi 

ratio and the groundwater pH in 12 subsurface iron 

removal wells at WTP Corle. At the normal oxygen 

concentration in the injection water of 0.28 mmol.L -1 

at WTP Corle, the V/Vi rises from 3 to 9 between pH 

7.0 and 7.4. The field experiments by van Beek (1983) 

show a similar correlation for different subsurface iron 

removal locations, with a V/Vi of 2-3 at pH of 6.3. With 

injecting higher oxygen concentrations, namely 0.55 

mmol.L -1 , the difference between the V/Vi at pH 7.0 

and 7.4 is even more pronounced as the V/Vi rose from 

9 to 24 in this pH range (Figure 2.6). It may therefore be 

concluded that the pH has, as expected, an enormous 

effect on subsurface iron removal. These results 

illustrate that without the proper pH (approximately 

>7.0) the Fe 2+ oxidation and/or adsorption cannot be 

achieved sufficiently during SIR. 

29 

2


2 

Successive cycles 

As reported before (Boochs and Barovic, 1981; Braester 

and Martinell, 1988; Grombach, 1985; Hallberg and 

Martinell, 1976; Mettler, 2002; Rott, 1985; van Beek, 

1985) the iron removal efficacy is known to improve 

with every injection-abstraction cycle. Figure 2.7 

illustrates that such an efficacy increase is also observed 

during the first cycles at WTP Lekkerkerk. Several 

mechanisms have been proposed to be responsible for 

this observation: 

• Incomplete Fe breakthrough enhances removal in future 

cycles (Appelo et al., 1999); 

• Pyrite oxidation during initial cycles; 

• Growth of iron oxidizing bacterial communities in the 

oxidation zone (Hallberg and Martinell, 1976); 

• Increased adsorptive surface area on the soil material 

[Fe] mmol.L -1 

through accumulation of precipitated iron oxides (Rott, 

1985; van Beek, 1985). 

0.12 

0.08 

0.04 

0.00 

0 2 4 6 8 10 12 14 16 

V/Vi 

Cycle 1 

Cycle 7 

Cycle 22 

Figure 2.7 Iron measurements during cycle 1, 7 and 22 at WTP 

Lekkerkerk 

A combination of above mechanisms may contribute 

to the improved efficacy of SIR and it is difficult 

to differentiate between them with field data. 

The first hypothesis, regarding the incomplete Fe 

breakthrough, is based on the assumption that lower 

Fe 2+ concentrations are present around a tube well at 

the end of the abstraction phase. These lowered Fe 2+ 

concentrations are subsequently pushed into the aquifer 

during injection, displacing the original groundwater 

even further. During abstraction these lowered Fe 2+ 

concentrations will load the oxidation zone to a lesser 

extent than the original groundwater, resulting in 

better retardation of Fe. As can be seen Figure 2.7, the 

subsurface iron removal well at WTP Lekkerkerk has 

been operated with complete breakthrough of Fe from 

the start. Nevertheless, the Fe removal efficacy increases 

with every cycle, right from the beginning. 

The second hypothesis may be responsible for 

that. The occurrence of pyrite oxidation (FeS 2 

), can be 

determined based on field data as sulphate is released in 

the reaction (Appelo and Postma, 2005): 

Equation 2.4 

15 7 

2 

FeS 4 

4 2 

− + 

2 

+ O2 

+ H 

2O 

→ Fe( 

OH ) 

3 

+ 2SO4 

+ H 

This equation summarizes the oxidation of both 

disulfide and ferrous iron, and subsequent hydrolysis 

of Fe 3+ to ferrihydrite. The oxidation of pyrite results in 

the mobilization of sulphate and a drop in pH. Sulphate 

concentrations have been monitored during the first 18 

cycles at WTP Lekkerkerk and after SIR was temporary 

30


[SO4] in mmol.L -1 

0.7 

0.6 

0.5 

0.4 

stop injection 

0 20 40 60 80 

abstracted volume (x10,000m 3 ) 

Figure 2.8 Sulphate measurements during and after the 

operation of subsurface iron removal 

stopped (Figure 2.8). The measurements show a slight 

increase in sulphate during SIR, from 0.5 mmol.L -1 

at the end of a cycle to 0.6 mmol.L -1 , indicating the 

occurrence of sulphate releasing processes, such as 

pyrite oxidation. Once SIR had stopped, sulphate 

concentrations slowly reduced to their original value. 

When combining this information with the observed 

mobilization of arsenic, it may be assumed that also 

some pyrite-bound arsenic or arsenopyrite (FeAs S 

) was 

present in the targeted aquifer. It should be noted that 

arsenic concentrations did not exceed 0.2 µmol.L -1 and 

after dry biosand filtration the concentrations were 

always below the provisional guideline of the World 

Health Organization (10 µg.L -1 ; 2006). Although iron 

is released from the pyrite as well, it is immediately 

oxidized and immobilized and will therefore not be 

found in elevated concentrations in the abstracted 

water. The pyrite oxidation process, however, limits the 

iron removal efficacy, as part of the oxygen is consumed 

for oxidation of disulfide (Equation 2.4). 

Co-removal of manganese, 

phosphate and arsenic 

Apart from subsurface iron removal subsurface 

manganese removal has also been operated in the 

Netherlands and other European countries (Braester 

and Martinell, 1988; van Beek 1983) and, more recently, 

in Egypt (Olsthoorn, 2000). Van Beek (1983) reported 

V/Vi efficiencies for manganese removal of 3.5 and 12, 

and indicated that operation for manganese removal 

is more sensitive to operational variations than iron. 

Abiotic Mn 2+ oxidation is a slower reaction than Fe 2+ 

oxidation and will start once Fe 2+ oxidation is more 

[Mn] mmol.L -1 

0.025 

0.020 

0.015 

0.010 

0.005 

0.000 

0 5 10 15 

V/Vi 

Cycle 1 

Cycle 7 

Cycle 22 

Figure 2.9 Manganese measurements during cycle 1, 7 and 22 at 

WTP Lekkerkerk 

31 

2


2 

[PO4 3- ] mmol.L -1 

0.05 

0.04 

0.03 

0.02 

0.01 

0.00 

Cycle 1 

Cycle 7 

Cycle 22 

0 5 10 15 

V/Vi 

Figure 2.10 Phosphate measurements during cycle 1, 7 and 22 at 


[As] µmol.L -1 

0.20 

0.15 

0.10 

0.05 

0.00 

0 5 10 15 

V/Vi 

Cycle 1 

Cycle 7 

Cycle 22 

Figure 2.11 Arsenic measurements during cycle 1, 7 and 22 at 


or less complete. Mn 4+ precipitates are found closer 

to the subsurface treatment well than Fe 3+ precipitates 

(Hallberg and Martinell, 1976; Rott, 1985; van Beek 

1985). The separation between iron and manganese 

deposits has also been observed in the dry filters at 

WTP Lekkerkerk (de Vet, 2011). In a completely 

different setting, namely in oxygen-diffusion around 

rice plants, the same Fe-Mn separation was found 

(Frommer et al., 2011). The manganese measurements 

at WTP Lekkerkerk during cycle 1, 7 and 22 are shown 

in Figure 2.9. Removal efficiencies increased slightly 

after multiple cycles, but not as distinctive as for iron. 

Anion co-removal during SIR has been reported 

in the literature before (Rott et al., 2002, Appelo and 

de Vet, 2003); the measurements for phosphate and 

arsenic are depicted in Figure 2.10 and Figure 2.11, 

respectively. Phosphate behaviour correlates well to 

the removal efficacy of iron, showing significant lower 

concentrations during cycle 22 than cycle 1 and 7. For 

arsenic the breakthrough curves are not as uniform, 

and during the first cycles arsenic concentrations do 

not seem to be very stable. It even seems as if arsenic 

is mobilized during cycle 1 and 7, whereas that process 

does not seem relevant anymore during cycle 22. An 

explanation for the release of arsenic during early cycles 

may be that the oxidized pyrite was arsenic-bearing, 

resulting in elevated arsenic concentrations in solution. 

This has also been observed during an experiment with 

injection of aerated water into a deep tube well at the 

nearby located Langerak (Wallis et al., 2010). 

Contaminant remobilization 

Fe 3+ oxides will accumulate in the aquifer around a 

tube well when subsurface iron removal is operated for 

several years (Mettler et al., 2001; van Halem et al., 2011). 

Although clogging of the aquifer has not been reported, 

32


there is still a concern about what happens to the 

accumulated deposits when operation is stopped. The 

aquifer will return to its anoxic or anaerobic state and 

reducing conditions will prevail. Reductive dissolution 

of the deposits may promote the mobilization of iron, 

manganese or even trace compounds, such as arsenic 

(Appelo and Postma, 2005): 

Equation 2.5 

[Fe], [Mn] in mmol.L -1 

0.1 

0.08 

0.06 

0.04 

0.02 

0 

0 100,000 200,000 300,000 400,000 

2 

CH O 

4 

2 

+ − 

+ H 

2O 

→ CO2 

+ 4H 

+ e 

Equation 2.6 

+ − 2+ 

Fe ( OH ) As + 3H 

+ e → Fe + H O + As 

3 

3 

Although this is a valid concern, especially at sites 

with high organic carbon content in the soil and/or 

groundwater, the mobilization of iron has not been 

observed to exceed its background value after stopping 

with subsurface iron removal (Figure 2.12). At WTP 

Lekkerkerk, the iron and manganese concentrations 

have been measured at the SIR well before and after 

the well was exposed to periodic injection. Dissolved 

organic carbon (DOC) concentrations in the abstracted 

groundwater fluctuated around 3 mg.L -1 (±0.5). This 

particular SIR well had been in operation for 18 

injection-abstraction cycles, and Figure 2.12 illustrates 

the results from the start of cycle 18. The measurements 

show an initial increase in Fe concentration, but after 

140,000 m 3 (approx. 10 months) the concentration 

stabilized at 0.08 mmol.L -1 . This is lower than the iron 

concentrations observed during cycle 1 and 7 (Figure 

2 

Pumped volume (m 3 ) 

Iron 

Manganese 

Figure 2.12 Iron and manganese measurements once subsurface 

iron removal had stopped after 18 injectionabstraction 

cycles (WTP Lekkerkerk) 

2.7), which supports the theory that pyrite oxidation 

may have occurred during the initial cycles. Manganese 

concentrations also increased initially, but stabilized 

at 0.015 mmol.L -1 , which corresponds closely to the 

background concentration in Figure 2.9. Based on 

these findings it can be concluded that accumulated 

iron and manganese oxides during subsurface iron 

removal were not remobilized to result in elevated 

concentrations to the groundwater at WTP Lekkerkerk 

once SIR operation had stopped. 

Iron or manganese remobilization has not been 

observed in the >2 years period that subsurface iron 

removal was stopped at WTP Lekkerkerk. Nevertheless, 

trace (heavy) metals incorporated or adsorbed to 

the Fe/Mn oxide matrix could be released through 

desorption, exchange or re-crystalization (Jessen et al., 

2005). Desorption may occur if groundwater conditions 

33


2 

[PO 4 

-3 ] in mmol.L-1 

0.05 

0.04 

0.03 

0.02 

0.01 

0 

0 

0 100,000 200,000 300,000 400,000 

0.1 

0.08 

0.06 

0.04 

0.02 

[As] in µmol.L -1 

did not exceed 0.01 mmol.L -1 , which is far below the 

concentrations measured during cycle 1 (Figure 2.10). 

Arsenic concentrations remained below 0.1 µmol.L -1 

and were thus also not found to remobilize after SIR 

was stopped at WTP Lekkerkerk. 

Pumped volume (m 3 ) 

Conclusions 

Phosphate Arsenic 

Figure 2.13 Phosphate and arsenic measurements once subsurface 

iron removal had stopped after 18 injectionabstraction 

cycles (WTP Lekkerkerk) 

suddenly change (e.g., pH), which is not likely in a 

normal well field. Exchange of the adsorbed anion with 

other passing anions with higher affinity could promote 

the release of retained arsenic or phosphate from the soil 

grain surface. However, once groundwater conditions 

stabilize around the stopped SIR well no large water 

quality variations are expected. Re-crystallization 

of (amorphous) iron or manganese oxides to oxides 

of higher crystallinity is a slow process, but could 

release small amounts of trace elements. Although 

Figure 2.13 shows some variations in the arsenic and 

phosphate concentrations, it is obvious that high peak 

concentrations are not observed. It may be concluded 

that the retained phosphate was not mobilized to 

elevated solute concentrations once injection was no 

longer applied. Phosphate concentrations generally 

During subsurface iron removal at WTP Lekkerkerk, 

iron, manganese, phosphate and arsenic were measured 

to be retained in the aquifer at varying efficacies. SIR 

technology was found to improve in efficacy with 

every successive cycle, which correlates well with 

findings in other literature. No peak concentrations 

of the studied constituents were observed in the 

abstracted groundwater once SIR was stopped and 

therefore not threatening the sustainability of the 

groundwater production well. pH was shown to be the 

most pronounced water quality parameter determining 

the efficacy of SIR, the operational V/Vi ratio more 

than doubled between pH 7.0 and 7.4 at WTP Corle. 

The experiences in the Netherlands have shown 

that subsurface iron removal is an effective, robust 

and sustainable iron removal technology with great 

potential for worldwide application. 

34


References 



Ground Water 37(6): 811-817. 







Boochs P. W. and G. Barovic (1981) Numerical-Model Describing 

Groundwater Treatment by Recharge of Oxygenated Water, 


Braester C, and R. Martinell (1988) The Vyretox and Nitredox 


and Technology 20: 149-163. 

Czekalla, C., W. Mevius and H. Hanert (1985) Quantative 

removal of iron and manganese by microorganisms in rapid 

sandfilters (in situ investigations). Water Supply 3(1): 111- 

123. 

de Vet W.W.J.M., I.J.T. Dinkla, G. Muyzer, L.C. Rietveld and 

M.C.M. van Loosdrecht (2009) Molecular characterization 

of microbial populations in groundwater sources and sand 

filters for drinking water production, Water Research 43(1): 

182-194. 

de Vet, W.W.J.M. (2011) Biological drinking water treatment 

of anaerobic groundwater trickling filters, PhD dissertation, 

Delft University of Technology. 



Geology 228(1-3): 6-15. 

Dzombak D. A. and F.M.M. Morel (1990) Surface complexation 

modeling: hydrous ferric oxide, Wiley. 

Frommer J., A. Voegelin, J. Dittmar, M.A. Marcus and R. 

Kretzschmar (2011) Biogeochemical processes and arsenic 

enrichment around rice roots in paddy soil: results from 

micro-focused X-ray spectroscopy, European Journal of Soil 

Science 62: 305-317. 

Grombach, P. (1985) Groundwater treatment in situ in the 

aquifer. Water Supply 3(1): 13-18. 


purification of groundwater, Ground Water 14(2): 88-93. 

Hiemstra T. and W.H. van Riemsdijk (2007) Adsorption and 

surface oxidation of Fe(II) on metal (hydr)oxides. Geochim. 

Cosmochim. Acta 71: 5913-5933. 

Jessen S., F. Larsen, C.B. Koch and E. Arvin (2005) Sorption 

and Desorption of Arsenic to Ferrihydrite in a Sand Filter, 

Environ. Sci. Technol 39: 8045-8051. 





Mettler, S. (2002) In-situ removal of iron from groundwater: 




Mouchet P. (1992) From conventional to biological removal of 

iron and manganese in France, Journal AWWA 84(4): 158- 

167. 

35 

2


2 

Olsthoorn T.N. (2000) Background of subsurface iron and 

manganese removal, Amsterdam Water Supply Research and 

Development, Hydrology Department. 

Olthoff R. (1986) Die Enteisenung und Entmanganung von 

Grundwasser im Aquifer, PhD thesis Hannover University. 


removal of iron and manganese underground, Water Supply 

3: 143-150. 


of arsenic, iron, manganese and ammonia from groundwater, 

Water Sci Technol: Water Supply 2(1): 17-24. 

Sharma, S.K. (2001) Adsorptive iron removal from groundwater. 

PhD dissertation, IHE Delft. 

Sharma S.K., J. Kappelhof, M. Groenendijk and J.C. Schippers 

(2001) Comparison of physiochemical iron removal 

mechanisms in filters, J. water Supply Res. Techn. Aqua 50(4), 

187-198. 

Smeets, P.W.M.H., G.J. Medema and J.C. van Dijk (2009) The 

Dutch secret: how to provide safe drinking water without 

chlorine in the Netherlands, Drink. Water Eng. Sci., 2: 1–14. 

Stumm, W. and J.J. Morgan (1996) Aquatic chemistry: chemical 

equilibrium and rates in natural waters. 3rd edition, Wiley 

Interscience, New York. 

Sung W. and J.J. Morgan (1980) Kinetics and products of ferrous 

iron oxygenation in aqueous systems, Environmental Science 

& Technology 14: 561-568. 

Tamura, H., S. Kawamura and M. Hagayama (1980) Acceleration 

of the oxidation of Fe 2+ ions by Fe(III)-oxyhhydroxides, 

Corrosion Science 20: 963-971. 

van Beek C.G.E.M. (1983) Ondergrondse ontijzering, een 

evaluatie van uitgevoerd onderzoek (in Dutch). KIWA 

mededeling 78. 


treatment in the Netherlands. Water Supply 3: 1-11. 

van Halem D., S. Olivero, W.W.J.M de Vet, J.Q.J.C Verbek, G.L. 

Amy and J.C. van Dijk (2010) Subsurface iron and arsenic 

removal for shallow tube well drinking water supply in rural 

Bangladesh. Water Research 44: 5761-5769. 

van Halem, D., W.W.J.M. de Vet, , J.Q.J.C Verbek, G.L. Amy 

and J.C. van Dijk (2011) Characterization of accumulated 

precipitates during subsurface iron removal, Applied 

Geochemistry 26: 116–124. 

von Oesten, G. (1900) Verfahren zur Enteisenung von 

Grundwasser im Untergrund selbst, Patentschrift Nr. 114709, 

Kaiserliches Patentamt Berlin. 

Wallis I., H. Prommer, C.T. Simmons, V. Post, and P.J. Stuyfzand 

(2010) Evaluation of Conceptual and Numerical Models 

for Arsenic Mobilization and Attenuation during Managed 

Aquifer Recharge, Environ. Sci. Technol. 44: 5035–5041. 

36

3 

Small-scale subsurface iron and 

arsenic removal in Bangladesh 


van Halem et al. (2010) Water Science and Technology (62): 2702-2709


3 

Introduction 

The most well-known and severe case of arsenic 

poisoning through drinking water is currently ongoing 

in Bangladesh. Two-thirds of the tube wells installed 

over the last three decades, roughly three million 

in total, have been reported to produce arsenic in 

concentrations above the permissible level of 10 

µg.L -1 set by the World Health Organization (BGS/ 

DPHE, 2001). It is estimated that 35 to 100 million 

people are at risk of drinking arsenic-contaminated 

water in Bangladesh above the WHO guideline (BGS/ 

DPHE, 2001; WHO, 2001; Chowdhury et al., 2006). 

In a bulletin (Smith et al., 2000), the World Health 

Organization reports that long-term exposure to 

arsenic in groundwater, at concentrations over 500 

µg.L -1 , causes death in 1 in 10 adults (including lung, 

bladder and skin cancers). The tube wells were installed 

with the firm conviction that they would contribute to 

a secure and reliable drinking water supply, in order 

to put an end to various contagious diseases caused 

by the use of (microbial unsafe) surface water. It is a 

bitter observation that it is this very approach that 

has led to widespread arsenic poisoning through 

drinking water. The large well-to-well variability in 

arsenic concentrations bears the consequence that in 

the villages all wells need to be tested. Currently an 

estimated 5 million wells have been tested and in 1.4 

million wells the arsenic concentration was measured 

above the national standard of 50 µg.L -1 . In Bangladesh, 

the source of arsenic in groundwater is deposited 

sediments from the Himalayas and strongly reducing 

conditions cause reductive dissolution of the arsenicrich 

iron hydroxides (Smedley and Kinniburgh, 2002). 

In the reducing environment arsenic predominantly 

occurs as arsenite or As(III), when oxidising conditions 

prevail the dominant species is generally arsenate or 

As(V). 

In industrialized countries arsenic contamination 

is also a recognized problem (van Halem et al, 2009). 

The removal of arsenic is achieved through commercial 

adsorption media, e.g. GFH (granular ferric hydroxide), 

or by means of coagulation/filtration, ion exchange or 

membrane filtration. These methods are expensive 

and therefore unavailable to the poor living in rural 

areas. Household water treatment systems with noncommercially 

available adsorption media, such as 

Sono and Alcan, show potential (Sutherland et al., 

2002). However, further research and documentation is 

needed on the long-term efficacy and potential risk of 

microbial contamination of the treated water. All filters 

will at one stage either clog and/or show breakthrough 

of arsenic. Some filters can be regenerated; however, 

many leave an arsenic-rich waste stream. The treatment 

option presented in this study can be operated without a 

waste stream, because the iron and arsenic are retained 

in the subsurface. The adsorption sites for arsenic are 

generated in the aquifer, providing a low-cost and 

robust subsurface filter. 

38

3 Small-scale subsurface iron and arsenic removal in Bangladesh 

The principle of subsurface iron and arsenic removal is 

that aerated water is periodically injected into an anoxic 

aquifer through a tube well (Figure 3.1, left), partially 

displacing the groundwater. The injected water oxidises 

adsorbed Fe 2+ on the soil grains, resulting in adsorptive 

surface area of iron hydroxides. When the flow is 

reversed, soluble Fe 2+ in the abstracted groundwater 

is adsorbed onto the Fe 3+ oxide coated soil grains and 

water with reduced iron concentrations is abstracted 

(Figure 3.1, right). Injection is started again once 

elevated iron levels arrive at the well. More water with 

reduced iron concentrations can be abstracted (volume 

V) than was injected (volume Vi), i.e., this volumetric 


The affected area in the subsurface around the tube well 

is referred to as the oxidation zone. 

Subsurface or in-situ iron removal has been used 

in central Europe for many decades (Hallberg and 

Martinell, 1976; Rott, 1985; van Beek, 1985; Appelo 

et al., 1999; Mettler, 2002), but the application of 

subsurface treatment for the removal of arsenic from 

groundwater is a relatively new approach (Sarkar and 

Rahman, 2001; Rott et al., 2002, van Halem et al., 2008). 

Arsenic is known to adsorb to Fe 3+ oxides (Dzombak 

and Morel, 1990), and since these are formed during 

subsurface treatment, arsenic can be retained in the 

subsurface. This technology has the potential to be a 

cost-effective way to provide safe drinking water in 

rural areas in decentralized applications. Existing 

shallow tube wells can be modified to be operated 

under infiltration and abstraction conditions. Such a 

small-scale set-up has been the focus of an earlier study 

in Bangladesh (Maijdee area), where 0.5m 3 of aerated 

3 

(a) injection phase 

(b) abstraction phase 





O2 front 

iron oxide 

adsorbed iron oxideFe 2+ 

and As(III) 

with adsorbed 

Fe 2+ and As(III) 

Injected water front 

Figure 3.1 Small-scale subsurface iron and arsenic removal in Bangladesh 

39


3 

water was injected into the shallow aquifer (Sarkar 

and Rahman, 2001). This study was executed at three 

different sites, with As concentrations ranging from 1.5 

to 17 μmol.L -1 and molar Fe:As ratios between 1 and 12. 

The design (Figure 3.2) consisted of a regular shallow 

tube well (screen depth 10-12m) and a storage tank with 

a plate aerator (up to 0.16 mmol O 2 

.L -1 ). After injection, 

the tube well was left undisturbed for 12 hours before 

abstraction was started. Though As concentrations were 

significantly lowered, concentrations below the WHO 

guideline were not reported. C/C 0 

=0.5 was reached 

at V/Vi=3 for all studied sites. A similar study was 

executed on the other side of the Indian border with 6 

small-scale SAR plants (Sen Gupta et al., 2009). In this 

study injection volumes were higher at 2m 3 and aeration 

was achieved with sprinklers/shower heads (Figure 

3.3). Arsenic concentrations varied between 1.2 and 

3.8 μmol.L -1 and molar Fe:As ratios range from 12 to 50 

(SAR Technology, 2011). The authors reported that As 

concentrations to below 0.027 μmol.L -1 were achieved 

at all sites, but it should be noted that volumetric ratios 

did not exceed V/Vi=2. Other researchers have studied 

the co-removal of As during SIR in Europe and found 

a reduction of As from < 0.5 μmol.L -1 to below the 

WHO guideline (Rott et al. 2002; Appelo and De Vet, 

2003). Miller (2006) combined the pulsed injection of 

aerated water and ferrous iron to remove As(V).Welch 

et al. (2000) investigated subsurface or in-situ arsenic 

removal with a combination of aerated water and the 

injection of ferric chloride. Preliminary results showed 

reduction of 1.3 μmol.L -1 As(V) to below the WHO 

guideline of 0.13 μmol.L -1 (10 μg.L -1 ; WHO, 2006). 

Previous studies have shown the potential 

of small-scale subsurface iron and arsenic removal, 

from tube well 

Aeration tray 

from tube well 

Shower aeration 

V i = 2m 3 V i = 2m 3 

to tube well 

V i = 0.5m 3 

iron sludge 

to tube well 

Injection tank 

Storage tank 

Figure 3.2 Injection facilities of Sarkar and Rahman (2001) 

Figure 3.3 Injection facilities of Sen Gupta et al.(2009) 

40


but have also indicated that the success of this 

technology may be site-specific. For this technology 

to be considered a safe arsenic mitigation solution it 

is inevitable that more knowledge needs to be gained 

about the operational conditions that control the 

efficacy of SIR/SAR. The objective of this study was to 

investigate the performance of small-scale SIR/SAR in 

rural Bangladesh. 

Materials and methods 

Household shallow tube wells with suction hand pumps 

are widely distributed in Bangladesh and the aim of the 

subsurface iron/arsenic removal technology is to use 

the existing infrastructure optimally. The Manikganj 

district, just west of Dhaka, was selected for this study, 

since this area is known to have high iron, arsenic and 

manganese concentrations in the groundwater. After 

testing numerous wells, two sites (A and B) were selected 

with the same arsenic concentrations (1.94 μg.L -1 ). The 

groundwater pH at the two test locations was 6.85 and 

the groundwater composition is illustrated in Table 

3.1. The initial iron concentrations at location A and 

B was 0.018 mmol.L -1 and 0.27 mmol.L -1 , respectively, 

resulting in molar Fe:As ratios of 9 and 140. Manganese 

concentrations also differed at the two locations, 

namely, (A) 0.045 mmol.L -1 and (B) 0.006 mmol.L -1 . 

The experimental set-up (Figure 3.4) was 

connected to an existing hand pump with tube well in 

the upper aquifer. The 1.5-inch tube-well had a depth 

of 31 meters and a perforated well length of 3 meter. 

As an added precaution, the set-up was placed with 

families who already had arsenic treatment since 2001 

(SIDKO system; BETV-SAM, 2011). For the purpose 

of subsurface treatment, the existing situation was 

modified with a pipe and valve for injection. After 

subsurface treatment, the groundwater was pumped 

(electrical suction pump) into the SIDKO system for 

aeration, sand filtration and Granular Ferric Hydroxide 

filtration (AdsorpAs, Harbauer GmbH). The treated 

water, low in arsenic (


3 

water meter 

pump 

valve 

inline 

monitoring 

sampling 

tap 

iron 

filter 

aeration 

tank 

storage 

tank 

SIDKO 

system 

arsenic 

filter 

oxygen concentrations were reached by aerating the 

water with a combination of a shower head (installed 

in the SIDKO system) and subsequent bubble aeration 

in the storage tank. Analysis of the water samples 

was done with field test kits (Wagtech International: 

Palintest and Arsenator) and confirmed in the 

laboratory (Perkin-Elmer Flame AAS 3110; Perkin- 

Elmer GF-AAS 5100PC). Duplicates or triplicates were 

taken to check the method of sampling and accuracy 

of analysis. Arsenic speciation was done with a field 

method (Clifford et al., 2004) using anion exchange 

resin columns (Amberlite IRA400). Multimeters 

(HACH 340i) were fixed inline to the experimental setup 

to monitor pH (WTW SenTix 41), dissolved oxygen 

(WTW Cellox 325), ORP (WTW SenTix ORP) and 

electric conductivity (TetraCon 325). Measurements 

were registered on a computer with Multilab Pilot 

v5.06 software. The injection and abstraction volumes 

were monitored using water meters. Operation started 

in October 2008, just after the monsoon season, and 

continued until May 2009. At both locations, the 

families shared their arsenic treatment facility with 

the community and the weekly water consumption 

was 5.7-8.7 m 3 and 2.4-2.9 m 3 for location A and B, 

respectively. Operational conditions, such as injection 

frequency and production discharge, varied due to 

irregular operation. Normally the set-ups were used 

for the families’ water production, however, during 

research periods the operation was intensified. In this 

paper, all results are presented using the ratio between 

abstracted volume (V) and injected volume (Vi), V/ 

Vi. This volumetric ratio determines the efficiency of 

the system. Every period of injection-abstraction starts 

at V/Vi=0 and is referred to as a cycle, with the first 

injection-abstraction period being cycle 1. Injection 

was done the night before, so abstraction was started at 

12-16 hours after injection. 

Figure 3.4 Experimental set-up for subsurface iron and arsenic 

removal at existing SIDKO system 

42


Results and discussion 

Typical breakthrough curve 

The graph in Figure 3.5 presents a typical breakthrough 

curve after injection. The electrical conductivity of 

the injection water was slightly lower than of the 

groundwater, and gives a good indication when the 

groundwater water arrives in the well and can thus be 

considered a conservative tracer. This was, as expected, 

at the moment when the volume of produced water was 

equal to the injected volume for this particular cycle, 

i.e., C/C 0 

=0.5 at around V/Vi=1. Although the tracer 

arrived at the well at this point, it is noteworthy that it 

took multiple V/Vi units before complete breakthrough 

C/C0 

1 

0.5 

0 

0 2 4 6 8 

V/Vi 

iron 

tracer 

oxygen 

arsenic 

Figure 3.5 Typical breakthrough curve for a tracer (electrical 

conductivity), dissolved oxygen, iron and arsenic. 

Cycle 6 at location B with an injection volume of Vi = 

744L 

was observed. The tailing of this curve suggests that 

injection water remains in the subsurface after V/Vi=1 

and a mixture of groundwater and injection water is 

withdrawn. It has been suggested before that stagnant 

zones, or “dead-end pores” in the oxidation zone 

influence the subsurface removal of iron (Rott et al., 

2002). Diffusion between these stagnant zones and the 

passing groundwater delays the complete breakthrough 

of a constituent. This would mean that although the 

majority of the injected water volume arrived at the well 

before V/Vi=1.5, there was still a small percentage in 

the stagnant zones at that moment and equilibrium was 

only found after V/Vi=5. The dissolved oxygen curve did 

not show similar tailing, since oxygen concentrations 

reduced immediately after the start of abstraction to


3 

this a mass balance can be calculated and, for this cycle, 

was calaculted to be an approximate removal of 0.66 

mmol and 0.77 µmol for iron and arsenic, respectively. 

One isolated cycle does however not describe the 

potential of subsurface iron and arsenic removal, since 

this technology is known for its increasing efficacy with 

every successive cycle. 

Influence of Fe:As ratio in the groundwater 

Subsurface iron and arsenic removal is applied directly 

at the shallow tube well and therefore depends greatly 

on the occurring groundwater geochemistry. One 

of the parameters that may enhance the effective coremoval 

of arsenic with the iron is the presence of 

significant iron levels in the groundwater. Figures 3.6 

and 3.7 show the breakthrough of arsenic and iron for 

cycle 6 at locations A and B. At location A, the initial 

iron concentration was on average 0.018 mmol.L -1 

(molar Fe:As = 9) and the dominant arsenic species 

in the groundwater was As(III), measured with anion 

exchange resin to be 80-83% of total arsenic. Under 

anoxic conditions, with a pH of 6.8 and temperature 

of 25°C, arsenic predominantly occurs as H 3 

AsO 3 

in 

the aquatic environment (Ferguson and Gavis, 1972). 

The injection water (657L for cycle 6) contained >90% 

arsenic(V) in concentrations below 0.1 μmol.L. -1 in all 

cycles. Subsurface iron removal at location A shows for 

cycle 6 that after V/Vi=6 the measured concentration 

was still 65% lower than the initial iron level. Although 

iron levels are significantly reduced in the subsurface, 

(C/C0)Fe 

1 

0.8 

0.6 

0.4 

0.2 

0 

location A 

location B 

0 1 2 3 4 5 6 7 8 

V/Vi 

Figure 3.6 Measurements of iron for cycle 6 at locations A 

(molar Fe:As = 9, Vi= 657L) and B (molar Fe: As = 

140, Vi=744L) 

(C/C0)As 

1 

0.8 

0.6 

0.4 

0.2 

0 

0 1 2 3 4 5 6 7 8 

V/Vi 

location A 

location B 

Figure 3.7 Measurements of arsenic for cycle 6 at locations A 

(molar Fe:As = 9, Vi= 657L) and B (molar Fe: As = 

140, Vi=744L) 

44


only small amounts of arsenic were removed from the 

groundwater. At V/Vi=2.5, an arsenic concentration 

of C/C 0 

=0.8 had been reached, which exceeds more 

than 10 times the WHO guideline of 10 µg.L -1 . It is 

noteworthy that the original arsenic concentration 

was not reached within the research period of V/Vi=6, 

potentially due to the long lasting effects of diffusion in 

the stagnant zones. 

At location B, the iron concentration was 

significantly higher at 0.27 mmol.L -1 (molar Fe:As = 

140), which could potentially result in better arsenic 

removal. Figure 3.6 shows that iron breakthrough was 

faster at this location, which was expected since the 

same amount of oxygen was injected at both locations. 

The breakthrough of arsenic is comparable with location 

A (Figure 3.7), and although the curve shows some 

tailing, it is clear that the arsenic concentration rose 

more or less simultaneously with the groundwater front 

(C/C0)Mn 

1 

0.8 

0.6 

0.4 

0.2 

0 

0 2 4 6 8 

V/Vi 

Figure 3.8 Manganese measurements during cycle 6 at location 

A (Vi= 657L) 

at V/Vi=1. This strongly indicates that the occurrence 

of high iron concentrations in the groundwater does 

not promote the effective co-removal of arsenic during 

subsurface treatment. The formed adsorptive iron 

hydroxide surface area is apparently either insufficiently 

available for arsenic adsorption, or other constituents, 

such as competing anions, are occupying the available 

adsorption sites. The removal of manganese during 

subsurface treatment was not an objective of this study, 

but during cycle 6 this inorganic contaminant was also 

measured at location A. Figure 3.8 illustrates that Mn 

breakthrough occurs immediately upon abstraction 

and that after V/Vi=6 the background concentration of 

0.046 mmol.L -1 was reached. 

Oxygen level of injection water 

Theoretically, 1 mole of oxygen can oxidise 4 moles 

of Fe 2+ . It might seem logical to assume that the more 

oxygen is injected into the subsurface, the higher 

the removal of iron will be. However, the absence of 

available Fe 2+ or the presence of other oxidizeable 

compounds may interfere with this assumption during 

subsurface treatment. During the injection phase, 

the oxygen front lags behind the injected water front, 

as illustrated in Figure 3.1. This results in a limited 

mixing of iron-rich groundwater and oxygen-rich 

injection water. Instead, the dissolved oxygen reaches 

the adsorbed Fe 2+ on the soil grains. Heterogeneous 

oxidation is extremely fast, so it may be assumed that 

consumption by other reducing compounds will only 

45 

3


3 

start once Fe 2+ oxidation is complete. This would mean 

that if more oxygen is injected into the subsurface than 

adsorbed Fe 2+ is available for oxidation, this oxygen 

will be consumed by other (a)biotic reactions. In other 

words, the injection of high oxygen levels may not only 

contribute to more oxidation of iron and is instead 

potentially consumed in other reactions. For the design 

of a community-scale subsurface treatment plant it 

is useful to know what oxygen concentrations are at 

least required in the injection water. Simple aeration 

tanks, as used in this study, have a limited capacity and 

aeration of anoxic/anaerobic (iron-rich) groundwater 

to saturation levels is difficult to reach. Figure 3.9 

shows the breakthrough curves for iron after injecting 

different oxygen concentrations into the subsurface. 

Increased iron retardation is visible for higher oxygen 

concentrations. The difference between


oxygen concentrations does enhance the subsurface 

removal of iron. At small injection volumes (500 m 3 , it is known that 

iron is retained in the subsurface (van Beek, 1985; 

van Halem, 2008). The objective of this study is, 

however, to apply this technology on a small-scale with 

injection volumes no more than 1 m 3 . As mentioned 

before, subsurface iron removal is known to increase 

in efficiency with every successive cycle. Figure 3.11 

shows the iron breakthrough curves at location B for 

cycle 1, 3, 6, 12, 16, and 20. It clearly shows that iron 

removal improved significantly in this period. During 

cycle 1, C/C 0 

=0.5 for iron was measured just after V/ 

Vi = 2, while during cycle 20 this was at V/Vi >> 5. 

When extrapolating these results linearly, C/C 0 

=0.5 was 


0.3 

0.2 

0.1 

0 

Cycle 1 Cycle 3 



0 1 2 3 4 5 

V/Vi 

Figure 3.11 Iron breakthrough curves during successive cycles 

(Location B). Injection volumes (Vi) were 864L, 980L, 

744L, 840L, 786L, and 827L for cycles 1, 3, 6, 12, 16, 

and 20, respectively 

potentially even as late as V/Vi > 15, but the duration 

of the experiment was not long enough to confirm this 

with measurements. Based on these findings it can 

therefore be concluded that subsurface iron removal 

can be applied at decentralized small-scale treatment 

plants with injection volumes as low as 1 m 3 . 

The breakthrough curves for arsenic were also 

monitored and did not show the same improvement as 

for iron (Figure 3.12). The curves for all cycles show 

more or less the same trend, with C/C 0 

=0.5 around V/ 

Vi = 1.5. The WHO drinking water guideline of 0.13 

μmol.L -1 was already exceeded before V/Vi = 1, which 

is clearly insufficient retardation for this system to be 

called an arsenic removal technique. An improved 

As removal efficiency had been reported in other 

publications (Rott et al., 2002; Appelo and de Vet, 2003), 

but does not seem to account for the sites in this study. 

47 

3


3 

The limited arsenic adsorption to the freshly formed Fe 3+ 

oxide surfaces could be caused by insufficient contact 

time in the subsurface during abstraction to reach 

equilibrium in the oxidation zone. Rough calculations 

show that the oxidation zone, with an injection volume 

of 0.9 m 3 is approximately 0.5 m around the well. At an 

average pumping rate of 1.2 m 3 .h -1 this would result in 

a contact time in the oxidation zone of approximately 

20 minutes. In other words, once the groundwater 

arrives at the oxidation zone, 0.5 m from the well, it 

takes the water a travel time of 20 minutes to reach 

the well. It is noteworthy that in reality the oxidation 

front lags behind the injected water front, so the actual 

oxidised zone may be even smaller. Though arsenic 

adsorption may not be complete within 20 minutes, it 

would theoretically suffice to adsorb part of the arsenic. 

A design consideration to increase the contact time 


2.0 

1.5 

1.0 

0.5 

0.0 




0 1 2 3 4 5 

V/Vi 

Figure 3.12 Arsenic breakthrough curves during successive cycles 

(Location B). Injection volumes (Vi) were 864L, 980L, 

744L, 840L, 786L, and 827L for cycles 1, 3, 6, 12, 16, 

and 20, respectively 

C/C0 

1 

0.5 

0 

0 2 4 6 8 

V/Vi 

Phosphate 

Arsenic 

Iron 

Figure 3.13 Phosphate, arsenic and iron breakthrough during 

cycle 8 at location B (Vi=643L) 

is to reduce the length of the well filter. Adsorption 

of zero-valent As(III) is known to be less effective to 

iron oxides than the negatively charged As(V). Since 

arsenic occurs at these sites predominantly as As(III), 

adsorption may be limited. Another explanation for 

the absence of arsenic adsorption may be sought in 

the competition with other anions (Stachowicz, 2007). 

Especially phosphate is present in concentrations high 

enough to interfere with arsenic adsorption (0.053 

mmol.L -1 ). Phosphate was measured to be delayed 

during abstraction, also when arsenic was already 

breaking through (Figure 3.13). The competitive role of 

phosphate during subsurface arsenic removal has been 

reported before by Appelo and de Vet (2003). 

48


Conclusions 

Subsurface treatment was successfully operated at 

small-scale (injection volumes below 1 m 3 ) for the 

removal of iron. For arsenic, however, the system did 

not prove to be very effective. Arsenic retardation was 

limited and breakthrough of 10 µg.L -1 (WHO guideline) 

was observed before V/Vi=1, which corresponds to the 

moment of groundwater arrival at the well. Iron removal 

was more effective after multiple injection-abstraction 

cycles and after injection of higher dissolved oxygen 

concentrations. Subsurface arsenic removal was also 

slightly more effective at higher oxygen concentrations, 

but did not show the same improvement after 

successive cycles. Two explanations are proposed to be 

responsible for this result, either (i) contact times with 

the freshly formed iron oxides were insufficient, or (ii) 

competing anions, such as phosphate, interfered with 

the adsorption process. 

References 



Ground Water 37(6): 811-817. 





BETV-SAM: Bangladesh Environmental Technology Verification 

– Support to Arsenic Mitigation, www.verification-unit.org. 

British Geological Survey/DPHE (2001). Arsenic contamination 

of groundwater in Bangladesh, Volume 2: Final report, BGS 

Technical Report WC/00/19. 

Chowdhury M.A.I., M.T. Uddin, M.F. Ahmed, M.A. Ali and 

S.M. Uddin (2006) How does arsenic contamination of 

groundwater cause severity and health hazard in Bangladesh, 

J Appl Sci 6(6): 1275-1286. 

Clifford D.A., S. Karori, G. Ghurye and S. Samanta (2004) Field 

speciation method for arsenic inorganic species, AWWA 

Research Foundation, Denver. 

Dzombak D. A. and F. M. M. Morel (1990) Surface complexation 

modeling: hydrous ferric oxide, Wiley. 

Ferguson J.F. and J. Gavis (1972) A review of the arsenic cycle in 

natural waters, Water Research 6: 1259-1274. 







Miller G. P. (2006) Subsurface treatment for arsenic removal. 

Denver, American Water Works Association Research 

Foundation: 59. 


removal of iron and manganese underground, Water Supply 

3: 143-150. 


49 

3


3 

of arsenic, iron, manganese and ammonia from groundwater, 

Water Sci Technol: Water Supply 2(1): 17-24. 






SAR Technology (2011) In-situ arsenic treatment. www. 

insituarsenic.org. 




arsenic removal method for community, Environmental 

Pollution 157: 3351–3353. 



Geochem 17: 517–568. 

Smith A.H., E.O. Lingas and M. Rahman (2000) Contamination 

of drinking-water by arsenic in Bangladesh: a public health 

emergency, Bulletin of the World Health Organization, 78 

(9): 1093-1103. 

Stachowicz M. (2007) Solubility of arsenic in multi-component 

systems: from the microscopic to the macroscopic scale, PhD 

thesis, Wageningen University, the Netherlands. 




of drinking water, Environmental Technology 23(12): 1385- 

1404. 




van Beek C. G. E. M. (1985). Experiences with underground 

water treatment in the Netherlands, Water Supply, 3(2), 1-11. 

van Halem, D., W. W. J. M. de Vet, G.L. Amy and J.C. van 

Dijk (2008) Subsurface iron removal for drinking water 

production: understanding the process and exploiting 

beneficial side effects, Water Quality Technology Conference, 

Cincinnati, American Water Works Association. 

van Halem D., S.A. Bakker, G.L. Amy and J.C. van Dijk (2009) 

Arsenic in drinking water: a worldwide water quality concern 

for water supply companies, Drinking Water Engineering and 

Science 2 (1): 29-34. 

Welch A.H., K.G. Stollenwerk, L. Feinson and D.K. Maurer 

(2000) Preliminary evaluation of the potential for insitu 

arsenic removal from ground water. In: Arsenic in 

Groundwater of Sedimentary Aquifers, 31st International 

Geological Congress, Rio de Janeiro, Brazil. 


report on arsenic in drinking water, Geneva 

World Health Organization (2006) Guidelines for Drinkingwater 

Quality, First addendum to third edition, Volume1, 

Recommendations, Geneva 

50

Simulation of adsorptive-catalytic 

oxidation in sand columns 

4 


van Halem et al. (2010) Water Research 44: 5761-5769


4 

Introduction 



health problem (British Geological Survey/DPHE, 

2001; World Health Organization, 2001; Smith et al., 

2002), disproportionately affecting the rural poor, i.e., 

those most reliant on this source of drinking water. 

Subsurface iron and arsenic removal, relies on the 

existing infrastructure of a hand-pump/shallow tubewell 

and retains iron and arsenic in the subsurface. As 

such, it has potential advantages over other household 

and community arsenic removal systems, such as 

SONO and Alcan (Sutherland et al., 2002): 

• no costly filter media and maintenance is needed; 

• the tube well is the 1st preferred option for drinking 

water in rural Bangladesh (WSP/Worldbank, 2003); 

and available to a majority of the rural poor in their 

household; 

• (minimal) additional hardware beyond the existing hand 

pump is affordable and locally available/repairable; 

• iron is also removed which improves colour and taste 

of the water; greatly enhancing potential for social 

acceptance; 

• iron could be a visible indicator for arsenic presence (and 

aid in post-deployment monitoring of water quality); 

• groundwater-irrigation leading to arsenic accumulation 

in crops (rice) may also be mitigated. 

By injecting oxygen-rich water into an anoxic aquifer, 

both homogenous and heterogeneous oxidation of 

ferrous iron will occur in the aquifer. Homogeneous 

oxidation of ferrous iron takes place in solution, and 

predominantly occurs at the interface of injected water 

and original, anoxic groundwater. Based on the large 

surface area of iron hydroxides on the soil grains in 

the subsurface, it is thought that the heterogeneous 

reaction of ferrous iron oxidation on the surface of 

ferric iron oxides is dominant during subsurface iron 

removal. In literature, the system’s efficacy is explained 

by adsorptive-catalytic oxidation (van Beek 1985; Rott, 

1985), where adsorbed ferrous iron is oxidized to form 

new adsorption sites. On its way into the aquifer, the 

injected water oxidizes adsorbed ferrous iron and thus 

“regenerates” the subsurface for adsorption during 

abstraction: 

Equation 4.1 


+ 0 + 

2 2 2 

Due to the rapid consumption of oxygen during 

injection of aerated water, the oxygen front will lag 

behind the injected water front. When heterogeneous 

ferrous iron oxidation is complete, the iron hydroxide 

surface is available for adsorption of ferrous iron and 

oxyanions, such as arsenic(III), during groundwater 

abstraction: 

Equation 4.2 and 4.3 

o 2+ + + 

S − OH + Fe →S − OFe( II ) + H 

S − OH + H3AsO3 →S − H 

2AsO3 + H 

2O 

52

4 Simulation of adsorptive-catalytic oxidation in sand columns 

Once the Fe 3+ oxide surface is exhausted, no more Fe 2+ 

or arsenic is adsorbed and iron/arsenic breakthrough 

is observed in the produced water. Hence, during 

abstraction the iron/arsenic front is delayed and more 

iron-free water can be produced than was injected. 

Every period of injection-abstraction is referred to 

as a cycle, with the first injection-abstraction period 

being cycle 1. More water with reduced iron/arsenic 

concentrations can be abstracted (volume V) than was 

injected (volume Vi), i.e., this volumetric ratio (V/Vi) 

determines the efficiency of the system. 

Subsurface or in-situ iron removal has been 

used in central Europe for many decades (Hallberg and 

Martinell, 1976; Boochs and Barovic, 1981; Jechlinger 

et al., 1985; Rott, 1985; van Beek, 1985; Appelo et al., 

1999; Mettler, 2002), but the application of subsurface 

treatment for the removal of arsenic from groundwater 

is a relatively new approach (Rott et al. 2002, van 

Halem et al. 2010). This technology has the potential 

to be a cost-effective way to provide safe drinking 

water in rural areas in decentralized applications. 

With minimal investments in additional equipment, 

the existing infrastructure (hand pumps/shallow tube 

wells) can be modified to be operated under injection 

and abstraction conditions. In literature, a reduction 

of arsenic concentrations from maximum 40 μg.L -1 to 

below the WHO guideline (10 μg.L -1 , WHO, 2006) has 

been reported with the injection of aerated water into 

the aquifer (Rott et al., 2002; Appelo and De Vet, 2003). 

In Bangladesh the subsurface removal of higher arsenic 

levels was investigated by Sarkar and Rahman (2001), 

namely, 500 - 1300 μg.L -1 . In that study concentrations 

as low as 10 μg.L -1 were not reached, nevertheless, 

more than 50% removal was observed. In the absence 

of naturally occurring soluble ferrous iron, other 

researchers have studied the simultaneous injection 

of aerated water with ferric or ferrous iron (Welch et 

al., 2000; Miller, 2006). Preliminary results showed 

reduction of 100 μg.L -1 arsenic(V) to below the WHO 

guideline. Although these results are promising, only 

little is known about the limitations of this technology 

in the diverse geochemical settings of Bangladesh. 

The focus of this study was to identify the dominant 

processes in subsurface iron and arsenic removal in 

order to assess the applicability for rural Bangladesh. 

The methodology included (1) a field study with a 

community-scale facility in Manikganj, Bangladesh to 

assess the potential of decentralized subsurface iron 

and arsenic removal, and (2) column experiments with 

natural groundwater to simulate the shifting redox 

conditions in the oxidation zone during subsurface 

iron and arsenic removal. The column experiments 

provided controlled conditions for the investigation of 

the adsorptive-catalytic oxidation mechanism, whereas 

the test facility enabled to study subsurface treatment 

in the complex subterranean environment. 

53 

4


4 


Test facility, Bangladesh 

Household shallow tube wells with suction hand pumps 

are widely distributed in Bangladesh and the objective 

of subsurface iron and arsenic removal is to use this 

existing infrastructure. The Manikganj district, 40 km 

west of Dhaka, was selected for this study, since the area 

is known to have high iron and arsenic concentrations 

in the groundwater. A site was selected with elevated 

iron concentrations (0.27 mmol.L -1 ) and arsenic 

concentrations (1.94 μmol.L -1 ). Unlike other parts of 

the Manikganj district, manganese concentrations were 

not found to be high (5.46 μmol.L -1 ) at this particular 

location. For phosphate, however, the groundwater did 

show elevated levels (52.6 μmol.L -1 ). The OR potential 

of the groundwater was measured to be on average 

-170mV and the pH of the groundwater was 6.85. 

The experimental set-up (Figure 4.1) was 

connected to an existing hand pump with tube well in 

the upper aquifer. The 1.5-inch tube-well had a depth of 

31 meters and a perforated well length of 3 meter. As an 

added precaution, the set-up was placed with a family 

who already had arsenic treatment since 2001 (SIDKO 

system, BCSIR 2003). For the purpose of subsurface 

treatment, the existing situation was modified with a 

pipe and valve for injection. After subsurface treatment, 

the groundwater was pumped (electrical suction pump) 

into the SIDKO system for aeration, sand filtration 

and Granular Ferric Hydroxide filtration (AdsorpAs, 

Harbauer GmbH). The treated water, low in arsenic 

and iron, was collected in a 1 m 3 storage tank and used 

for injection into the aquifer. The maximum injection 

volume was therefore limited to 1 m 3 . Analysis of the 

water samples was done with field test kits (Wagtech 

International: Palintest and Arsenator) and confirmed 

in the laboratory (Perkin-Elmer Flame AAS 3110; 

Perkin-Elmer GF-AAS 5100PC). Duplicates or 

triplicates were taken to check the method of sampling 

and accuracy of analysis. Arsenic speciation was 

done with a field method (Clifford et al., 2004) using 

anion exchange resin columns (Amberlite IRA400). 

Multimeters (HACH 340i) were fixed inline to the 

water meter 

pump 

valve 

inline 

monitoring 

sampling 

tap 

iron 

filter 

aeration 

tank 

storage 

tank 

SIDKO 

system 

arsenic 

filter 

Figure 4.1 Experimental set-up for subsurface iron and arsenic 

removal at existing SIDKO system 

54


experimental set-up to monitor pH (WTW SenTix 41), and phosphate concentration of 33 μmol.L -1 . The 

dissolved oxygen (WTW Cellox 325), OR potential groundwater was pumped onto the columns during 

(WTW SenTix ORP) and electric conductivity the abstraction phase of a cycle, and the injection phase 

(TetraCon 325). Measurements were registered on consisted of drinking water. The drinking water had 

a computer with Multilab Pilot v5.06 software. The an oxygen concentration of 0.28 mmol.L -1 , a slightly 

injection and abstraction volumes were monitored higher pH of 7.4, and iron, manganese and arsenic were 

using water meters. Operation started in October 2008, below detection limits. 

just after the monsoon season, and continued until 

The experimental set-up (Figure 4.2) consisted 

May 2009. The family shared their arsenic treatment of duplicate transparent PVC columns with a length of 

facility with their community and the weekly water 80cm and an inner diameter of 36mm (wall thickness 

consumption was 2.4-2.9 m 3 . Operational conditions, 2mm). The columns were filled with washed (24h 

such as injection frequency and production discharge, with 3% HCl) filter sand (grain size = 0.6-1.2mm, 

varied due to irregular operation. Normally the set-ups D 10 

=0.75mm) that contained 48.4 μmol Fe.g -1 .ds 

were used for the families’ water production, however, after total iron extraction with 5M HCl. The pushpull 

operational mode of injection-abstraction at 

during research periods the operation was intensified. 

Injection was done the night before, and abstraction the test facility was simulated in the 1D plug-flow 

was started at least 12 hours after injection. 

environment of the columns with down flow (1.0 L.h -1 

±0.05) for both injection and abstraction. At the start 

Anoxic column experiments 

of the experiments the columns were conditioned 

The raw groundwater of Oasen Drinking Water with groundwater, until complete breakthrough of 

Company drinking water treatment plant Lekkerkerk iron occurred. Anoxic conditions were maintained 

in the Netherlands was used as influent for the column in the columns by using an airtight FESTO system (6 

experiments. In addition, arsenic(III) (NaAsO 2 

, Fisher) x 1 PUN, I.D. 4mm) with matching connectors and 

was added to simulate high arsenic conditions as found valves. The flow rate in the columns (2.16 m.h -1 ±0.11) 

in Bangladesh. In order to evaluate the effect of different was controlled with a multi-channel pump and PVC 

Fe:As ratios in the groundwater, several lower arsenic tubing with low gas permeability. The set-up remained 

concentrations were dosed. During the research period under constant positive hydrostatic pressure to prevent 

the groundwater had an average pH of 7.1, a nearly oxygen. An injection-abstraction cycle started with 1.5 

constant temperature of 12 °C, iron concentration of pore volume of (oxic) injection water and subsequently 

95 μmol.L -1 , manganese concentration of 11 μmol.L -1 the influent was switched to (anoxic) groundwater for 

4 

55


4 

multiple pore volumes. Electrical conductivity was used 

as a conservative tracer from which the pore volume 

could be calculated to be on average 0.37L (±0.005). 

For the columns, the V/Vi was calculated by dividing 

the produced water (V) by one pore volume (Vi), since 

the latter corresponds to the actual oxidized volume of 

sand in the column. The water quality parameters were 

monitored until at least C/C 0 

= 0.8 was reached for 

iron and arsenic (C= measured concentration, and C 0 

= 

original concentration), and runtimes of the columns 

per cycle varied between 9.2 and 16.1 pore volumes. 

During the experiments samples were taken for iron 

analysis (Perkin-Elmer Flame AAS 3110) and arsenic 

analysis (GF-AAS; Perkin-Elmer 5100PC). Arsenic 

speciation was done with a field method (Clifford 

et al., 2004) using anion exchange resin columns 

(Amberlite IRA400). On-line measurements were 

done for dissolved oxygen (WTW Cellox 325), E h 

potential (WTW SenTix ORP), pH (WTW SenTix 

41), and electrical conductivity WTW (TetraCon 325). 

Measurements were registered on a computer with 

Multilab Pilot v5.06 software. 

pH ORP EC DO 

Waste/Sample 

Point 

H 2 

O 

+ 

O 2 

Discharge Pump 

Groundwater 

Dosing Pump 

N 2 

Column 1 

Column 2 

NaAsO 2 

Figure 4.2 Experimental column set-up at WTP Lekkerkerk, the Netherlands 

56


Results and discussion 

2 

1.6 

0.2 

0.16 

Breakthrough curves at test facility 

A dimensionless retardation factor (R) has been defined 

to represent the delayed arrival of iron or arsenic in 

the well compared to the original groundwater. R+1 

is equal to the V/Vi when the C/C 0 

(C= measured 

concentration, and C 0 

= original concentration) for 

iron or arsenic equals 0.5 divided by the V/Vi for a 

conservative tracer, e.g., electrical conductivity, at C/C 0 

= 0.5: 

C/C0 

R 

Well 

1 

0.5 

0 

TracerC/ C0 

0 1 2 3 4 5 6 7 8 

V/Vi 

⎛ V ⎞ 

⎜ ⎟ 

⎝Vi 

⎠Fe 

C / C0= 

0.5 4.5 

RFe 

= − 1 = = 3.5 

⎛ V ⎞ 

1 

⎜ ⎟ 

⎝Vi 

⎠ 

TracerC / C0= 

0.5 

Equation 4.4 

⎛V 

⎞ 

⎜ ⎟ 

⎝Vi ⎠ 

( PV ) 

FeC/ C 

Fe 

0 C/ 

C0 

= −1⇒ RColumn 

= −1 

⎛V 

⎞ 

( PV ) 

TracerC/ C0 

⎜ ⎟ 

⎝Vi 

⎠ 

iron 

tracer 

oxygen 

Figure 4.3 Typical breakthrough curve for electrical conductivity, 

dissolved oxygen, and iron, including determination 

of the retardation factor 


1.2 

0.8 

0.4 

0 

0 1 2 3 4 5 6 7 8 

V/Vi 

total As As(III) total Fe 

Figure 4.4 Development of arsenic and iron at the test facility in 

Manikganj, Bangladesh (cycle 20) 

The determination of the retardation factor is 

illustrated in Figure 4.3 for cycle 6 at the communityscale 

test facility in Manikganj, in this case the R Fe 

for iron is 4.5. Figure 4.4 depicts the breakthrough 

of total arsenic, arsenic(III) and iron during cycle 20 

at the test facility in Bangladesh. The graph clearly 

shows that iron breakthrough was delayed significantly, 

since the background concentration of 0.27 mmol.L -1 

was not reached at V/Vi = 7.5. The retardation factor 

(R Fe 

) for iron has an estimated value of 12. It can be 

calculated that the total amount of removed iron would 

in that case be approximately 2.6 moles. The volume 

of injected water for this particular cycle was 827L 

and had an oxygen concentration of 0.17 mmol.L -1 , 

which adds up to a total amount of injected oxygen of 

±0.14 moles. In the case that all injected oxygen was 

consumed by subterranean adsorbed ferrous iron, and 

0.12 

0.08 

0.04 

0 


4 

57


4 

thus used for the formation of new iron hydroxide 

surfaces, the measured iron removal does not even 

closely correspond to the equation that 1 mol of oxygen 

can oxidize 4 moles of ferrous iron (Equations 4.1 and 

4.2). In other words, iron removal at this particular site 

was much more effective than can be explained by the 

theory of adsorptive-catalytic oxidation. 

Arsenic breakthrough started immediately at 

V/Vi=0 and reached complete breakthrough before 

V/Vi=5. During this cycle, the retardation factor for 

arsenic did not even reach 1. In the initial stage of the 

cycle all arsenic that breaks through is arsenic(III), 

but after V/Vi=4 arsenic(V) also arrived at the well. In 

total, 2.6 mmol of arsenic is removed during this cycle, 

of which 1.1 mmol is arsenic(V). This gave an arsenic 

adsorption ratio of 1.0 mmol As/mol of removed iron. 

It may be concluded that the efficient iron removal does 

not promote the equivalent co-removal of arsenic. Also, 

iron does not provide a visible indicator for arsenic 

presence at this site – which could have been an aid in 

post-deployment monitoring of the water quality. 

Breakthrough curves for anoxic column 

experiments 

In the columns, oxygen-rich drinking water was dosed 

to the columns for 1.5 pore volumes and remained in 

the columns overnight (16 hours). In the morning, 

columns were re-started with natural groundwater 

and monitored for the retardation of arsenic and iron. 

Since the columns were operated under strict plug 

EC (C/C0) 

pH 

O2 (mmol.L -1 ) 

Eh (mV) 

1 

0.8 

0.6 

0.4 

0.2 

0 

7.2 

7.1 

7 

0.25 

0.2 

0.15 

0.1 

0.05 

0 

0 

-50 

-100 

-150 

0.0 0.5 1.0 1.5 

Pore volume 

0.0 0.5 1.0 1.5 

Pore volume 

0.0 0.5 1.0 1.5 

Pore volume 

0.0 0.5 1.0 1.5 

Pore volume 

Figure 4.5 Development of the tracer (electrical conductivity), 

pH, dissolved oxygen and Eh potential at the start of 

an injection cycle in the columns 

58


flow conditions, it can be assumed that homogeneous 

oxidation and precipitation were very limited and that 

heterogeneous oxidation and adsorption processes, 

and thus adsorptive-catalytic oxidation, dominated. 

The injection of aerated water into the Fe 2+ saturated 

columns onsets the oxidation of adsorbed Fe 2+ , 

resulting in a pH drop. Figure 4.5 illustrates the 

development of the tracer (electrical conductivity), 

pH, oxygen concentration and E h 

potential during the 



4 

3 

2 

1 

0 

0.1 

0.08 

0.06 

0.04 

0.02 

0 

total As 

As(III) 

Initial - total As 

Initial - As(III) 

0 2 4 6 8 10 12 14 

0 2 4 6 8 10 12 14 

V/Vi 

total Fe 

Initial Fe 

Figure 4.6 Breakthrough of arsenic and iron in the columns 

(cycle 14) 

start of an injection cycle. The amount of consumed 

oxygen can be registered by the retardation of the 

oxygen curve compared to the conservative tracer, 

electrical conductivity. The total oxygen consumption 

during injection was 0.05 mmol (results not shown), 

which corresponds to approximately ¼ of the amount 

of removed iron (Equation 4.1). Based on this mass 

balance, it can be concluded that the oxygen retardation 

was indeed caused by heterogeneous oxidation of 

ferrous iron in the column. 

The typical breakthrough curves of arsenic and 

iron are depicted in Figure 4.6 for one of the columns 

(cycle 14). The arsenic concentration spiked to the 

influent consisted of 3.7 μmol.L -1 , of which 2.8 μmol.L -1 

was arsenic(III). The graph shows that the original 

arsenic concentration was reached just before V/Vi = 

7, with a retardation factor (R As 

) of 1. Like in the test 

facility, arsenic(V) was initially completely removed, 

but passed the columns around V/Vi = 4. The iron 

content of the natural groundwater was 95 μmol.L -1 and 

in the columns this concentration was reduced with a 

retardation factor of 6. The total amount of removed 

iron was 0.21 mmol, which yields an arsenic adsorption 

ratio of 24.8 mmol As per mol of removed iron. 

Retardation factors at community-scale test 

facility 

Subsurface iron removal has been frequently reported 

to increase in efficacy with every successive cycle 

(Hallberg and Martinell, 1976; Rott, 1985; van Beek, 

59 

4


4 

1985; Jechlinger et al., 1985; Boochs and Barovic, 1981; 

Mettler, 2002; Braester and Martinell, 1988). Figure 

4.7 shows that this was also the case for the test facility 

in Bangladesh, hence R Fe 

increased from 1 to 14 after 

multiple cycles. It is noteworthy that the injection 

volume at this facility was limited to only 1 m 3 , which 

can be considered very small-scale compared to 

existing treatment plants in Europe where injection 

volumes typically vary between 500 and 1,000 m 3 (van 

Beek, 1985; Appelo and de Vet, 2003). It is therefore an 

important finding that, even at small-scale, subsurface 

iron removal is effective and could provide iron-free 

water in decentralized facilities in rural areas. R As 

did 

not increase with every successive cycles, it remained 

stable at around 1, illustrating that the process which is 

responsible for the increasingly effective iron removal 

Retardation factor 

20 

15 

10 

5 

0 

arsenic 

iron 

0 2 4 6 8 10 12 14 16 18 20 

Cycle # 

Figure 4.7 Retardation factors for iron and arsenic during 

successive cycles at the test facility in Manikganj, 

Bangladesh 

during subsurface treatment did not promote an 

equally effective co-removal of arsenic. 

There is a general consensus in the literature 

that adsorptive-catalytic oxidation is the dominant 

mechanism in subsurface iron removal, but the 

increasing efficacy of subsurface iron removal has 

also been attributed to bacterial activity (Hallberg 

and Martinell, 1976; Jechlinger et al., 1985; Rott, 1985; 

Grombach, 1985), occurrence of dead-end pores or 

stagnant zones (Boochs and Barovic, 1981), growth of 

the oxidation zone with every cycle (Appelo et al., 1999), 

and oxidation of reductants other than Fe 2+ during 

initial cycles (van Beek, 1985). Heterogeneous 

oxidation of Fe 2+ is extremely fast, so oxidation of 

other reductants is unlikely to be favored. Nevertheless, 

oxidation of, e.g., pyrite could be a secondary reaction, 

resulting in elevated iron (and sulphate) concentrations 

in the produced water. Such mobilization of iron 

could underestimate the actual iron removal efficiency 

through adsorptive-catalytic oxidation during initial 

cycles. In the columns, only the adsorptive-catalytic 

oxidation mechanism within the oxidation zone was 

simulated. The clean filter sand did not contain other 

reductants that consume oxygen during initial cycles 

and changing transport mechanisms are unlikely to be 

relevant in the columns, since the conservative tracer 

tests confirmed stable plug conditions after multiple 

cycles. Bacterial activity was checked by Phase Contrast 

Microscopy of the produced water during cycle 21 and 

no iron oxidizers like Gallionella spp. were found. In 

60


other words, the adsorptive-catalytic oxidation process 

during subsurface iron and arsenic removal is expected 

to be the dominant process in the columns. 

Retardation factors for anoxic column 

experiments 

The retardation factors for iron and arsenic for the 

different injection-abstraction cycles in the columns are 

depicted in Figure 4.8. It should be noted that dosing of 

arsenic to the natural groundwater was started after 7 

cycles; therefore the arsenic retardation factors for the 

initial cycles are not included in the graph. Successive 

cycles in the columns show similar retardation factors 

for arsenic as the test facility. R As 

remained around 

Retardation factor 

10 

8 

6 

4 

2 

0 

0 2 4 6 8 10 12 14 16 18 20 22 

Cycle # 

arsenic iron arsenic-C2 iron-C2 

Figure 4.8 Retardation factors for iron and arsenic during 

successive cycles in the duplicate columns (C1 and 

C2) 

1 and did not increase after multiple cycles. In the 

columns, the removal of arsenic is expected to be 

purely adsorptive, and the strong correlation with the 

results from the test facility suggests that those results 

were also achieved with adsorptive arsenic removal. 

Iron retardation in the columns did increase 

initially, but remained more or less stable at R Fe 

=7 

after the first 6 cycles. The efficacy did not improve 

significantly with every cycle as was found in the test 

facility, and since the dominant mechanism in the 

columns was adsorptive-catalytic oxidation it can be 

concluded that this mechanism was not responsible for 

increasing efficacies at the test facility. Apparently this 

mechanism does not provide sufficient new adsorptive 

surface area through freshly formed iron hydroxides to 

improve the system’s efficacy with every cycle. Hence, 

the mechanism which was responsible for improved 

iron removal with every successive cycle in the field 

situation does not prevail in the columns. Unlike in 

the columns, at the test facility bacterial activity and/or 

occurrence of stagnant zones may control the improved 

iron removal efficacy with every cycle. 

Fe:As ratio of the groundwater 

The total amount of arsenic removed per cycle in the 

duplicate columns varied between 1.6 and 3.6 μmol. 

cycle -1 and did not increase with every successive cycle. 

It appears that the sites available for arsenic adsorption 

are regenerated during every injection phase, but the 

number of sorption sites does not seem to increase 

61 

4


4 

due to the freshly retained iron in the columns. On 

average, the amount of removed arsenic per mol of 

removed iron is 8.4 and 10.0 mmol As per mol Fe 

for the duplicate columns. This is valid for an arsenic 

concentration of 3.1 μmol.L -1 and an iron concentration 

of 95 μmol.L -1 . The amount of available sorption sites 

appears stable with every cycle, so the breakthrough of 

arsenic would, in theory, be delayed in case of lower 

arsenic concentrations (Equation 4.3). To study the 

effect of the molar Fe:As ratio in groundwater on the 

adsorptive removal efficiency of arsenic, the column 

experiments were repeated with different arsenic(III) 

concentrations: 3.1, 1.5, and 0.9 μmol.L -1 . 

Iron concentrations in the groundwater 

remained constant at 95 μmol.L -1 , thus the Fe:As 

ratios were 28, 63, and 103. The results for arsenic 


4 

3 

2 

Fe:As = 28 

Fe:As = 63 

breakthrough in the duplicate columns are depicted in 

Figure 4.9. It clearly shows that, independent of arsenic 

concentration, the breakthrough trend and retardation 

factor (R As 

=~1) were the same at the different Fe:As 

ratios. In other words, the percent breakthrough 

curves for arsenic at different Fe:As ratios matched 

exactly. This implies that although the amount of 

removed iron was equal per cycle, the total amount of 

removed arsenic was not. At a Fe:As ratio of 28, 63, and 

103, the total amount of removed arsenic was 3.9, 1.6, 

and 1.1 μmol, respectively. The sorption sites on the 

freshly formed iron hydroxide surfaces were apparently 

only available for arsenic adsorption up to a V/Vi of 

7-9, independent of the Fe:As ratio. This indicates that 

the available sorption sites may have been occupied by 

other sorbates in the groundwater, such as competing 

anions (e.g., phosphate), which limited the adsorption 

of arsenic. It is noteworthy that the Fe:As ratio at the test 

facility was even higher than in the columns (Fe:As = 

140) and also complete arsenic breakthrough occurred 

around V/Vi of 5-6. 

1 

Fe:As = 103 

General discussion 

0 

0 2 4 6 8 10 12 

V/Vi 

Figure 4.9 Arsenic breakthrough curves for duplicate columns 

with Fe:As ratios of 28, 63 and 103 

The total amount of iron that was removed per cycle 

was stable over time at an average of 0.30 and 0.29 

mmol for the duplicate columns. The retardation curve 

for oxygen showed that only a portion of the injected 

62


oxygen was consumed for oxidation of adsorbed 

ferrous iron. After one pore volume, with a travel time 

of approximately 22 minutes, the oxygen concentration 

had reached 82% of its original concentration again. 

Of the total amount of injected oxygen (155 μmol) 

around 75-105 μmol passed the column. Only 50- 

80 μmol O 2 

/cycle was consumed during the injection 

phase in the columns, corresponding to the molar 

Fe:O 2 

removal ratio of 4 and thus consistent with the 

theory of adsorptive-catalytic oxidation (Equation 4.1 

and 4.2). In other words, the oxidation of adsorbed 

ferrous iron was complete during the injection phase, 

illustrating that the iron oxidation reaction was fast and 

the adsorbed ferrous iron reacts with only a portion of 

the total amount of injected oxygen. 

For operational purposes this is an important 

finding, since, above a certain threshold value, the 

oxygen concentration does not control the adsorptioncatalytic 

oxidation, but rather the injection volume. 

The key is to oxidize as much soil grain surface area 

as possible, since this will provide new sorption sites 

for iron and arsenic. It is thus unlikely that injection of 

chemical oxidants (such as permanganate) will improve 

subsurface iron removal efficiencies. Such chemicals 

may even inhibit any bacterial activity that could be 

responsible for the improved iron removal efficiency 

with every successive cycle. Although only a portion 

of injected oxygen is used for rapid heterogeneous 

iron oxidation, the surplus oxygen in the oxidation 

zone is available for consumption by other adsorbed 

components – such as iron-oxidizing bacteria. 

Freshly formed iron oxides usually have high 

(adsorptive) surface areas (Cornell and Schwertmann, 

1996) and enhance the removal of ferrous iron. 

Interfacial Electron Transfer (IET, Jeon et al., 2001; Jeon 

et al., 2003) has been proposed to describe the “loss” of 

ferrous iron in an ferrous/ferric iron system. IET entails 

the transport of an electron to adsorbed ferrous iron 

from the incorporated iron hydroxide, creating new 

sorption sites at the surface. The theory of IET could 

provide an explanation for the improved subsurface 

iron removal efficiency at the test facility; however, 

these results were not reproduced by the adsorptivecatalytic 

oxidation in the columns. It is therefore more 

likely that other processes, such as bacterial activity or 

transport phenomena, are responsible for the enhanced 

iron removal in full-scale facilities. 

The community-scale test facility in Manikganj 

has shown that iron removal increases after multiple 

cycles, but arsenic removal remains stable at a 

retardation factor of 1. Hence, the amount of arsenic 

removed per mol of removed iron reduces with every 

successive cycle. This indicates that arsenic adsorption 

during subsurface treatment is controlled by the 

amount of oxidized iron, and not by the amount of 

removed iron. In the columns, the arsenic adsorption 

is also stable at R As 

=1 but, unlike at the test facility, iron 

removal does not improve with every successive cycle. 

The mechanism of adsorptive-catalytic oxidation is 

isolated in the columns from other potential removal 

63 

4


4 

processes, showing that subsurface arsenic removal is, 

indeed, controlled by the amount of oxidized iron per 

cycle. In the field, enhancement of subsurface arsenic 

removal can therefore only be achieved by increasing 

the oxidation zone, i.e. the volume of injected water. A 

high molar Fe:As ratio of the groundwater has not been 

shown to promote improved co-removal of arsenic 

with the iron. A proposed explanation for this finding is 

that arsenic removal is limited by the presence of other 

constituents in the natural groundwater, competing 

for the same adsorption sites. The process of arsenic 

adsorption is not limited by ferrous iron sorption 

(Dixit and Hering, 2006), but may be limited by the 

presence of competing anions in the multi-component 

environment, such as phosphate (Stachowicz, 2008). 

In the columns, phosphate concentrations were 10 

to 37 times higher than the arsenic concentrations, 

and phosphate adsorption may thus dominate over 

arsenic adsorption. In practice, competing anions will 

frequently co-occur in the groundwater with arsenic and 

could therefore locally threat the efficacy of subsurface 

arsenic removal. One could overcome this limitation by 

improving the current design to reach larger injection 

volumes through utilizing rainwater for injection. In 

areas with heavy rainfall during the monsoon season(s), 

it could even be considered to combine subsurface 

arsenic removal with artificial rainwater recharge and 

recovery. With such a design, the oxidation zone is 

scaled-up and arsenic/iron adsorption would occur in 

a much larger area around the well. 

Conclusions 

At the community-scale test facility in Bangladesh, 

subsurface iron removal showed great potential 

for decentralized application in rural areas. The 

efficacies were much higher than could be explained 

by the adsorptive-catalytic oxidation in the column 

experiments and therefore other ((a)biotic or transport) 

processes must contribute to the system’s efficacy. 

Unlike iron removal, subsurface arsenic removal did 

not increase after multiple cycles, illustrating that 

the process which is responsible for the effective iron 

removal did not promote an equally effective coremoval 

of arsenic. 

The strong correlation between field and 

column results indicates that arsenic adsorption during 

subsurface treatment is controlled by the amount of 

adsorbed iron that is oxidized, and not by the amount 

of removed iron. For operational purposes this is 

an important finding, since apparently the oxygen 

concentration of the injection water does not control 

the arsenic removal, like it does for iron. On the other 

hand, increasing the injection volume may be a better 

approach to enhance arsenic removal. No relation has 

been observed in this study between the amount of 

removed arsenic and the Fe:As ratio of the groundwater. 

It is proposed that the removal of arsenic is limited 

by the presence of other anions, such as phosphate, 

competing for the same adsorption sites. 

64


References 



Ground Water 37(6): 811-817. 





BCSIR (2003) Performance evaluation and verification of five 

arsenic removal technologies, ETV-AM Field Testing and 

Technology Verification Program, Dhaka. 

Boochs P. W. and G. Barovic (1981) Numerical-model describing 

groundwater treatment by recharge of oxygenated water. 



methods of in situ treatment of groundwater. Water Science 


British Geological Survey/DPHE (2001) Arsenic contamination 

of groundwater in Bangladesh, Volume 2: Final report, BGS 

Technical Report WC/00/19. 

Clifford D.A., S. Karori, G. Ghurye and S. Samanta (2004) Field 

speciation method for arsenic inorganic species, American 

Water Works Association Research Foundation, Denver. 

Cornell R. M. and U. Schwertmann (1996) The iron oxides - 

structure, properties, reaction, occurrence and uses, VCH- 

Germany and USA. 



Geology 228(1-3): 6-15. 


aquifer, Water Supply 3(1): 13-18. 


purification of groundwater, Ground Water 14(2): 88-93. 

Jechlinger G., W. Kasper, F. Scholler and F. Seidelberger (1985) 

The removal of iron and manganese in groundwaters through 

aeration underground, Water Supply 3(1): 19-25. 

Jeon B. H., B.A. Dempsey, W.D. Burgos and R.A. Royer (2001) 

Reactions of ferrous iron with hematite, Colloids and 

Surfaces A: Physicochemical and Engineering Aspects 191(1- 

2): 41-55. 

Jeon B. H., B.A. Dempsey and W.D. Burgos (2003) Kinetics and 

mechanisms for reactions of Fe(II) with iron(III) oxides. 

Environmental Science and Technology 37(15): 3309-3315. 





Miller G. P. (2006) Subsurface treatment for arsenic removal, 

American Water Works Association Research Foundation: 

59, Denver. 



3(2): 143-150. 



Water Science and Technology: Water Supply 2(1), 17-24. 




65 

4


4 



Smith A.H., P.A. Lopipero, M.N. Bates and C.M. Steinmaus 

(2002) Arsenic epidemiology and drinking water standards, 

Science 296, 2145-2146. 

Stachowicz M., T. Hiemstra and W. H. Van Riemsdijk (2008) 

Multi-competitive interaction of As(III) and As(V) oxyanions 

with Ca 2+ , Mg 2+ 3− 2− 

, PO 4 

4 , and CO 3 

ions on goethite, Journal 

of Colloid and Interface Science 320, 400–414. 




of drinking water. Environmental Technology 23(12): 1385- 

1404. 

van Beek C. G. E. M. (1985) Experiences with underground 

water treatment in the Netherlands, Water Supply 3(2): 1-11. 

van Halem D., S. G. J. Heijman, G. L. Amy and J. C. van Dijk 

(2010) Subsurface arsenic removal for small-scale application 

in developing countries. Desalination 251: 241-248. 

Welch A.H., K.G. Stollenwerk, L. Feinson and D.K. Maurer 

(2000) Preliminary evaluation of the potential for insitu 

arsenic removal from ground water. In Arsenic in 

Groundwater of Sedimentary Aquifers, 31st International 

Geological Congress, Rio de Janeiro, Brazil. 


report on arsenic in drinking water. Geneva. 

World Health Organization (2006) Guidelines for Drinkingwater 

Quality, First addendum to third edition, Volume1, 


WSP/Worldbank (2003) Fighting Arsenic: Listening to Rural 

Communities, Willingness to Pay for Arsenic-Free Safe 

Drinking Water in Bangladesh. Geneva. 

66

5 

Cation exchange during subsurface 

iron removal 


van Halem et al. (2010) Water Research: under review


5 

Introduction 

Based on the premise that heterogeneous oxidation 

prevails during Subsurface Iron Removal (SIR), the 

adsorptive-catalytic oxidation theory summarizes the 

processes during an injection-abstraction cycle (Rott, 

1985; van Beek, 1985). Besides the adsorptive-catalytic 

oxidation theory, it has also been proposed that the 

injection of O 2 

-rich water onsets the exchange of 

adsorbed Fe 2+ with other cations, such as Ca 2+ and Na + 

(Appelo et al., 1999; Appelo and Postma, 2005): 

Equations 5.1 and 5.2 

XCa + Fe ⇔ XFe + Ca 

2+ 2+ 

2+ + 

2XNa + Fe ⇔ X 

2Fe + 2Na 

With X being the exchange sites and exchange 

coefficients, following the Gaines-Thomas convention, 

of K Ca\Fe 

= 0.7 and K Na\Fe 

=0.6 (Appelo and Postma, 2005). 

Once Fe 2+ has exchanged/desorbed, the dissolved O 2 

in the injection water oxidizes the soluble Fe 2+ . In the 

presence of high Fe 3+ oxide concentrations, as generally 

found in the aquifer, this oxidation reaction can also 

occur heterogeneously. Subsequent hydrolysis results in 

the formation of immobile iron oxides or even mobile 

colloidal material (Wolthoorn, 2003). During the 

abstraction phase, these iron hydroxides provide new 

surface sites for Fe 2+ adsorption and/or cation exchange. 

The process of cation exchange (CIEX) during SIR has 

been schematized in Figure 5.1 for a simplified system 

containing Fe 2+ during abstraction and Ca 2+ , Na + and 

O 2 

during injection. Before the start of injection, both 

soluble and adsorbed Fe 2+ are present (Figure 5.1A). 

During injection the cations in the injection water, Ca 2+ 

and Na + , exchange with the adsorbed Fe 2+ on the soil 

grains (Figure 5.1B). The desorbed Fe 2+ is then flushed 

deeper into the aquifer, partially mixing with O 2 

in the 

injection water, resulting in hydrolyzed Fe 3+ precipitates 

(Fe(OH) 3 

; Figure 5.1C). When abstraction is started, 

the flow is reversed and the Fe 2+ in the groundwater 

is retained on the mineral surface and on the Fe 3+ 

hydroxides, either through CIEX or adsorption (Figure 

5.1D). 

The contribution of cation exchange to the 

system’s efficiency depends on the water composition 

of the injection water and groundwater, but also on 

the exchangeable Fe 2+ on the aquifer material. The 

cation exchange capacity (CEC) of the soil depends, 

in order of importance, on the clay, organic carbon 

and iron hydroxide content (Appelo and Postma, 

2005). The occurrence of CIEX is difficult to extract 

from field data, as iron oxidizes after it has exchanged, 

and will not reach the well in its ferrous, soluble 

state. However, for the purpose of optimizing the 

subsurface iron removal process or the application 

for other inorganic constituents, such as arsenic 

(Rott et al., 2002; van Halem et al., 2010) it is vital to 

understand the underlying mechanisms. The objective 

of this study was therefore to simulate the process of 

oxidation, adsorption and cation exchange in injectionabstraction 

column studies, and to investigate to what 

68

5 Cation exchange during subsurface iron removal 

A 

Fe 2+ 

Fe 2+ Fe 2+ 

Fe Fe 2+ 

2+ 

Fe 2+ Fe 2+ 

Fe 2+ 

Fe 2+ 2+ 

Fe 2+ Fe 2+ 

Fe 2+ 

Fe 2+ 

Fe 2+ 

Fe 2+ 

Mineral Surface 

Fe 2+ Fe 

Fe 2+ Fe 2+ Fe 2+ 

Fe 2+ Fe 2+ 

Fe 2+ 

Fe 2+ 

Fe 2+ 

Na + Ca 2+ 

O Na + 

2 

O 2 

O 

Ca Na Ca 2+ 

+ 

2+ 

2 

Na + 

Na Ca + 

2+ 

Na + Ca 2+ Ca 2+ Ca 2+ 

Na + 

Ca 2+ 

Na + Na + 

Na + 

Ca 2+ Ca 2+ 

Na + 

Fe 2+ 

Fe 2+ 

Mineral Surface 

Fe Fe Fe 2+ 

2+ 

2+ 

Fe 2+ 

Fe 2+ 

Fe 2+ Fe 2+ Fe 2+ 

B 

5 

O 2 

Ca 2+ 

Na + 

Ca 2+ 

Na O + 

2 

Ca 2+ O Na + 

2 

O 2 

O 

Na + 

Ca 2+ 2 

Ca 2+ Na 

Na + 

+ 

Ca 

Ca 2+ 

2+ 

Na 

Na + + 

Ca Ca Fe(OH) 2+ 2+ Fe 2+ 

Ca 2+ 

3 

Fe 2+ Fe 2+ 

Na + 

Fe 2+ 

Ca 2+ Mineral Surface Ca 2+ 

Na 

Fe 2+ Mineral Surface 

Fe 2+ 

+ 

Na + 

Fe 2+ 

Na + 

Fe 2+ Fe 2+ 

Ca 2+ 

O Na Fe + 2+ 

2 Fe(OH) 

Fe 3+ 

Fe 3+ 3 Fe(OH) 

3 

Fe 2+ 

Fe 2+ 

Fe 2+ Fe2+ 

Na 

Fe 2+ 

+ 

Na Fe Fe 2+ 

2+ 

+ 

O 2 

O 2 

O 2 

Fe(OH) 

Ca 2+ 3 

Fe Fe 2+ 2+ 

Fe Fe(OH) Fe 3+ 2+ 

3 

O 2 

C 

Fe 3+ 

Fe 2+ 

Fe 2+ 

Fe Fe 2+ 2+ 

Figure 5.1 Schematic presentation of cation exchange during subsurface iron removal on the sand grain surface 

D 

69


5 

extent the occurrence of cation exchange influences the 

subsurface iron removal process. The column studies 

have been performed with a Na + exchanger during 

injection, either in the presence or absence of O 2 

, 

with synthetic groundwater and in the more complex 

environment of natural groundwater. 


The experimental set-up (Figure 5.2) consisted of 

duplicate transparent PVC columns with a length of 

30cm and an inner diameter of 36mm (wall thickness 

2mm). During all experiments, the columns were 

wrapped in aluminium foil to exclude light. The columns 

were filled with washed (24h with 1M HCl) filter sand 

(500g; grain size = 0.5-0.8mm; D 10 

=0.58). The absence 

of other mineral structures than quartz on the sand 

material was checked with X-ray Powder Diffraction 

(Bruker D5005; Brain PSD). The push-pull operational 

mode of injection-abstraction was simulated in the 1D 

plug-flow environment of the columns with down flow 

for injection and up flow abstraction (1.1 L.h -1 ±0.05). 

An injection-abstraction cycle started with 14 (±0.5) 

pore volumes of injection water, to allow for complete 

breakthrough of dissolved O 2 

. Subsequently the influent 

was switched to groundwater to allow retention of Fe 2+ . 

Electrical conductivity was used as a conservative tracer 

from which the pore volume could be calculated to be 

on average 0.12L (±0.002). The flow rate in the columns 

(2.7 m.h -1 ) was controlled with a multi-channel pump 

and PVC tubing with low gas permeability. Anoxic 

conditions were maintained in the columns by using 

an airtight FESTO system (6 x 1 PUN, I.D. 4mm) with 

matching connectors and valves. 

The experiments were executed (i) in the 

laboratory with synthetic groundwater and (ii) 

at a drinking water treatment plant with natural 

groundwater. At the start of each experiment the 

columns were conditioned with synthetic or natural 

groundwater, until complete breakthrough of iron 

occurred, and the E h 

potential stabilized. A normal 

injection mode consisted of demineralized water 

containing a pH buffer (5mM NaHCO 3 

) and 0.28mM 

O 2 

. However, to study the role of cation exchange, 

injection cycles have also been performed in the 

absence of O 2 

and/or Na + . The water quality during the 

abstraction phase was constant during the experiments, 

being either synthetic or natural groundwater. 

The synthetic groundwater was produced by 

sprinkling demineralized water on a 6m gas stripping 

column containing stainless steel Pall Rings. From the 

bottom pure N 2 

was blown into the degassing column 

to sparge out all O 2 

. Before entering the sand columns, 

the water was checked for O 2 

with the Orbiphere 

(HACH Lange; M1100 Sensor; 410 Analyser) to 

ensure concentrations below 1.5 µmol.L -1 . Addition of 

stock solutions for FeSO 4 

, NaHCO 3 

and/or NaCl was 

done with a dosing pump followed by a static mixer. 

70


pH correction was achieved by addition of HCl or 

NaOH and all stock solutions were sparged with N 2 

in order to ensure the absence of O 2 

. The experiments 

were performed with synthetic groundwater of pH 6.9 

(±0.02), a temperature of 20°C (±0.1), Fe concentration 

of 0.1 mmol.L -1 (±0.01), pH buffer of 5mM NaHCO 3 

and ionic strength buffer of 1.6mM NaCl. To study the 

occurrence of cation exchange in the multicomponent 

groundwater matrix, the column set-up was transported 

to a groundwater treatment plant (WTP Loosdrecht, 

Vitens Water Supply Company). The groundwater, 

naturally containing Fe 2+ , was used as feed water for 

the column experiments. During the research period 

the groundwater had an average pH of 7.2 (±0.01), a 

constant temperature of 11 °C, 0.1 mmol Fe.L -1 (±0.01), 

3.3 μmol Mn.L -1 , 1.07 mmol Ca.L -1 , 0.52 mmol Na.L -1 , 

0.28 mmol Si.L -1 , and 0.14 mmol SO 4 

.L -1 . 

Fe analysis of the water samples was done with 

an Atomic Absorption Spectrometer (Perkin-Elmer 

Flame AAS 3110). In-line measurements were done for 

dissolved oxygen (Orbisphere and WTW Cellox 325), 

Eh potential (WTW SenTix ORP), pH (WTW SenTix 

H 2 

O 

5 

pH ORP EC DO 

Waste/Sample 

Point 

N 2 

Stainless steel Pall Rings 

H 2 

O 

+ 

O 2 


Dosing Pump 

Column 2 

Column 1 

Fe/As 

NaHCO 3 

Figure 5.2 Schematic overview of experimental column set-up 

71


5 

41), and electrical conductivity (WTW TetraCon 325). 



Results 

Injection-abstraction cycles 

A regular injection phase during subsurface iron 

removal consists of the injection of aerated water, in 

most cases drinking water from the clean water reservoir. 

In the experiments, demineralized water was used 

for injection containing a pH buffer (5mM NaHCO 3 

) 

and a 0.28mM O 2 

concentration. During injection 

the tracer passed the column at 1 pore volume (PV), 

but O 2 

concentrations were delayed in the columns, 

corresponding to an approximate O 2 

consumption 

of 0.15mmol in the synthetic groundwater columns 

(results not shown). Based on the stoichiometric ratio 

for Fe 2+ oxidation of 4, this could result in a total Fe 

removal of 0.6mmol for one cycle. 

To what extent iron is delayed compared to the 

groundwater determines the efficacy of the subsurface 

Fe removal technology. In general, groundwater 

treatment plants operate a V/Vi ratio based on the 

moment Fe starts to arrive at the well, thus (C/C 0 

) Fe 

>0. 

In the experiments, [Fe 2+ ] is allowed to breakthrough, in 

order to calculate the dimensionless retardation factor 

R. R Fe 

+1 is calculated from a Vi corresponding to the 

pore volume when the tracer (electrical conductivity) 

is C/C 0 

= 0.5, and V is the number of pore volumes that 

can be abstracted with iron concentrations below (C/ 

C 0 

) Fe 

=0.5. In the case of a column study, the oxidation 

zone is limited to the size of the column, i.e., 1 pore 

volume, making the calculation of V/Vi redundant: 

R 

Well 

TracerC/ C0 

Equation 5.3 

⎛V 

⎞ 

⎜ ⎟ 

⎝Vi ⎠ 

( PV ) 

FeC/ C 

Fe 

0 C/ 

C0 

= −1⇒ RColumn 

= −1 

⎛V 

⎞ 

( PV ) 

TracerC/ C0 

⎜ ⎟ 

⎝Vi 

⎠ 

The Fe retardation during abstraction for the regular 

injection-abstraction cycle is shown for both the 

synthetic and natural groundwater columns in Figure 

5.3. The tracer passed the columns at PV=1, but 

elevated Fe concentrations were not observed until 

30 and 8 pore volumes for the synthetic and natural 

columns, respectively. The Fe retardation in the natural 

groundwater columns (R Fe 

= 9) was measured to be 

lower than for synthetic water (R Fe 

= 42). The reduction 

in removal efficiency can be explained by the presence 

of competing cations in the natural groundwater, 

binding to the same sites as Fe 2+ . For instance, Ca 2+ is 

known to inhibit the Fe 2+ adsorption onto virgin and 

iron hydroxide coated sand (Sharma, 2001). 

72


(C/C0)Fe 

1 

0.5 

0 

R=9 R=42 

0 20 40 60 80 

pore volumes 

synthetic 

natural 

synthetic 

natural 

Figure 5.3 Fe breakthrough in duplicate columns after injection 

of water containing 0.28mM O 2 

and 5mM NaHCO 3 

for columns loaded with synthetic and natural 

groundwater (C 0 

=0.1mM Fe 2+ ) 

When considering the synthetic groundwater column 

results it was calculated from the retardation factor 

(R Fe 

= 42) that the Fe retention was approximately 0.6 

mmol/cycle. This value correlates well to the 0.6 mmol 

Fe calculated from the O 2 

consumption in the columns. 

Based on this finding it may be concluded that all O 2 

consumption was used for Fe 2+ oxidation, however, 

there is still the question of whether cation exchange 

plays a (catalyzing) role. Adsorbed Fe 2+ could either 

directly (heterogeneously) oxidize on the surface, 

like proposed with the catalytic adsorptive-oxidation 

mechanism, or it could exchange with other cations 

before oxidizing to Fe 3+ hydroxides. During a regular 

injection-abstraction cycle, the Fe:O 2 

mass balance 

does not differentiate between these two mechanisms, 

as Fe 2+ will always oxidize in the presence of O 2 

and be 

retained in the columns. 

Fe 2+ -Na + exchange 

After reaching steady state with the synthetic 

groundwater the columns were loaded for an injection 

phase without O 2 

, but with elevated Na + concentrations. 

Na + was added as 0.01M NaHCO 3 

or 0.5M NaCl. 

Additionally a control cycle was tested without the 

addition of any Na + . It is noteworthy that the pH during 

injection with water containing low buffer capacities 

(absence of HCO 3- 

) remained above pH 7.5 and the pH 

drop lasted for a maximum of 3 pore volumes. Figure 

5.4 shows the Fe concentrations that were measured 

during the injection phases. The dotted lines represent 

the results from the geochemical surface complexation 

model PHREEQC (v2.15; Parkhurst and Appelo, 1999). 

Fe desorption was measured from the column material 

at concentrations 1.5 to 25 times the C 0 

of 0.1mM. The 

PHREEQC model results even show a higher desorbed 

concentration, indicating that during sampling the 

high peak concentration was missed. The amount 

of exchanged Fe was higher in the case of 0.5M Na + , 

resulting in subsequent higher removal efficiencies 

during abstraction (Figure 5.4). R Fe 

was measured to 

be 11 and 30, for 0.01M and 0.5M Na + , respectively. 

The same experiment was conducted with 0.1M NaCl, 

showing the same retardation as for 0.5M NaCl (not 

shown, confirmed with PHREEQC), which means 

5 

73


5 

that maximum Na+ regeneration has occurred at 0.1M 

NaCl. 

The retardation of Fe during this particular cycle 

provides information on the exchangeable Fe 2+ capacity 

of the sand, as retention of approximate 0.4 mM Fe can 

be recalculated to a Cation Exchange Capacity (CEC) 

of 1.55 meq.kg -1 (0.78 mmol Fe.kg -1 ) for the column 

A 

mM Fe 

B 

(C/C0)Fe 

0.01M NaHCO3 0.01M NaHCO3 Phreeqc 

no Na no Na Phreeqc 

0.5M NaCl 0.5M NaCl Phreeqc 

0.8 

0.6 

0.4 

0.2 

0 

1 

0.8 

0.6 

0.4 

0.2 

0 

Injection 

0 2 4 6 8 10 

pore volumes 

Abstraction 

0 20 40 60 

pore volumes 

Figure 5.4 Fe measurements during injection of 0.01M NaHCO 3 

, 

Na-free or 0.5M NaCl without O 2 

and abstraction of 

synthetic groundwater (C 0 

=0.1mM Fe 2+ ), dotted lines 

represent results from PHREEQC (v2.15) 

8 

6 

4 

2 

0 

mM Fe (0.5M NaCl) 

sand. This CEC value is lower than for average aquifer 

sand (10 meq.kg -1 ; Appelo and Postma, 2005), as can be 

expected based on the absence of clay, organic carbon 

and iron oxides on this clean (silica) filter sand. The CEC 

value correlates well with the Fe 2+ adsorption capacity 

of virgin sand of 0.74 mmol Fe.kg -1 observed by Sharma 

et al. (1999). In that study the adsorption of Fe 2+ onto 

clean sand (D=0.7-1.25mm) at pH 7.0 in the absence 

of oxygen was investigated. When using the measured 

exchange value of 1.55 meq.kg -1 in PHREEQC, the 

measured results were simulated well. These results 

show that also in the absence of O 2 

, but in the presence 

of abundant Na + , a retardation of Fe can be achieved to 

a certain level. The desorption of Fe during injection 

shows the exchange of attached Fe 2+ with soluble Na + , 

and subsequent retardation of Fe points towards the 

vise versa exchange of Fe 2+ and Na + . The occurrence 

of Fe 2+ -Na + exchange, even on clean filter sand, can 

potentially play an important role during subsurface 

iron removal. However, this depends strongly on cation 

concentrations in the injection water and on the CEC 

of the aquifer material. The control cycle with Na + -free 

water confirmed that the Fe 2+ exchange was very low 

during injection (Figure 5.4). Fe was subsequently not 

significantly delayed in the columns, with an R Fe 

of 1. 

CIEX in natural groundwater 

The role of CIEX in the multi-component environment 

of natural groundwater was studied in experiments 

using groundwater from a water treatment plant 

74


(WTP Loosdrecht) instead of synthetic groundwater. 

Under these natural conditions the injection cycles 

in the presence and absence of O 2 

and/or Na + were 

investigated. Figure 5.5A depicts the Fe desorption 

during the injection in the absence of oxygen. A peak 

concentration almost reaching the initial concentration 

of 0.1mM was detected. At 0.1M NaCl the Fe leaching 

A 

mM Fe 

0.1 

0.08 

0.06 

0.04 

Injection 

5mM NaHCO3 

5mM NaHCO3 

0.1M NaCl 

0.1M NaCl 

no Na 

no Na 

from the columns was higher than for 5mM Na + , 

resulting in a slightly better Fe retardation during the 

abstraction phase (Figure 5.5). The Fe desorption after 

injection in the absence of Na + was most likely caused 

by the cations in the natural groundwater present in the 

columns just before injection. It is noteworthy that the 

leaching in these natural groundwater columns was up 

to 20 times lower than in the synthetic groundwater 

columns, as can be seen from Figure 5.4 and Figure 

5.5, not exceeding the background Fe concentration of 

0.1mM. 

The Fe retardation factors that were measured 

in all experiments for both synthetic and natural 

5 

B 

(C/C0)Fe 

0.02 

0 

1 

0.8 

0.6 

0.4 

0.2 

0 

0 2 4 6 8 10 12 

pore volumes 

Abstraction 

0 5 10 15 

pore volumes 

Retardation factor (R Fe) 

(A) 5mM NaHCO3/0.28mM O2 

50 

synthetic natural 

40 

30 

20 

10 

0 

(B) 5mM NaHCO3* (no O2) 

(C) 0.1M NaCl (no O2) 

(D) no Na/O2 

(E) 0.28mM O2 (no Na) 

Injection water quality 

* for synthetic 10mM NaHCO3 was used 

Figure 5.5 Fe measurements during (A) injection of 0.01M 

NaHCO 3 

, 0.5M NaCl or Na-free water without O 2 

and (B) abstraction with natural groundwater 

Figure 5.6 The measured Fe retardation factors for cycles with 

different injection modes with abstraction of either 

synthetic or natural groundwater 

75


5 

groundwater are summarized in Figure 5.6, showing 

the decrease in efficiency in the natural groundwater 

columns compared to the synthetic groundwater 

columns. The R Fe 

after injection with or without 

oxygen show a reduced value in the multi-component 

environment of natural groundwater (injection modes 

A, B, C, E). The Fe 2+ exchange/adsorption is clearly 

limited by the presence of other cations, such as Ca 2+ , 

resulting in retardation values below 12. In the absence 

of oxygen, the R Fe 

was somewhat similar for both 

5mM and 0.1M Na + , i.e., CIEX between Na + and Fe 2+ 

is not significantly enhanced at higher Na + . Also in the 

absence of Na + during injection, the Fe 2+ retardation was 

limited in natural groundwater compared to synthetic 

water (injection mode E), as the R Fe 

dropped from 33 to 

8 in the multi-component environment of the natural 

groundwater columns. It should be noted that although 

R Fe 

between injection modes A, C and E correlate well, 

the Fe breakthrough trend looks different (not shown). 

After injection of Na-free water, the Fe concentrations 

never reached below the detection limit, i.e., some 

Fe always passed the columns. This was observed 

in the synthetic and natural groundwater columns, 

indicating that some CIEX is needed to reach ultra low 

Fe concentrations at the beginning of an abstraction 

phase. 

In Figure 5.7 the measurements for Na and 

Ca are depicted, showing some desorption of Ca 

during injection (Figure 5.7A). For the duration of 9 

pore volumes, the Na concentration increased during 

injection from the concentration in the groundwater 

(C/C 0 

=0.12) to the concentration in the injection water 

(C/C 0 

=1). The tailing of both curves indicates that 

CIEX proceeds between these cations. The abstraction 

period shows the reversed trend (Figure 5.7B), as the Na 

concentration of the groundwater (C 0 

) is reached after 

approximately 8 pore volumes. The Ca concentration 

is already restored after 6 pore volumes. It should be 

A 

B 

(C/C0)Na,Ca 

(C/C0)Na,Ca 

1 

0.8 

0.6 

0.4 

0.2 

0 

1 

0.8 

0.6 

0.4 

0.2 

0 

Na 

Na 

Ca 

Ca 

0 2 4 6 8 10 

pore volumes 

0 5 10 15 

pore volumes 

Figure 5.7 Na and Ca measurements during an (A) injection 

and (B) abstraction cycle with natural groundwater 

76


noted that Ca 2+ complexation with natural organic 

matter (NOM) may have occurred as well, but because 

of the low binding contact time can be considered 

insignificant. It may be concluded that CIEX occurs, 

in the multi-component environment of natural 

groundwater, between a wide range of cations, resulting 

in retardation during the abstraction phase. The effect 

of CIEX is most predominant on Fe retardation, as Fe 

concentrations in the groundwater are relatively low 

compared to Ca 2+ and Na + . 

Fe 2+ desorption theory 

The column studies in the absence of O 2 

and/or Na + 

confirmed that, in addition to the adsorptive-oxidation 

process (van Beek, 1985; Rott, 1985), CIEX occurs 

during subsurface iron removal. The Na + concentrations 

of actual injection water will always be lower than the 

concentrations in the experiments. The experiments 

therefore overestimate the actual occurrence of CIEX 

during subsurface iron removal. However, the CEC of 

clean filter sand in the columns is much lower than that 

of actual aquifer sand (on average 10 times) resulting in 

lower CIEX. At the beginning of a full-scale injection 

cycle, the CIEX will happen in the presence of O 2 

, 

resulting in immediate oxidation. While injection 

proceeds, the injection water will penetrate further into 

the aquifer, but the O 2 

front will lag behind, caused by 

the consumption of O 2 

during Fe 2+ oxidation. In other 

words, the injected water front does not contain any O 2 

, 

but contains cations for exchange with Fe 2+ . The further 

the injection water flows into the aquifer, the larger the 

distance between the injected water front and the O 2 

front becomes. In this moving zone where O 2 

is absent, 

the process of CIEX prevails and Fe 2+ desorbs from the 

soil material and travels deeper into the aquifer (Figure 

5.4A and Figure 5.5A). 

In theory, these elevated Fe 2+ concentrations 

never come in contact with the O 2 

in the injection water 

and will not oxidize in this cycle (Figure 5.8). When 

abstraction starts and the flow direction is reversed, 

the desorbed Fe 2+ passes the available adsorption sites 

on the soil grains closer to the production well and is 

removed through either adsorption or CIEX. In other 

words, the injection phase of subsurface iron removal 

mobilizes a part of the adsorbed Fe 2+ through CIEX 

but does not subsequently oxidize the Fe 2+ . One may 

state that this proportion of mobilized Fe 2+ limits the 

system’s efficacy during the abstraction phase, as part 

of the adsorption sites will be occupied by the desorbed 

C 

C 0 

1 

0.5 

Separation of Fe 2+ and O2 

O2 front 

Injection 

water 

Fe 2+ 

desorption 


Figure 5.8 Schematic representation of the separation between 

O 2 

front and Fe 2+ desorption peak 

77 

5


5 

Fe 2+ and not by the Fe 2+ present in the groundwater. 

On the other hand, the exchange may also enhance 

SIR, as the desorbed Fe 2+ is pushed deeper into the 

aquifer and concentrations may be lowered through the 

buffering capacity of the aquifer material. Whether the 

Fe 2+ truly enhances or limits the efficacy of SIR is yet 

to be determined, because it all depends on the actual 

separation between the Fe 2+ desorption front and the 

O 2 

front. The investigation of this front separation 

was not the focus of these column experiments and is 

recommended for future research. 

Injection elevated O 2 

concentrations 

CIEX may play a role during subsurface iron removal, 

but it is the supply of O 2 

to the (im)mobile Fe 2+ that 

determines the efficiency of the system. Appelo et 

al. (1999) concluded that increasing the oxidant 

concentration of the injected water would be useless 

as long as the efficiency is limited by the amount of 

exchangeable Fe 2+ capable of consuming the oxidant. 

The observation that injected oxygen will not be used 

in the absence of available oxidizeable Fe 2+ is valid, 

but there is also a reason why injection of higher O 2 

concentrations can increase the system’s efficacy. 

Namely, if the injection water contains higher O 2 

concentrations, the O 2 

front will not lag far behind the 

injected water front. The desorbed Fe 2+ will then come 

in contact with O 2 

for oxidation and not leach out of the 

oxidation zone into the aquifer. Thus by injecting higher 

oxidant concentrations, the exchangeable Fe 2+ fraction 

V/Vi 

25 

20 

15 

10 

5 

0 

1 2 3 4 5 6 7 8 9 10 11 12 

well # 

0.28mM O2 

0.55mM O2 

Figure 5.9 The operational V/Vi ratio of 12 wells at WTP Corle 

with (0.55mM) and without (0.28mM) the injection 

of 47mol.m -3 pure oxygen 

on the soil material can be utilized for oxidation and 

will contribute as hydrolyzed Fe 3+ hydroxides with 

their exchangeable/adsorptive surface area during the 

following abstraction phase. 

At water treatment plant Corle (Vitens Water 

Supply Company) there has been extensive experience 

with the injection of elevated oxygen concentrations 

into subsurface iron removal wells. In the past they 

injected 3,000m 3 of drinking water containing 0.28mM 

O 2 

, but they have changed the operational mode to 

the injection of 2,000m 3 with an O 2 

concentration of 

0.55mM. Although the O 2 

concentration increased with 

a factor of 1.9, the total O 2 

injection increased only with 

a factor 1.3, from 0.84 to 1.1 *10 3 mol. As a result of this 

operational change, the volumetric ratio for abstraction 

(V) and injection (Vi) has increased from an average 

7 to 16. This immense and sudden efficiency increase 

78


was clearly caused by the operational change, as an 

increasing efficacy caused by successive cycles (Appelo 

et al., 1999) may not be expected at stablized subsurface 

iron removal wells. WTP Corle calculates V/Vi based 

on the moment when Fe breakthrough starts, so the 

moment [Fe]>2µM is registered. 

Figure 5.9 depicts the relation of the operational 

parameter V/Vi for 12 subsurface iron removal 

wells after injection of 0.28mM O 2 

and 0.55mM O 2 

. 

Although there was some variation between the results, 

considering these are operational data from 12 different 

wells, it can be concluded that on average the V/Vi 

increases by an approximate factor 2. This indicates that 

not the available (adsorbed and/or exchangeable) Fe 2+ 

is limiting during injection, but the supply of O 2 

to the 

(in)soluble Fe 2+ . The field measurements support the 

theory that increasing the oxidant concentration has a 

positive effect on the subsurface removal of Fe, but they 

do not prove that it is actually the Fe 2+ desorption front 

that is targeted by the higher O 2 

dose. The field results 

point towards the assumption that Fe 2+ is abundantly 

available on the soil material. In conclusion, CIEX may 

occur in the aquifer during injection, but does not seem 

to limit the system’s efficiency when injecting elevated 

oxygen concentrations. 

Conclusions 

In sand column experiments with synthetic and natural 

groundwater it was found that cation exchange (Na + - 

Fe 2+ ) occurs during the injection-abstraction cycles of 

subsurface iron removal. The Fe 2+ exchange increased 

at higher Na + concentration in the injection water, 

but decreased in the presence of other cations in the 

groundwater. Field results with injection of high O 2 

concentrations indicated that not the exchangeable 

Fe 2+ on the soil material is the limiting factor during 

injection, but it is the supply of O 2 

to the available Fe 2+ . 

References 


(1999) Modeling in situ iron removal from ground water. 

Ground Water 37(6): 811-817. 



Parkhurst D.L. and C.A.J. Appelo (1999) User’s guide to phreeqc 

(version 2) – a computer program for speciation, batchreaction, 

one-dimentional transport, and inverse geochemical 

calculations. Water-Resources Inverstigation Report 99-4259, 

US Geological Survey. 

Rott U., (1985) Physical, chemical and biological aspects of the 


3(2): 143-150. 

79 

5


5 



Water Science and Technology: Water Supply 2(1): 17-24. 

Sharma S.K., M.R. Greetham, J.C. Schippers (1999) Adsorption 

of iron(II) onto filter media. J Water SRT-Aqua 48(3): 84-91. 

Sharma S.K. (2001) Adsorptive iron removal from groundwater. 

PhD dissertation, Wageningen University. 


water treatment in the Netherlands. Water Supply 3(2): 1-11. 

van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk, 

G.L. Amy and J.C. van Dijk (2010) Subsurface iron and 

arsenic removal for shallow tube well drinking water in rural 


Wolthoorn A. (2003) Subsurface aeration of anaerobic 

groundwater; iron colloid formation and the nitrification 

process. Ph.D. dissertation, Wageningen University. 

80

6 

Characterization of accumulated 

deposits 


van Halem et al. (2011) Applied Geochemistry 26: 116-124


6 

Introduction 

Subsurface iron removal has been implemented since 

the 1970s in Europe (Hallberg and Martinell, 1976; 

Mettler, 2002), nevertheless the technology has not 

yet found widespread application elsewhere. The iron 

removal occurs in the aquifer and is therefore not 

as visible as conventional above-ground rapid sand 

filtration, resulting in concerns regarding the longterm 

sustainability. Clogging of the aquifer is an issue 

that is raised frequently, however, in literature there is 

an agreement that clogging of the aquifer does not pose 

a serious threat to subsurface iron removal (Appelo 

et al., 1999; Braester and Martinell, 1988; Grombach, 

1985; van Beek, 1985). The absence of aquifer clogging 

has been proposed to be caused by (Appelo et al., 1999; 

Rott et al., 2002; van Beek 1985): 

• decreasing porosity with time, leading to further 

infiltration of injection water into the aquifer; 

• iron precipitation at different infiltration distances, i.e., 

variable size of the oxidation zone; 

• iron deposits predominantly formed in dead-end pores 

or stagnant zones; 

• transformation of voluminous amorphous iron 

hydroxides to less voluminous, crystalline hydroxides. 

Although previous publications have not reported 

aquifer clogging around subsurface iron removal wells, 

doubts on the sustainability of this technology still limit 

the widespread application. However, this inexpensive 

technology holds great promise for the developed 

and developing country context, with the potential 

application for arsenic (van Halem et al., 2010) and 

manganese removal. 

Iron hydroxide deposits near subsurface 

treatment wells have been previously investigated 

in literature. Braester and Martinell (1988) have 

been unable to microscopically identify any severe 

precipitation at the oldest Swedish plant built in 1971. 

In some samples precipitates were interpreted as 

bacterial iron stalks, but also mineral transformation 

into haematite has been observed. Oxalate/dithionite 

extractions of aquifer material near subsurface treatment 

wells have shown that half of the iron hydroxides were 

poorly crystalline and X-ray diffraction showed traces 

of lepidocrocite (Olthoff, 1986). Mettler et al. (2001) 

characterized the iron precipitates close to a subsurface 

iron removal well in Switzerland. With chemical 

extraction and 57 Mössbauer spectroscopy they found 

that iron was mainly deposited as ferric hydroxides and 

consisted for 50-100% of goethite. 

The mechanism of adsorptive-catalytic oxidation 

would suggest the formation of neatly ordered Fe 3+ 

hydroxides, rather than voluminous amorphous iron 

sludge. Also, cation exchange (Appelo, et al., 1999), 

Interfactial Electron Transfer (Jeon et al., 2001; Mettler 

2002) and recrystallization (Pederson et al., 2005) have 

been proposed to occur during subsurface iron removal 

resulting in different hypotheses on iron precipitate 

accumulation with time. The main objective of this study 

was to identify and characterize iron accumulation 

82

6 Characterization of accumulated deposits 

deposits in the aquifer near subsurface iron removal 

wells to assess the sustainability regarding clogging of 

the aquifer. An additional objective was to investigate 

the potential co-accumulation of other groundwater 

constituents, such as arsenic and manganese. 


Soil sample collection 

Oasen Drinking Water Supply Company, in 

the Netherlands, has extensive experience with 

groundwater-related operational modes, such as 

subsurface treatment and (river) bank filtration (de Vet 

et al., 2009). The production wells 03 and 04 at Oasen 

Water Treatment Plant (WTP) “De Put” have been 

intermittently used for subsurface iron removal since 

1996. Injection of 1,500-2,200 m 3 drinking water was 

done every other month, resulting in cycles of 2 months. 

Both wells were in operation for one month and stood 

still for the other month, while the other well was in 

operation. During operation, water is abstracted at a 

rate of ±1,250-1,600 m 3 .day -1 , for subsequent treatment 

with biosand filters. The average groundwater quality of 

the reference wells is summarized in Table 6.1. 

The accumulation of iron and other constituents 

onto aquifer soil material during subsurface iron 

removal was assessed by analysis of drilled samples 

from aquifer material. Bore holes (SonicMast with 

SonicDrill 2x7) were drilled at 5m from injection/ 

abstraction wells 03 and 04, and between the two wells 

(at ±50m from both wells, Figure 6.1). The borehole 

between well 03 and 04 can be used as a reference, 

since, at 50m, it was outside the affected oxidation zone. 

800mL samples were taken every meter at the depths 

from 10 to 28 meters below ground level. Classification 

of the drilled material showed that at approximately -9 

and -12m for well 03 and 04, respectively, the clay layer 

ended and coarse sand material (brownish) started. The 

drillings showed similar variations, with greyish coarse 

sand from a depth of 20-25 and 22-28 m for wells 03 and 

04, respectively. The lower clay layer started at 26-27 

m, showing that the wells were drilled into a confined 

aquifer. The depth of the perforated well filters was 12- 

6 

Table 6.1 Average water quality parameters of groundwater at Oasen WTP “De Put” (reference wells 1992-2008) 

pH Fe Mn Ca Cl Mg Na HCO 3 

SO 4 

NH 4 

Si As Pb Ni 

µmol.L -1 mmol.L -1 nmol.L -1 

7.3 47 11 2.1 2.7 0.46 2.0 3.9 0.5 0.2 0.2 44 1.3 26 

83


6 

5m 

-12m 

-27m 

injection/abstraction wells 

50m 

bore holes 

50m 

Well 03 Well 04 

-15m 

-26m 

Figure 6.1 Schematic overview of injection/abstraction wells and 

drilled bore holes 

27 m and 15-26 m for wells 03 and 04, respectively. The 

soil samples were taken in September 2008, subsurface 

treatment started in 1996 and had been operated for 12 

years at the moment of sampling. 

5m 

Chemical extraction 

Depending on the subsequent analyses, the samples 

were oven dried at 40°C or 105°C. The organic matter 

was measured after ignition at 450°C (Craft, 1988), with 

an average of all samples of 0.21%(±0.15), 0.25%(±0.13), 

and 0.23%(±0.14), for the reference well, well 03 and 

well 04, respectively. Samples from every meter depth 

were analysed in a certified laboratory with ICP-MS 

scan after chemical extraction with: Acid-Oxalate for 

amorphous iron and manganese; or, Nitric Acid for 

total iron and a wide range of elements (including, 

Na, Mg, Al, Si, K, Ca, Mn, Sr, Ni, Sc, Ti, As). Acid- 

Oxalate extraction consisted of reductive dissolution 

of amorphous iron/manganese with an oxalate buffer 

solution which consisted of 0.2M ammonium oxalate 

and 0.2M oxalic acid (at pH=3). During digestion the 

sample was agitated and light was excluded. 

The samples at 19-20, 20-21 and 21-22m depth 

in the reference well and near well 04 were oven dried at 

40°C and sieved (mesh sizes 00.63, 0.106, 0.125, 0.150, 

0.250, 0.50, 1.0, and 2.0mm). Duplicate samples between 

0.02 to 0.7 mg were taken from every sieve fraction for 

chemical extraction (1) with 5M Hydrochloric Acid for 

24 hours to extract total iron (Heron et al., 1994), and 

(2) with 1M Hydrochloric Acid addition to 50mL of 

demineralised water to dissolve carbonates at pH 3.5- 

4.0 (Kroetsch and Wang, 2006). Once the samples were 

diluted they were filtered with 0.2μm filters (cellulose 

acetate) for total iron analysis with Flame Atomic 

Absorption Spectrometer (Perkin-Elmer AAS 3110). 

84


ESEM-EDX 

Sieved soil samples (fraction


6 

production. The stoichiometric ratio of injected oxygen 

versus removed iron is theoretically 1:4, at WTP “De 

Put” this ratio was between 1:3.3 and 1:3.8 in the year 

2006-2007, meaning that 81 to 94% of the injected 

oxygen was consumed for iron oxidation. The amount 

of iron that has been retained in the subsurface during 

12 years of operation can be estimated based on the 

historical water quality data. The average iron removal 

per cycle is an estimated 1.4•10 3 mol/cycle and 1.6•10 3 

mol/cycle for well 03 and 04, respectively. Although 

there is no difference in operation between the wells, 

well 04 performs better than well 03. Nevertheless, it 

can be roughly calculated that in a period of 12 years 

approximately 1•10 5 mol of iron has been retained in 

the subsurface for each well. 

Figure 6.3 shows the results for total iron 

extraction with Nitric Acid on the soil samples from 

[Fe] mmol Fe.L -1 

0.06 

0.04 

0.02 

0.00 

21.8.06 29.11.06 9.3.07 17.6.07 25.9.07 3.1.08 12.4.08 21.7.08 

Well 03 Well 04 Reference 

Figure 6.2 Fluctuating iron concentration pattern in the 

produced groundwater in wells 03 and 04 (year: 

2006-2008) 

a depth of 12 to 28m at a distance of 5m from wells 

03 and 04 and in the reference bore hole. The graph 

indicates that iron accumulation was found in distinct 

soil layers near both wells. Near well 03, elevated iron 

levels were found at a depth of 12 to 15m and from 23 

to 26m. The difference between the reference value and 

the sample at 5m from this well varied between 11.5 

and 271.7 mmol Fe.(kg.ds) -1 . Compared to well 04, less 

iron seems to have accumulated near well 04, which 

corresponds to the water quality data in Figure 6.2. 

The iron accumulation near well 04 was mainly found 

at a depth from 19 to 21m, where iron concentrations 

were between 12.1 and 390.8 mmol.(kg.ds) -1 higher 

than the reference bore hole. The accumulation of iron 

at specific depths could indicate that preferred flow 

lines were followed by the injected water, e.g., due to 

higher permeability. Certain soil layers may have a 

lower resistance both during injection and abstraction, 

resulting into a greater contribution to the subsurface 

treatment system than layers with other properties. 

However, based on the classification of the aquifer 

material (coarser sand or more silt/clay; data not 

shown) no relation was found between soil layers and 

iron accumulation. 

The Acid-Oxalate extraction method was used 

to dissolve only the amorphous or poorly crystalline 

iron hydroxides in the drilled soil samples. The results 

in Figure 6.3 show increased iron concentrations 

compared to the reference level near both wells. The 

soil layers where iron accumulated were more or less 

86


similar to Figure 6.3, indicating that indeed the injected 

water affected mainly certain soil layers. Compared to 

the reference samples, amorphous iron has accumulated 

up to 35.0 and 62.2 mmol.(kg.ds) -1 for well 03 and 04, 

respectively. 

Based on the assumption that all iron accumulates 

as iron (oxy)hydroxides, it can be calculated what 

portion of accumulated iron is amorphous or crystalline: 

Fe crys 

=Fe total 

-Fe amorph 

. The percentage of amorphous and 

crystalline iron that has accumulated over 12 years in 

12 

mmol Fe.(kg.ds) -1 

0 200 400 600 800 

the affected layers was found to be 56-100% crystalline 

near well 03 and, 67-100% near well 04. In order to 

ascertain the limited contribution of iron in carbonates 

on the soil grains, the carbonates were dissolved at pH 

3.5-4.0 with Hydrochloric Acid at the different mesh 

sizes (Kroetsch and Wang, 2006). All samples taken 

in the reference well and near well 04 between 19 and 

22m confirmed that the iron release during carbonate 

removal was very small at


of crystalline deposits excludes the accumulation of 

massive iron sludge volumes in the subsurface. The 

presence of amorphous deposits indicates that iron 

initially (partially) precipitates as amorphous iron, but 

transforms to more crystalline iron hydroxides with 

time. 

Grain size vs. iron accumulation 

The samples at depths 19-20, 20-21 and 21-22m near 

well 04 and in the reference well were selected for 

sieve analyses and subsequent extraction with 5M 

Hydrochloric Acid, as these soil layers were clearly 

affected by subsurface iron removal. All samples were 

classified as sandy aquifer, with brownish sand. The 

samples near well 04 were more orange coloured, than 

the samples from the reference well. Sieve analyses 

showed a very similar size distribution in all six 

samples with 95% of the total weight between mesh 

sizes 0.125 and 0.250mm. Additional analyses with 

the Hydrometer Method showed that the clay and silt 

500 

400 

S = 

φ 

6 

s s ρ 

i 

j 

25 

20 

6 

mmol Fe.(kg.ds) -1 

300 

200 

Well 04 

Reference 

15 

10 

mmol Fe.m -2 

100 

5 

0 

0 

0.063-0.106 0.106-0.125 0.125-0.150 0.150-0.250 0.250-0.50 0.50-1.0 1.0-2.0 

Grain size (mm) 

Reference Well 04 

Figure 6.4 Relation between grain size and iron per geometric surface area (lines) and per weight dry solids 

(bars) for samples taken at 19-22m in reference well and near well 04 

88


content of the samples was no more than 3%w/w and 

the clay content was below 0.5-0.6 %w/w. 

Figure 6.4 depicts the average iron concentrations 

for 19-22m depth, distributed over the different 

sand grain sizes. The smallest fractions (


[C04]-[CREF] 

5 

4 

3 

2 

1 

mmol/kg.ds 

*10e-1 mmol/kg.ds 

mmol Ca.(kg.ds) -1 

0 100 200 300 400 

12 

14 

16 

6 

0 

Na Mn Sc Ti V As Sr 

Figure 6.5 Accumulated trace compounds near well 04 at depths 

between 19 and 22m 

was not sufficiently regular or accurate to monitor a 

reduction in the abstracted water due to subsurface 

iron removal. For arsenic, however, the concentrations 

were very low (


the majority of the Cation Exchange Capacity (CEC) 

in the groundwater, with 4.2 meq.L -1 of the total 7.2 

meq.L -1 (Ca+Mg+Na+Fe+Mn). The organic carbon 

and clay content can be used to estimate the CEC of the 

soil with the following empirical formula (Appelo and 

Postma, 2005): 

Equation 6.1 

CEC( meq / kg) 

= 7 ⋅ (% clay) 

+ 35⋅ 

(% organicC) 

The clay content has been determined with the 

Hydrometer Method and the organic carbon can be 

calculated from the organic matter concentration. 

When calculating the organic carbon from the organic 

matter content, the bound water needs to be taken into 

account when significant proportions of clay minerals 

are present. The clay content in the studied sediments 

was Fe > Na > Si > Al > Ca, 

with an average iron content of 7.6 mol% and an O:Fe 

mol% ratio of 5.3 (±2.25). The O:Fe mol% ratio in the 

reference coatings at 19-22m depth were much lower 

with an average of 1.3 (±0.57), the iron content was 

slightly higher at 9.9 mol%. The reduction of carbon 

near well 04 and the increase in oxygen indicates a 

change in coating composition to iron phases high 

in oxygen, like iron (oxy)hydroxides (Fe(OH) 3 

or 

α-FeOOH). 

Although based on the acid digestion results the 

iron content increased significantly near well 04, the 

ESEM-EDX results do not reflect this. This indicates 

that the coatings themselves have not gained more 

percent iron, but the overall amount of coating(s) must 

have increased. As the results from chemical extraction 

had already indicated, both sodium and calcium were 

found to be present in the iron coatings. The Fe:Ca 

91 

6


(a) 

(b) 

6 

(c) 

(d) 

1 Figure 6.7 ESEM pictures of iron coatings (a) iron coated grain at 19-20m; (b) iron precipitate on sand grain at 21-22m; (c) calcium 

sulphate coating at 23-34m; and (d) pyrite at 23-24m 

92


%mol ratio was on average 4.8 (±1.65) and 5.4 (±1.26) 

near well 04 and the reference well, respectively. The 

precipitates in Figure 6.7(c) and (d) have been found 

in the deeper layer at 23-24m. ESEM-EDX showed 

that the Ca:S ratio in coating (c) is 0.85, which roughly 

corresponds to CaSO 4 

(gypsum). Calcium sulfate has 

not been observed in any of the samples taken between 

19 and 22m depth, neither near well 04 nor in the 

reference well. The same was the case for the occurrence 

of the framboidal pyrite (FeS 2 

, (d)). In the presence of 

oxygen, pyrite is known to oxidize and release sulphate 

and iron into the groundwater. In addition, pyrite 

oxidation is an acidifying reaction and therefore not 

promoting the (rapid) oxidation of Fe 2+ . The presence 

of sulphide minerals is a vital difference between the 

depths with iron accumulation (19-22m) and without 

iron accumulation (deeper than 22m). 

X-Ray Powder Diffraction and 57 Fe 

Mössbauer Spectroscopy 

The same ten soil samples (


6 

hyperfine values of haematite. The Fe 2+ contribution, 

doublet with a Quadrupole Splitting (Q.S.) of ~ 2.39 

mm.s -1 , can be assigned to iron in phyllosilicates/ clay 

minerals like kaolinite and illite (Murad and Cashion, 

2004) or smectites (Heller-Kallai and Rozenson, 1981). 

Since the Fe 3+ doublets of most iron hydroxides can 

overlap irresolvable, it is recommended to measure 

Mössbauer spectra also at 77 and/or 4.2 K to allow a 

definitive identification of the iron structures and the 

quantitative determination of their proportions. 

The XRPD peaks for calcite allowed for the 

calculation of molar weight percentages by quantitative 

analyses with the Reference Intensity Ratio method. 

Calcite concentrations in the pre-treated samples 

(ground coating at


specific type of well clogging, intermittent abstraction 

to prevent well clogging has also been proposed by van 

Beek et al. (2010). In case of subsurface iron removal 

wells, the well is stopped and the flow direction is 

reversed, resulting in an even more effective reduction 

of the particle accumulation on the well screen. The 

contribution of particles in clogging is apparently more 

important than the possible effect of subsurface iron 

removal, indicating that clogging of the aquifer and/ 

or well is no threat to the sustainable operation of this 

technology. 

Clogging of the well and aquifer is thus not 

noticeably occurring, even after removal of more than 

100,000 moles of iron. If this amount of iron would 

have been removed by conventional rapid sand filters, 

it is most likely that clogging by the voluminous iron 

sludge would have occurred. Water treatment sludge 

retains excessive amounts of water: e.g., 85% (w/w) 

after 5 months of sedimentation (Georgaki et al., 2004). 

Assuming formation of ferrihydrite (approximate 

composition, 5Fe 2 

O 3 

•9H 2 

O), the total amount of iron 

sludge would add up to 1.28•10 6 kg in 12 years of 

subsurface iron removal. Pure ferrihydrite has a density 

of 4,000 kg.m -3 (Cornell and Schwertmann, 1996), and 

thus a very conservative estimation of the sludge density 

is 1,450 kg.m -3 . After 12 years of subsurface iron removal 

800 

600 

rehabilitation 

rehabilitation 

6 

Drawdown (cm) 

400 

normal production 

wells 

subsurface iron 

removal well 

200 

0 

2.7.06 10.10.06 18.1.07 28.4.07 6.8.07 14.11.07 22.2.08 1.6.08 9.9.08 

Figure 6.9 Drawdown development in normal production wells and a subsurface iron removal well at WTP 

De Put (year 2006-2008) 

95


6 

a sludge volume of >880 m 3 would have been produced. 

This is more than half of the injection (pore)water 

volume, and, even with a large (pore) storage capacity 

it is unthinkable that such a sludge amount remains 

unnoticed by the operator, i.e., clogging of the aquifer 

should have been observed. The finding that 56-100% 

of the iron has accumulated as compact, crystalline 

iron (oxy)hydroxides indicates that no voluminous 

water-containing iron sludge is produced subsurface. 

The presence of amorphous iron hydroxides, however, 

indicates that iron initially precipitates as ferrihydrite 

and subsequently transforms to iron crystal phases. 

This process in itself reduces the volume of the iron 

precipitates significantly, but could occur together with 

one or more of the previously proposed processes: iron 

precipitation in dead-end pores or deeper infiltration 

into the aquifer with time. 

More interesting is the existence of actively 

contributing soil layers where iron has accumulated 

versus soil layers where subsurface iron removal seems 

not to have occurred. Whether it was due to preferred 

flow paths or preferred geochemical mineralogy 

conditions (e.g., calcite); subsurface iron removal clearly 

favoured certain soil layers. The occurrence of such soil 

layers and the overall contribution of these layers in 

the production capacity of the well is site-specific and 

greatly influences the well’s performance. It is in this 

observation that the exact physical distribution of the 

iron precipitates, e.g., in dead-end pores or distance 

from the well, is not relevant for the clogging of the 

aquifer. The assumed dominant process of adsorptivecatalytic 

oxidation during subsurface iron removal, 

with or without additional cation exchange (Appelo 

et al., 1999) or interfactial electron transfer (Mettler, 

2002), cannot be further supported based the presented 

data. Nevertheless, there seems to be a close correlation 

between the occurrence of iron and calcium (as calcite) 

in the analysed samples. This finding points out the 

proposed catalyzing role of calcite in the oxidation of 

Fe 2+ during subsurface iron removal (Mettler et al., 

2009), which needs further investigation. 

The co-occurrence of iron in the subsurface 

in three phases, soluble Fe 2+ in the groundwater, 

poorly ordered amorphous iron hydroxides and 

crystalline iron hydroxides, supports existing theory 

on recrystallisation of amorphous iron hydroxides. In 

the presence of soluble Fe 2+ , freshly formed ferrihydrite 

was found to transform to the more crystalline goethite 

after 5 days (Jeon et al., 2003). Pederson et al. (2005) 

observed complete ferrihydrite transformation into 

new and more stable phases such as lepidocrocite and 

goethite within 5 days, and it was hypothesized that this 

was induced by the catalytic action of soluble Fe 2+ . In 

the subsurface iron removal system with soluble Fe 2+ , 

amorphous iron hydroxides and stable (oxy)hydroxides 

it is likely that recrystallization of the freshly formed 

precipitation occurs, resulting in a more crystalline 

accumulation of iron near injection/abstraction wells. 

96


Conclusions 

After 12 years of operation, iron has accumulated 

at specific depths near the subsurface iron removal 

wells. Whether it was due to preferred flow paths 

or geochemical mineralogy conditions; subsurface 

iron removal clearly favoured certain soil layers. The 

majority of accumulated iron was characterized as 

crystalline, suggesting that precipitated amorphous 

iron hydroxides have transformed to iron hydroxides 

of higher crystallinity. These crystalline, compact 

iron hydroxides have not noticeably clogged the 

investigated well and/or aquifer between 1996 and 

2008. The subsurface iron removal wells even need less 

frequent rehabilitation, as drawdown increases slower 

than in normal production wells. Other groundwater 

constituents, such as manganese, arsenic and strontium 

were found to co-accumulate with iron. Acid extraction 

and ESEM-EDX showed that calcium occurred together 

with iron and with X-Ray Powder Diffraction it was 

identified as calcite. 

References 



Ground Water 37(6): 811-817. 




methods of in situ treatment of groundwater. Water Science 


Cornell R. M. and U. Schwertmann (1996) The iron oxides - 

structure, properties, reaction, occurrence and uses. VCH- 

Germany and USA. 

Craft C.B., S.W. Broome and E.D. Seneca (1988) Nitrogen, 

phosphorus and organic carbon pools in natural and 

transplanted marsh soils. Estuaries 11(4): 272-280. 

de Vet W. W. J. M., C.C.A. van Genuchten, M.C.M. van 

Loosdrecht and J.C. van Dijk (2009) Water quality and 

treatment of river bank filtrate. Drinking Water Engineering 

and Science Discussions 2: 127–159. 

Georgaki I., A.W.L. Dudeney and A.J. Monhemius (2004) 

Monhemius characterisation of iron-rich sludge: correlations 

between reactivity, density and structure. Minerals 

Engineering 17: 305–316. 


aquifer. Water Supply 3: 13-18. 


purification of groundwater. Ground Water 14(2): 88-93. 

Heron G., C. Crouzet, C.M. Bourg and T.H. Christensen 

(1994) Speciation of Fe(II) and Fe(III) in contaminated 

aquifer sediments using chemical extraction techniques. 


Jeon B. H., B.A. Dempsey, W.D. Burgos and R.A. Royer (2001) 

Reactions of ferrous iron with hematite. Colloids and 

Surfaces A: Physicochemical and Engineering Aspects 191(1- 

2): 41-55. 

97 

6


6 

Jeon B. H., B.A. Dempsey and W.D. Burgos (2003) Kinetics and 

mechanisms for reactions of Fe(II) with iron(III) oxides. 


Kroetsch D. and C. Wang (2006) Chapter 55: Particle size 

distribution. In:Soil sampling and methods of analysis. 

Canadian Soil Society, Taylor & Francis Group, LLC. 

Mettler S., M. Abdelmoula, E. Hoehn, R. Schoenenberger, P.G. 



treatment plant. Groundwater 39(6): 921-930. 



calcareous aquifer. PhD dissertation, Swiss Federal Institute 


Mettler S., M. Wolthers, L. Charlet and U. von Gunten (2009) 

Sorption and catalytic oxidation of Fe(II) at the surface of 

calcite. Geochimica et Cosmochimica Acta 73: 1826-1840. 

Murad E. and J.H. Johnston (1987) Iron oxides and oxyhydroxides, 

in: Mössbauer Spectroscopy Applied to Inorganic Chemistry, 

G.J. Long, Ed. Plenum, New York, 507-582. 

Murad E. and U. Wagner (1994) The Mössbauer Spectrum of 

illites. Clay Minerals 29, 1-10. 

Murad E. and J.D. Cashion (2004) Mössbauer Spectroscopy of 

environmental materials and their industrial utilization, 

Kluwer, Boston. 

Olthoff R. (1986) Removal of iron and manganese in the aquifer. 

Inst. Siedlungswasserwirtschaft Abfalltechn. 63. University of 

Hannover, Germany. 

Pedersen H. D., D. Postma, R. Jakobsen and O. Larsen (2005) 

Fast transformation of iron oxyhydroxides by the catalytic 

action of aqueous Fe(II). Geochimica et Cosmochimica Acta 

69(16): 3967-3977. 



3(2): 143-150. 




Timmer H., J.D. Verdel and A.G. Jongmans (2003) Well clogging 

by particles in Dutch well fields. Journal AWWA 95(8): 112- 

118. 



van Beek C. G. E. M., R. Breedveld, M. Tas, and R. Kollen 

(2010) Prevention of wellbore clogging by intermittent 

abstraction. Groundwater Monitoring & Remediation, DOI: 

10.1111/j.1745-6592.2010.01307.x. 

van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verbek., G.L. 




98

7 

Catalysis by accumulated deposits 


van Halem et al. (2011) Water Science and Technology: submitted


7 

Introduction 

Subsurface iron removal relies on the presence of 

(adsorbed) Fe 2+ available for oxidation during the 

injection phase (van Halem et al., 2011). The amount 

of adsorbed and/or exchangeable Fe 2+ depends, 

apart from the groundwater matrix, on the aquifer 

material composition. In general, the more adsorptive 

surface sites on the material, the more adsorbed Fe 2+ 

is available for oxidation. Adsorption/exchange surface 

sites are mostly provided by organic material, clay 

and Fe 3+ oxides. Apart form adsorptive surface sites, 

the Fe 3+ mineral surfaces are also known to catalyze 

the heterogeneous oxidation of Fe 2+ (Tamura et al., 

1976), which is relevant to subsurface iron removal. 

In addion to Fe 3+ oxides, also the presence of calcite 

has been proposed to enhance Fe 2+ immobilization 

(Mettler, 2002). Both mineral surfaces may be found in 

the aquifers and could locally enhance subsurface iron 

removal. 

Once subsurface iron removal has been in 

operation for multiple cycles, the Fe 3+ oxide mineral 

surface will increase. In the affected area around the 

tube well, the so-called oxidation zone, these Fe 3+ 

oxides have been found to accumulate (Mettler et al., 

2001; van Halem et al., 2011). Such accumulated iron 

deposits have not been reported to clog the aquifer, but 

they may have a positive effect on the immobilization 

of both Fe 2+ and As(III). An improved efficacy over 

successive cycles has already been observed at many 

full scale wells, mainly for the removal of iron. 

Subsurface arsenic removal has not shown the same 

improving trend after multiple cycles (van Halem et al., 

2010), but experiments have never lasted longer than 

a year (or 20 cycles). Investigation of arsenic retention 

during injection-abstraction cycles by accumulated 

deposits could provide information on the long-term 

performance of subsurface arsenic removal. 

Accumulated Fe 3+ oxides act as a sorbent for Fe 2+ 

during the abstraction phase, while during injection 

these mineral surfaces enhance heterogeneous 

oxidation. The amount of Fe 2+ adsorbed during 

abstraction, theoretically determines the amount of 

O 2 

consumption during injection. Thus, the more 

accumulated Fe 3+ , the more O 2 

consumption, Fe 2+ 

oxidation and subsequent Fe 2+ removal. This is only 

the case if the amount of O 2 

is not limiting and if the 

adsorbed Fe 2+ is reached by the passing O 2 

. The latter 

could be limited by the occurrence of cation exchange 

during injection, as exchangeable Fe 2+ may be released 

from the mineral surface (e.g., Ca 2+ -Fe 2+ ; Appelo et al., 

1999; van Halem et al., 2011). These complex mineralwater 

interaction reactions occur simulteously during 

surbsurface iron and arsenic removal. The objective 

of this study was to isolate these reactions in order 

to investigate the influence of the accumulated (Fe 3+ ) 

deposits on the behaviour of Fe 2+ and As(III). Column 

experiments were executed with sediments obtained 

100

7 Catalysis by accumulated deposits 

from within and outside the oxidation zone nearby a 

12-year-old iron removal well of Oasen Drinking Water 

Company (van Halem et al., 2011). Subsurface injectionabstraction 

cycles were simulated in the columns with 

natural and synthetic groundwater to investigate the 

contribution of Fe 2+ oxidation, adsorption and exchange 

on these sediments. In addition, the experiments with 

these sediments are compared to clean filter sand for 

the co-removal of As(III) during injection-abstraction 

cycles. 


Composition of sediments 

Bore holes (SonicMast with SonicDrill 2x7) were drilled 

at WTP De Put at 5m from the injection/abstraction 

wells and 50m from the wells (van Halem et al., 2011). 

The borehole at 50m was considered to be outside 

the affected oxidation zone. 800mL sediment samples 

were taken every meter at the depths from 10 to 28 

meters below ground level. Classification of the drilled 

material showed that at approximately -12m the clay 

layer ended and coarse sand layer (brownish) started. 

Greyish coarse sand was found at a depth of 22-26 m 

and the lower clay layer started at 26-27 m, showing 

that the well was into a confined aquifer. The depth of 

the perforated well filters was from 15 to 26 m. The soil 

samples were taken in September 2008, subsurface iron 

removal started in 1996 and had been operated for 12 

years at the moment of sampling. 

Samples from every meter depth were analysed 

in a certified laboratory with ICP-MS scan after 

chemical extraction with: Acid-Oxalate for amorphous 

iron and manganese; or, Nitric Acid for total iron and 

a wide range of compounds (including, Na, Mg, Al, Si, 

K, Ca, Mn, Sr, Ni, Sc, Ti, As). Acid-Oxalate extraction 

consisted of reductive dissolution of amorphous iron/ 

manganese with an oxalate buffer solution which 

consisted of 0.2M ammonium oxalate and 0.2M oxalic 

7 

Table 7.1 

Chemical composition of the sediments (A), (B) and (C) 

Sediment depth distance from well Fe am-Fe Mn Ca Na 

m m mmol.kg -1 

A 21-22 5 419 65 3.1 265 1.8 

B 23-24 5 210 24 1.8 49 1.8 

C 21.22 50 241 19 1.9 178 1.6 

101


acid (at pH=3). During digestion the sample was 

agitated and light was excluded. 

For the column studies 3 different samples were 

selected, namely (A) one from within the oxidation zone 

at 21m depth; (B) one from within the oxidation zone 

at 23m depth; and (C) one from outside the oxidation 

zone at 21m depth. The chemical composition of these 

sediments is summarized in Table 7.1. Additionally, 

clean silica filter sand (washed for 24h with 1M HCl) 

was used in one of the columns as a reference. 

Column studies 

For the experiments two different column set-ups 

were used: (i) 30cm columns and (ii) 9cm columns 

(Figure 7.1). The small columns were used for the 

experiments with specific sediment samples, as the 

sample volume was too small to fill the larger columns. 

As a reference filter sand was also used in one of these 

columns. All columns had an inner diameter of 38mm 

and were wrapped in aluminium foil to exclude light. 

The experiments were executed at a drinking water 

treatment plant, WTP Loosdrecht of Vitens Drinking 

Water Supply, with natural groundwater. During the 

pH ORP EC DO 

Waste/Sample 

Point 


7 

H 2 

O 

+ 

O 2 

Groundwater 

Dosing Pump 

N 2 

Column FS 

Column A 

Column B 

Column C 

NaAsO 2 

Figure 7.1 Experimental column set-up with filter sand column (FS) and columns with sediments A, B and C 

102


research period the groundwater had an average pH of 

7.2 (±0.01), a constant temperature of 11 °C, 0.1 mmol 

Fe.L -1 , 3.3 μmol Mn.L -1 , 1.07 mmol Ca.L -1 , 0.52 mmol 

Na.L -1 , 0.28 mmol Si.L -1 , 0.5 μmol PO 4 

3- 

.L -1 , and 0.14 

mmol SO 4 

.L -1 . 

At the start of each experiment the columns 

were conditioned with the groundwater, until complete 

breakthrough of iron occurred, and the redox potential 

stabilized. In order to simulate the injection-abstraction 

cycles of subsurface iron removal, the flow was reversed 

from up flow to down flow to start an injection phase. 

A normal injection mode consisted of injection with 

drinking water originating from the same water 

treatment plant. However, to measure the exchangeable 

Fe 2+ , injection cycles have also been performed in the 

absence of O 2 

and/or Na + . In addition to the experiments 

with natural groundwater, some experiments were 

also executed with synthetic groundwater, namely 

experiments to study the absence/presence of Fe 2+ and 

experiments to study the behaviour of As(III). These 

experiments were conducted in 30cm columns with 

filter sand and a mixture of sediment samples. The 

synthetic groundwater was produced by distributing 

demineralized water through a 6m gas stripping column 

containing stainless steel Pall Rings. From the bottom 

pure N 2 

was blown into the degassing column to sparge 

out all O 2 

. Before entering the sand columns, the water 

was checked for O 2 

(Orbisphere; HACH Lange; M1100 

Sensor; 410 Analyser) to ensure concentrations below 

1.5µmol.L -1 . Addition of stock solutions for FeSO 4 

, 

NaHCO 3 

and/or NaCl was done with a dosing pump 

followed by a static mixer. pH correction was achieved 

by addition of HCl or NaOH and all stock solutions 

were sparged with N 2 

in order to ensure the absence 

of O 2 

. The experiments were performed with synthetic 

groundwater of pH 6.9 (±0.02), a temperature of 20°C 

(±0.1), Fe concentration of 0.1 mmol.L -1 (±0.01), pH 

buffer of 5mM NaHCO 3 

and ionic strength buffer of 

1.6mM NaCl. 

The push-pull operational mode of injectionabstraction 

was simulated in the plug-flow environment 

of the columns with down flow for injection and up 

flow for abstraction (1.2 L.h -1 ±0.05). An injectionabstraction 

cycle started with 14 (±0.5) pore volumes 

of injection water to achieve >80% breakthrough of O 2 

. 

Subsequently the influent was switched to groundwater 

for multiple pore volumes to allow breakthrough of Fe 

and As. Electrical conductivity was used as a conservative 

tracer from which the pore volume could be calculated 

to be 0.04-0.05L, depending on the sediment type. 

The weight of the sediment material per column was 

0.16kg and 0.17kg for filter sand and natural sediment, 

respectively. The flow rate in the columns (2.0-2.5 m.h - 

1 

) was controlled with a multi-channel pump and PVC 

tubing with low gas permeability. Anoxic conditions 

were maintained in the columns by using an airtight 

FESTO system (6 x 1 PUN, I.D. 4mm) with matching 

connectors and valves. 

Fe and As analyses of the water samples were 

done with an Atomic Absorption Spectrometer 

103 

7


(Perkin-Elmer Flame AAS 3110; Perkin-Elmer GF- 

AAS 5100PC). In-line measurements were done for 

dissolved oxygen (Orbisphere and WTW Cellox 325), 

ORP(WTW SenTix ORP), pH (WTW SenTix 41), 

and electrical conductivity (WTW TetraCon 325). 



(C/C0)Fe 

1 

0.8 

0.6 

0.4 

0.2 

0 

filter sand 

sediment A 

sediment B 

sediment C 

0 10 20 30 40 

7 

104 

Results 

Regular injection-abstraction cycle 

Figure 7.2 shows the breakthrough curves for Fe in the 

groundwater after injection with tap water (0.28mM 

O 2 

) in the columns with the different sediments. The 

Fe breakthrough is the fastest for the clean filter sand 

column, with a retardation to C/C 0 

=0.5 of 6.5 pore 

volumes (R Fe 

=6.5). The Fe retardation factor for the 

reference sample from outside the oxidation zone, 

sediment (C), is slightly higher at 9. The presence of 

iron and manganese oxides, natural organic matter and 

clay content will enhance the Fe removal compared to 

clean filter sand. The columns filled with sand from 

inside the oxidation zone (A, B) showed better Fe 

removal efficacies with retardation factors 20 and 14, 

respectively. The column with the highest Fe content in 

the sediment (A) was most effective in the retention of 

Fe 2+ during a regular injection-abstraction cycle. The 

pore volumes 

Figure 7.2 Fe behaviour during injection-abstraction cycle in 

columns containing filter sand, sediments A, B or C 

R Fe 

of this column was 3 times higher than of the clean 

filter sand column, illustrating that the presence of 

(accumulated) iron hydroxides enhanced the removal 

of Fe. Although the Fe content in the sediment can 

account for the enhanced performance of column A, 

it does not differentiate between columns B and C; 

both columns have more or less the same Fe content. 

Between these two columns a major difference can be 

found in the amount and type of calcium mineral. The 

calcium in the sediment of column C has been identified 

as calcite (XRPD; van Halem et al., 2011), whereas the 

calcium in column B was linked to the presence of 

calcium sulphates (SEM-EDX; van Halem et al., 2011). 

Column C did not perform much better than the filter 

sand column, indicating that the presence of high 

calcite concentrations did not enhance the subsurface 

iron removal process.


O 2 

consumption 

The laminar plug flow in the columns permits only very 

little mixing of the injection water with the groundwater, 

meaning that heterogeneous oxidation will prevail, i.e., 

adsorptive-catalytic oxidation reaction. The columns 

do not simulate processes that occur on the boundary 

of the oxidation zone, but merely a fragment of soil that 

is flushed by injection water within the zone. The O 2 

in the injection water, or tap water, passes the small 

columns after a few pore volumes to allow complete 

oxidation of all Fe 2+ on the soil material. The retardation 

of O 2 

during injection shows the O 2 

consumption for 

the Fe 2+ oxidation reaction: 

Equation 7.1 


[O2] mmol.L -1 

0.4 

0.3 

0.2 

0.1 

0 

+ 0 + 

2 2 2 

0 2 4 6 8 10 

pore volumes 


sediment A 

Figure 7.3 O 2 

concentrations during an injection phase at WTP 

Loosdrecht for the filter sand and sediment A 

Figure 7.3 shows the O 2 

consumption for the column 

with filter sand and the column containing sediment 

A. Clearly, the sediment from within the oxidation 

zone consumes more O 2 

, illustrating the abundance 

of adsorbed Fe 2+ on the sediment. The more adsorbed 

Fe 2+ is oxidized during injection, the more adsorption 

sites will be available during the abstraction phase. The 

enhanced performance of sediment A has already been 

demonstrated in Figure 7.2. 

The amount of adsorbed Fe 2+ that is available 

during injection depends on both the number of 

available adsorption sites (= type of sediment; Figure 

7.2) and the saturation of this surface with Fe 2+ from 

solution. The saturation of the soil material in the 

columns with Fe 2+ has been achieved by loading the 

columns with groundwater from WTP Loosdrecht 

until complete breakthrough of Fe, and stabilization 

of pH and ORP. Only once this steady state situation 

had been reached, the columns were injected with 

tap water. In order to investigate the difference in O 2 

consumption between columns saturated with Fe 2+ and 

columns unsaturated with Fe 2+ , the filter sand columns 

were loaded with synthetic groundwater either with or 

without 0.1mM Fe 2+ . Subsequently in both columns a 

regular injection-abstraction cycle was performed with 

0.28mM O 2 

. Figure 7.4 depicts the O 2 

concentration 

during injection and abstraction in both experiments. 

O 2 

breakthrough was significantly more delayed during 

injection in the columns saturated with Fe 2+ than the 

columns without Fe 2+ , i.e., more O 2 

was consumed for 

105 

7


7 

A 


0.4 

0.3 

0.2 

0.8 

0.1 

no Fe2+ 

0.6 

Fe2+ saturated 

0.4 

0 

R Fe =7 

0.2 

0 2 4 6 8 10 

0 

in columns saturated or unsaturated with Fe 2+ 

B 

0.4 

pore volumes 

-10 -5 0 5 10 15 20 25 

no Fe2+ 

pore volumes 

0.3 

Fe2+ saturated 

Figure 7.5 Fe behaviour during injection of 0.1 M NaCl and 

subsequent abstraction of natural groundwater 

0.2 

(sediment C) 

0.1 

0 

0 2 4 6 8 10 

pore volumes 

Figure 7.4 O 2 

behaviour during (A) injection and (B) abstraction 


adsorbed Fe 2+ oxidation. During abstraction (Figure 

7.4B) more O 2 

was observed to leave the column with 

no Fe 2+ , which corresponds to the finding that, in the 

absence of adsorbed Fe 2+ , less O 2 

will be consumed. It 

is noteworthy that O 2 

can be found in the abstracted 

water for nearly 10 pore volumes. 

injection phase 

C/C0 

1.6 

1.4 

1.2 

1 

abstraction phase 

Exchangeable Fe 2+ 

The Fe 2+ exchange on the material in the columns can 

be calculated from the Fe retention measurements after 

injection with a solution of 0.1M NaCl, in the absence of 

O 2 

. During such an injection phase, abundant Na + will 

be available to exchange with the exchangeable Fe 2+ on 

the sand material. The amount of exchangeable Fe 2+ on 

the sand material can be estimated either from the Fe 2+ 

leaching from the columns during injection, or from 

the Fe 2+ retention during abstraction. It is, however, 

practically almost impossible to measure the relatively 

short Fe 2+ peak passing after injection of 1 pore volume. 

The sample size (30mL) dilutes the initial peak and the 

Fe concentration in the first sample will therefore be 

lower than the actual Fe peak. During abstraction the 

breakthrough process is much slower, making it feasible 

106


Table 7.2 

Retardation factor, Fe retention and exchangeable Fe 2+ for filter sand and sediments A, B and C 

Column Retardation factor Fe retained Exchangeable Fe 2+ 

0.1M NaCl tap water mM Fe mM Fe.kg -1 meq.kg -1 

filter sand 7 6 0.04 0.20 0.40 

A 14 14 0.07 0.41 0.83 

B 10 11 0.05 0.30 0.60 

C 7 7 0.04 0.22 0.44 

to accurately measure the total amount of Fe retained in 

the columns. Figure 7.5 depicts the Fe measurements 

during injection (negative pore volumes) and during 

abstraction (positive pore volumes) for column C. The 

retardation factor for this cycle was 7 and the total 

amount of retained Fe was calculated to be 0.04mM, 

corresponding to 0.22mM.kg -1 of sediment. The sites 

available for exchangeable Fe 2+ on this particular 

sediment with this particular groundwater type can be 

calculated to be 0.44 meq.kg -1 . 

It should be noted that the experiments 

were executed with natural groundwater, thus the 

exchangeable Fe 2+ on the sediments are limited by the 

presence of other cations, such as Ca 2+ . An overview of 

the retained Fe in the other columns is given in Table 

7.2, both for injection of O 2 

-free tap water and 0.1M 

NaCl demineralized water. The table shows that Fe 2+ 

exchange was higher in sediment (A) from within 

the oxidation zone, reaching up to 0.83 meq.kg -1 . The 

presence of high iron oxides is known to enhance Fe 2+ 

exchange (Appelo and Postma, 2005). The presence of 

calcite did not have the same effect on the exchangeable 

Fe 2+ fraction, as both clean sand and calcite-containing 

sediment (C) have the same Fe 2+ retention in the 

columns. 

Fe 2+ oxidation 

In order to differentiate between a catalytic oxidation 

reaction and the occurrence of cation exchange, the 

columns have been loaded with O 2 

-containing water, 

either with or without 0.1M NaCl. The injectionabstraction 

cycle without 0.1M NaCl simulates the 

oxidation-adsorption process during subsurface iron 

removal without the interference of cation exchange. 

The results in Figure 7.6 show Fe retardation factors of 

6, 13, 16 and 17 for the filter sand, and sediments C, 

B, A, respectively. R Fe 

for the filter sand, and sediments 

C, B, A, were measured to be 10, 13, 22 and 31 after 

injection with a solution containing 0.1M NaCl and 

0.28mM O 2 

. These retardation factors are higher 

than the regular injection-abstraction cycle with O 2 

- 

containing tap water for all sediments, except sediment 

107 

7


7 

108 

RFe 

35 

30 

25 

20 

15 

10 

5 

0 


0.28mM O2 - no NaCl 

0.28mM O2 - 0.1M NaCl 

Sediment A Sediment B Sediment C 

Figure 7.6 Fe retardation factors after injection with 

demineralised water and 0.1M NaCl solution, both 

containing 0.28mM O 2 

C. This sediment originates from outside the oxidation 

zone and shows less affinity for cation exchange during 

subsurface iron removal than the other sediments. 

However, the exchangeable Fe 2+ of sediment C was the 

same as for filter sand, whereas the column with filter 

sand performed better at higher Na concentrations. An 

explanation for this deviation may be the fact that the 

clean filter sand did not contain any catalyzing material, 

such as calcite. Sediment C contained significant calcite 

content which may have caused rapid heterogeneous 

Fe 2+ oxidation/incorporation, and therefore limiting 

the potential role of cation exchange. Sediments A and 

B have a higher exchangeable Fe 2+ fraction, resulting 

in better performance after injection of high Na 

concentrations. 

(C/C0)As 

1.0 

0.8 

0.6 

0.4 

0.2 

0.0 

0 10 20 30 40 50 60 70 80 

pore volumes 

filter sand 1 

filter sand 2 

sediments 1 

sediments 2 

Figure 7.7 As(III) behaviour during an injection-abstraction 

cycle in duplicate columns with filter sand and a 

sediment mixture from within the oxidation zone 

Co-removal of As(III) 

A mixture of the sediments A and B was used to 

study the behaviour of As during regular injectionabstraction 

cycles. Before injection with 0.28mM O 2 

solution, the columns were saturated with synthetic 

groundwater, containing 0.1mM Fe 2+ and 2µM As(III). 

Figure 7.7 depicts the results for arsenic measurements 

during the abstraction phase. The behaviour of arsenic 

is very similar for the columns containing clean filter 

sand or a mixture of sediments A and B. Clearly, the 

breakthrough of arsenic is not delayed by sediments 

from within the oxidation zone compared to the clean 

filter sand. Arsenic breakthrough starts immediately 

upon abstraction and C/C 0 

=0.5 is reached at 5 to 

10 pore volumes. Based on these results subsurface 

arsenic removal does not seem to be enhanced by iron 

deposits nearby subsurface removal wells. Also, the


enhanced O 2 

consumption (Figure 7.3) in the columns 

does not translate in better arsenic retention. The 

adsorptive capacity of the sediments was insufficient 

to significantly enhance the As removal, which shows 

that while subsurface iron removal may benefit from 

accumulated deposits (Figure 7.5), the same does not 

apply to As(III). 

Conclusions 

Sediments obtained inside the oxidation zone of a 

12-year-old subsurface iron removal well showed 

better Fe 2+ retention during injection-abstraction 

cycles in the column environment than clean filter 

sand or sediments from outside the oxidation zone. 

This finding was supported by the fact that more 

O 2 

was consumed during injection in columns with 

sediments from within the oxidation zone. In addition, 

the exchangeable Fe 2+ fraction for sediments was twice 

as high as for sediments that had been unaffected by 

subsurface iron removal, resulting in a doubling of the 

Fe retardation factor. The same cannot be concluded 

for As(III) retardation, as sediments from within the 

oxidation zone did not show better removal than clean 

filter sand. 

References 



Ground Water 37(6): 811-817. 







aquifer, Water Supply, 3(1): 13-18. 











Mettler S., M. Wolthers, L. Charlet and U. von Gunten (2009) 

Sorption and catalytic oxidation of Fe(II) at the surface of 

calcite. Geochimica et Cosmochimica Acta 73: 1826-1840. 



3(2): 143-150. 



109 

7




treatment in the Netherlands, Water Supply 3: 1-11. 

van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verbek, G.L. 



Bangladesh. Water Research (44): 5761-5769. 

van Halem D., W. de Vet, J. Verberk, G. Amy and H. van Dijk 

(2011) Characterization of accumulated precipitates during 

subsurface iron removal. Applied Geochemistry 26: 116–124. 

van Halem, D., D.H. Moed, J. Q.J.C. Verberk, G.L. Amy and 

J.C. van Dijk (2011) Cation exchange during subsurface iron 

removal. Water Research: under review. 

7 

110

8 

Influence of groundwater 

composition 


van Halem et al. (2011) Science of the Total Environment: submitted


8 

112 

Introduction 

In Bangladesh, elevated iron and arsenic concentrations 

have been found to often co-occur in anoxic and 

anaerobic groundwater (Nickson et al., 2000). The 

presence of both iron and arsenic are a prerequisite for 

the operation of Subsurface Iron and Arsenic Removal 

(SIR/SAR; van Halem et al., 2010)). The injection of 

aerated water promotes the oxidation of Fe 2+ , resulting 

in the formation of iron oxides nearby subsurface 

treatment wells. More water with reduced iron and 

arsenic concentrations can be abstracted (volume V) 

than was injected (volume Vi), i.e., this volumetric 


The deposited iron oxides provide adsorptive surface 

area for both Fe 2+ and arsenic during abstraction (Table 

8.1). The adsorptive capacity of iron oxides has been 

found to be limited by the presence of other inorganic 

ions, such as phosphate, silicate and bicarbonate 

(goethite; Stachowicz, 2008). Also Fe 2+ adsorption 

and precipitation is affected by the presence of other 

inorganic ions, including calcium (Sharma, 2001; 

Ciardelli et al., 2006). The majority of these compounds 

are commonly found in Bangladeshi ground- and 

drinking water (BGS/DPHE, 2001; Ciardelli et al., 2006). 

It is hypothesized that the groundwater composition 

will greatly influence the site-specific effectiveness of 

SIR/SAR under diverse geochemical conditions. 

There have been several pilot studies to 

investigate the potential of SAR, with various results. In 

Bangladesh three small-scale injection facilities (0.5m 3 ) 

were constructed with a plate aerator, allowing injection 

of oxygen-containing water with 0.16 mmol.L -1 (Sarkar 

and Rahman, 2001). Arsenic concentrations were 

found to be reduced from 1.5 µmol.L -1 to below 0.67 

µmol.L -1 (national guideline; 50 µg.L -1 ) until V/Vi 

reached 4. At higher arsenic concentrations, namely 6.7 

µmol.L -1 and 17.0 µmol.L -1 , the reduction of arsenic did 

not reach below 2.7 µg.L -1 and 6.7 µg.L -1 , respectively. 

Across the Indian border in the same Bengal Delta, 

Sen Gupta and co-workers (2009) reported that at 

6 different community SAR plants a V/Vi ratio of 4 

to 6 could be operated providing arsenic-free water 

to the people. Background arsenic levels were not 

reported in this publication, but the surrounding wells 

Table 8.1 

As(III), arsenite 

As(V), arsenate 

Fe 2+ , ferrous iron 

Adsorption reactions and constants for As(V), As(III) 

and Fe 2+ onto the iron oxide goethite (Dixit and 

Hering, 2003; Dixit and Hering, 2006) 

Reactions 

S − OH + AsO + 3H →S − H AsO + H O 

S − OH + AsO + H →S − HAsO + H O 

S − OH + AsO 

S − OH + AsO 

3− + 

3 2 3 2 

3− + − 

3 

2 

3 2 

3− 

4 

2 

3− 

4 

− + → − + 

+ 

+ H → S − HAsO4 

+ H 

2O 

+ 

2− 

+ H → S − AsO + H O 

2+ + + 

S OH Fe S OFe( II ) H 

2+ + 

S − OH + Fe + H 

2O →S − OFe( II ) OH + 2H 

4 

− 

2

8 Influence of groundwater composition 

had concentrations exceeding 6.7 µmol.L -1 . Arsenic 

removal during subsurface treatment has also been 

observed by Rott and co-workers (2002), reducing 

arsenic concentrations from 0.5 µmol.L -1 to below 0.13 

µg.L -1 after 20 cycles. Van Halem et al. (2010) found 

less encouraging results at sites in Bangladesh with 1.9 

µg As(III).L -1 , with immediate arsenic breakthrough 

upon abstraction (Vi=1m 3 ). These pilot studies have 

indicated that there is potential for the technology 

of SAR, nevertheless at some sites the efficacy seems 

to be much better than at others. Unfortunately, only 

very little is reported on the soil and groundwater 

composition at these field sites, making it difficult to 

extract information on the influence of limiting and/ 

or catalyzing compounds. Appelo and de Vet (2003) 

reported the behaviour of a wide range of groundwater 

compounds during subsurface treatment, proposing 

that phosphate was responsible for sudden arsenic peaks 

in the abstracted water. The arsenic concentrations 

at the studied site were, however, many times lower 

(80% breakthrough of dissolved O 2 

. Subsequently the 

influent was switched to groundwater to allow retention 

of Fe 2+ and As(III). Electrical conductivity was used 

as a conservative tracer from which the pore volume 

could be calculated to be, on average, 0.12 L (±0.002). 

The flow rate in the columns (2.7 m.h -1 ) was controlled 

with a multi-channel pump and PVC tubing with low 

gas permeability. Anoxic conditions were maintained 

113 

8


H 2 

O 

pH ORP EC DO 

Waste/Sample 

Point 

N 2 

Stainless steel Pall Rings 

H 2 

O 

+ 

O 2 


Dosing Pump 

Column 2 

Column 1 

Fe/As 

NaHCO 3 

Figure 8.1 Schematic overview of experimental column set-up 

8 

114 

in the columns by using an airtight FESTO system (6 

x 1 PUN, I.D. 4 mm) with matching connectors and 

valves. All injection-abstraction experiments were 

performed twice in the duplicate columns, with freshly 

filled columns before each experiment. At the start of 

each experiment the columns were conditioned with 

synthetic groundwater (containing the compounds 

of interest), until complete breakthrough of iron 

occurred, and the Eh potential stabilized. An injection 

mode consisted of demineralized water containing 

a pH buffer (5mM NaHCO 3 

) and 0.28mM O 2 

. The 

abstraction phase consisted of synthetic groundwater 

with pH 6.9, a temperature of 20°C, 0.07mM Fe 2+ , 2.7µM 

As(III), pH buffer (5mM NaHCO 3 

) , ionic strength 

buffer (1.6mM NaCl) and additionally the inorganic 

compound of interest (phosphate, silicate, nitrate, 

calcium or manganese). The chemicals (reagent grade, 

Sigma-Alldrich) were dosed as FeSO 4 

.7H 2 

O, FeCl 2 

, 

NaAsO 2 

, NaHCO 3 

, NaCl, Na 2 

HPO 4 

, Na 2 

SO 3 

.5H 2 

O, 

NaNO 3 

, Ca(NO 3 

) 2 

.4H 2 

O and MnSO 4 

.H 2 

O.


The synthetic groundwater was produced by 

distributing demineralized water through a 6m gas 

stripping column containing stainless steel Pall Rings. 

From the bottom pure N 2 

was blown into the degassing 

column to sparge out all O 2 

. Before entering the sand 

columns, the water was checked for O 2 

(Orbisphere; 

HACH Lange; M1100 Sensor; 410 Analyser) to ensure 

concentrations below 1.5 µmol.L -1 . Addition of stock 

solutions was done with a dosing pump followed by a 

static mixer. pH correction was achieved by addition of 

HCl or NaOH and all stock solutions were sparged with 

N 2 

in order to ensure the absence of O 2 

. 

Iron analysis of the water samples was done 

with an Atomic Absorption Spectrometer (Perkin- 

Elmer Flame AAS 3110). Arsenic analysis with 

Graphite Furnace Atomic Absorption Spectrometer 

(Perkin-Elmer 5100PC) with an electrode discharge 

lamp (EDL) and a Ni(NO 3 

) 2 

.6H 2 

O matrix modifier. 

C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

R Fe = 42 

0 20 40 60 80 

pore volumes 

Fe Column 1 

Fe Column 2 

oxygen 

tracer 

Figure 8.2 Behaviour of O 2 

, tracer and Fe 2+ during the reference 

injection-abstraction cycles 

In-line measurements were done for dissolved oxygen 

(Orbisphere and WTW Cellox 325), E h 

potential (WTW 

SenTix ORP), pH (WTW SenTix 41), and electrical 

conductivity (WTW TetraCon 325). Measurements 

were registered on a computer with Multilab Pilot v5.06 

software. 

Results 

Fe 2+ -O 2 

system 

A regular injection phase during full scale subsurface 

iron or arsenic removal consists of the injection of 

aerated water, in most cases drinking water from the 

clean water reservoir. In the experiments described in 

the paper, demineralized water was used for injection 

containing a pH buffer (5mM NaHCO 3 

) and a 0.28mM 

O 2 

concentration. After injection, the influent was 

switched to synthetic groundwater for the abstraction 

phase, containing 0.07mM Fe 2+ , 2.7µM As(III), pH 

buffer (5mM NaHCO 3 

) and ionic strength buffer (1.6 

mM NaCl). The measurements in Figure 8.2 illustrate 

that once abstraction has started first the tracer passed 

the column (electrical conductivity), with a C/C 0 

at 

1 pore volume. The initial oxygen concentration of 

0.28mM was pushed out of the column and reached 


8 

116 

the delay of Fe front arrival at the well compared to 

the injection water front. The more water with low Fe 

concentrations can be abstracted from the well/column, 

the better the efficacy of SIR. 

In general, groundwater treatment plants 

operate a V/Vi ratio based on the moment Fe starts to 

arrive at the well, thus C/C 0 

>0. In the experiments, Fe 2+ 

is allowed to breakthrough, in order to calculate the 

dimensionless retardation factor R. R Fe 

-1 is calculated 

from a Vi corresponding to the pore volume when the 

tracer (electrical conductivity) is C/C 0 

= 0.5, and V is 

the number of pore volumes that can be abstracted with 

iron concentrations below (C/C 0 

) Fe 

=0.5. The reference 

injection-abstraction cycles in Figure 8.2 show a R Fe 

of 

42, meaning that 43 times more water with [Fe]


Clearly, the presence of Fe 2+ in the changing redox 

conditions in the column improved As(III) retention. It 

should be noted that, unlike for Fe, the As breakthrough 

starts almost immediately upon abstraction. The As 

breakthrough showed extensive tailing, resulting in 

a high R As 

. The tailing during As(III) breakthrough 

may indicate incomplete adsorption kinetics in the 

columns, which is not surprising as residence times 

of the synthetic groundwater in the column was no 

more than 10 minutes (empty bed contact time). Such 

residence times correlate well with actual retention 

times in small-scale applications of SIR/SAR (Sarkar 

and Rahman, 2001; van Halem et al., 2010), but reaching 

As(III) adsorption equilibrium on iron oxides may take 

longer than that (Raven et al., 1998). Irregular operation 

of hand pump wells may increase the residence time 

of groundwater in the oxidation zone, potentially 

increasing removal efficacies. Additionally it should 

be taken into account that although As breakthrough 

is observed immediately, the arsenic concentration 

passed the WHO and Bangladeshi guideline at 4 and 

8 pore volumes, respectively. An SIR/SAR efficiency 

of V/Vi=4 without the need of any operational costs 

(chemicals, electricity) may well be very effective for 

household application. 

Phosphate 

The results in Figure 8.2 and Figure 8.3 of the reference 

injection-abstraction cycles in the Fe 2+ -As(III)-O 2 

system show promising results for the household 

application of SIR/SAR. Nevertheless, the presence 

of other inorganic groundwater constituents may 

interfere, reducing Fe and As removal efficacies. In 

Bangladesh, PO 4 

3- 

concentrations are measured in 

the shallow aquifer between 0.01mM PO 4 

3- 

at pH>7 inhibits the precipitation of Fe 3+ (Ciardelli et 

al., 2008; Guan et al., 2009). The adsorbed H 2 

PO 4 

- 

and 

HPO 4 

2- 

maintains the iron oxide surfaces as negatively 

charged preventing the particles from hydrolyzing 

and subsequently precipitating. Sharma et al. (2001) 

observed reduced adsorptive capacity for Fe 2+ of virgin 

sand in the presence of 0.01mM PO 4 

3- 

, 23% at pH 

6.8. Nevertheless, the results in Figure 8.4 illustrate 

that subsurface iron removal in the sand columns is 

not affected by the presence of PO 4 

3- 

, Fe retardation 

C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

R As = 4 

R Fe = 44 

0 20 40 60 80 

pore volumes 

Fe Column 1 

As Column 1 

Fe Column 2 

As Column 2 

Figure 8.4 Fe and As retardation in the presence of 0.01 mM 

PO 4 

3- 

117 

8


8 

118 

remains almost the same at 44. This may be explained 

by the fact that the sand grains (partially) retained the 

non-hydrolysed iron oxide particles through filtration, 

adsorption or exchange, whereas the batch experiments 

from the literature did not include filtration. 

The effect of phosphate on the removal of As(III) 

during subsurface arsenic removal is two-fold, namely 

through (i) inhibition of Fe 2+ oxidation-precipitation 

and (ii) competitive adsorption. Competitive 

adsorption between As(III) and PO 4 

3- 

onto iron oxides 

has been studied in detail (Jain and Loeppert, 2000; 

Ciardelli et al., 2007; Stachowicz et al., 2008), making 

PO 4 

3- 

the most important competing anion for arsenic 

(Su and Puls, 2001; Guan et al., 2009). As(III) behaviour 

during oxidation and precipitation of soluble Fe 2+ was 

studied by Ciardelli et al. (2008), reporting an As(III) 

removal reduction from 73% to 63% in the presence 

of 0.07mM PO 4 

3- 

-P at pH 7.2. The combination of its 

ability to occupy the same adsorption sites as arsenic 

and its elevated concentrations compared to trace 

amount of arsenic makes PO 4 

3- 

a strong competitor in 

natural waters. This observation correlates well with 

the results from the column experiments (Figure 8.4), 

showing that As retardation was reduced from 30 to 4 

in the presence of 0.01mM PO 4 

3- 

. 

Silicate 

Silicon concentrations of groundwater in Bangladesh 

are commonly high, with median concentrations in the 

shallow aquifer of 0.71 mmol.L -1 (BGS/DPHE, 2001). 

C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

R As = 5 

R Fe = 57 

0 20 40 60 80 100 

pore volumes 

Fe Column 1 

As Column 1 

Fe Column 2 

As Column 2 

Figure 8.5 Fe and As retardation in the presence of 0.2mM SiO 2 

Silicon, or in groundwater found as silicate, is known 

to interact with Fe 3+ to form soluble polymers and higly 

dispersed colloids (Iler, 1979; Meng et al., 2000) and 

hinder the Fe 3+ iron oxide precipitation and aggregation 

(Lui et al., 2007a; Lui et al., 2007b). The negative effect 

of silicate on the precipitation of iron oxides is greater 

at higher pH (7-9). This dispersed colloidal material 

cannot be removed by filtration (Davis et al., 2001), 

so it may be found to pass the injection-abstraction 

columns. Fe 2+ adsorption onto virgin sand has also 

been reported to be inhibited by the presence of silicate, 

reducing by 14 and 27% in the presence of 10 and 1.4 

mmol.L -1 Si, respectively (Sharma, 2001). The results in 

Figure 8.5 do not show a clear inhibition of iron removal 

after the addition of 0.2mM SiO 2 

, the retardation factor 

R Fe 

is even higher at 57. It is, however, noteworthy 

that the iron concentrations rise immediately after 

one pore volume to C/C 0 

=0.06, which correlates to 

an approximate 5.4 µM Fe.L -1 . This instant increase


in iron concentration has not been observed in the 

absence of silicate and continues until 40 pore volumes. 

It is proposed that the formation of mobile colloidal 

material prevents the adsorption of this iron fraction 

onto the column sand, resulting in measurements of 

total iron after the columns. 

Below pH 7 silicate is found to occur as H 2 

SiO 3 

, 

whereas at higher pH it may be found as HSiO 3- 

. At 

higher pH the affinaty of the negatively charged silicate 

for adsorption onto iron oxides increases, resulting in 

more competition with arsenic (Meng et al., 2000; Guan 

et al., 2009). Ciardelli et al. (2008) found a reduction 

in As(III) removal from 73% to 67% during oxidationprecipiation 

of Fe 2+ after addition of 1.1 mmol.L -1 . 

Figure 8.5 illustrates that subsurface arsenic removal 

is negatively affected by the presence of 0.2mM SiO 2 

. 

The retardation factor R As 

was reduced from 30 to 65 

which is very much comparable to the impact of 0.01 

mM PO 4 

3- 

. 

Nitrate 

In anaerobic groundwater, an NO 3- 

-reducing 

environment, one expects to find dissolved NH 4 

+ 

rather 

than NO 3- 

. Nevertheless, the aeration and storage of 

anaerobic groundwater for the purpose of injection 

may lead to nitrification of NH 4 

, resulting in elevated 

NO 3 

- 

concentrations. Additionally, in Bangladesh, 

shallow aquifers have been measured to contain nitrate 

concentrations up to 1.2 mM (Majumder et al., 2008). 

The effect of 1mM NO 3 

- 

in the synthetic groundwater 

C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

R As = 4 

R Fe = 47 

0 20 40 60 80 100 

pore volumes 

Fe Column 1 

As Column 1 

Fe Column 2 

As Column 2 

Figure 8.6 Fe and As retardation in the presence of 1mM NO 3 

- 

on the iron and arsenic retardation was studied during 

the injection-abstraction cycles in the columns. Figure 

8.6 shows that Fe removal did not differentiate much 

compared to the reference cycle, with a slightly higher 

R Fe 

of 47. Previous studies showed that addition of 

NaNO 3 

had no effect on Fe 2+ adsorption onto virgin 

sand and iron oxides (Hayes and Leckie, 1987; Dzombak 

and Morel, 1990; Sharma, 2001). 

Su et al. (2001) investigated the influence of 

nitrate on As(III) removal and found that when adding 

1 mM NaNO 3 

to a solution containing As(III) and 

zerovalent iron oxides, As(III) removal was reduced 

by 55%. Push-pull column studies by Johnston (2008), 

where a 0.1M NaNO 3 

ionic strength buffer was used, 

showed As(III) retardation of R As 

=6-10 at pH 7 with 

goethite coated sand columns. The R As 

in that study was 

significantly lower than the reference cycles presented 

in Figure 8.3, but corresponds more or less to the R As 

in Figure 8.6. In the presence of 1mM NO 3 

- 

the arsenic 

119 

8


8 

120 

retardation factor was reduced from 30 to 4, illustrating 

the significant effect of this constituent on subsurface 

arsenic removal. 

Calcium 

One of the major cations in groundwater is calcium, 

especially in the calcium-carbonate groundwaters of 

Bangladesh (Ciardelli et al., 2007). Ca 2+ concentrations 

are generally much higher than Fe 2+ concentrations, 

with a median and maximum in Bangladesh of 

0.88mM and 9.1mM, respectively (BGS/DPHE, 2001). 

Competition/exchange between these divalent cations 

during SIR may therefore be expected, with Ca 2+ being 

in favour because of its concentration advantage. A 

reduction in Fe 2+ adsorption onto virgin and iron oxide 

coated sand of 42% and 12%, respectively, was observed 

in the presence of 1.2mM Ca 2+ at pH 6.8 (Sharma, 2001). 

Calcium was added in this column study as Ca(NO 3 

) 2 

, 

resulting in 1.2mM Ca 2+ , but also 2.4mM NO 3- 

. The 

C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

R As = 4 

R Fe = 15-24 

0 10 20 30 40 

pore volumes 

Fe Column 1 

As Column 1 

Fe Column 2 

As Column 2 

Figure 8.7 Fe and As retardation in the presence of 1.2mM Ca 2+ 

previous experiment illustrated that NO 3 

- 

had almost 

no effect on Fe removal during SIR, but there was a 

strong effect on As removal. Therefore conclusions can 

be drawn based on the results presented in Figure 8.7 

regarding the influence of Ca 2+ on Fe removal, but not 

on the removal of As. Stachowicz et al. (2008) did not 

observe an effect of Ca 2+ on As(III) adsorption onto 

goethite. It is shown in Figure 8.7 that the R As 

is almost 

the same after addition of Ca(NO 3 

) 2 

as after addition 

of NaNO 3 

, thus not providing interesting information 

on the effect on Ca 2+ . The R Fe 

was, however, clearly 

lower than in earlier injection-abstraction cycles, 

with a variation between 15 and 24 for the duplicate 

experiments. The variation between the duplicate cycles 

cannot be explained at this stage, but the reduced R Fe 

confirms that the presence of Ca 2+ inhibited the removal 

of Fe 2+ during SIR. Fe concentrations start to rise in the 

water after approximately 5 to 8 pore volumes, resulting 

in an operational V/Vi ratio of 5 to 8, which is much 

lower than the V/Vi=25 of the reference cycle. 

Manganese 

Subsurface iron removal has frequently been associated 

with co-removal of manganese (Hallberg and Martinell, 

1976; van Beek 1985; Rott, 1985). Mn 2+ oxidation is 

slower than abiotic Fe oxidation and it is proposed that 

manganese removal does not start in the subsurface 

until iron removal is complete. This results in a spatial 

separation between the formation of manganese and 

iron oxides, with the latter further away from the


C/C0 

1 

0.8 

0.6 

0.4 

0.2 

0 

Fe Column 1 

As Column 1 

Fe Column 2 

As Column 2 

R As = 28 

0 10 20 30 40 50 

pore volumes 

R Fe = 34 

Figure 8.8 Fe and As retardation in the presence of 0.06 mM 

Mn 2+ 

well. Mn 2+ oxidation is known to be catalyzed in the 

presence of Mn 4+ oxides, and also As(III) oxidation 

and adsorption is suggested to be enhanced by Mn 4+ 

oxides. Nevertheless, Mn 2+ is not expected to compete 

with Fe 2+ for the injected O 2 

in the columns, as the Mn 2+ 

oxidation kinetics are much slower and the contact time 

in the columns is relatively short. Mn 2+ may, however, 

inhibit the Fe 2+ adsorption. Sharma (2001) observed a 

reduction in Fe 2+ adsorption in the presence of 0.03mM 

Mn onto virgin and iron coated sand with 44% and 6%, 

respectively. 

During injection-abstraction cycles with 

0.06mM Mn 2+ , the Fe removal was slightly lower 

compared to the reference cycles, with a R Fe 

of 34 

(Figure 8.8). The moment that Fe 2+ was detected did 

not differ from the reference cycles, showing that the 

operational V/Vi ratio is not affected by the presence of 

0.06mM Mn. As(III) behaviour remained the same in 

the presence of soluble Mn 2+ , resulting in R As 

=28. 

Conclusions 

It was found that during injection-abstraction cycles 

in the Fe 2+ -As(III)-O 2 

column system it was possible 

to reach very effective iron and arsenic retardation, 

though arsenic breakthrough started immediately 

upon abstraction. Arsenic removal was found to be 

significantly inhibited by the presence of 0.01mM 

phosphate, 0.2mM silicate, and 1mM nitrate, illustrating 

the vulnerability of this Subsurface Arsenic Removal 

technology in diverse geochemical settings. Subsurface 

Iron Removal was found not be as sensitive to other 

inorganic groundwater compounds, though iron 

retardation was found to be limited by 1.2mM calcium 

and 0.06mM manganese. Total iron removal was 

enhanced by the presence of 0.2mM silicate, but low 

iron concentrations (5µM) were found to breakthrough 

immediately upon abstraction. 

References 





Ciardelli M.C., H. Xu and N. Sahai (2008) Role of Fe(II), 

phosphate, silicate, sulfate, and carbonate in arsenic uptake 

by coprecipitation in synthetic and natural groundwater 

121 

8


8 

Water Research 42(3): 615-624. 

Davis C.C., W.R. Knocke and M. Edwards (2001) Implications 

of aqueous silica sorption to iron hydroxide: mobilization of 

iron colloids and interference with sorption of arsenate and 

humic substances. Environ. Sci. Technol. 35: 3158-3162. 

Dixit S. and J. G. Hering (2003) Comparison of arsenic(V) and 

arsenic(III) sorption onto iron oxide minerals: implications 

for arsenic mobility. Environ. Sci. Technol. 37: 4182-4189. 



Geology, 228(1-3): 6-15. 

Dzombak D.A. and F.M.M. Morel (1990) Surface complexation 

modelling – hydrous ferric oxides. John Wiley & Sons. 

Guan X., H. Dong, J. Ma and L. Jiang (2009) Removal of arsenic 

from water: effects of competing anions on As(III) removal in 

KMnO 4 

-Fe(II) process. Water Research 43: 3891-3899. 


purification of groundwater. Ground Water 14(2): 88-93. 

Hayes K.F. and J.O. Leckie (1987) Modelling ionic strength effects 

on cation adsorption ad hydrous oxide/solution interfaces. 

Journal of Colloid and Interface Science (115): 564-572. 

Iler R. K. (1979) The Chemistry of Silica. Wiley-Interscience, 

New York. 

Johnston R. (2008) Chemical interactions between iron and 

arsenic in water, PhD dissertation, University of North 

Carolina at Chapel Hill. 

Liu R.P., J.H. Qu, S.J. Zia, G.S. Zhang and G.B. Li (2007a) Silicate 

hindering in situ formed ferric hydroxide precipiation: 

inhibing arsenic removal from water. Environ. Eng. Sci. 24: 

707-715. 

122 

Liu R.P., X. Li, S.J. Zia, Y.L. Yang, R.C. Wu and G.B. Li (2007b) 

Calcium-enhanced ferric hydroxide co-precipitation of 

arsenic in the presence of silicate. Water Environ. Res. 79: 

2260-2264. 

Majumder R.K., M.A. Hasnat, S. Hossain, K. Ikeue, M. 

Machida (2008) An exploration of nitrate concentrations in 

groundwater aquifers of central-west region of Bangladesh. 

Journal of Hazardous Materials 159(2-3): 536-543. 

Meng X., S. Bang and G.P. Korfiatis (2000) Effects of silicate, 

sulfate, and carbonate on arsenic removal by ferric chloride. 

Water Research 34(4): 1255-1261. 

Nickson R., J. McArthur, P. Ravencroft, W.G. Burgess and K.M. 

Ahmed (2000) Mechanism of arsenic release to groundwater, 

Bangladesh and West Bengal, Applied Geochemistry 15(4): 

403-413. 

Raven K.P., A. Jain and R.H. Loeppert (1998) Arsenite and 

arsenate adsorption on ferrihydrite: kinetics, equilibrium, 

and adsorption envelopes. Environ. Sci. Technol. 32: 344-349. 



3(2): 143-150. 









arsenic removal method for community, Environmental


Pollution 157: 3351–3353. 

Sharma, S.K. (2001) Adsorptive iron removal from groundwater. 

PhD dissertation, Wageningen University. 

Stachowicz M., T. Hiemstra and W. H. Van Riemsdijk (2008) 

Multi-competitive interaction of As(III) and As(V) oxyanions 

with Ca 2+ , Mg 2+ 3− 2− 

, PO 4 

, and CO 3 

ions on goethite, Journal of 

Colloid and Interface Science 320, 400–414. 

Su C., and R.W. Puls (2001) Arsenate and arsenite removal by 

zerovalent iron: effects of phosphate, silicate, carbonate, 

borate, sulphate, chromate, molybdate, and nitrate, relative 

to chloride. Environmental Science and Technology 

35(22):4562-4568. 



van Halem, D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk, 

G.L. Amy and J.C. van Dijk (2010) Subsurface iron and 

arsenic removal for shallow tube well drinking water in rural 


8 

123


8 

124

Concluding remarks9


9 

126 

Main conclusion 

In this concluding chapter the results from this thesis 

are combined to address the main problem definition 

as stated in Chapter 1. The combination of sand column 

studies, field experiments, sediment characterization 

and past experiences in the Netherlands have gained 

knowledge on the applicability of SIR/SAR for smallscale 

drinking water treatment. The considerations per 

knowledge gap have contributed to the formulation of 

the main conclusion of this thesis: 

The results presented in this thesis have shown 

that Subsurface Iron Removal is a safe and 

sustainable small-scale drinking water treatment 

solution. Subsurface Arsenic Removal has been 

found to be less suitable for hand pump application 

and vulnerable to diverse geochemical conditions. 

This conclusion underlines the great potential of SIR 

for household water treatment and storage (HWTS), 

but also points out the variable performance of SAR. 

This two-fold conclusion does not mean that the 

application of SAR should no longer be considered a 

safe option, but indicates that this novel technology is 

no silver bullet for arsenic mitigation. The efficacy of 

SAR is sensitive to geochemical variations, which is 

in conflict with the decentralized character of HWTS. 

This vital observation, however, applies not just for SAR 

but for the majority of (arsenic) HWTS solutions. In 

general, no household water treatment option has been 

designed to suit all conditions, illustrating the urgent 

need to investigate the (technical) boundary conditions 

in which a HWTS solution can succeed. In this thesis 

the efficacy of SIR/SAR has been described as the ratio 

between the injected volume (Vi) and the abstracted 

volume (V). The more water that can be abstracted 

with reduced iron and arsenic, the lower the frequency 

of injection and the higher the efficacy. This thesis has 

shown that removal of iron was more encouraging 

than for arsenic, meaning that for SAR operation users 

need to inject more frequently (depending on the 

groundwater composition). It therefore depends on 

the users’ willingness to inject whether SIR/SAR can be 

considered an appropriate and safe alternative to current 

arsenic mitigation solutions. The SIR/SAR technology 

has many socio-economic advantages (Chapter 1), 

including the removal of iron. The odour, taste and 

colour of the drinking water are significantly improved 

by SIR/SAR, which could potentially increase the social 

acceptance and willingness to operate the technology 

at low V/Vi ratios. A study on the socio-economic 

feasibility of this technology, within the formulated 

technological boundary conditions, is therefore highly 

recommended. 

The three identified knowledge gaps in Chapter 1 

are addressed in this concluding chapter: (1) Fe 2+ /As(III) 

immobilization processes; (2) site-specific effectiveness 

in diverse (geochemical) conditions; and (3) safe and 

sustainable application. The findings of every individual 

chapter are placed in a broader perspective to support 

the main conclusion of this thesis. Additionally an

9 Concluding remarks 

outlook is given on the recommended research and the 

future of small-scale SIR/SAR. 

Fe 2+ /As(III) immobilization processes 

A complex combination of (heterogeneous) oxidation, 

precipitation, adsorption and cation exchange reactions 

simultaneously occur in the subsurface, making it 

difficult to differentiate between these processes in 

field experiments. In sand column experiments the 

processes of adsorptive-catalytic oxidation and Fe 2+ 

exchange were isolated. During adsorptive-catalytic 

oxidation, the consumption of 1 mole of O 2 

resulted in 

the stoichiometric removal of 4 moles of Fe 2+ (Chapter 

4 and 7). Columns containing sediments from within 

a 12-year-old oxidation zone consumed more oxygen 

during injection than clean filter sand and subsequently 

retained more Fe 2+ during abstraction (Chapter 7). 

Although the adsorptive-catalytic oxidation process 

could describe the regeneration of the subsurface 

by injection of dissolved O 2 

, it could not account for 

the improved efficiency after multiple cycles. The 

retardation of Fe was found to increase significantly per 

cycle at the small-scale SIR/SAR set-ups in Bangladesh. 

After multiple cycles, the Fe 2+ removal efficacies at these 

sites exceeded the stoichiometric Fe 2+ :O 2 

ratio for Fe 2+ 

oxidation (Chapter 3). This efficiency increase can be 

described based on the memory of the SIR/SAR system, 

i.e., previous injection-abstraction cycles have not used 

the freshly formed adsorptive surface sites completely 

because the injection phase was started before complete 

breakthrough of Fe 2+ . Hence, these adsorption sites 

are available for future cycles, resulting in increasing 

efficacy (modelled by Appelo et al., 1999). The size 

of the oxidation zone should therefore be considered 

dynamic in size and water quality. 

This is also illustrated by the behaviour of 

exchangeable Fe 2+ during injection. In the absence of O 2 

and in the presence of high cation concentrations the 

adsorbed Fe 2+ was found be released from the sand grain 

surface and pushed deeper into the aquifer (Chapter 

5). The role of this cation exchange process in the 

efficacy of SIR/SAR may nevertheless be insignificant, 

as exchanged Fe 2+ will either be oxidized or adsorbed 

in- or outside the oxidation zone. It was hypothesized 

by Appelo et al. (1999) that injection of higher oxidant 

concentrations would not enhance SIR performance, as 

the available exchangeable Fe 2+ determines the efficacy. 

Field results with injection of high O 2 

concentrations, 

0.55mM instead of 0.28mM, increased the operational 

V/Vi ratio from an average 7 to 16 (WTP Corle, Chapter 

5). This indicates that the exchangeable Fe 2+ on the soil 

material is not necessarily the limiting factor during 

injection, but rather the supply of O 2 

to the subsurface 

Fe 2+ . Apart from the O 2 

concentration in the injection 

water, also the pH of the groundwater strongly affects 

the Fe 2+ immobilization. At higher pH (>6.5) Fe 2+ 

removal was found to be enhanced (Chapter 2), which 

is clearly based on the catalysed Fe 2+ oxidation and 

adsorption reaction in higher pH ranges (Tamura et al., 

1980; Sharma, 2001; Dixit and Hering, 2006). Of the 

127 

9


9 

128 

investigated inorganic groundwater constituents, only 

Ca 2+ was found to significantly inhibit the removal of 

Fe 2+ during injection-abstraction cycles (Chapter 9). 

As(III) behaviour during adsorptive-catalytic 

oxidation in the columns proved to be independent of 

the sediment source, Fe:As ratio and successive cycles 

(Chapter 3, 4 and 7). The latter was confirmed at the 

small-scale SIR/SAR set-up in Bangladesh, where 

arsenic was found in the abstracted water immediately 

upon abstraction (Chapter 3). Unlike iron removal, 

arsenic removal did not benefit from enhanced O 2 

consumption in the columns (Chapter 7), so co-removal 

of As(III) during SIR was not evident. As(III) retention 

is a secondary process during SIR/SAR, depending 

completely on the (chemical) affinity of As(III) for the 

freshly formed Fe 3+ oxides (amorphous; Chapter 6). 

Speciation of As(III) and As(V) in the abstracted water 

revealed that As(V) was slightly more delayed than 

As(III), illustrating the charge advantage of As(V) over 

As(III) at near neutral pH (Chapter 3 and 4). In general 

it can be concluded that the adsorptive process of As(III) 

removal depends strongly on the presence of other (oxy) 

anions in the groundwater (PO 4 

3- 

, NO 3- 

, SiO 2 

; Chapter 

9). Investigations with natural groundwater, in column 

and full-scale experiments, have shown that although 

SIR could perform under these different geochemical 

conditions, As(III) removal was negatively influenced 

by these complex water matrices. Independent of the 

available surface sites, either in the aquifer sediment or 

freshly formed by Fe 2+ oxidation, the presence of other 

anions -usually in much higher concentrations than 

As(III)- limited the As(III) adsorption. 

Site-specific effectiveness in diverse 

geochemical conditions 

The presence of elevated arsenic concentrations in 

the groundwater shows a very strong spatial (wellto-well) 

variability, even if Fe 2+ concentrations are 

constant (Smedley and Kinniburgh, 2002). These local 

geochemical variations require that an arsenic mitigation 

solution performs in a wide range of conditions. In 

the absence of limiting (inorganic) constituents, the 

removal of Fe 2+ and As(III) was extremely effective in 

columns with synthetic groundwater (R Fe 

=42; R As 

=30) 

at pH 6.9 (Chapter 8). At the small-scale SIR/SAR setup 

in Bangladesh, with the same pH, the retardation 

of Fe and As was much lower (R Fe 

=14; R As 

=1; Chapter 

3). In order to reduce the risk of chronic arsenic 

poisoning, concentrations below the WHO guideline 

(10 µg.L -1 ) are desirable, but those were only reached 

during abstraction of the As-free injection water. The 

effect of groundwater composition on SIR/SAR was 

investigated with sand columns and it was found that 

subsurface arsenic removal was seriously inhibited by 

the presence of PO 4 

3- 

, NO 3- 

, and SiO 2 

, illustrating the 

vulnerability of this technology in diverse geochemical 

settings (Chapter 8). Subsurface iron removal was 

observed not to be as sensitive to other inorganic 

groundwater constituents, although iron retardation 

was limited by Ca 2+ . From three decades of experience


with SIR in the Netherlands it can be concluded that pH 

is a vital parameter. The operational V/Vi ratio more 

than doubled between pH 7.0 and 7.4 at WTP Corle 

(Chapter 2), showing that a low pH can inhibit effective 

SIR operation. A pH buffer (as HCO 3- 

) to maintain 

the pH during the acidifying Fe 2+ oxidation reaction is 

desired for operation of SIR, though this may entail the 

presence of elevated Ca 2+ concentrations. 

At the investigated sites in Bangladesh a higher 

Fe:As(III) ratio did not significantly enhance As removal 

(Chapter 3), which was also confirmed in column 

experiments (Chapter 4). From the perspective of 

adsorptive-catalytic oxidation this makes sense, because 

the formation of Fe 3+ oxides (for subsequent adsorption 

of As) depends on the supply of oxygen to the adsorbed 

Fe 2+ . As long as the soil grains are saturated with Fe 2+ , 

the presence of higher Fe 2+ concentrations will not 

necessarily lead to more adsorptive surface area for As. 

Arsenic removal efficiencies from natural groundwater 

were limited by the presence of competing anions. The 

retention of Fe 2+ during injection-abstraction cycles 

was found to be successful in all experiments, both in 

the laboratory and field. The presence of accumulated 

deposits, mainly as Fe 3+ oxides, was found to enhance 

the adsorptive-catalytic oxidation process. For As(III) 

these deposits did not stimulate the removal process, as 

sediments from within the oxidation zone performed the 

same as clean filter (silica) sand. As(III) removal seems 

to rely primarily on the oxidation of Fe 2+ during a cycle, 

independent of the sediment composition. Though 

geochemical composition of sediments was not found 

to significantly influence As(III) removal, the presence 

of arsenic-bearing sulphide minerals may threaten 

the SAR technology during initial cycles. At WTP 

Lekkerkerk, arsenic peaks (


9 

130 

great potential of SIR, as the removal of iron improves 

colour and taste of the water; greatly enhancing social 

acceptance. From a technological perspective, the 

effective removal of Fe 2+ has an additional advantage, 

as the injection procedure start before arrival of Fe 2+ at 

the well. Aeration is more effective in the absence of 

Fe 2+ , making it easier to achieve high O 2 

concentrations 

without the production of iron sludge. 

HWTS may never threaten the long-term 

availability of a water source. For SIR/SAR this means 

that a shallow tube well may not produce water of lesser 

quality or quantity before, during or after use of this 

technology. Clogging of the aquifer around a subsurface 

iron removal well was addressed in Chapter 6, showing 

that after 12 years of operation, iron had accumulated 

at specific depths near the subsurface iron removal 

wells. Whether it was due to preferred flow paths 

or geochemical mineralogy conditions, subsurface 

iron removal clearly favoured certain soil layers. The 

majority of accumulated iron was characterized as 

crystalline, suggesting that precipitated amorphous 

iron oxides had transformed to iron oxides of higher 

crystallinity. These crystalline, compact iron oxides 

had not noticeably clogged the investigated well and/or 

aquifer. The subsurface iron removal wells even needed 

less frequent rehabilitation, as drawdown increase 

was slower than in normal production wells. Other 

groundwater constituents, such as manganese, arsenic 

and strontium were found to co-accumulate with iron. 

The accumulation of groundwater constituents 

around a SIR/SAR well is a direct consequence of this 

technology. And although it may seem undesirable 

from a sustainability perspective, these accumulated 

deposits do not harm human health or environment, 

as long as they remain immobile. Both Fe and As are 

abundantly available in sediments throughout the world 

(Smedley and Kinniburgh, 2002), but it merely depends 

on the local groundwater conditions whether these 

constituents will dissolve. The remobilization of iron, 

manganese, arsenic and phosphate was investigated 

when a SIR well was stopped after 18 injectionabstraction 

cycles (Chapter 2). At this particular 

site, none of the studied constituents was found to 

be released in the groundwater to concentrations 

exceeding the background value for the operational 

period of 2 years. The application of SIR at that site had 

therefore not threatened the long-term operation of 

this groundwater production well. 

The future of small-scale SIR/SAR 

This thesis has shown the potential of SIR as a household 

drinking water treatment solution, nevertheless a 

HWTS solution will only prove smart once it sustains 

in the users’ environment. This elementary observation 

comprises the real challenge for researchers who 

design technologies for developing countries. Some 

technologies might not yet be ready for implementation 

in the near future, but can form a base for upcoming 

trends. (Drinking water) engineers and researchers


have the tendency to focus on the technology, rather 

than on the social boundary conditions. In addition, 

enthusiasm about a new treatment system can be the 

driving force to implement the technology fast. In the 

process implementers can then forget the need for 

objective scientific research. Only when the advantages 

and, perhaps more importantly, the limitations of 

a technology are identified is it safe to scale-up the 

implementation. The research presented in this 

thesis can be seen as an example where a promising 

technology, according to the involved universities and 

national partners in Bangladesh, has been critically 

evaluated on its technical feasibility. The role of 

drinking water researchers can be vital in this, because 

during development of new technologies researchers 

must (i) remain critical of their own work, despite 

enthusiasm for the technology under investigation, and 

(ii) show flexibility and patience to let their findings 

be slowly adopted and, ultimately, be improved by 

the users. Currently rural Bangladesh seems to be 

a playground for researchers to test their ‘products’, 

and communication with local authorities is minimal. 

Even though this seems to be the fast way forward 

scientifically, working together will eventually give the 

best long-term solutions. Only by joining modern and 

endogenous knowledge true sustainable solutions will 

develop. 

In this perspective the main conclusion of 

this thesis should be re-formulated, as only longterm 

application of small-scale SIR will prove its 

sustainability in Bangladesh. Nevertheless, the results 

in this thesis have shown that SIR is technically feasible 

in a wide range of (geochemical) settings and holds 

therefore great promise for decentralized application. 

Subsurface arsenic removal did not show the same 

promise at household level. The performance of SAR at 

small-scale may be enhanced by enriching the injection 

water with higher O 2 

concentrations, e.g., with pure O 2 

, 

though this may be practically infeasible, for logistic, 

safety and economic reasons. It may also be, however, 

all a matter of scale. At larger injection volumes the 

oxidation zone is expanded and higher retention times 

of arsenic-contaminated groundwater in this zone can 

be reached. In favourable groundwater composition 

conditions (low competing anions concentrations, high 

pH), the subsurface removal of arsenic may still be 

feasible at community and/or municipality level. Such 

medium- or large-scale applications do not benefit 

from the existing hand pump infrastructure, but may 

be a safe arsenic mitigation solution in more densely 

populated areas, as frequently found in Bangladesh. 

In rural settings the scale issue may be overcome by 

combining SAR with Subsurface Rainwater Storage, 

where monsoon rain is utilized as injection water. 

Rainwater harvesting has been promoted as a safe 

drinking water alternative, but above-ground storage 

adds the risk of microbial contamination. Subsurface 

storage could overcome this problem, but research is 

recommended to investigate the effect of rain water 

(low in pH and alkalinity) on the mobility of arsenic. 

131 

9


9 

132 

It is also recommended to investigate the relationship 

between the size of the oxidation zone (contact time) 

and the removal of arsenic, which could be achieved 

by monitoring larger injection-abstraction columns, 

in combination with solute transport modelling. In 

addition, the threshold concentrations of competing 

anions for which SAR could still be effectively 

operated are yet to be determined, which can be done 

by expanding the column studies as presented in this 

thesis. 

Iron contamination of groundwater is not a 

health hazard like arsenic, but the iron removal benefits 

the users as well. Not just for aesthetic improvement 

of drinking water quality, but for all water-consuming 

activities (cooking, washing, laundry, etc.). Arsenic 

removal is primarily needed for drinking (and cooking) 

water, making the application of SIR in combination 

with above-ground arsenic removal very viable. SIR may 

thus be very useful for families in rural areas, just not as 

an arsenic mitigation solution. Subsequent (household) 

water treatment is needed in order to remove arsenic 

from the groundwater, which may be even more 

effective with subsurface treated water because: (i) iron 

is removed, reducing the clogging of the filters; and 

(ii) phosphate concentrations are reduced, leaving the 

available adsorption sites on the media available for 

arsenic adsorption. Some HWTS, however, rely on the 

presence of Fe 2+ , either for coagulation or regeneration 

of the adsorptive surfaces, making those systems less 

promising for combination with SIR. 

Subsurface manganese removal was not within the 

scope of this thesis, but over the past years elevated 

manganese concentrations have been reported 

in Bangladeshi groundwater. Chronic exposure 

to manganese through drinking water may have 

neurological effects above the WHO guideline of 0.4 

mg.L -1 (WHO, 2006). Subsurface manganese removal 

has been successfully operated in several European 

countries (Hallberg and Martinell, 1976; van Beek 1983) 

and, more recently, in Egypt (Olsthoorn, 2000). It may 

also be feasible in decentralized, rural settings, though 

previous experiments have indicated its sensitivity to 

operational fluctuations (van Beek, 1983), making it 

less attractive for hand pump application. 

References 



Ground Water 37(6): 811-817. 



Geology, 228(1-3): 6-15. 



Olsthoorn, T.N. (2000) Background of subsurface iron and 

manganese removal. Amsterdam Water Supply Research and 

Development, Hydrology Department.


Sharma S.K. (2001) Adsorptive iron removal from groundwater. 

PhD dissertation, IHE Delft. 



Geochem 17: 517–568. 

Tamura H., S. Kawamura and M. Hagayama (1980) Acceleration 

of the oxidation of Fe 2+ ions by Fe(III)-oxyhydroxides, 

Corrosion Science 20, 963-971. 




World Health Organization (2006) Guidelines for drinkingwater 

quality, First Addendum to Third Edition, Volume 1 


9 

133


9 

134

Summary 


- Doris van Halem 

Summary 



health problem. This problem disproportionately 

affects the rural poor, i.e., those most reliant on this 

source of drinking water. Current arsenic mitigation 

solutions, including (household) arsenic removal 

options, do not always provide a sustainable alternative 

for safe drinking water in rural Bangladesh. A novel 

household water treatment and storage (HWTS) 

technology, Subsurface Arsenic Removal (SAR), 

relies on the existing technology of Subsurface Iron 

Removal (SIR), or in-situ iron removal. The principle 

of SIR is that aerated water is periodically injected 

into an anoxic or anaerobic aquifer through a tube 

well, partially displacing the original Fe 2+ -containing 

groundwater. The O 2 

-rich injection water oxidizes the 

adsorbed Fe 2+ in the subsurface environment around 

the tube well. The flow direction is reversed during 

groundwater abstraction, resulting in adsorption 

of dissolved Fe 2+ (and arsenic) on the oxidized Fe 3+ 

surface. Subsequently groundwater with reduced iron 

and arsenic concentrations can be abstracted. However, 

for SIR/SAR to be considered a viable arsenic mitigation 

tool, the following problem needs to be addressed: 

Problem description 

Subsurface Iron and Arsenic Removal is a 

promising safe water solution, however, there is the 

need for better understanding of the (subsurface) 

processes determining the sustainable operation 

in diverse geochemical settings. 

The research methodology consisted of four 

components: (1) data analysis of existing SIR plants 

in the Netherlands, (2) monitoring of a test facility in 

135 

S


S 

136 

Bangladesh, (3) sand column studies in laboratory and 

field, and (4) sediment characterization nearby SIR 

wells. The methodology allowed to address the main 

problem definition by focusing on the following: 

• past experiences in the Netherlands (chapter 2); 

• small-scale applicability of SIR/SAR (chapter 3); 

• adsorptive-catalytic oxidation of Fe 2+ (chapters 4 and 7); 

• co-removal of As(III) (chapters 3,4, and 7); 

• cation exchange during injection (chapter 5); 

• clogging of the aquifer (chapter 6); 

• influence of groundwater composition (Chapter 8). 

The field and laboratory data have gained insight in the 

occurring (geo)chemical processes under the shifting 

redox conditions during injection-abstraction cycles, 

resulting in two main conclusions. 

Conclusion #1 

The results in this thesis have shown that 

Subsurface Iron Removal is technically feasible 

in a wide range of (geochemical) settings and 

has therefore great potential for hand pump 

application. 

The experiences in the Netherlands have shown that 

subsurface iron removal is an effective, robust and 

sustainable iron removal technology with great potential 

for worldwide application. pH showed to be the most 

pronounced water quality parameter determining the 

efficacy of SIR, as the operational V/Vi ratio more 

than doubled between pH 7.0 and 7.4 at WTP Corle. 

Apart from iron, it was measured that also manganese, 

phosphate and arsenic were retained in the aquifer. In 

general, successive cycles resulted in improved removal 

efficacies, illustrating the technology’s potential for safe 

and long-term operation. Additionally, when SIR was 

stopped for several years, no peak concentrations of the 

studied constituents were observed in the abstracted 

groundwater. 

For hand pump application of SIR, the injection 

volumes, as operated in the Netherlands were reduced 

from over 1,000m 3 to below 1m 3 . The field study 

in Manikganj, Bangladesh, showed that SIR could 

be successfully operated at small-scale. The system 

gained in efficiency after multiple injection-abstraction 

cycles, and after injection of water with higher O 2 

concentrations. 

The injection-abstraction cycles were simulated 

in sand column experiments, where the observed 

(stoichiometric) 1:4 ratio for O 2 

consumption and Fe 2+ 

removal confirmed the occurrence of the adsorptivecatalytic 

oxidation mechanism. Also, O 2 

consumption 

and subsequent Fe 2+ removal were enhanced in the 

columns by accumulated Fe 3+ deposits. Nevertheless, 

the adsorptive-catalytic oxidation mechanism could 

not account for the efficacy increase after multiple 

cycles. The O 2 

consumption by secondary processes 

during initial cycles (e.g., pyrite oxidation) could 

provide an explanation, but it is more likely that the 

spatial (volume) increase of the subsurface oxidation 

zone enhances the Fe 2+ immobilization per cycle.

Summary 

Apart from adsorptive-catalytic oxidation, also cation 

exchange has been proposed to occur during injectionabstraction 

cycles. In sand column experiments with 

synthetic and natural groundwater it was found that 

cation exchange (Na + -Fe 2+ ) occurs during subsurface 

iron removal. The Fe 2+ exchange increased at higher 

Na + concentration in the injection water, but decreased 

in the presence of other cations in the groundwater. 

Field results at WTP Corle with injection of high O 2 

(0.55 mmol.L -1 ) concentrations showed better removal 

than the normal 0.28 mmol.L -1 O 2 

, indicating that not 

the exchangeable Fe 2+ on the soil material is the limiting 

factor during injection, but it is the supply of O 2 

to the 

available Fe 2+ . From a groundwater quality perspective, 

it was found in the column studies that SIR was not 

significantly limited or enhanced by other inorganic 

groundwater constituents, though iron retardation was 

lower in the presence of 1.2 mM calcium or 0.06 mM 

manganese. 

One of the concerns for sustainable application 

of SIR is the risk of clogging of the aquifer by longterm 

operation of SIR. Therefore sediments nearby 

12-year-old SIR wells were characterized for their 

chemical composition. At the investigated site (WTP 

De Put), iron has accumulated at specific depths near 

the subsurface iron removal wells. Whether it was due 

to preferred flow paths or geochemical mineralogy 

conditions; subsurface iron removal clearly favoured 

certain soil layers. The majority of accumulated iron 

was characterized as crystalline, suggesting that 

precipitated amorphous iron oxides have transformed 

to iron oxides of higher crystallinity. These crystalline, 

compact iron oxides had not noticeably clogged the 

investigated well and/or aquifer between 1996 and 2008. 

The subsurface iron removal wells even needed less 

frequent rehabilitation, as drawdown increases slower 

than in normal production wells. Other groundwater 

constituents, such as manganese, arsenic and strontium 

were found to co-accumulate with iron. 

Conclusion #2 

Subsurface Arsenic Removal has been found to 

be less suitable for hand pump application and 

vulnerable to diverse geochemical conditions. 

The hand pump field study in Bangladesh did not prove 

to be very effective for the removal of arsenic. Arsenic 

(as As(III)) retardation was limited and breakthrough 

of 10 µg.L -1 (provisional guideline of the World 

Health Organization) was observed before V/Vi=1, 

which corresponds to the moment of groundwater 

arrival at the well. Unlike iron removal, subsurface 

arsenic removal did not increase after multiple cycles, 

illustrating that the process which is responsible for 

the effective iron removal did not promote an equally 

effective co-removal of As(III). No relation has been 

observed in this study between the amount of removed 

As(III) and the Fe 2+ :As(III) ratio of the groundwater. 

And because arsenic is observed in the abstracted 

groundwater amply before iron, the latter cannot be 

used as a visible indicator for arsenic presence, i.e., aid 

137 

S


S 

138 

in post-deployment monitoring. 

During injection-abstraction cycles in sand columns 

with a synthetic Fe 2+ -As(III)-O 2 

system it was possible 

to reach significantly better arsenic removal than at 

the field study in Bangladesh and in the columns with 

groundwater (WTP Lekkerkerk and WTP Loosdrecht). 

This can be explained by the competition with other 

anions in the natural groundwater. In the presence of 

0.01mM phosphate, 0.2mM silicate, or 1mM nitrate, 

arsenic removal was seriously inhibited, illustrating 

the vulnerability of this SAR technology in diverse 

geochemical settings. 

This vital observation, however, applies not 

just for SAR but for the majority of (arsenic) HWTS 

solutions. In general, no household water treatment 

option has been designed to suit all conditions, 

illustrating the urgent need to investigate the (technical) 

boundary conditions in which a HWTS solution can 

succeed. In favourable groundwater composition 

conditions (low competing anions concentrations, 

high pH), the subsurface removal of arsenic may still 

be feasible. However, the decentralized character of 

HWTS does not promote extensive knowledge on the 

groundwater composition at hand pump level, making 

community- and/or municipality scale facilities more 

attractive. 

A newly recognized groundwater quality 

concern in Bangladesh is contamination of groundwater 

by elevated manganese concentrations. Chronic 

exposure to manganese through drinking water may 

have neurological effects above the WHO guideline of 

0.4 mg.L -1 . The (small-scale) application of subsurface 

manganese removal would be an interesting approach 

to investigate. Iron contamination of drinking water 

is not a health hazard, but the iron removal benefits 

the users as well. Not just for aesthetic improvement 

of drinking water quality, but for all water-consuming 

activities (cooking, washing, laundry, etc.). Arsenic 

removal is primarily needed for drinking (and cooking) 

water, making the application of SIR in combination 

with subsequent (household) arsenic removal very 

viable. Arsenic filters may even perform better as iron 

and phosphate levels are reduced by SIR, reducing both 

clogging and adsorptive competition. 

A HWTS solution will only prove smart 

once it sustains in the users’ environment. This 

elementary observation comprises the real 

challenge for researchers who study technologies 

for developing countries. Some technologies might 

not yet be ready for implementation in the near 

future, but can form a base for upcoming trends. 

(Drinking water) engineers and researchers have 

the tendency to focus on the technology, rather than 

on the social boundary conditions. In addition, 

enthusiasm about a new treatment system can be 

the driving force to implement the technology fast. 

In the process, implementers can then forget the 

need for objective scientific research. Only when 

the advantages and, perhaps more importantly, 

the limitations of a technology are identified is it 

safe to scale-up the implementation.

Samenvatting 

Ondergrondse ijzer-en arseenverwijdering voor drinkwaterzuivering in Bangladesh 

- Doris van Halem 

Samenvatting 

Arseenverontreiniging van ondiep grondwater is 

een groot ontwikkelings- en gezondheidsprobleem 

in Bangladesh. Dit probleem raakt de armen op 

het platteland onevenredig, omdat zij het meest 

afhankelijk zijn van deze bron van drinkwater. Huidige 

oplossingen voor het arseenprobleem, waaronder 

arseenverwijdering op huishoudschaal, vormen niet 

altijd een duurzaam alternatief voor veilig drinkwater 

op het platteland van Bangladesh. Een nieuwe household 

water treatment and storage (HWTS) technologie 

combineert ondergrondse arseenverwijdering (SAR) 

met de bestaande technologie van ondergrondse 

ijzerverwijdering (SIR). Het principe van SIR is dat 

belucht water periodiek wordt geïnjecteerd in een 

zuurstofloze of anaërobe watervoerende laag, waardoor 

het originele Fe 2+ -bevattende grondwater gedeeltelijk 

verplaatst wordt. Het O 2 

-rijke injectie water oxideert 

het geadsorbeerde Fe 2+ in de ondergrond rondom de 

put. Tijdens grondwaterontrekking keert de stroming 

om waardoor opgelost Fe 2+ (en arseen) kan adsorberen 

aan het geoxideerde Fe 3+ oppervlak. Hierdoor kan 

vervolgens grondwater met verlaagde ijzer en arseen 

concentraties onttrokken worden. Voordat SIR/SAR 

beschouwd kan worden als een haalbaar hulpmiddel ter 

bestrijding van het arseenprobleem, moet het volgende 

probleem opgelost worden: 

Probleembeschrijving 

Ondergrondse ijzer- en arseenverwijdering is een 

veelbelovende oplossing voor veilig drinkwater, 

maar er is de noodzaak voor beter inzicht in 

de (ondergrondse) processen die de duurzame 

werking onder verschillende geochemische 

condities bepalen. 

139 

S


S 

140 

De onderzoeksmethodiek bestond uit vier 

componenten: (1) data analyse van bestaande SIR 

zuiveringen in Nederland, (2) het monitoren van 

een proefopstelling in Bangladesh, (3) zandkolom 

studies in het veld en laboratorium, en (4) sediment 

karakterisering nabij SIR putten. Met hulp van deze 

methodes is er gericht onderzoek gedaan naar: 

• ervaringen uit het verleden in Nederland (hoofdstuk 2); 

• kleinschalige toepassing van SIR/SAR (hoofdstuk 3); 

• adsorptieve-katalytische oxidatie van Fe 2+ (hoofdstukken 

4 en 7); 

• co-verwijdering van As(III) (hoofdstukken 3,4 en 7); 

• kationuitwisseling tijdens injectie (hoofdstuk 5); 

• verstopping van het watervoerend pakket (hoofdstuk 6); 

• invloed van de grondwatersamenstelling (hoofdstuk 8). 

De veld- en laboratoriumresultaten hebben inzicht 

gegeven in de (geo)chemische processen die 

plaatsvinden onder de veranderende redox condities 

tijdens de injectie-onttrekkingscycli, resulterend in 

twee hoofdconclusies. 

Conclusie #1 

De resultaten in dit proefschrift hebben 

aangetoond dat ondergrondse ijzerverwijdering 

technisch haalbaar is in een breed scala van 

(geochemische) condities en heeft daarom veel 

potentie voor de toepassing op handpomp schaal. 

De ervaringen in Nederland hebben aangetoond 

dat ondergrondse ijzerverwijdering een effectieve, 

robuuste en duurzame technologie is met een groot 

potentieel voor wereldwijde toepassing. pH bleek de 

meest uitgesproken waterkwaliteitsparameter voor het 

bepalen van de effectiviteit van SIR. De operationele V/ 

Vi-verhouding verdubbelde ruimschoots tussen pH 7,0 

en 7,4 op PS Corle. Naast ijzer, werden ook mangaan, 

fosfaat en arseen vastgelegd in het watervoerende 

pakket. Opeenvolgende cycli resulteerden in een 

verbeterde verwijderingsefficiëntie, wat het potentieel 

van deze technologie voor veilig en langdurig gebruik 

illustreert. Nadat SIR werd gestopt voor een aantal jaren, 

zijn er bovendien in die periode geen piekconcentraties 

van de onderzochte bestanddelen waargenomen in het 

onttrokken grondwater. 

Voor handpomp toepassing van SIR werden 

de injectievolumes teruggebracht van meer dan 1.000 

m 3 , zoals gebruikelijk in Nederland, tot onder de 1m 3 . 

Het veldonderzoek in Manikganj, Bangladesh, toonde 

aan dat SIR met succes zou kunnen worden toegepast 

op kleine schaal. De systeemefficiëntie had profijt van 

opeenvolgende injectie-ontrekkingscycli en van injectie 

met hogere O 2 

concentraties. 

De injectie-ontrekkingscycli zijn gesimuleerd 

in zandkolom experimenten, waarbij de waargenomen 

(stoichiometrische) 1:4 verhouding voor O 2 

verbruik en Fe 2+ verwijdering het optreden van 

het adsorptieve-katalytische oxidatie mechanisme 

bevestigde. Aanvullend werden het O 2 

verbruik en 

de daaropvolgende Fe 2+ verwijdering versterkt in de 

kolommen door aanwezigheid van geaccumuleerde

Samenvatting 

Fe 3+ oxiden. Niettemin is het adsorptieve-katalytische 

oxidatie mechanisme niet verantwoordelijk voor de 

verbeterde efficiëntie na meerdere cycli. Het O 2 

verbruik 

van secundaire processen tijdens de eerste cycli 

(bijvoorbeeld pyriet oxidatie), kan wel een verklaring 

zijn, maar het is waarschijnlijker dat de ruimtelijke 

(volume) vergroting van de ondergrondse oxidatie 

zone de Fe 2+ immobilisatie per cyclus verhoogt. 

Naast adsorptieve-katalytische oxidatie, is de 

hypothese dat ook kationenuitwisseling een rol speelt 

tijdens injectie-ontrekkingscycli. In de zandkolom 

experimenten met synthetisch en natuurlijk 

grondwater bleek dat kationuitwisseling (Na + -Fe 2+ ) 

optreedt tijdens ondergrondse ontijzering. De Fe 2+ 

uitwisseling verhoogde bij hogere Na + -concentraties 

in het geïnjecteerde water, maar daalde in de 

aanwezigheid van andere kationen in het grondwater. 

Veldresultaten op PS Corle met het injecteren van 

water van hoge O 2 

(0,55 mmol/l) concentraties toonde 

een betere verwijdering dan de normale 0,28 mmol/l 

O 2, 

wat aangeeft dat niet de uitwisselbare Fe 2+ op het 

bodemmateriaal de beperkende factor is tijdens injectie, 

maar de aanvoer van O 2 

aan de beschikbare Fe 2+ . Met 

betrekking tot het effect van de grondwaterkwaliteit, 

bleek in de kolomstudies dat SIR niet significant 

beperkt of versterkt wordt door de aanwezigheid van 

andere anorganische grondwater bestanddelen, hoewel 

de ijzer retardatie lager was in de aanwezigheid van 1,2 

mmol/l calcium of 0,06 mmol/l mangaan. 

Een aandachtspunt voor de duurzame toepassing 

van SIR is het risico van verstopping van het 

watervoerend pakket door langdurig gebruik van SIR. 

Sedimenten nabij 12-jarige SIR putten zijn daarom 

geanalyseerd op hun chemische samenstelling. 

Op de onderzochte locatie (PS De Put), is ijzer 

geaccumuleerd op specifieke dieptes in de buurt van 

de ondergrondse ontijzeringsputten. Veroorzaakt 

door voorkeursstromingen of geochemische condities; 

ondergrondse ontijzering had duidelijk een voorkeur 

voor bepaald bodemlagen. De meerderheid van het 

geaccumuleerde ijzer is gekarakteriseerd als kristallijn 

dat suggereert dat neergeslagen amorfe ijzeroxiden 

zijn getransformeerd naar ijzeroxiden van hogere 

kristalliniteit. Deze kristallijne en compacte ijzeroxiden 

hebben de onderzochte put en/of het watervoerend 

pakket niet merkbaar verstopt tussen 1996 en 2008. 

De ondergrondse ontijzeringsputten hoefden zelfs 

minder vaak gereinigd te worden, vanwege het verschil 

in peilhoogte dat langzamer toenam dan in de normale 

onttrekkingsputten. Uit metingen is ook gebleken 

dat andere grondwaterbestanddelen, zoals mangaan, 

arseen en strontium accumuleerden samen met het 

ijzer. 

Conclusie #2 

Ondergrondse arseenverwijdering is minder 

geschikt bevonden voor handpomp toepassing en 

is kwetsbaar in diverse geochemische condities. 

S 

141


S 

142 

Uit het handpomp veldonderzoek in Bangladesh is 

SAR niet effectief gebleken voor de verwijdering van 

arseen. Arseen (als As(III)) retardatie was beperkt en 

de doorbraak van 10 μg/l (voorlopige richtlijn van de 

Wereldgezondheidsorganisatie) werd waargenomen 

nog voor V/Vi =1, wat overeenkomt met het moment 

van grondwater aankomst in de put. In tegenstelling 

tot de verwijdering van ijzer, nam ondergrondse 

arseenverwijdering niet toe na meerdere cycli. Dit 

illustreert dat het proces dat verantwoordelijk is voor 

de effectieve verwijdering van ijzer niet een evenredige 

efficiëntie garandeert voor As(III). In deze studie 

is geen relatie waargenomen tussen de hoeveelheid 

verwijderde As(III) en de Fe 2+ :As(III) verhouding van 

het grondwater. Bovendien kan ijzer niet gebruikt 

worden als zichtbare indicator voor de aanwezigheid 

van arseen, aangezien arseen waargenomen wordt 

in het grondwater voordat opgelost ijzer bij de put 

arriveert. 

Tijdens de injectie-ontrekkingscycli in de 

zandkolommen met een synthetisch Fe 2+ -As(III)- 

O 2 

-systeem was het mogelijk om significant 

betere arseenverwijdering te bereiken dan bij het 

veldonderzoek in Bangladesh en bij de kolommen 

met grondwater (PS Lekkerkerk en PS Loosdrecht). 

Dit kan verklaard worden door de competitie met 

andere anionen in het natuurlijke grondwater. In de 

aanwezigheid van 0,01mmol/l fosfaat, 0,2 mmol/l 

silicaat of 1 mmol/l nitraat, was de arseenverwijdering 

erg beperkt, welke de kwetsbaarheid van de SAR 

technologie in verschillende geochemische condities 

aantoont. 

Deze essentiële waarneming geldt echter 

niet alleen voor SAR, maar voor de meerderheid 

van (arseen) HWTS oplossingen. Geen enkele 

huishoudelijke waterbehandelingsoptie is ontworpen 

om aan alle mogelijke voorwaarden te voldoen en er 

is dus een dringende noodzaak om de (technische) 

randvoorwaarden waarin een HWTS oplossing 

kan slagen te onderzoeken. Met een gunstige 

grondwatersamenstelling (lage concentraties 

van concurrerende anionen, hoge pH), kan de 

ondergrondse arseenverwijdering nog steeds haalbaar 

zijn. Echter, het decentrale karakter van HWTS is niet 

bevorderlijk voor de aanwezigheid van voldoende 

kennis van de grondwatersamenstelling ter plaatse, 

waardoor zuivering op community- of gemeentelijke 

schaal aantrekkelijker is. 

Een nieuw waterkwaliteitsaandachtspunt in 

Bangladesh is de verontreiniging van grondwater 

met verhoogde mangaan concentraties. Chronische 

blootstelling aan mangaan via drinkwater kan 

neurologische effecten hebben boven de richtlijn 

van 0,4 mg/l (Wereldgezondheidsorganisatie). 

De (kleinschalige) toepassing van ondergrondse 

mangaanverwijdering zou daarom een interestante 

onderzoeksrichting zijn. IJzerverontreiniging van het 

drinkwater vormt geen gevaar voor de gezondheid, maar 

de gebruikers hebben wel profijt van ijzerverwijdering. 

Niet alleen voor esthetische verbetering van de

Samenvatting 

drinkwaterkwaliteit, maar voor alle water verbruikende 

activiteiten (koken, wassen, etc.). Verwijdering van 

arseen is alleen nodig voor drink- en kookwater, maar 

huidige (huishoud) arseenfilters worden beperkt in 

hun efficiëntie door de aanwezigheid van ijzer en 

fosfaat in het grondwater (verstopping en adsorptieve 

competitie). De combinatie van SIR gevolgd door een 

arseenfilter kan deze beperking wegnemen, aangezien 

tijdens SIR zowel ijzer als fosfaat concentraties verlaagd 

worden. 

Een HWTS oplossing is pas een ‘smart 

solution’ als het duurzaam standhoudt in 

de gebruikersomgeving. Deze elementaire 

observatie omvat de echte uitdaging voor 

wetenschappers die onderzoek doen naar 

technologieën voor ontwikkelingslanden. 

Sommige technologieën zijn misschien nog niet 

klaar voor implementatie in de nabije toekomst, 

maar kunnen een basis vormen voor toekomstige 

trends. (Drinkwater) ingenieurs en onderzoekers 

hebben de neiging om zich te concentreren op 

de technologische mogelijkheden, in plaats van 

op de maatschappelijke randvoorwaarden. 

Daarnaast kan enthousiasme over een nieuw 

zuiveringssysteem de drijvende kracht worden 

om de technologie te snel te implementeren. In 

het proces van realisatie kunnen betrokkenen 

de noodzaak van objectief wetenschappelijk 

onderzoek vergeten. Alleen wanneer de voordelen 

en, misschien nog belangrijker, de beperkingen 

van een technologie zijn geïdentificeerd is het 

veilig om de implementatie op te schalen. 

S 

143


S 

144

List of publications 


International refereed journals 

van Halem D., W. de Vet, J. Verberk, G. Amy, H. van Dijk 

(2011) Characterization of accumulated precipitates during 

subsurface iron removal, Applied Geochemistry 26: 116-124. 

van Halem D., S. Olivero, W.W.J.M. de Vet, J.Q.J.C. Verberk, G.L. 



Bangladesh, Water Research 44: 5761-5769. 

van Halem D., R. Johnston, I.M. Huq, S.K. Ghosh, J.Q.J.C. 

Verberk, J.C. van Dijk and G.L. Amy (2010) Subsurface iron 

and arsenic removal: low-cost technology for communitybased 

drinking water supply in Bangladesh. Water Science 

and Technology: Water Supply 62(11): 2702-2709. 

van Halem D., H. van der Laan, S.G.J. Heijman, J.C. van Dijk 

and G.L. Amy (2009) Assessing the sustainability of the 

silver-impregnated ceramic pot filter for low-cost household 

drinking water treatment, Physics and Chemistry of the 

Earth 34: 36-42. 

van Halem D., S.G.J. Heijman, G.L. Amy, and J.C. van Dijk 

(2009) Subsurface arsenic removal for small-scale application 

in developing countries, Desalination 248(1-2): 241-248. 

van Halem D., S.A. Bakker, G.L. Amy, and J.C. van Dijk (2009) 

Arsenic in drinking water: a worldwide water quality concern 

for water supply companies, Drinking Water Engineering and 

Science 2: 29-34. 

van Halem D., S.G.J. Heijman, J.C. van Dijk and G.L. Amy 

(2007) Ceramic silver-impregnated pot filters for household 

drinking water treatment in developing countries: material 

characterization and performance study, Water Science & 

Technology: Water Supply 7(5-6): 9-17. 

Publications in preparation 

van Halem D., D.H. Moed, J.Q.J.C. Verberk, G.L. Amy and J.C. 

van Dijk (2011) Cation exchange during subsurface iron 

145 

P


P 

removal, Water Research: under review. 

van Halem D. D.H. Moed, J.Q.J.C. Verberk, G.L. Amy and J.C. 

van Dijk (2011) The influence of groundwater matrix on 

subsurface arsenic and iron removal, Science of the Total 

Environment: submitted. 

van Halem J.Q.J.C. Verberk, G.L. Amy and J.C. van Dijk (2011) 

Accumulated deposits near subsurface iron removal wells: 

do they catalyze the Fe 2+ removal process? Water Science and 

Technology: submitted. 

van Halem D., W.W.J.M. de Vet, M. de Jonge, G.L. Amy and J.C. 

van Dijk (2011) Subsurface iron removal in the Netherlands, 

Journal of American Water Works Association: to be 

submitted. 

Editor of proceedings 

van Halem D. and W. M. Savenije (2009). International Ceramic 

Pot Filter Conference: Proceedings, International Pot Filter 

Conference, Atlanta, WEF Disinfection. 

Conference proceedings 

van Halem D.H. Moed, J.Q.J.C. Verberk, G.L. Amy and J.C. van 

Dijk (2011) Subsurface iron and arsenic removal for drinking 

water production: influence of the multi-component 

groundwater matrix, IWA Groundwater, Belgrade. 

van Halem J.Q.J.C. Verberk, G.L. Amy and J.C. van Dijk (2011) 

Accumulated deposits near subsurface iron removal wells: do 

they catalyze the Fe 2+ removal process? IWA Groundwater, 

Belgrade. 

van Halem D., J.Q.J.C. Verberk and J.C. van Dijk (2011) 

Decentralised subsurface iron and arsenic removal for 

146 

drinking water production: field and laboratory investigations, 

Conference on Integrated Water Management, Perth. 

van Halem D., P. Smeets, et al. (2010) Natural Engineered 

Systems: a sustainable alternative for drinking water 

treatment. AEESP session: Water Sustainability, WEFTEC 

2010, New Orleans: 1-13. 

Li Z., D. van Halem, et al. (2010) Review of high arsenic 

groundwater in China, 4th International Conference on 

Bioinformatics and Biomedical Engineering, Chengdu, 

China. 

van Halem D., R. Johnston, et al. (2009) Subsurface iron and 

arsenic removal: low-cost technology for community-based 

drinking water supply in Bangladesh, 1 st IWA Development 

Congress, Mexico City. 

van Halem D., H. van der Laan, et al. (2009) Subsurface 

iron removal: modelling the process, BeNeLux Young 

Water Professionals, Eindhoven, International Water 

Association. 

Bloem S. C., D. van Halem, et al. (2009) Silver impregnated 

ceramic pot filter: Flow rate versus the removal efficiency 

of pathogens, WEF Disinfection: International Ceramic Pot 

Filter Conference, Atlanta, WEF. 

van Halem D., W. W. J. M. de Vet, et al. (2008) Subsurface iron 

removal for drinking water production: understanding the 

process and exploiting beneficial side effects, Water Quality 

Technology Conference, Cincinnati, American Water Works 

Association. 

van Halem D., S. G. J. Heijman, et al. (2008) Subsurface arsenic 

removal for small-scale application in developing countries, 

Water and Sanitation in International Development and


Disaster Relief, Edinburgh. 

van Halem D., S. G. J. Heijman, et al. (2008) Design considerations 

for small-scale subsurface arsenic removal in developing 

countries, Arsenic in the environment: Arsenic from nature 

to humans, Valencia. 

van Halem D., S. G. J. Heijman, et al. (2007) Low-cost ceramic 

membrane filtration for application in developing countries: 

the ceramic silver-impregnated pot filter, Water Quality 

Technology Conference, Charlotte, American Water Works 

Association. 

van Halem D., S. G. J. Heijman, et al. (2007). Ceramic silverimpregnated 

pot filters for household drinking water 

treatment in developing countries: material characterization 

and performance study, International Conference on Water 

Management and Technology Applications in Developing 

Countries, Kuala Lumpur, IWA Specialist Conference. 

National publications 

Bakker S.A., D. van Halem, J.C. van Dijk, and G.L. Amy (2008) 

Arseen in drinkwater: niet alleen een probleem in Bangladesh, 

H 2 

O: Tijdschrift voor watervoorziening en waterbeheer. 

van Halem D. (2008) Drinking water research for all, 

Vakantiecursus, Delft University of Technology, J.C. van Dijk 

and N. Krikke (eds), Delft. 

van Halem D., A.I.A. Soppe, J. Kroesbergen and H. van der 

Jagt (2006) Effectiviteit van de met zilver geïmpregneerde 

keramische potfilter, H 2 

O: Tijdschrift voor watervoorziening 

en waterbeheer. 

P 

147


Biography 

Doris van Halem (Eindhoven, 1981) graduated in 2000 

from secondary school (Eindhoven, VWO/IGCSE). 

Her interest in water and development started during 

her studies in Civil Engineering at Delft University of 

Technology, where she was involved in water projects 

in Benin and Sri Lanka. For her MSc thesis she 

investigated the efficacy of low-cost ceramic filters, 

locally produced around the world. As a follow-up she 

organized, on behalf of Aqua for All, the 1 st international 

CPF conference at WEF Disinfection in Atlanta. 

After graduating cum laude in 2006 she started her 

PhD research under the guidance of Prof. Hans van 

Dijk (TU Delft) and Prof. Gary Amy (UNESCO-IHE). 

The research entitled “Subsurface Iron and Arsenic 

Removal for drinking water treatment in Bangladesh” 

was presented at many (international) IWA, AWWA 

and WEF conferences. Her articles have been published 

in international refereed journals and she is active as a 

reviewer for several journals, including Water Science 

and Technology and Water Research. 

B 

148

Subsurface Iron and Arsenic Removal

Create successful ePaper yourself

Delete template?

Save as template?