NATURAL ARSENIC IN GROUNDWATER: OCCURRENCE, REMEDIATION AND MANAGEMENT - CHAPTER 21 ppsx

Section 4: Remediation of arsenic-rich
groundwaters
Copyright © 2005 Taylor & Francis Group plc, London, UK
Technologies for arsenic removal from potable water
W. Driehaus
GEH Wasserchemie GmbH & Co. KG, Osnabrueck, Germany
ABSTRACT: Arsenic (As) contamination of ground water resources used for potable water sup-
ply is an emerging problem throughout the world. Except for the use of alternative, uncontami-
nated water sources, water treatment for As removal is often the only solution to meet the
standards. The effective application of As removal processes requires the knowledge of its chem-
istry in natural water. Before making a choice of a suitable treatment technique it is to ascertain
whether As(III) is present and to evaluate the need for an oxidation technique for As(III). The
techniques existing for the removal of As include conventional processes like ion exchange and
coagulation/filtration and also emerging processes with iron oxide based adsorbents like granular
ferric hydroxides. While most of the removal techniques suitable for water works operation rely on
sorption processes, they exhibit big differences in investment, chemical cost, maintenance and the
type of arsenic bearing waste. Different options for disposal are discussed.
1 INTRODUCTION
Arsenic (As) became a serious health problem in several countries around the world. This is also
attributed to new findings about the toxicity of arsenic, especially for long term exposure to low
levels in drinking water and food. Thus, a number of countries have lowered their drinking water
standards within the last 13 years (WHO 2001). The lowered standards, mostly to 10 ␮g/L and the
“As calamity” in India and Bangladesh induced world wide activities in research of available treat-
ment methods for As-removal and the development of new technologies. The situation of drinking
water supply in the affected countries and areas are different, from more centralized treatment
plants and supply i.e. in the UK and the USA, to smaller decentralized waterworks in Germany
and France to a supply by hand pumps, missing any distribution network in the affected areas of
West Bengal and Bangladesh.
This paper gives an introduction to the basics of aquatic chemistry of As and an overview on As
removal techniques in potable water treatment, including oxidation techniques for As(III).
2AQUATIC CHEMISTRY OF ARSENIC

Arsenic is a semimetal and occurs in natural groundwater mostly occurring as oxyanions as triva-
lent As (H
3
AsO
3
) or pentavalent As (H
3
AsO
4
). They are commonly named arsenite or As(III) and
arsenate or As(V). The occurrence of organic As compounds, especially methylated species is only
reported from surface water (Anderson & Bruland 1991), but they rarely, if ever occur in ground-
water. Elevated As concentrations at a level relevant for human health are mostly caused by
inorganic As species in groundwater. Figure 1 shows the stability diagram of inorganic As.
Natural groundwater is commonly in the field between pH 5–9 and Eh Ϫ0.50 V to ϩ0.50 V.
Under this conditions there are two dominating species, As(III) in more reducing environments
and As(V) under oxidizing conditions. In strongly reducing environments with hydrogen sulfide
present form also a couple of As-sulfides, most of them being solids. As this conditions are rare in
189
Natural Arsenic in Groundwater: Occurrence, Remediation and Management –
Bundschuh, Bhattacharya and Chandrasekharam (eds)
© 2005, Taylor & Francis Group, London, ISBN 04 1536 700 X
Copyright © 2005 Taylor & Francis Group plc, London, UK
groundwater bodies used for drinking water supply, these species are excluded from the following
description.
The dissociation constants of As(III) are pK
S1
ϭ 9.22; pK
S2
ϭ 12.10; pK

S3
ϭ 13.40 (Pierce
1981). That means that at pH 9.2 the As(III) is by 50% dissociated. At lower pH, most of As(III)
exists as a neutral molecule.
The dissociation constants of As(V) are pK
S1
ϭ 2.22; pK
S2
ϭ 6.96; pK
S3
ϭ 11.5. That means,
that a pH 6.96 about 50% of As(V) exists as a monovalent anion and 50% as an divalent anion.
The difference in charge has important effects on the removal characteristics of both As species,
because neutral, uncharged molecules can not or less effectively be removed by most treatment
techniques (Jekel 1994). As(III) can be oxidized to As(V) at a relatively low Eh-potential of
0.1–0.2 V and, from the energetic point of view, dissolved oxygen is sufficient as an oxidant.
Unfortunately, the oxidation of As(III) by oxygen is very slow with conversion rates of a few per-
cent per day. Thus, even oxidizing groundwater with high oxygen concentrations may contain
some As(III). As(III) in its typical stability range is very often associated with dissolved iron and
manganese. As(V) is the stable species in oxidizing water and has a great similarity to ortho-phosphate
in terms of dissociation, precipitation, adsorption and ion exchange (Jekel 1994). Thus phosphate
is identified as one of the major competing ions in As(V) adsorption.
3 TREATMENT TECHNIQUES
Nearly all existing treatment technologies have been examined for their ability to remove As from
water. In addition, new techniques were developed to avoid the shorts of existing techniques. The
principle goals for As removal techniques are:
•
Efficiency: output concentrations below the standard of 10 ppb should be safely achieved,
•
Reliability and maintenance: Techniques should be reliable and adapted to the skills of the staff.

•
Residuals: Take into account the handling, treatment (if necessary) and disposal of any treat-
ment residuals, they are suspected to contain high amounts of As.
•
Costs: Investment as well as operating costs define whether the treatment technique is
economical.
190
Figure 1. Stability diagram for inorganic As (according to Ferguson & Gavis 1972, Baldauf 1995).
Copyright © 2005 Taylor & Francis Group plc, London, UK
The more promising treatment techniques for As removal, which are currently applied, are
(USEPA 2000):
(I) Technologies based on adsorption reactions:
•
Coagulation with iron or alum coagulants
•
Iron and manganese removal
•
Subterranean removal
•
Adsorption on activated alumina
•
Adsorption on iron oxide based adsorbants
(II) Other technologies:
•
Ion exchange
•
Membrane processes, i.e. reverse osmosis and nanofiltration.
Figure 2 gives schematically an overview for the classification of As-removal processes.
We distinguish between the sorption processes and the separation processes. Sorption processes
means here, that chemically active solids are involved in the process and that the contaminant is

bound by electrostatic or chemical forces to the surface. Ion exchange generally requires a charged
ion and thus, it fails completely in removing As(III) at neutral pH. The adsorption techniques are
all based on the great affinity of As species to metal oxide surfaces, to form stable surface com-
plexes (chemisorption). Some surfaces have the potential to bind As(III), but mostly to a minor
extend compared to As(V).
Arsenic removal by adsorption on metal oxides and hydroxides is the most important removal
mechanism in natural environments (Soils, lacustrine and marine sediments) and also in technical
applications. The adsorption reaction is understood as a surface complexation reaction, where As
forms inner-sphere complexes.
Metal oxide surfaces are usually protonated and contain positive and negative electrostatic
charges at the different surface sites, according to the following reactions (M stands for metals like
Fe or Al).
191
Figure 2. Classification of As-removal techniques.
Copyright © 2005 Taylor & Francis Group plc, London, UK
Protonation and deprotonation of metal oxides:
Adsorption of As(V) on metal oxide surfaces:
Protonation and deprotonation reactions are pH dependent and at a distinct pH, the isoelectric
point, the surface contains negative and positive charges to the same extend and the overall charge
is zero. The pH of the isoelectric point varies for different metal oxides. Iron(III) hydroxides and
oxihydroxides have their isoelectric point generally around pH 7–8. At greater pH, the overall sur-
face charge is negative, below, the overall charge is positive. This has a big impact on the adsorpt-
ion of As(V): At neutral pH, As(V) as an anion is negatively charged. It will adsorb very good to
a positively charged surface, but the adsorption is hindered by a negatively charged surface. The
specific adsorption of As takes place via an oxygen bridge between the metal atom and As. The
reaction given above shows a monodentate complex, but also bidentate complexes occur, which
are using two sites for one As atom (Waychunas et al. 1993).
At high pH adsorption and removal efficiency is reduced and vice versa. Thus, removal techn-
iques for As(V), which rely on adsorption work far better at low pH. In the practical consequence
one need to dose more coagulant for the same removal result, or, with fixed bed techniques, one

need to change out or regenerate the sorbent more frequently.
The membrane processes are separation processes and remove nearly all dissolved substances
from the water. They are also uneven in the removal of As(III) or As(V). Usually, As(V) is much
better removed than As(III). Uncharged molecules have, in any way, the ability to pass membranes
to some extend.
3.1 Oxidation techniques for As(III)
Most of the treatment techniques are much more effective in removing As(V) rather than As(III),
because they partially rely on electrostatic forces. Thus, there is a great discussion on oxidation
techniques for As(III). With As(III) present in a raw water, usually other reduced species are pres-
ent: ferrous iron and manganese(II). This co-occurrence is important for the evaluation of oxidat-
ion processes because iron and manganese are also subject to treatment goals. As described
previously, As(III) can not be oxidized simply by dissolved oxygen, because this reactions are
kinetically hindered and very slow. In this way, stronger oxidants have to be applied.
The following oxidants are applied successfully for As(III) oxidation (Jekel 1994):
•
Chlorine, hypochlorite
•
Ozone
•
Permanganate
•
Hydrogen peroxide
•
Manganese oxide
Chlorination is one method for As(III) oxidation, either by dosing gaseous Cl
2
or by dosing a
hypochlorite solution. The required dose is between 0.3–1 mg/L of Cl. The dose depends mainly
on the Cl-consumption by other reduced species like ferrous iron, manganese(II) and organics.
Restrictions and concerns occur about the formation of chlorination byproducts. They evolve by

the reaction of natural organic matter (NOM) with chlorine and are generally called
THM ϭ Trihalomethanes. The oxidation of As(III) by chlorine is a fast and complete reaction.
This and the fact, that chlorination is a very common chemical in the preparation and treatment of
potable waters, makes it to the most commonly applied oxidation method for As(III).
192
Copyright © 2005 Taylor & Francis Group plc, London, UK
Ozone (O
3
) is another potent oxidant, effective for As(III) oxidation. Same as Chlorine, ozone
oxidizes also other reduced species, like ferrous iron and manganese(II). The reaction mechanism
is via hydroxyl radicals. Despite the different mechanism, there are similar concerns about the side
effects as with chlorine, the oxidation byproducts. This byproducts are produced during partial
oxidation of NOM. Ozone has to be prepared on site by a high voltage discharge process and can-
not be stored in tanks. Ozone is applied in water treatment processes for nearly all oxidation
processes, also those to correct taste deteriorations by oxidizing organic compounds.
The oxidants permanganate and hydrogen peroxide are used in a low number of treatment facil-
ities to oxidize As(III). Especially permanganate has a good potential to be used for As(III) oxidat-
ion. It reacts fast, leads to an almost complete conversion and the As(III) oxidation is preferred
over other specific oxidation processes like ferrous to ferric iron (Borho 1996). The application of
permanganate requires a filtration step to remove the manganese oxide particles which are gener-
ated during its reaction, that can simultaneously be used to remove the oxidation product arsen-
ate(V) after dosing of coagulants.
A single hydrogen peroxide application cannot oxidize As(III) quick enough. If ferrous iron is
present in the raw water, hydrogen peroxide reacts with Fe
2ϩ
, which is known as the Fenton’s reac-
tion. The Fenton’s reaction generates hydroxyl radicals, which act as oxidants for As. Since some
iron hydroxide is also generated by this reaction, As(III) can be oxidized and partially adsorbed to
the suspended ferric hydroxide in one step. Same as with Ozone, the Fenton’s reaction produces
undesired byproducts by partially oxidizing NOM.

The need for an As(III) oxidation has to be carefully evaluated for several reasons:
•
The application of oxidants has sometimes undesired side effects.
•
Arsenic occurs more frequently in its oxidized form as As(V) and an oxidation is completely
useless with respect to As removal performance.
•
Some of recently developed removal techniques are also effective in removing As(III).
•
Oxidation by dosing chemicals makes the process more complex. This may be unacceptable,
when using maintenance-free fixed bed processes.
Oxidation reactions of As(III) occur also as a side effect in conventional treatment for iron and
manganese removal: As(III) is oxidized together with iron and manganese during filtration and
removed by adsorption on solid ferric hydroxide. This mechanism is reported from a number of
conventional treatment plants in Germany (Haase, personal communication), which contain high
As amounts in the residual sludge. There is the evidence of a biological aided oxidation during the
oxidation and removal of ferrous iron and/or manganese(II) (Seith & Jekel 1997, Driehaus et al.
1995). One promising technique for As(III) oxidation is also the filtration over manganese oxide
media: That can be manganese oxide coated sand (byproduct from groundwater treatment
processes) or commercially available mined manganese oxide products. For the recently devel-
oped As treatment by adsorption on iron based adsorbents, it is reported that they are also effect-
ive for removing As(III), thus eventually avoiding the need of a special oxidation treatment
(Driehaus et al. 1998).
4 ARSENIC REMOVAL TECHNIQUES
4.1 Conventional techniques
4.1.1 Coagulation/filtration
Coagulation, followed by filtration is a widely used treatment, not only for As, but especially for
the removal of colloidal substances. There exist also the terms “flocculation” and “precipitation”
for this technique, but all mean the same. The coagulation or precipitation technique relies on the
dosing of alume or iron salts to form precipitates of iron hydroxide which takes up arsenate by

adsorption. The coagulation is applied in combination with a separation technique to remove
As loaded particles from the water, i.e. by filtration over granular media, or microfiltration. The
193
Copyright © 2005 Taylor & Francis Group plc, London, UK
residual of this technique is a backwash sludge with a high water content. The sludge needs fur-
ther treatment for dewatering for easy shipping and disposal.
The coagulants of choice are iron salts like ferric chloride (FeCl
3
), ferric sulphate (Fe
2
(SO
4
)
3
),
ferrous chloride (FeCl
2
) or ferrous sulphate (FeSO
4
). The use of alum based coagulants is not rec-
ommended, because they are far less effective in removing As (Jekel 1994).
A scheme of a coagulation/filtration plant for As removal is given in Figure 3. This plant has
installed four filter vessels, with two vessels used as a two layer filter for coagulation/filtration.
Jekel & Seith (1999) reported a coagulation process at this site being effective for a groundwater
source with about 20 ␮g/L As. Arsenic was nearly completely in the oxidized form and no oxida-
tion process was applied. The performance of As removal as a function of the coagulant dose is
displayed in Figure 4 (a) for a raw water with about 50% As(III) at a pH of 6–6.2. The removal of
As(III) without oxidation is, even at high coagulant doses, very low. Fig 4 (b) shows the percentage
194
Figure 3. Scheme of the full scale As removal plant in Germany, where coagulation/filtration and adsorpt-

ion techniques were examined side by side. (acc. Jekel & Seith 1999).
0
20
40
60
80
100
120
140
024
Arsenic conc., µg/L
As conc. Without chlorine
As conc with 1 ppm chlorine
(a)
0
10
20
30
40
50
60
70
80
90
100
percent As removal
(b)
Ferric iron dose, mg/L
Ferric iron dose, mg/L
control value 5 ppb

pH: 7.9–8.0
pH: 6.0–6.2
13
0213
5
Figure 4. (a) Arsenic removal depends on coagulant dose and oxidation (after Jekel 1994); (b) Iron dose and
percent removal at the plant shown in Figure 3 (acc. Jekel & Seith 1999).
Copyright © 2005 Taylor & Francis Group plc, London, UK
of As removal versus iron dose from the full scale pilot plant illustrated in Figure 3. An acceptable
removal of 85% was achieved with a dose of 1 mg/L as Fe.
The coagulant dose depends on the raw water profile, especially on the pH and on concentrat-
ions of adsorption competitors like phosphate and silica. For ferric salts, the minimum dose can be
calculated from the As concentration. At low pH between 6 and 7 a minimum dose of ferric iron
of 10–20 times of the As concentration is required. At higher pH у 8 the minimum dose is 40–50
times the As concentration. This is only a rough estimation, and it is recommended to test the
required coagulant dose by jar tests in the laboratory.
The filter(s) separate the As bearing coagulant flocs from the liquid stream. Filters have to be
backwashed frequently to remove particles and to avoid a breakthrough of As-loaded particles.
The backwash water can be stored in a tank to allow the particles to settle down. After sedimentat-
ion the supernatant solution can be discharged or fed again in the process.
The main critical step for the implementation and optimization of the conventional coagulat-
ion/filtration treatment is the separation of the As loaded particles from the liquid stream. The filt-
ration properties of the particles depend on the chosen coagulant, the dose and the raw water
parameters like pH, hardness, alkalinity.
The residual from the coagulation process is an As bearing sludge with a high water content.
This residual needs a special treatment, at least dewatering, to be handled and transported safely.
4.1.2 Iron and manganese removal, subterranean removal
Arsenic removal during iron and manganese removal can be understood as a variation of a coagu-
lation process, because the raw water contains soluble iron, which is oxidized by aeration or by
chemical means and precipitated. This technique also needs a filtration step to remove the precip-

itates from the liquid; mostly applied is the filtration over granular media. There exists the same
dependence with regard to pH and competing ions. In general, a sufficient As removal is achieved
at a mass ratio of iron to As above 50:1. Some complication arise from the fact, that ground waters
with elevated iron concentrations mostly have reduced As(III). This leads to reduced efficiency of
the As removal, if it can not be oxidized during treatment.
Subterranean removal is a special case of iron removal. The iron containing raw water is aerated
and reinjected to the ground, where iron precipitates, adsorbs As and is filtered of by the porous
aquifer. The subterranean removal is rarely applied, because it needs special aquifer conditions
and it is thought to influence and block the aquifer’s porous structure.
4.1.3 Ion exchange and membrane techniques
Ion exchange (IX) and the membrane techniques are less important for water works operation and
the application is restricted to special conditions like household treatment in Point-Of-Use and
Point-Of-Entry devices. IX is applied as a simple filter technique and needs to be regenerated fre-
quently, the capacity is restricted and other (an)-ions are also removed.
Ion exchange (IX), especially anion exchange can effectively remove As, generally using strong
base anion exchange resins in the chloride form. However, sulfate and other anions compete with
As and can greatly reduce run length. Regeneration is inevitable for ion exchange systems. In nat-
ural water, the runlength of IX columns are only 300–1000 bed volumes. Pilot testing indicates
that the brine regeneration solution could be reused as about 20 times with no impact on As
removal provided that some salt was added to the solution to provide adequate chloride levels for
regeneration. This mode of operation would reduce the amount of waste for disposal, but increases
the ultimate As concentration of the spent brine. Disposal routes for the spent brine become criti-
cal in assessment of the viability of IX for specific treatment situations. The brine can be treated
with ferric coagulant to remove the As from the liquid waste. The additional complexity intro-
duced by brine handling and disposal may make IX unattractive for small systems. IX is widely
applied and accepted in the USA for potable water treatment. Thus it seems quite attractive to use
this technique also for As removal, despite of its shorts as restricted capacity and generation of liq-
uid wastes. There are also a couple of efforts to make improvements to this process. Kim et al
(2003) suggested an IX process with a complete brine recycling and treatment, which might
195

Copyright © 2005 Taylor & Francis Group plc, London, UK
overcome the shorts with liquid waste handlings, but could not improve the run length of less than
100 bed volumes.
Membrane techniques are not selective to As and other toxic contaminants, but remove most of
the dissolved substances and reduce the TDS and conductivity drastically. The membrane techn-
iques remove As(III) to a minor extend and produce a constant stream of waste water, where all
removed substances are concentrated. Reverse osmosis (RO) and nanofiltration (NF) provide
effective removal of As(V) and nearly all other dissolved species. These technologies are interest-
ing options for point of use and point of entry applications at low flowrates, particularly when As
is just one of several water quality parameters requiring treatment. For larger flows, the 15% to
30% feed flow lost as reject (concentrate waste liquid) is an important consideration, as is the
50–150 psi operating pressure required even for modern high efficiency low pressure RO operat-
ing on low total dissolved solids feed water. For sites with several water quality issues to address,
particularly if these are related to dissolved solids, reverse osmosis or nanofiltration have the
attraction of providing complete treatment in a single process step.
4.1.4 Activated alumina
The adsorption techniques in the stricter sense rely on a simple filtration process over granular
adsorbents like activated alumina or granular iron oxides and ferric hydroxides. Dosing of chem-
icals is usually not required. These techniques do not produce a backwash sludge, the residual is
the As loaded adsorbent itself. Activated alumina is known since more than 20 years as a good
adsorbent for arsenate containing waters, but needs a regeneration, due to the restricted lifetime of
the media until exhaustion. It is reported that the optimal pH for As removal with activated alu-
mina is around 6.0. Thus a pH-adjustment by adding mineral acids or CO
2
could increase the treat-
ment capacity drastically. Figure 5 shows the treatment capacities of activated alumina with a
model raw water containing As(III) or, after oxidation, As(V). Obviously is activated alumina not
effective in adsorbing As(III).
Activated alumina can be regenerated by rinsing with diluted caustic soda at a concentration of
2–4%, followed by rinse with 2% sulfuric to equilibrate to neutral pH. This causes some problems

because it produces a significant loss in capacity and, more important, liquid waste streams highly
enriched with As.
Recently, various activated aluminas were pilot tested with ground water from Arizona. They
exhibited treatment capacities between 1000 and 4000 bed volumes at raw water pH of 7.5–9
(Chang et al., in press). The advantages of activated alumina, simple filter operation, and the shorts,
restricted capacity, lead to the development of iron oxide based adsorbents in granular form, as it
was expected that they have a much higher capacity, while being suitable for fixed bed operation.
4.2 Emerging techniques: Iron oxide based adsorbents
The technique with iron oxide based adsorbents was in 1991–1994 developed at the Technical
University of Berlin, Department of Water Quality Control, to meet the new treatment goals of the
196
0
20
40
60
80
100
120
0 5000 10000 15000 20000 25000 30000 35000
Effluent arsenic, µg/L
Bed volumes
As(III)
As(V)
Figure 5. Process life of activated alumina with 100␮g/L As(III) and As(V) at pH 6 (Frank & Clifford 1986).
Copyright © 2005 Taylor & Francis Group plc, London, UK
lowered As standard in Germany (Driehaus et al. 1998). The technique provides a simple filtration
process over granular adsorbent medias, commonly without any dose of chemicals and without pH
adjustment. The following description focuses on GEH, one of several available types of iron
oxide based adsorbents. Other iron oxide based adsorbents are currently under development and
examination (Zeng 2002).

GEH is a pure ferric hydroxide in granular form. The grain size is 0.32–2 mm and the specific
surface is 250–300 m
2
/g. The maximum As adsorption density is 55 g/kg, the typical adsorption
from drinking water applications is 1–10g/kg. The adsorption densities, that are calculated from
batch tests at different pH values are given in Figure 6. The adsorption density at a residual con-
centration of 10 ␮g/L is plotted against pH for both As(III) and As(V). At low pH, the adsorption
density of As(V) is much higher than of As (III), but at slightly alkaline pH, adsorption is nearly
equal for both oxidation states of As.
The adsorption density and the lifetime until exhaustion in a treatment plant increases with
decreasing pH for As(V). It is quite constant for As(III). The batch tests were prepared simulating
a typical groundwater chemistry with an electrical conductivity of 480␮S/cm.
Natural water has some constituents which interact with the ferric hydroxide surface and lead
more or less to a reduction in adsorption density for As. The most important interfering substances
are phosphate, dissolved organic matter and at low pH. Also high silica amounts may interfere
with As adsorption, reducing the adsorption kinetics and the treatment capacities (Waltham &
Eick 2002). For practical evaluations, for the comparison of different media and for economic cal-
culations, the treatment capacity, expressed in bed volumes, rather than the adsorption density is a
useful expression. Figure 7 shows a typical scheme of an installation and an installation in a 32
year old building. Currently more than 80 plants with granular ferric hydroxide are installed
worldwide. We give an overview on operational data for 24 plants in EU states.
The design flow rates of the yet installed treatment plants are between 4 and 800m
3
/h and the
annual supply is from 10,000m
3
to 7,000,000 m
3
. All plants usually work in a non-continuous
mode, i.e. with supply by night to a reservoir. The empty bed contact time (EBCT) can be

expressed in two values: (a) hydraulic EBCT at design flow and (b) average EBCT which includes
also the regular down time of the plant. This value is calculated from the annual supply and the bed
volume. Figure 9 displays both EBCT values from 24 treatment plants.
The hydraulic EBCT at design flow of the plants varies between 3 and 10 minutes, the mean
value is 4.2 minutes. The average EBCT varies between 4 and 17 minutes, with a mean of 10 min-
utes. The small EBCT leads to quite small plant designs and footprints, which allow to reduce
installation costs.
The As concentrations are between 10–40 ␮g/L in the raw water. The raw water pH values are
in the range of 6.5–8 for the drinking water applications. The treatment capacities, expressed as
197
0,1
1
10
100
5,5 6,5 7,5 8,5 9,5
pH
Adsorption density, g/kg
q(As V)
q(As III)
Figure 6. Adsorption densities with granular ferric hydroxide for As (III) and As (V) in a typical ground-
water at different pH values.
Copyright © 2005 Taylor & Francis Group plc, London, UK
bed volumes (BV) treated until exhaustion vary between 50,000 BV and more than 200,000 BV.
The term exhaustion, as used in this paper, is not a unique parameter for all plants but depends on
the preferences of the clients. Some prefer to exchange the media 5␮g/L As, others operate the
plants close to the standard of 10 ␮g/L.
Figure 9 shows the treatment capacities plotted against raw As and against raw water pH. Lower
pH-values correlate well with high treatment capacities, but there is obviously only a small ten-
dency of decreasing capacities at high As concentrations.
198

Figure 7. Scheme for an As removal plant with granular ferric hydroxide (GEH
®
) and As treatment
plant with granular ferric hydroxide, design flow 45m
3
/h, installation with two vessels in parallel operation.
0,0
2,0
4,0
6,0
8,0
10,0
12,0
14,0
16,0
DEH
VIL
KSO
TRH
SUH
BOW
OSW
LIN
MIS
ELK
BGA
WAK
SGL
MAH
MAA

NOB
WHA
SK-B
RGL
M3
M4
M9
MFH
EGM
Hydraulic EBCT
Average EBCT
Minutes
Figure 8. Empty bed contact time at design flow (hydraulic EBCT) and the yearly average value (average
EBCT) (After Driehaus2002, updated).
Copyright © 2005 Taylor & Francis Group plc, London, UK
4.2.1 Performance of arsenic treatment with granular ferric hydroxide
In order to predict treatment capacities and performance of the applications with granular ferric
hydroxide, we developed a mathematical tool to calculate this data from the raw water quality. It
was the aim of this tool to use only a small set of quality parameters, which were always and eas-
ily available. From the view of practicability, including more parameters does not necessarily lead
to more accurate results, because parameters are not always reliable in terms of analytical and
sampling errors. For example, there is no detailed knowledge how dissolved organic carbon
(DOC) affects As adsorption. On the other hand, we have reliable data about the important comp-
etitors in As adsorption. From this competing ions, phosphate as the most important competitor is
included in the tool, whereas the influence of sulphate and fluoride are far behind and are not
included. The influence of silica as a competing and interfering substance is not yet fully under-
stood, as either adsorption or precipitation reactions are involved.
Results from the kinetic model and monitoring data from treatment plants are presented in
Figures 10 and 11. Figure 10 shows the data from plant OSW. Plant OSW has a raw water pH of
7.1, As 14 ␮g/L, phosphate 280␮g/L and a total hardness of 90mg/L CaCO

3
. The hydraulic
EBCT is 2.8 minutes, and the plant is in operation for 10 h per day. The initial breakthrough occurs
at 50,000 BV and is somewhat earlier than predicted by the kinetic model, but the further increase
is less steep. After 140,000BV, the effluent concentration remains constant at 5 ␮g/L, which
corresponds to 60% removal. This is in sharp contrast to the mathematical prediction, which
199
0
50000
100000
150000
200000
250000
300000
01020304050
Treatment Capacity in BV
0
50000
100000
150000
200000
250000
300000
6 6,5 7 7,5 8
Treatment Capacity in BV
Raw arsenic, µg/L Raw water pH
Figure 9. Treatment capacities versus raw As and versus raw water pH of 24 treatment plants with granular
ferric hydroxide.
0
5

10
15
20
0 50.000 100.000 150.000 200.000 250.000
Arsenic, µg/L
Raw Water
Treated Water
Calculated
Bed Volumes
Figure 10. Treatment data from the full scale plant OSW at a raw water pH of 7.1 and calculated exhaustion.
Copyright © 2005 Taylor & Francis Group plc, London, UK
shows the steepest increase at around 60 to 50% removal. Thus the treatment capacity is much
higher than predicted. One lifetime cycle with the granular ferric hydroxide lasts between 2.5 and
3 years, as the municipality decided to keep the As in potable water below 6 ␮g/L.
This behavior of a steady state adsorption can not be explained by adsorption alone, which is
here understood as monomolecular covering of the internal surface. If an adsorbent has a fixed
number of adsorption sites, the ongoing operation and removal consumes free adsorption sites and
the number of remaining sites decreases. Thus, the further uptake should decrease and lead to
increasing effluent concentrations. We conclude, that there is another mechanism behind the
adsorption in a monomolecular layer on the adsorption sites. We assume either multilayer adsorpt-
ion, surface precipitation or an uptake of adsorbed species into the hydroxide structure and the
forming of a new mineral phase.
Treatment results from two lifetime cycles in Plant KSO are displayed in Figure 11, together
with a predicted exhaustion curve. Plant KSO has a raw pH of 7.9, average raw As of 18␮g/L and
a raw phosphate of 200 ␮g/L. The plant has an EBCT at design flow of 7.4 minutes and is oper-
ated for 14 h per day.
Initial effluent data are not completely reported from this plant, but the initial breakthrough is
at 40,000 BV. This is followed by an increase up to 10–11 ␮g/L at 90,000 BV. Afterwards, the efflu-
ent concentration remains constant at this level, showing a steady state adsorption with 40%
removal. The mathematical model predicts a breakthrough for 10 ␮g/L at 80,000 BV and a further

increase of the effluent As. Although the operational and raw water data are quite different from
plant OSW, we have the similar effect at plant KSO at a higher concentration. Future research on
this steady state adsorption is needed, to understand this effect. Up to date, this effect has been
seen also at other localities using granular ferric hydroxide and in some lab scale column tests.
This examples show, that a kinetic model with a small number of input parameters is able to
describe the exhaustion in full scale applications, but with a tendency to underestimation.
5WASTE STREAM CONSIDERATIONS
The element As can not disappear during treatment and will be accumulated in sludges, adsor-
bents, ion exchangers or membrane concentrates. All treatment techniques must produce waste
streams, that are enriched in As. Table 1 gives an overview of the waste producing characteristics
and the usual way of handling and disposing.
Residuals from the conventional coagulation with ferric salts, thickened sludge, and exhausted
iron oxide based adsorbents contain As which is bound to the iron hydroxide surface. Iron hydroxide
is stable in oxidic environments, but will be dissolved and decomposed in reducing environments,
and the toxic ingredients will be dissolved. Following also the pH-dependence of As uptake of iron
200
0
5
10
15
20
0 50.000 100.000 150.000
Arsenic, µg/L
Average Raw Water
Treated Run 1
Treated Run 2
Calculated
Bed Volumes
Figure 11. Treatment data from the full scale plant KSO at a raw water pH of 7.9 and calculated exhaustion.
Copyright © 2005 Taylor & Francis Group plc, London, UK

oxide based adsorbents, As will be desorbed, if the pH of the surrounding solution is increased.
Förstner & Haase (1998) concluded, that the consequence for a safe disposal is to keep this mater-
ials under oxidizing and neutral conditions. A disposal together with municipal waste, rich in
organics will undoubtedly produce As-rich leachates, as well will do the co-disposal with caustic
materials like fly ash.
There is no common rule how to handle with the As treatment wastes. The conventional and
alternative disposals given above are only a suggestion for more or less safe disposal or discharge.
As it seems that the decision for one As treatment technique will be quite difficult, it will be as dif-
ficult to define the way of a safe and environmentally friendly waste disposal. In case of failing,
one may create a bigger problem anywhere, either downstream in a supply chain or “downstream”
in the environment.
6 CONCLUSION
The introduction into the aquatic chemistry of As showed that oxidation techniques for As(III)
have to be evaluated in each individual case. Most of the presented, conventional and new treatm-
ent techniques have a better performance with As(V) in the raw water. Several proven oxidation
methods exist, either by addition of chemicals or as fixed bed technologies. Application of strong
oxidants like chlorine and ozone have side effects like formation of oxidation byproducts. One has
to decide whether this is acceptable for a given application. Chlorine and hypochlorite seem to be
the most frequently used oxidants for As(III) oxidation, because most waterworks staff are used to
201
Table 1. Waste streams of As-removal techniques, their properties and ways of discharge and disposal.
Toxic
Type of waste
As-removal waste (passes Conventional Alternative
technique stream Properties TCLP test)* Treatment disposal disposal
Coagulation/ Ferric Fluids Unknown Dewatering, Sewer, Brick manu-
filtration sludge Water content drying landfill after facturing**)
up to 97% dewatering
Sensitive for
redox changes

Activated Alkaline Fluids Not Neutralization, Treated
alumina with and acidic applicable precipitation liquids: Sewer
regeneration fluids with ferric Residual:
salts landfill?
Iron oxide Exhausted Solids, Passes None Landfill Brick manu-
based adsorbent water Ͻ50% TCLP test facturing*)
adsorbents Sensitive for Immobilization
redox changes and utilization
Ion exchange Liquid, Fluids Not Precipitation Treated Efforts for brine
high saline applicable with ferric liquids: Sewer recycling to
brines salts or brine minimize
discharge line amounts
Residuals:
landfill?
Membrane Concentrate Fluids Not Not used Sewer or
techniques: liquids applicable brine
RO, NF discharge?
* TCLP . Toxicity Characteristic Leaching Procedure (US-EPA).
** Rouf & Hossain 2003.
Copyright © 2005 Taylor & Francis Group plc, London, UK
handle chlorine. On the other hand, some countries (i.e. Germany) refuse to use chlorine for oxid-
ation purposes.
The decision on a suitable treatment method for As removal has to consider not only the raw
water profile, treatment goals and flow rate, but also the infrastructure, availability of trained staff
and raw materials, as well as disposal of waste streams. Usually this concentrates the choice to 2
or 3 different removal technologies.
Coagulation followed by filtration is one of the conventional treatment techniques, which is
especially suitable for large drinking water plants with well trained staff. Coagulation is based on
dosing ferric or ferrous salts as coagulant, which precipitates, adsorbs As and builds up flocs,
which are removed by filtration. Coagulation is in most cases effective to remove As, and typical

doses of iron are in the range of 1–10 mg/L. As(III) is less efficiently removed and requires oxid-
ation upstream. This technique requires an additional treatment for the As loaded backwash
sludge, at least a dewatering. Coagulation could achieve low operation costs, as the ferric solutions
are usually quite cheap. It is rarely applicable for small remote facilities, because the handling and
dosing of corrosive chemicals requires a minimum infrastructure and trained staff. Nevertheless,
there is a “manual” variation of coagulation applied in Bangladesh. This technique is applied by
batchwise adding the chemicals packed for such a small volume to raw water in a bucket, stirring
and waiting, and then filtering the water through sand in a second bucket.
Ion exchange is given some attention in the USA, but bed life cycles of about 300–1000 BV
between regenerations cause a lot of costly maintenance work for regeneration. Ion exchange as
well as membrane techniques remain a domain of small household units, which could be operated
automatically. Both are not selective to As and remove other anions or, with reverse osmosis, lead
to a demineralised water.
Activated alumina has been applied for As removal since more than 20 years. It exhibits bed life
cycles of 1000–10000BV until it needs to be regenerated or disposed. As activated alumina is
sometimes available from local manufacturers at reasonable cost, it could be the method of choice
for small facilities, especially those with neutral to acidic raw waters and As(V).
The new techniques with iron oxide based adsorbents, which were developed and introduced into
the market during the last 8 years could overcome the shorts of the previously mentioned techniques.
The actually best medias have bed life cycles of 100,000 to 300,000 BV until exhaustion. Exhausted
adsorbents are disposed of and replaced by factory fresh adsorbent. After 7 years of experience with
granular ferric hydroxide they have proven their high performance under various conditions in
Europe and overseas. The operation and maintenance of treatment plants is quite simple and fulfills
all expectations of a simple adsorption process. No chemicals need to be dosed to the water and
the only task for maintenance is backwashing of the adsorber vessel on a monthly frequency. All
drinking water standards were met in all plants and the breakthrough of As in the treated water is
very slow. The operational cost for treatment plants with iron based adsorbents are mostly compar-
able to coagulation plants, and for small facilites they could be the most economic solution.
An interesting variation of iron based adsorbents is the use of natural and locally occurring iron
oxides and hydroxides. Singh et al. (1988) investigated As(III) removal by hematite ores. Another

variation is the use of metallic iron, as proposed by Khan et al. (2000). Metallic or zerovalent iron
oxidizes in the raw water and the developed iron hydroxides acts as an adsorbent for As.
During the last years, especially driven by the implementation of a new As standard in the USA,
came up a number of new adsorption medias for As removal. Some are still under development or only
lab tested, others are commercially available. Also some of the offered adsorbent medias are not really
suitable for As removal, others have been proven under pilot plant conditions to have a suitable per-
formance. Iron oxide based adsorbents will probably play an important role in As removal in the future.
REFERENCES
Anderson, L.D.C. & Bruland, K.W. 1991. Biogeochemistry of arsenic in natural waters: The improtance of
methylated species. Environ. Sci. Technol. 25(3): 420–427.
202
Copyright © 2005 Taylor & Francis Group plc, London, UK
Baldauf, G. 1995. Aufbereitung arsenhaltiger Grundwässer. GWF Wasser Special 136(14): 99–110.
Borho, M. 1996. Arsenentfernung in Grundwasserwerken durch optimierte Kopplung von Oxidations und
Fällungs-/Flockungsverfahren. Bericht aus der Wassergüte- nd Abfallwirtschaft der Technischen
Universität München, 127 (Doctoral thesis).
Chang, J. 2004. Demonstration of emerging technologies for arsenic removal. Research Foundation Report.
119p. AWWARF-AWWA, Denver.
Driehaus, W., Seith, R. & Jekel, M. 1995. Oxidation of arsenate(III) with manganese oxides in water treat-
ment. Water Research 29(1): 297–305.
Driehaus, W., Jekel, M. & Hildebrandt, U. 1998. Granular ferric hydroxide–a new adsorbent for the removal
of arsenic from natural water. Jour. Water SRT-Aqua 47: 30–35.
Ferguson, J.F. & Gavis, J. 1972. A review of the arsenic cycle in natural waters. Water Res. 6: 1259–1274.
Förstner, U. & Haase, I. 1998. Geochemical demobilization of pollutants in solid waste–implications for
arsenic in waterworks sludges. Journal Geochemical Exploration 62: 29–36.
Frank, P. & Clifford, D. 1986. Arsenic(III)-oxidation and removal from drinking water. USEPA 600-52-86/021.
Jekel, M.R. 1994. Arsenic removal in water treatment. In: Nriagu J. (ed.) Arsenic in the Environment, Part 1,
Cycling and Characterization: pp 433–446, John Wiley, New York.
Jekel, M. & Seith, R. 1999. Comparison of conventional and new techniques for the removal of arsenic in a
full scale water treatment plant.–World Water Congress 1999, Buenos Aires, Proceedings.

Khan, A.H., Rasul, S.B., Munir, A.K.M., Habibuddowla, M., Alauddin, M., Newaz, S.S., Hussam, A. 2000.
Appraisal of a simple arsenic removal method for groundwater of Bangladesh. J. Environ. Sci. Health,
Part A: Toxic/Hazardous Substances and Environmental Engineering 35(7): 1021–1041.
Kim, J., Benjamin, M.M., Kwan, P. & Chang, Y. 2003. A novel ion exchange process for As removal. Journal
American Water Works Association 95(3): 77–85.
Pierce, M.L. 1981. Chemical modeling of arsenic in aqueous systems: Ph.D. Thesis, Arizona State University.
Rouf, M.A. & Hossain, M.D. 2003.Effects of using arsenic iron sludge in brick making. Presented at the
BUET-UNU International Symposium Fate of Arsenic in the Environment, February 2003 Dhaka,
Bangladesh. URL: />Seith, R. & Jekel, M. 1997. iologische Oidation von Arsen(III) in Festbettreaktoren. Vom Wasser 89: 283–296.
USEPA 2000.Technologies and costs for removal of arsenic from drinking water, EPA 815-R-00-028.
Singh, D.B., Prasad, G., Rupainwar, D.C. & Singh, V.N. 1988. As(III)-removal from aqueous solution by
adsorption. Water Air Soil Pollution 42: 373–386.
Waltham, C.A. & Eick, M.J. 2002. Kinetics of arsenic adsorption on goethite in the presence of sorbed silicic
acid. Soil Sci Am J. 66: 818–825.
Waychunas, G.A., Rea, B.A., Fuller, C.C. & Davies, J.A. 1993. Surface chemistry of ferrihydrite: Part 1.
EXAFS studies of the geometry of coprecipitated and adsorbed arsenate. Geochim. Cosmochim. Acta 57:
2251–2269.
WHO 2001. United Nations synthesis report on arsenic in drinking water. .
Zeng, L. 2002. Preparation of an iron(III) oxide adsorbent for arsenic removal. URL: http://www.
globe2004.com/2002/presentations/zeng.doc (Accessed on October 4, 2004).
203
Copyright © 2005 Taylor & Francis Group plc, London, UK