Removal of inorganic anions from drinking water supplies by membrane bio/processes

21
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/226421354 Removal of inorganic anions from drinking water supplies by membrane bio/processes Article in Reviews in Environmental Science and Bio/Technology · December 2004 DOI: 10.1007/s11157-004-4627-9 CITATIONS 48 READS 178 3 authors, including: Svetlozar Velizarov New University of Lisbon 90 PUBLICATIONS 1,314 CITATIONS SEE PROFILE Joao Crespo New University of Lisbon 248 PUBLICATIONS 4,619 CITATIONS SEE PROFILE All content following this page was uploaded by Svetlozar Velizarov on 01 December 2016. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

Transcript of Removal of inorganic anions from drinking water supplies by membrane bio/processes

Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/226421354

Removalofinorganicanionsfromdrinkingwatersuppliesbymembranebio/processes

ArticleinReviewsinEnvironmentalScienceandBio/Technology·December2004

DOI:10.1007/s11157-004-4627-9

CITATIONS

48

READS

178

3authors,including:

SvetlozarVelizarov

NewUniversityofLisbon

90PUBLICATIONS1,314CITATIONS

SEEPROFILE

JoaoCrespo

NewUniversityofLisbon

248PUBLICATIONS4,619CITATIONS

SEEPROFILE

AllcontentfollowingthispagewasuploadedbySvetlozarVelizarovon01December2016.

Theuserhasrequestedenhancementofthedownloadedfile.Allin-textreferencesunderlinedinbluearelinkedtopublicationsonResearchGate,lettingyouaccessandreadthemimmediately.

Removal of inorganic anions from drinking water supplies

by membrane bio/processes

Svetlozar Velizarov*, Joao G. Crespo & Maria A. ReisCQFB/REQUIMTE, Department of Chemistry, FCT, Universidade Nova de Lisboa, P-2829-516 Caparica,Portugal (*author for correspondence, e-mail: [email protected])

Received 29 June 2004; accepted 8 October 2004

Key words: Donnan dialysis, drinking water, electrodialysis, inorganic anionic pollutants, integratedprocesses, membrane bioreactors, nanofiltration, ultrafiltration, reverse osmosis

Abstract

This paper is designed to provide an overview of the main membrane-assisted processes that can be used forthe removal of toxic inorganic anions from drinking water supplies. The emphasis has been placed onintegrated process solutions, including the emerging issue of membrane bioreactors. An attempt is made tocompare critically recently reported results, reveal the best existing membrane technologies and identify themost promising integrated membrane bio/processes currently being under investigation. Selected examplesare discussed in each case with respect to their advantages and limitations compared to conventionalmethods for removal of anionic pollutants. The use of membranes is particularly attractive for separatingions between two liquid phases (purified and concentrated water streams) because many of the difficultiesassociated with precipitation, coagulation or adsorption and phase separation can be avoided. Therefore,membrane technologies are already successfully used on large-scale for removal of inorganic anions such asnitrate, fluoride, arsenic species, etc. The concentrated brine discharge and/or treatment, however, can beproblematic in many cases. Membrane bioreactors allow for complete depollution but water quality,insufficiently stable process operation, and economical reasons still limit their wider application in drinkingwater treatment. The development of more efficient membranes, the design of cost-effective operatingconditions, especially long-term operations without or with minimal membrane inorganic and/or biologicalfouling, and reduction of the specific energy consumption requirements are the major challenges.

Abbreviations: D – dialysis; DD – Donnan dialysis; DMB – dialysis membrane bioreactor; ED – electro-dialysis; IEMB – ion exchange membrane bioreactor; MCL – maximum contaminant level; MF – micro-filtration; NF – nanofiltration; RO – reverse osmosis; TOC – total organic carbon; UF – ultrafiltration; USEPA – United States Environmental Protection Agency; WHO – World Health Organization

1. Introduction

A number of inorganic anions have been found inpotentially harmful concentrations in numerousdrinking water sources (DeZuane 1997; Smithet al. 2002; Petrovic et al. 2003). The maximumallowed concentrations of these compounds aregenerally set by the drinking water quality regu-latory standards in the relatively low lgÆl)1–mgÆl)1

range; therefore, the majority of them can bereferred to as micropollutants. The internationallyaccepted standards and guidelines, regarding themaximum allowed levels of these compounds, areproposed by the World Health Organization(WHO). In addition, the European Union and theUS Environmental Protection Agency have issuedsimilar health and environmental standards and aconsiderable number of regulatory methods have

Reviews in Environmental Science & Bio/Technology (2004) 3: 361–380 � Springer 2005DOI: 10.1007/s11157-004-4627-9

been published worldwide for the analysis ofinorganic anions in drinking water (US EPA 1998;Jackson 2001). Table 1 lists the current status forpotentially toxic inorganic anions along with someinformation about the main sources of pollutionand the potential health risks, associated with theiringestion in drinking water

Since there are usually no organoleptic changesin drinking water that can be attributed to thepresence of toxic inorganic anions in trace levels, itis rather possible that some of them may remainundetected, thus increasing the possible healthrisks. An example of such recently found com-pound is perchlorate, an important ingredient ofsolid rocket propellants, which may interfere withthe ability of the thyroid gland to utilize iodine inhormones production (Richardson 2003; Minet al. 2004). Since perchlorate is a serious problemin some US regions, a provisional drinking watergoal of 1 lg/l has been suggested (US EPA 2002).

A number of inorganic anionic contaminantscan be present at the same time in rather differentlevels (e.g. nitrate and perchlorate), thus leading tothe emerging issue of their control and simulta-neous removal from drinking water supplies.Finally, water of defined ion composition isrequired in the manufacturing of a number of foodproducts, pharmaceuticals and in the fresh waterfisheries and sea aquariums.

Several common treatment technologies arenowadays used for removal of inorganic contam-inants from water supplies. Large-scale plantsusually apply coagulation with aluminium or ironsalts followed by filtration but a number of anions(e.g. nitrate) have very little tendency to coordi-nate with metal ions and low potential forco-precipitation (Duan & Gregory 2003). Small-scale treatment facilities often use ion exchangeand/or adsorption due to their ease of handlingand compactness; however, regeneration andadditional costs, associated with the disposal ofthe regenerants used, represent serious problems.Moreover, release of undesirable organics (such asstyrene, divinylbenzene, trimethylamine, etc.) fromsome synthetic resins to the treated water stillhinders a larger application of ion exchange fordrinking water production (Kapoor & Viraragha-van 1997).

Membrane separation processes such as reverseosmosis (RO), nanofiltration (NF), ultrafiltration(UF), microfiltration (MF), dialysis (D), Donnan

dialysis (DD) and electrodialysis (ED), if properlyselected, offer the advantage of producing highquality drinking water. In many cases, one mem-brane process can be integrated with another toproduce water of even higher quality. In theseprocesses, the membrane can be viewed as a bar-rier between contaminated and purified waterstreams. The separation of the two streams oftenallows for operation with no or minimal chemicalwater pre-treatment, which otherwise can formdeleterious by-products (Bergman 1995; Jacangeloet al. 1997). However, in physical membrane pro-cesses, inorganic anions are not destroyed butnormally concentrated and the concentrate dis-posal can be costly and difficult to permit in manycases; therefore, post-treatment of the concentratestream or hybrid membrane-assisted technologiescapable of converting anionic contaminants toharmless products are highly desirable.

Chemical, electrochemical or biochemicaltreatment processes are able to deal efficiently withanions (Daub et al. 1999; Carraro et al. 2000;Centi & Perathoner 2003). Among them, biologi-cal reduction is especially appropriate since itoffers selective removal of the target anion fromwater due to anoxic bacteria, which under appro-priate conditions (pH, oxidation–reductionpotential, temperature, etc.) can use anions aselectron acceptors and organic (heterotrophicmicroorganisms) or inorganic (autotrophicmicroogranisms) compounds as electron donorsfor their growth. However, the main concern ofusing bioprocesses is the risk of secondary waterpollution by cells, incompletely degraded nutrientsand metabolic by-products, which can promotemicrobial growth in water distribution systems,thus requiring extensive post-treatment in order toproduce safe and biologically stable water. Theseproblems can be overcome, or at least reduced, byintroducing a membrane unit as a pre- or post-treatment stage in the water production process.When membrane(s) are integrated with an appro-priate bioprocess in a single process, the configu-ration is generally referred to as ‘‘membranebioreactor’’.

In this paper, discussion on useful applicationsof membrane processes for removal of inorganicanionic contaminants from drinking water, with aspecial emphasis on the emerging issue of mem-brane bioreactors, is presented. Since it is ratherdifficult to select which membrane applications

362

Table

1.Standardsandguidelines

forinorganic

anionsin

drinkingwater(W

HO

2003),main

sources

ofcontaminationandpossible

healthrisks

Anion

Chem

icalform

ula

Main

sources

ofcontamination

Maxim

um

acceptable

standard

orguidelineb

level,mg/l

Possible

adverse

healtheff

ects

ofhigher

levels

Arsenate

aH

2AsO

4);HAsO

42);

AsO

43)

Naturalerosion(m

ostly

presentin

surface

waterunder

aerobic

conditions)

0.01b(asAs)

Kidney

damage;

skin

cancer

Arsenitea

H2AsO

3);HAsO

32);

AsO

33)

Dissolutionofrocksandminerals(m

ostly

presentin

groundwater)

0.01b(asAs)

Kidney

damage;

skin

cancer

Bromate

BrO

3)

Disinfectionby-product

ofbromide

(ifnaturallypresent)generatedduring

ozonation

0.025b

Possibly

cancerogenic

Chlorite

ClO

2)

Disinfectionby-product

inwaters

treated

withchlorinedioxide

0.2

bNervoussystem

effects;anem

ia

Chromate

CrO

42)

Metal-workingandminingindustries

0.05

b(asCr)

Liver

damage;

kidney

damage

Cyanide

CN

)NNaturalsources;somefoods;

electroplating,steelandmining

industries;certain

fertilizersandpesticides

0.07

Weightloss;thyroid

effects;Neurological

effects

Fluoride

F)

Naturalsources;productionofaluminium,

phosphate

fertilizers,whichmaycontain

up

to4%

fluorine.

1.5

Pittingofteeth,bonefluorosis

Nitrate

NO

3)

Naturaleutrophication;inorganic

fertilizers;humanandanim

alwastes

50

Mainly

attributable

toitsreduction

to

NO

2)

Nitrite

NO

2)

Chloramination;nitrificationin

water

distributionsystem

s

0.2

bMethaem

oglobinaem

iac,whichcould

be

fatalto

infants

Molybdade

MoO

42)

Naturalsources;nearsomeminingsites

0.07(asMo)

Anaem

ia;Diarrhoea;Highlevelsofuric

acidin

theblood

Selenate

SeO

42)

Naturalsources;somefoods

0.01(asSe)

Skin,hair,andnaildamage;

canreplace

sulphurin

biomolecules

Selenite

HSeO

3)

Naturalsources;leached

from

certain

agriculturalsoils

0.01(asSe)

Perchlorate

dClO

4)

Military

operations;somefertilizers

0.001b

Inhibitsiodide

uptake

inhumansand

anim

als

aTheprevailingform

dependsonpH.

bProvisionalvalue(thisterm

isusedforconstituents

forwhichthereissomeevidence

ofapotentialhazard

butwheretheavailable

inform

ationonhealtheff

ects

islimited).

cOxidationofnorm

alhaem

oglobin

(Hb)to

methaem

oglobin

(metHb),whichisunable

totransport

oxygen

tothetissues.

dStillnotregulatedbytheWHO;thisprovisionalvalueisissued

bytheUSEPA

(EnvironmentalProtectionAgency).

363

might be referred as specifically anion removalones (other water constituents are often alsoremoved), the authors’ opinion, in some instancesmight be subjective. In most cases, only recentexamples are referred to since the articles selectedoften familiarize the interested reader with thegeneral history of a specific problem. Where it isappropriate, the discussion starts with a givenmembrane separation used as a single process andthen moves on to its integration in membranebioprocesses. An attempt is made to identify theadvantages, limitations and future research needsin each case. Some of the processes discussed at thepresent time could be considered still far frompractical implementation, but nevertheless, theymight provide useful information, on which futuredevelopments could be based.

2. Pressure-driven membrane processes and

membrane bioreactors

Pressure-driven membrane processes use pressuredifference between the water to be treated and apermeate side as the driving force to transportwater across the membrane. These membraneprocesses include RO, NF, UF and MF. Theoperating trans-membrane pressure ranges varysignificantly but are usually: 20–100 bar for RO,5–20 bar for NF, 2–5 bar for UF, and 0.1–2 barfor MF. It has to be mentioned that while MFand UF membranes have well-defined porousstructure, for RO and NF membranes the term‘‘pores’’ may be better associated with the intra-molecular voids within the polymeric matrix.Another possible classification can be based onthe molecular mass of the solute to be separatedby a given membrane process that is usually of upto 100 Da (RO), 100–500 Da (NF) and 500–10,0000 Da (UF). Obviously, UF and MF arenot suitable for the direct removal of inorganicanions from solution; however they can beimplemented in hybrid processes, which producelarger aggregates (subsequently filtered) (Hanet al. 2002; Yoon et al. 2003) or in bioprocesseswith the purpose to retain microbial cells, usinganions as electron acceptors. On the other hand,RO and NF can be used for removal of inorganicanions as single processes. The advantages andlimitations of each process will be discussed in thefollowing sections.

2.1. Reverse osmosis

Reverse osmosis is a well-established technologyused for many years in water desalination. Reverseosmosis membranes are asymmetric, i.e., consist ofa thin polymer (nowadays, mostly polyamide)layer combined with a porous support to guaran-tee the membrane mechanical stability. The ROmembranes discriminate on the basis of molecularsize and due to the dense properties of the sepa-rating layer very high (often close to 100%)retention of low-molecular mass compounds andions (total desalination) can be achieved. More-over, the process can be easily automated andcontrolled.

The treated water stream, however, may lackthe right balance of minerals and has unpleasingtaste because of the retention of all ions (Nicoll2001). Another disadvantage of RO is the highenergy consumption needed to maintain therequired pressure difference. Reverse osmosismembranes are very sensitive to polarizationphenomena (ions accumulation) at the membranesurface contacting the concentrate (retentate) side.In addition, the solubility products of some salts inthe retentate can be exceeded, forming precipitates(mineral fouling) besides possible biological foulingdue to natural organic matter and microogranismspresent. The presence of non-toxic components,such as hardness (Ca2+, Mg2+) and SO4

2) anions,can interfere with the separation of toxic anionicspecies due to problems that these components maycause with water recovery and ionic strength(osmotic pressure) (Ritchie & Bhattacharyya 2002).Therefore, the contaminated water usually requirespre-treatment (other membrane processes like MFand/or UF are nowadays more and more used)before entering the RO modules.

Concerning the RO applications in the case oftoxic inorganic anions, few studies have been re-cently performed mostly aiming at the removal ofarsenic species and nitrate (Table 2). Brandhuberand Amy (1998) showed that if the main arsenicspecies are present as As(III) only RO membraneswould be effective. In a pilot-scale study, removalefficiencies between 96–99% for As(V) and 46–84% for As (III) have been reported (Ning 2002).Pre-oxidation of As(III) to As(V) could guaranteebetter removal.

It was demonstrated that RO could be effec-tively applied for removal of nitrate along with

364

water desalination in a rural area (Schoeman &Steyn 2003). The nitrate removal efficiency wasclose to 98% and, although the total dissolvedsolids (TDS) in the treated water stream werestrongly reduced (from 1292 mg/l in the sourcewater to 24 mg/l in the RO permeate), the authorssuggested that the water could be used directly forpotable purposes. Preliminary estimates showedthat for an approximately 2 m3/h output plant, thecapital and operating cost were about USD 29,900and 0.50/m3, respectively.

It can be concluded that RO is a highly efficientprocess for removal of inorganic anions fromdrinking water, which guarantees a secure detoxi-fication of the water supply. However, total desa-lination is undesired due to possible corrosiveproblems if water hardness is reduced to very lowlevels. Water with hardness values under 50 mg/l isexpected to be corrosive (lead, copper, iron, zinc,etc.) (DeZuane 1997). Therefore, modifications,allowing for selective toxic anions removal alongwith a sufficient retention of water salinity arerequired.

2.2. Nanofiltration

Nanofiltration (NF) uses membranes, which canprovide selective desalination, and is usuallyapplied to separate multi-valent ions from mono-valent ones; however, it is also possible to achievea certain separation of ions of the same valence byselecting the proper membrane and operatingconditions (Lhassani et al. 2001). NF membranesare sometimes designated as ‘‘loose’’ RO mem-branes (Ho & Sirkar 1992), since they providehigher water fluxes at lower trans-membranepressures. These membranes are usually asym-metric and negatively charged at neutral andalkaline drinking water pH. Therefore, separationof anions is based not only on different rates oftheir diffusion through the membrane (at lowpressure), convection (at high pressure), but alsoon repulsion (Donnan exclusion) between anionsin solution and the surface groups, which is obvi-ously higher for multi-valent anions (Levensteinet al. 1996). The advantage of introducing thisadditional mechanism of ion exclusion (in additionto the size-based exclusion) is that high ion sepa-ration degrees (ion rejections) similar to those inRO can be achieved but at higher water fluxesthrough the membrane. On the other hand, the NFT

able

2.Pressure

)driven

mem

braneprocesses

forremovalofinorganic

anionsfrom

drinkingwater

Process

Mem

braneandmanufacturer

Target

anion(s)

Waterorigin

Reference

RO

4040-LHA-C

PA2(H

ydranautics)

NO

3)

Naturalwith188mgNO

3)/l(South

Africa)

Schoem

anandSteyn(2003)

RO,NF

Differentmem

branes

(Osm

onics)

NO

3)

Tapwater(Poland)

Bohdziew

iczet

al.(1999)

RO,NF,UF

Differentmem

branes

andmanufacturers

Asspecies

Pilotstudiesatvarioussites(U

SA)

Brandhuber

andAmy(1998)

RO

Differentmem

branes

andmanufacturers

Asspecies

Review

onpilotstudiesatvarioussites(U

SA)

Ning(2002)

NF

ES-10(N

itto-D

enko)

Asspecies

Groundwaterspiked

with0.6

mgAs/l(Japan)

Urase

etal.(1998)

NF

Differentmem

branes

(Nitto-D

enko)

Asspecies

Model

water

Sato

etal.(2002)

NF

FilmtecNF45(D

ow

Chem

ical)

Asspecies

Model

water

Vrijenhoek

andWaypa(2000)

NF

Nanomax50(M

illipore)

NO

3)

Model

water

Paugam

etal.(2002)

NF

Differentmem

branes

(Nitto

)Denko)

NO

3)

MFpre

)treatedsurface

water(Japan)

Thanuttamavonget

al.(2002)

NF

FilmtecNF70(D

ow

Chem

ical)

NO

3)

Groudwaterspiked

withNO

3)salts(Belgium)

Vander

Brugen

etal.(2001)

NF

NF300(O

smonics)

NO

3),F

)Groudwater(C

alifornia,USA)

Cohen

andConrad(1998)

NF

FilmtecNF45(D

ow

Chem

ical)

F)

Model

water

Diawara

etal.(2003)

NF

FilmtecNF70(D

ow

Chem

ical)

F)

Model

water

Lhassaniet

al.(2001)

NF

TFCS(Fluid

System

s)CrO

42)

Model

water

Hafianeet

al.(2000)

Surfactants

+UF

GM

(Desal);8kDacutt

)off

ClO

4)

River

waterspiked

withClO

4)(C

olorado,USA)

Yoonet

al.(2003)

Flocculation

+MF

Cellulose

acetate

mem

brane;

0.22

lm

pores

Asspecies

Naturalsource,

0.042mgAs/l(C

olorado,USA)

Hanet

al.(2002)

365

process is much more sensitive than RO to theionic strength and pH of source water. The mem-brane surface charge is mainly due to anionadsorption from water rather than to fixed chargedgroups (as in the case of ion exchange mem-branes), therefore it depends strongly on bulkanion concentration (Hagmeyer & Gimbel 1998).Furthermore, it changes from negative to zero netcharge at the membrane isoelectric point and thento positive at lower pH values (usually <4) due tocation adsorption. This pH dependence canstrongly affect the target anion separation.Therefore, the selection of adequate operatingconditions is more critical for NF applications.Despite these challenges, a number of studiesdealing with the removal of toxic anions fromdrinking water have been performed (Table 2) andmost of them have showed promising results.

Experiments with groundwater, to which arse-nate As(V) and arsenite As(III) were added, wereperformed by Urase et al. (1998), who showed thatthe As(V) rejection between 90 and 97% from thenegatively charged NF membrane used was almostnot influenced by water pH; however, the As(III)rejection increased with pH, being 50% at pH 3and 89% at pH 10. At pH 10, most of arsenite is ina mono-valent anion form, while at low pH theneutral form dominates because the pKa value ofarsenite is 9.1. For the same reasons, As(III) wasnot effectively removed in two other studies (Vri-jenhoek & Waypa 2000; Sato et al. 2002) while As(V) removal reached 90 and 95%, respectively. Itwas found that the rejection of As(III) decreasedwith increasing bulk concentration, an effect thatwas attributed to its enhanced diffusion and con-vection through the membrane under such condi-tions (Vrijenhoek & Waypa 2000).

Fluoride removal by NF has also been testedfor model water (Lhassani et al. 2001) and con-taminated groundwater (Cohen & Conrad 1998).In a model study, NF showed potential to selec-tively separate the following single halide salts:NaF, NaCl, NaI, LiF and LiCl (Diawara et al.2003). The fluoride selectivity was higher at lowfluxes, where the chemical parameters (e.g.hydration energy, partition coefficient) are pre-dominant and larger ions were less retained.Therefore, fluoride, with a hydration energyhigher than those of chloride and iodide, wasbetter retained despite the fact that it is thesmallest anion.

Nitrate removal from drinking water by NFhas been the subject of several studies (Table 2).A reasonably high nitrate rejection of 76% wasobserved by Van der Bruggen et al. (2001), whoalso performed preliminary cost analysis, showingthat for a water treatment capacity of 2000 m3/hthe operating cost would be approximately0.13/m3. However, the hardness rejection was

very high, up to 95%, therefore the treated watershould be re-mineralised to obtain a hardness ofapproximately 2 mmol/l. It was suggested thatthe development of new membranes with thesame rejection for neutral molecules but withlower rejection for charged compounds wouldallow to obtain a permeate with hardness readyfor distribution.

Nanofiltration is rapidly becoming more andmore attractive alternative to the traditional ROwater treatment. To a great extent, this is due tothe introduction of highly efficient NF membranesand module configurations, which allow lowerinvestment and operating costs. Nowadays, theworld’s largest NF plant treating surface water(from the river Oise) is located in France andcomprises over 9000 Filmtec NF200 membranemodules able to produce 140,000 m3 of drinkingwater per day (Nicoll 2001).

2.3. Pressure-driven membrane bioreactors

The concept of integration of biological treatment(aerobic or anoxic) of polluted water with its fil-tration across a porous membrane, driven by apressure difference was first applied for the sepa-ration of activated sludge by an UF membraneand its recycling to the aeration tank (Smith et al.1969), and since that time has been exploitedextensively in urban and industrial wastewatertreatment (Bouillot et al. 1990; Brindle & Ste-phenson 1996; Gander et al. 2000; Ben Aim &Semmens 2002; Cicek 2003; Wintgens et al. 2003).The polluted water and microbial culture are in thedirect contact, while the product water is ‘‘forced’’through the membrane due to pressure difference(Figure 1). The main advantage is the possibilityof achieving high biomass concentrations withinthe bioreactor; therefore, the plant size can bereduced. As a result of membrane separation, thebiomass retention time is independent from thehydraulic retention time, thus allowing slow-growing microorganisms to be maintained in the

366

bioreactor. This feature is of particular importancefor toxic compounds and/or micropollutants,which usually need long periods for their completebiological degradation/transformation to harmlessproducts. Previous articles dealing with pressure-driven membrane bioreactors categorized suchsystems in terms of the manner in which themembrane module is integrated (e.g. an externalmembrane module, or membranes directlyimmersed in the culture medium, a configurationknown as an ‘‘submerged membrane bioreactor’’).The latter configuration appears to be less energy

consuming and less detrimental for the biomass(Ben Aim & Semmens 2002).

Following the successful application of pres-sure-driven membrane bioreactors in wastewatertreatment, the first attempts to extend thisapproach to the production of drinking water weredone in the 1990s (Table 3). Chang et al. (1993)studied denitrification of tap water supplementedwith nitrate salts, which was continuously pumpedto an agitated anoxic bioreactor coupled to anexternal hollow fibre UF membrane (a nominalporosity of 0.01 lm) module, through which theculture medium was recycled. The carbon andphosphorus needed for bacterial growth were pro-vided by ethanol and phosphoric acid, respectively.Efficient denitrification (NO3

) and NO2) concen-

tration bellow the respective MCLs) was achievedat a maximum permeate flow rate of more than100 l/(m2 h) and nitrate removal rate of 11 g/(m2 h). The most serious detected problem wasrelated to the strong decrease of water filtrationrate from 120 to 45 l/(m2 h) after 10 days ofoperation because of membrane fouling. Therefore,a backwashing procedure (for 12 seconds every12 min) had to be implemented, thus increasingthe process complexity and energy demands.Furthermore, a secondary pollution of the treatedwater by 1.5–2.1 mg C/l (as total organic carbon)was found. While this TOC level was much lowerthan the one in the bioreactor (40–50 mg/l), it wasstill high enough to promote secondary microbialgrowth in the distribution systems.

Using the same concept but with a plane UFmembrane module (cut-off diameter of 200 kDa),Delanghe et al. (1994) also obtained a highly effi-cient denitrification of tap water, supplementedwith nitrate salts, and reached a maximum per-meate flow rate of about 21 l/(m2 h) and nitrateremoval rate of 3 g/(m2 h). They dealt with thefouling problem by washing weekly the mem-branes with 3% NaClO solution pumped throughthe modules. However, the TOC concentrations inthe treated water varied between 5–10 mg C/l.Since no ethanol was found, this TOC increase wasattributed to low-molecular mass organic com-pounds, lower than the membrane cut-off diame-ter. The authors concluded that whatever itsorigin, this TOC level made the water unsuitablefor drinking and needs further treatment.

Barreiros et al. (1998), using acetate as thecarbon source and electron donor, succeeded to

η

Water

µ

Water

P

Water

Dialysis membrane bioreactor

Ion exchange membrane bioreactor

Gas-permeable membrane bioreactor

Pressure-driven membrane bioreactor

H2

Water

Gas-

-

Figure 1. Membrane bioreactor configurations. DP – Pressurediffence; Dl – chemical potenital difference; Dg – Electro-chemical potential difference. The bigger circles represent toxicanions and the smaller ones – driving counter-ions (non-toxicanions, e.g. Cl)). The ellipses illustrate microorganisms present.

367

denitrify efficiently naturally contaminatedgroundwater. They used a hollow fibre polysulfoneUF membrane module with a cut-off of 500 kDaand obtained reasonably high water permeate flowrate of about 30 l/(m2 h) at a nitrate removal rateof 4.5 g/(m2 h). TOC values of 1.5–2.0 mg C/lwere determined in the treated water, which weresimilar to those measured by Chang et al. (1993).

The dosage of electron donor to the contam-inated water has to be carefully controlledin response to the electron acceptor (anion) con-centration. On-line monitoring of the target anionconcentration in the feed water and adding thecarbon source as a function of that concentrationby means of an adaptive control system couldsolve this problem. Such control is importantsince electron donor limitation would lead toincomplete anion reduction, while overdosingcould promote microbial growth in water distri-bution systems. However, the capability of apressure-driven membrane bioreactor to provideTOC free water would be still questionable becauseof possible transport through the membrane oflow-molecular organic compounds (other thatthe carbon source) as noted by Delanghe et al.(1994).

Recently, the removal of nitrate from syntheticfeed water using a sulfur-based autotrophic deni-trification in a pressure-driven membrane biore-actor utilizing a rotating UF membrane disks witha cut-off of 750 kDa was tested (Kimura et al.2002). While the obtained water permeate flowrate of about 20 l/(m2 h) is within the expectedrange, this study is interesting due to the use ofmembranes directly immersed in the bioreactor.Membrane fouling was not severe and cleaning byincreasing the disk rotation velocity from 200 to400 rpm for 15 min every 3 days allowed for astable process operation for more than 3 months.However, assimilable organic carbon was stilldetected in the treated water. Fluorescence spec-troscopy data showed differences between feedwater and filtrate, however, the authors could notidentify or characterize the organic matter fromthe contour plots obtained.

Besides their direct application for drinkingwater production, as discussed above, pressure-driven membrane bioreactors have been occa-sionally used to treat concentrated water streams,obtained after applying other processes. Anadvantage of such ‘‘associations’’ is that biologicalT

able

3.Pressure-driven

mem

branebioreactors

forremovalofinorganic

anionsfrom

drinkingwater

Process

description

Mem

branetypeandmanufacturer

Electrondonor

Waterorigin

Reference

Waterdenitrification

+UF

Derivatizedcellulose

(Aquasource)

;

0.01

lmpores

Ethanol

Tapwaterspiked

with

NO

3)(France)

Changet

al.(1993)

Waterdenitrification

+UF

UFP2(Tech-Sep);200kDa

cutt-off

Ethanol

Tapwaterspiked

with

NO

3)(Japan)

Delangheet

al.(1994)

Waterdenitrification

+UF

Polysulfone;

500kDacutt-off

Acetate

Groundwater(Portugal)

Barreiroset

al.(1998)

Waterdenitrification

+UF

Polysulfone;

Rotatingsubmerged;

750kDacutt-off

Sulfur

Model

solutions

Kim

ura

etal.(2002)

ED

brinedenitrification+

UF

Ceramic

mem

brane;

0.05

lmpores

Ethanol

Groundwater(France)

Wisniewskiet

al.(2002)

368

treatment can be applied to solutions whosecompositions and temperature can be controlledand is independent of possible changes in thecharacteristics of water to be treated. For example,nitrate-containing brine, obtained after an elec-trodialysis treatment of groundwater, contami-nated by agricultural and farm activities, wasstudied (Wisniewski et al. 2002). A ceramic mem-brane with an average pore size of 0.05 lm, peri-odically regenerated by chemical means, was used.An almost total removal of nitrates (99%) wasachieved by the mixed culture despite the presenceof other ions in relatively high concentrations. Thedrinking water quality is determined by the pro-cess used (electrodialysis in the case studied).

Overall, in pressure-driven membrane bioreac-tors, fouling was found to affect the process per-formance either due to the deposit of a layer at themembrane surface and/or by partial or completeblockage of the pores. However, fairly efficientsolutions (e.g., periodic membrane backwashing)can be implemented so that relatively highamounts of water may be treated per unit area ofmembrane. With the reduction of the price ofcommercial membranes, the process energy con-sumption is usually the most important economicfactor.

A general limitation of the pressure-drivenmembrane bioreactors studied until now is thetreated water quality. While contamination ofwater with microbial cells and biopolymers can beavoided, the retention of ions and low molecularmass compounds (such as some metabolicby-products) by porous membranes is generallyinsufficient to meet the stringent drinking watercriteria; therefore either process modifications orwater post-treatment are necessary.

3. Dialytic membrane bio/processes

Dialysis uses a semi-permeable membrane to sep-arate compounds due to their different rates ofdiffusion in the membrane. Since the dialysis pro-cess driving force is the chemical potential differ-ence (in opposition to pressure difference for RO,NF, UF and MF), one of the main differencesbetween pressure-driven membrane processes anddialyitic processes is that the solvent (water) passesthrough the membrane with more or less selectivesolutes (ions) retention in pressure-driven

processes, while the solutes pass through themembrane with more or less selective transfer indialysis. In practice, dialysis is used for separationof compounds, which differ significantly in size inorder to guarantee a large difference in diffusionrates; the classical example is hemodialysis (artifi-cial kidney) for purifying human blood byremoving small solutes such as urea, chloride, etc.,while retaining large proteins and other compo-nents in the blood. Therefore, the ability of dialysisto discriminate between different ions in water islimited. Furthermore, the flux of a given com-pound depends on its concentration gradientthrough the membrane, thus dialysis is character-ized by considerably lower flux rates in compari-son to pressure-driven membrane separations.Therefore, despite its great pharmaceuticalimportance, dialyisis has not been applied as asingle process in drinking water treatment. On theother hand, attempts have been made to separateanaerobic mixed microbial culture from water tobe treated using microporous hydrophobic mem-branes, based on a dialysis mode of operation, aconfiguration, which might be designated as adialysis membrane bioreactor. Besides the chemi-cal potential (instead of pressure) difference usedas the driving force, another important differencebetween this type of bioreactor and the pressure-driven membrane bioreactor is that the water to betreated and the microbial culture are separated bythe membrane in different compartments(Figure 1).

3.1. Dialysis membrane bioreactors

Nitrate removal was studied in a system, in whichmodel water and denitrifying microorganisms wereseparated by a microporous membrane made of anexpanded polytetrafluoroethylene material with amean pore size of 0.02 lm and an effectiveporosity of 50% (McCleaf & Schroeder 1995;Mansell & Schroeder 1999). Because the mem-brane material is hydrophobic, its wetting withmethanol was required to allow the pores to fillwith water. Nitrate was removed by diffusionthrough the membrane and into the culture,forming a biofilm on the membrane surface. Anitrate removal efficiency approaching 90% wasachieved at a nitrate removal rate of 0.7 g/(m2 h)and a treated water production rate of 9 l/(m2 h).Since nitrate was the only ion analysed in these

369

studies, the treated water quality with respect toother ions, which are expected to enter the waterstream from the more concentrated culture med-ium due to their concentration gradients cannot beevaluated. Particularly, a possible nitrite appear-ance would be highly undesirable since it is muchmore toxic than nitrate itself (Table 1).

Methanol (used as the electron donor andcarbon source) caused secondary contamination ofthe treated water, which showed an average TOCvalue of 4 mg C/L (Mansell & Schroeder 1999). Inorder to avoid the addition of an organic electrondonor and the associated treated water secondarycontamination, an hydrogenotrophic culture wasapplied in a recent study of the same group(Mansell & Schroeder 2002); however, membranewetting with methanol was still required.

Since in a simple dialysis mode of operationusing a microporous membrane, the rate of iontransport is controlled by the size of the water-filled pores together with a given ion concentrationgradient across the membrane, the process selec-tivity towards target anionic pollutant(s) isexpected to be low. Therefore, processes utilisingmembranes capable of separating ions not onlyaccording to their size (sieving effect) but alsousing other differences in their physical and/orchemical properties may provide better waterquality.

3.2. Donnan dialysis

Ion exchange membranes combine the ability toact as a separation barrier between two solutions,with the chemical and electrochemical propertiesof ion exchange materials, the most important onebeing their permeability to ions of the oppositecharge (counter-ions) while excluding ions of thesame charge (co-ions). The two different types ofion exchange membranes used are known as‘‘homogenous’’ and ‘‘heterogeneous’’ membranes.While the homogenous ones are coherent gels,usually in planar form, the heterogeneous mem-branes consist of ion exchange particles embeddedin an inert matrix (polystyrene, polyethylene, etc.).The recent synthesis of homogenous membraneswith greatly improved mechanical membranestogether with their high conductivity makes thempreferable in most applications.

Donnan dialysis (DD) is a process that uses anion exchange membrane without applying an

external electric potential difference across themembrane (Cwirko & Carbonell 1990). The ions,which are permeable to the membrane, will equil-ibrate between the two solutions until the Donnanequilibrium is satisfied. Therefore, in Donnandialysis (opposite to simple dialysis), the separa-tion achieved between the solutions is not lost evenif the system is closed to the surrounding, a featurecharacteristic of equilibrium separation processessuch as absorption, extraction and distillation (Ho& Sirkar 1992). On the other hand, in simple rate-controlled membrane separation processes, such asdialysis, all separation achieved is lost after sometime if the system is closed to the surrounding. TheDonnan dialysis type of operation requires theaddition of a so-called driving counter-ion to thestripping solution, which is transported in adirection opposite that of the target ion in order tomaintain electroneutrality. Since concentrationratios determine the Donnan equilibrium, notconcentration differences, Donnan dialysis allowsfor transport of charged micropollutants againsttheir concentration gradients, which is of utmostimportance for drinking water supplies, whichusually contain only trace levels of polluting ions.Furthermore, the hydraulic residence time can beindependently adjusted in the two compartmentsto optimise the degree of extraction of the targetpollutant. Due to these characteristics, the removalof inorganic anions from drinking water, especiallynitrate and fluoride, by Donnan dialysis hasreceived attention (Table 4).

Salem et al. (1995) studied nitrate removal in aDonnan dialysis system utilising an ADS-Morg-ane anion-exchange membrane and chloride asthe driving counter-ion in the strip solution. Themaximum nitrate removal rate reached 9 g/(m2 h); however bicarbonate and sulfate wereco-extracted from the drinking water, thus sig-nificantly reducing the process selectivity towardsnitrate. Therefore, the authors recommended theuse of anion exchange membranes of highernitrate selectivity.

Fluoride removal by Donnan dialysis wasinvestigated and mathematically modelled for abi-ionic system (NaF as the feed and NaCl as thestrip) by Dieye et al. (1998) and later on tested insynthetic drinking waters with compositions closeto those found in some natural waters in Africa(Hichour et al. 2000). A fluoride concentration inagreement with the norm ( <1.5 mg/l) was

370

reached in the latter study and the addition of acomplex-forming ion such as Al3+ (to obtainfluoride complexes, which are not able to crossthe membrane) to the strip solution allowed tomaintain a low free fluoride concentration in thestrip and a reasonably high process driving force.A pilot-scale plate-and-frame DD module, con-sisting of eleven cells (five feed and six strip cells)separated by DSV anion exchange membranes(Asahi Glass was tested and the maximum treatedwater production rate reached 2.5 l/(m2 h); how-ever, the treated water salinity was increased byabout 25% due to electrolyte leakage from thestrip to the feed solution. It may be concludedthat the Donnan co-ion exclusion provided bythis type of membrane was not sufficiently high.The use of a Neosepta ACS membrane (Tokuy-ama) was able to solve this problem in a sub-sequent study by Garmes et al. (2002), who alsoinvestigated a combination of a Donnan dialytictransport of fluoride with its adsorption on Al2O3

or ZrO2, added to the strip solution, instead ofcomplex-forming agents.

Overall, Donnan dialysis has shown potentialfor removal of inorganic anions from drinkingwater, especially from waters of low salinity. Theuse of ion exchange membranes offers the advan-tage of steady state operation without regenerationcycles typical for ion exchange processes with ionexchange beads, which produce large volumes ofrinse water that has to be discarded. The choice ofmembrane (preferably non-porous and, if possible,selective towards the target polluting anion) andthe driving counter-ion type (chloride appears tobe a good candidate) and its concentration in thestripping solution are of major importance.Selective anion exchange membranes, having sig-nificantly different transport properties for mono-valent and multi-valent ions are already available(Sata 2000). However, the mono-valent anionfluxes through these membranes are relatively low.Thus, the higher quality of the treated waterstream, retaining or only slightly changing its ori-ginal composition with respect to multi-valentions, is obtained on the price of a low mono-valentpollutant(s) transport rate. Hybrid processesallowing for the transport of a target anion to thestripping solution, where it can form complexes,precipitate, or adsorb on a suitable carrier, couldprovide higher driving forces and thereforeimprove the anion flux.T

able

4.Donnandialysis(D

D)andelectrodialysis(ED)processes

usedforremovalofinorganic

anionsfrom

drinkingwater

Process

Mem

braneandmanufacturer

Target

anion

Waterorigin

Reference

DD

Neosepta

AFN;Neosepta

AFX

(TokuyamaSoda)

F)

Model

water

Dieyeet

al.(1998)

DD

Selem

ionDSV

(AsahiGlass)

F)

Model

water

Hichouret

al.(2000)

DD

+adsorption

Neosepta

ACS(TokuyamaSoda)

F)

Groundwater(M

orocco)

Garm

eset

al.(2002)

ED

Neosepta

AFN;Neosepta

ACS(TokuyamaSoda)

F)

Brackishwater

Amoret

al.(2001)

DD,ED

ADS(M

organ)

NO

3)

Waterfrom

Montpellier

(France)

Salem

etal.(1995)

ED

Monovalentanionperm-selective(N

otreported)

NO

3)

Groundwater(A

ustria)

Hellet

al.(1998)

ED

Neosepta

ACS(TokuyamaSoda)

NO

3)

Groundwater(M

orocco)

Elm

idaouiet

al.(2002)

ED

Neosepta

ACS(TokuyamaSoda);Selem

ionAMV

(AsahiGlass)

NO

3)

Model

water

Vander

Brugen

etal.(2004)

ED

Notreported

(Ionics)

ClO

4)

Groundwater(U

tah,USA)

Roquebertet

al.(2000)

DD

Neosepta

ACM;AFN

(TokuyamaSoda);SB-6407(G

elman)

Cr 2O

72)

Model

water

Cengeloglu

etal.(2003)

371

3.3. Ion exchange membrane bioreactor

A new membrane bio/process has been proposed(Crespo et al. 1999; Crespo & Reis 2001) for theremoval and bioconversion of ionic micropollu-tants from water streams. In this process, the ionicmicropollutant is transported from the waterstream through a non-porous ion-exchange mem-brane into a biological compartment where it issimultaneously converted by a suitable microbialculture into harmless products. The essential dif-ference of this configuration from the dialysismembrane bioreactor is that the driving force forpollutant transport through the membrane is theanion electrochemical potential difference insteadof its chemical (concentration) difference. There-fore, as discussed previously for Donnan dialysis,pollutant transport against its concentration gra-dient is possible due to the presence of drivingcounter-ions in a higher concentration (Figure 1).The co-ions (cations) are excluded from the posi-tively charged membrane and the target anion(s)transport is combined with its bioconversion byanoxic mixed microbial culture fed with an ade-quate carbon source and other required nutrientsin a continuous mode. The selection of a non-ionizable carbon source is preferable in order tominimise its transfer through the membrane. Inaddition, the bioconversion of the pollutant in theIEMB keeps its concentration at low levels andguarantees an adequate driving force for trans-port.

This concept was first demonstrated in syn-thetic waters (Table 5) for the removal and bio-conversion of nitrate to harmless nitrogen gasusing Neosepta ACS mono-anion permselectivemembrane and ethanol as the carbon source(Fonseca et al. 2000; Velizarov et al. 2000). Due toits very low diffusion coefficient (three orders ofmagnitue lower than that in water) through thisnon-porous type of membrane, and the develop-ment of an ethanol-consuming biofilm on themembrane surfface contacting the biocompart-ment, carbon source penetration into the treatedwater was avoided. The mechanism of anionicpollutant transport in the IEMB was identified anda simplified bi-ionic transport model, which maybe applied as a first step in the process analysis,was elaborated using chloride as the major coun-ter-ion for nitrate exchange. The effect of the mostimportant process parameters were assessed by the

model and experimentally validated for a steady-state process operation (Velizarov et al. 2003a).Within the concentration domain relevant tonitrate polluted water (50–350 mg NO3

) l)1),complete denitrification was achieved withoutaccumulation of NO3

) and NO2) ions in the bio-

compartment. A maximum water production rateof 33 l/(m2 h) and a nitrate removal rate of 3.5 g/(m2 h) were obtained.

The biofilm contribution to the improvement ofthe nitrate flux to the biocompartment was time-dependent being most significant in the early stageof the process, corresponding to a thin and activelymetabolizing biofilm. Excessive biofilm growthmay become a problem in later stages because ofthe increase of the corresponding mass transferresistance, thus reducing the transport rate of thepollutant. Therefore, periodical removal of theaccumulated biofilm is necessary.

The process performance when a third counter-ion, in addition to nitrate and chloride, is presentin the IEMB system and the respective control ofthe anion fluxes was also investigated (Velizarovet al. 2002). Bicarbonate was chosen since it is themajor anion metabolically produced in the systemdue to ethanol oxidation. The possibility to regu-late the magnitude and direction of the net flux ofbicarbonate, which depends on the chloride con-centration in the biocompartment was demon-strated. When chloride is not added or itsconcentration is sufficiently low, bicarbonate maybe used as a driving counter-ion for transport ofnitrate to the biocompartment. With increasingchloride concentration, bicarbonate penetration tothe treated water can be totally prevented and, ifnecessary, its co-extraction with nitrate from thetreated water can be achieved. In this way, thewater composition with respect to bicarbonate(which also affects the treated water pH) can beadjusted to the desired level.

High-salinity water after oceanarium fish cul-tivation, polluted with nitrates was also shown tobe effectively denitrified according to the IEMBconcept (Velizarov et al. 2003b).

Recently, the possibility of treating waterstreams, contaminated with several ionic micro-pollutants, was proved for the case of simulta-neous removal and bioconversion of nitrate (in themg/l range) and perchlorate (in the lg/l range)(Velizarov et al. 2004). Despite these very differentconcentrations, both contaminants were success-

372

fully reduced to nitrogen gas and chloride,respectively. The provisional low perchlorate limitof 4 lg/l in the treated water was achieved with theimportant observation that any temporal accu-mulation of perchlorate in the biocompartmentdid not influence the treated water quality. Thisbeneficial effect was due to the Donnan dialyticcontrol of perchlorate transport (flux against itsconcentration gradient) and sufficiently high driv-ing ion (chloride) concentration in the IEMBbiocompartment.

The main advantage of the simultaneoustransport of a target ionic micropollutant througha dense membrane followed by its bioconversion inan isolated compartment is that it allows for theisolation of the microbial culture from the feedstream behind the membrane barrier, thus avoid-ing the contamination of the treated water withcells, metabolic by-products and excess carbonsource, which, in turn, prevents the secondarycontamination of the treated water. While a pres-sure-driven membrane bioreactor generally allowsfor highest treated water production rate per unitof membrane area, the IEMB system is particu-larly indicated if high water quality is envisageddue to the complete retention of the biomass andavoiding treated water secondary pollution byresidual organics.

Another advantage of the IEMB system, incomparison to the pressure-driven membranebioreactor configuration, is the lower energy con-sumption (only necessary for pumping the solu-tions).

Reduction of the price of the currently avail-able anion exchange membranes, especially thosehaving mono-anion permselectivity, which areexpensive in comparison with membranes used inpressure-driven processes, would make the IEMBprocess economically compatible. Moreover, dueto their typical use in electrodialysis, these mem-branes are generally available as flat sheets, whileother configurations would lead to more compactsystems allowing for better fluid dynamic condi-tions in the water compartment.

4. Electrodialytic membrane processes

As already discussed in Section 3.2., Donnandialysis is a process, that can be used to extractanions from drinking water by means of an anionT

able

5.Dialysismem

branebioreactors

(DMB)andionexchangemem

branebioreactor(IEMB)forremovalofnitrate

from

drinkingwater

Process

description

Mem

branetypeandmanufacturer

Electrondonor

Waterorigin

Reference

DMB

PTFE(W

.L.Gore

&Assoc.);0.2

lmpores

Methanol

Tapwaterspiked

withNO

3)(U

SA)

McC

leafandSchroeder

(1995)

DMB

PTFE(G

ore-Tex);0.02

lmpores

Methanol

Model

water

MansellandSchroeder

(1999)

DMB

PTFE(G

ore-Tex);0.02

lmpores

Hydrogen

Model

water

MansellandSchroeder

(2002)

IEMB

Neosepta

ACS(TokuyamaSoda)

Ethanol

Model

water

Fonseca

etal.(2000)

IEMB

Neosepta

ACS(TokuyamaSoda)

Ethanol

Model

water

Velizarovet

al.(2000)

IEMB

Neosepta

ACS(TokuyamaSoda)

Ethanol

Tapwaterspiked

withNO

3-(Portugal)

Velizarovet

al.(2004)

373

exchange membrane. However, since the mecha-nism of ion transport is governed solely by theDonnan equilibrium principle, anion separationsbeyond that thermodynamic limit are not possible.Since the driving force is determined by thedeparture of the system from the equilibrium, inDonann dialysis the anion fluxes achieved may below for certain applications. In electrodialysis(ED), the transport of ions present in contami-nated water is accelerated due to an electricpotential difference applied externally. In thisprocess, besides anion exchange membranes, cat-ion exchange membranes are also applied in orderto transport cations to the cathode. Process dete-rioration due to membrane scaling is a frequentlyobserved problem, therefore, the ED systems areusually operated in the so-called electrodialysisreversal mode, in which the polarity of the elec-trodes is reversed several times per hour to changethe direction of ion movement. The external elec-tric potential driving force allows to obtain higheranion fluxes than those in DD, but a differentdegree of demineralisation (cations are alsoremoved from the water) depending on the voltageand type of the membranes used is obtained. Whenthe purpose is the removal of inorganic toxicanion(s), reduction in water hardness could be adesired side effect in some cases but in others maycause a too deep softening (as in RO treatment);therefore the suitability of ED depends strongly onthe polluted water ionic composition. Successfulapplications of ED include removal of variousanions, e.g. fluoride, perchlorate and especiallynitrate (Table 4).

An ED reversal pilot unit was installed at anuncontaminated well, providing groundwater withhigh hardness, alkalinity, and total dissolved solids(TDS) (Roquebert et al. 2000). Various perchlo-rate levels (of up to 130 lg/l) were spiked into thefeed water to test the ED process performance. Itwas found that the removal efficiency of perchlo-rate was high (�90%) but similar to that of chlo-ride. If higher perchlorate removal is required,additional stages of ED reversal units (up to four)provided a more cost-effective alternative than ionexchange polishing. Some sorption of perchlorateto the membranes occurred. The operating costwas between USD 0.30 and 0.40/m3.

ED studies were performed to remove fluoridefrom brackish water containing 3000 mg/l of TDSand 3 mg/l of fluoride (Amor et al. 2001). The use

of a mono-anion permselective membrane (Neo-septa ACS) allowed to maintain the water sulfateconcentration close to its original value (only 5%was removed). The membrane transported theanions in the following order: Cl) > F) >HCO3

) > SO42).

The use of a mono-valent anion perm-selectivemembrane proved successful in a full-scale EDplant designed to remove nitrate from groundwa-ter in Austria (Hell et al. 1998). The nitrate con-centration in the raw water was 120 mg NO3

)/land the plant removal efficiency (66%) was ad-justed to obtain a product concentration of 40 mgNO3

)/l. Under these conditions, the total desali-nation degree was about 25%, therefore the nitrateselectivity was reasonably high. Sulfate was prac-tically not removed and water pH changed by0.1 units only (7.4–7.5). The concentrate brinedischarge, however, represented a challenge. Theauthors suggested that irrigation would be a pos-sible solution. Since the chloride concentrationwas 270 mg Cl)/l, the brine should be suitable fordirect irrigation, at least for chloride tolerableplants, or blended with other waters, before use.The option tested in practice was a discharge to thelocal sewage plant, which required less effort.The results showed that the total nitrate load ofthe brine (889 mg/l) was degraded in the sewer.

In summary, electrodialysis can provide anefficient removal of inorganic anions from drink-ing water. Since most known toxic anions aremono-valent (Table 1), the use of mono-valentanion perm-selective exchange membranes isespecially attractive. Situations, in which EDappears to be less applicable are for waters of verylow salinity (conductivity of less than 0.5 mS), forwhich Donnan dialysis can be a better solution,and, in cases when besides ions, removal oflow-molecular mass non-charged compounds (towhich ED is obviously ineffective) from the water isdesired. In the latter case, pressure-driven mem-brane processes as RO or NF may be preferable.The brine discharge/treatment issue, however, re-mains important for all these separation processes.

5. Gas-transfer membrane bioreactors

Originally, gas-permeable membranes were sug-gested to provide bubbleless aeration in aerobicbioprocesses (Cote et al. 1988; Semmens 1991).

374

The advantage is that if no gas bubbles areformed, medium foaming and gas stripping effectscan be avoided. Gas mass transfer can be accom-plished by gas-permeable dense membranes or byhydrophobic microporous membranes, in whichthe pores remain gas filled (Ahmed & Semmens1992). Composite membranes made of differentmaterials can provide even more flexible opera-tion.

Since inorganic anions serve as electronacceptors in anaerobic microbial processes, oxy-genation (aeration) must be prevented. Hydrogencan serve as a non-polluting and non-toxic elec-tron donor for reducing different oxy-anions byautotrophic microorganisms. However, hydrogeneconomy concept must overcome tremendousobstacles and challenges due to the high cost of H2

production from renewables, and more impor-tantly, the need to develop the required H2 infra-structure (Simbeck 2004).

Moreover, direct H2 injection into contami-nated water supplies is not acceptable because ofits low solubility and high flammability. In suchapplications (as in the case of oxygen mass trans-fer) gas-permeable membranes can offer theadvantage of efficient hydrogen mass transferwithout bubble formation, which prevents boththe waste of excess H2 and the accumulation ofexplosive amounts of H2 in the headspace abovegroundwater. Since the anionic pollutant(s) andmicrobial culture are in contact, in this aspect theconfiguration is more similar to pressure-drivenmembrane bioreactors (Figure 1).

A promising application is believed to be an insitu water remediation, i.e., at the contaminatedsite. Stimulation of naturally occurring anaerobicprocesses in situ can be an effective alternativetreatment for inorganic contaminants in aquiferssince they can be anaerobically transformed to lesstoxic and/or immobilized forms (Coates &Anderson 2000). Therefore, hydrogen gas-transfermembrane bioreactors, in which H2 is supplied tothe lumen of hollow fibre membranes and diffusesto the shell side, where it is used by microrganismsas an electron donor for the reduction of inorganicanions, have received considerable attention(Table 6).

Lee and Rittmann (2000) studied autohydro-genotrophic denitrification of drinking water in acomposite hollow-fibre membrane bioreactor. Themembrane used was made up of two materials.

The outer and inner layers were of microporouspolyethylene and, in between, a 1-mm thick layerof non-porous polyurethane allowed the creationof a high driving force for gas dissolution withoutbubble formation. Accumulation of a biofilm(thicker near the hydrogen gas inlet) on the surfaceof the hollow fibres was observed due to the partialpressure drop inside the fibres. An almost totalH2-utilization efficiency within the biofilm wasachieved at a nitrate removal rate of 0.41 g/(m2 h)and a treated water production rate of about 8 l/(m2 h). However, an increase of 0.9 mg C/l (asdissolved organic carbon) in the treated watereffluent was detected. The authors pointed out thatthis was a level high enough to result in bacterialgrowth in a distribution system, therefore, a pro-cess removing this ‘‘biological instability’’, such asrapid filtration or activated carbon adsorption,was suggested as a post-treatment step.

Later, it was shown that the most importantfactor for nitrogen removal efficiency is thehydrogen pressure, which controls the hydrogenflux into the biofilm (Lee & Rittmann 2002).Under hydrogen-limiting conditions, incompletenitrate removal degrees between 39 and 92%, andpossible appearance of nitrite in the treated waterwere observed. The optimum pH for autotrophicdenitrification was found to be in the range 7.7–8.6, with a maximum at 8.4 (Lee & Rittmann2003). Increasing the pH above 8.6 caused a sig-nificant decrease in nitrate removal rate and adramatic increase in nitrite accumulation; there-fore it was suggested that the process might requirepH control.

A hydrogen gas-transfer membrane bioreactorcontaining hydrophobic polypropylene hollowfibres with a pore size 0.05 lm was tested fornitrate removal from contaminated water by Ergasand Reuss (2001), who also observed the build upof a biofilm layer on the surface of the membranes.A pressure difference (of about 30 kPa) betweenthe shell (liquid) and lumen (H2 gas) was requiredto maintain the proper conditions for systemoperation. A bigger pressure difference resulted inwater penetration into the membranes while alower one led to bubble formation. The maximumnitrate removal rate was 0.46 g/(m2 h), whichcompares well with the values obtained by Lee andRittmann (2000); however, a very large increase inthe effluent dissolved organic carbon from 11 to31 mg C/l was detected. The authors attributed

375

this to soluble microbial products, such as proteinsand polysaccharides, leaking from the microbialcells and concluded that further treatment is nec-essary to remove biological products from thewater prior to distribution.

Haugen et al. (2002) developed a novel flow-through reactor designed to simulate groundwaterflowing through a nitrate-contaminated aquifer.This process guaranteed almost complete NO3

)

removal once H2 limitation was corrected. With amembrane module, made of silicone-coated rein-forced fibreglass fibres, a nitrate removal rate of0.8 g/(m2 h) and a treated water production rate ofabout 10 l/(m2 h) were achieved. Effluent wateranalysis indicated that the total organic carbonrose by only about 0.5 mg C/l. Since the testreactor contained aquarium rock media in order tomimic natural environment, filtration andadsorption of organic matter may had improvedthe water quality. Again, biofilm formationencouraged by the delivery of the electron donor(hydrogen) on the membrane surface wasobserved. Although, in a previous study, Semmensand Essila (2001) found that the biofim couldincrease the flux of a gaseous substrate out themembrane due to the driving force improvementeffect, this phenomenon would impend gas transferto the bulk liquid, thus decreasing the reactionzone around the membranes. This problemappears to be more important for in situ remedi-ation since natural convection in groundwater maybe poor. Furthermore, mineral precipitates on themembrane surface (e.g. ferric oxide) mightdecrease gas transfer both to the biofilm and to thebulk liquid (Roggy et al. 2002).

Following the promising results obtained withnitrate, perchlorate removal in the gas-transfermembrane bioreactor developed by Lee and Ritt-mann (2000) was studied by Nerenberg et al.(2002). The perchlorate concentration was suc-cessfully reduced to below 4 lg/l when the influentperchlorate concentration was as high as 100 lg/l.Perchlorate reduction occurred immediately uponexposure to this pollutant in nitrate-reducing sys-tem, and no inoculation with specialized perchlo-rate reducing bacteria was necessary. However,nitrate in the water slowed perchlorate reductionrate, therefore, the authors suggested that pHcontrol (at around pH 8) and/or complete nitrateremoval may be required in order to increase theperchlorate reduction rates.T

able

6.Hydrogen

gas-transfer

mem

branebioreactors

forremovalofinorganic

anionsfrom

drinkingwater

Mem

braneconfiguration

Target

anion(s)

Waterorigin

Reference

Composite

hollow

fibres

NO

3)

Tapwatersupplementedwithnitrate

salts(U

SA)

Lee

andRittm

ann(2000)

Polypropylenehollow

fibres(0.05mm

pore

size)

NO

3)

Groundwater(M

assachusetts,USA)

Ergass

andReuss

(2001)

Siliconetube

NO

3)

Model

water

Hoet

al.(2001)

Silicone-coatedhollow

fibres

NO

3)

Model

water

Haugen

etal.(2002)

Composite

hollow

fibres

ClO

4),NO

3)

Spiked

groundwater(C

alifornia,USA)

Nerenberget

al.(2002)

Composite

hollow

fibres(Pilotscale

system

)ClO

4)

Groundwater(C

alifornia,USA)

Rittm

annet

al.(2004)

376

A pilot-scale system, consisting of two gas-transfer membrane bioreactors in series, was testedto treat a perchlorate-contaminated groundwater(60 lg/l of perchlorate) in La Puente, California(Rittmann et al. 2004). The perchlorate removalwas high (95%) and the treated water perchlorateconcentration below 4 lg/l. The water flow ratewas set to 66 l/h and effluent recycle at 1020 l/hwas performed to improve mass transfer andreduce clumping of the fibres; however, periodicalair scouring to remove excessive biofilm accumu-lation was still necessary. For a polluted ground-water, containing 30 lg/l of ClO4

) and about58 mg/l of NO3

) at a water flow rate of 228 m3/h,the preliminary cost estimations are of USD933,000 total capital cost and USD 0.21/m3 oper-ating cost. The authors pointed out that the con-centration of nitrate controls the operating costsince it is the dominant electron acceptor and itsreduction is the largest H2 consumer.

It may be summarised that the gas-transfermembrane bioreactor is a promising technologyfor the bioremediation of drinking water supplies.Until now, research has been carried out withex situ laboratory-scale reactors, but the possibil-ity for in situ implementation is rather attractive.The most serious concerns for future researchseems to be prevention, or at least minimization,of the membrane inorganic and microbial foulingand treated water quality, which has still notreached acceptably low TOC levels.

6. Conclusions

In this paper, we described some of the features ofmembrane processes, with an emphasis on mem-brane bioreactors, making their inclusion indrinking water production attractive. We alsodiscussed some current problems that still providea challenge for research.

The use of membranes in drinking watertreatment is a developing technology. NF and EDcan provide more or less selective removal of thetarget pollutants, especially when separationsbetween mono- and multi-valent anions aredesired. In NF, this is a consequence of both, ionsize and charge exclusion effects, while in ED is dueto the use of ion exchange membranes withmono-anion perm-selectivity. Higher membraneaffinity for a given anion in ED can additionally

increase the selectivity discriminating also betweenmono-anions (nitrate and chloride are goodexample).However, the concentrate brine dischargeand/or treatment can be problematic in many cases.

Combining the advantages of membrane sepa-ration with biological reactions for the treatmentof polluted water supplies has resulted in thedevelopment of three major membrane bio/pro-cesses: pressure-driven membrane bioreactors, gas-transfer membrane bioreactors, and ion exchangemembrane bioreactors. In the first case, mem-branes are essentially regarded as micro/ultraporous barriers to promote high biomass forprocess intensification and avoid contamination ofthe treated water with microbial cells. However,secondary water pollution by an incompletelydegraded organic carbon source and other low-molecular mass compounds is possible. Hydrogengas-transfer membrane bioreactors appear espe-cially attractive for in situ water remediation, whileion exchange membrane bioreactors can provide ahighly selective target ion removal and avoid sec-ondary pollution of the treated water.

What next? The immense potential for usingperm-selective membranes – acting not exclusivelybased on size exclusion – has not been yet com-pletely explored. The opportunity for improvingthe performance of the existing and developmentof novel types of membrane bioreactors is clear:the demand for selective removal of defined pol-lutants (not only ionic but polar and non-polarcompounds as well) opens new directions with theultimate goal of producing safe and high qualitydrinking water.

Acknowledgements

The financial support from Fundacao para aCiencia e a Tecnologia (Lisbon, Portugal) throughprojects POCTI/BIO/43625/2000, POCTI/EQU/39482/2001 and grant SFRH/BPD/9467/2002 isgratefully acknowledged.

References

Ahmed T & Semmens MJ (1992) The use of independentlysealed microporous membranes for oxygenation of water:Model development. J. Membr. Sci. 69: 11–20

Amor Z, Bariou B, Mameri N, Taky M, Nicolas S & ElmidaouiA (2001) Fluoride removal from brackish water by electro-dialysis. Desalination, 133: 215–223

377

Barreiros AM, Rodrigues CM, Crespo JPSG & Reis MAM(1998) Membrane bioreactor for drinking water denitrifica-tion. Biop. Eng. 18: 297–302

Ben Aim RM& Semmens MJ (2002) Membrane bioreactors forwastewater treatment and reuse: A success story. Water Sci.Technol. 47: 1–5

Bergman RA (1995) Membrane softening versus lime softeningin Florida: A cost comparison update. Desalination 102:11–24

Bohdziewicz J, Bodzek M & Wasik E (1999) The application ofreverse osmosis and nanofiltration to the removal of nitratesfrom groundwater. Desalination 121: 139–147

Bouillot P, Canales A, Pareilleux A, Huyard A & Goma G(1990) Membrane bioreactors for the evaluation of mainte-nance phenomena in wastewater treatment. J. Ferment.Bioeng. 69: 178–183

Brandhuber P & Amy G (1998) Alternative methods formembrane filtration of arsenic from drinking water. Desali-nation 117: 1–10

Brindle K & Stephenson T (1996) The application of membranebiological reactors for the treatment of wastewaters of specialinterest. Biotechnol. Bioeng. 49: 601–610

Carraro E, Bugliosi EH, Meucci L, Baiocchi C & Gilli G (2000)Biological drinking water treatment processes, with specialreference to mutagenicity. Water Res. 34: 3042–3054

Cengeloglu Y, Tor A, Kir E & Ersoz M (2003) Transport ofhexavalent chromium through anion-exchange membranes.Desalination 154: 239–246

Centi G & Perathoner S (2003) Remediation of water contam-ination using catalytic technologies. Appl. Catal. B: Environ.41: 15–29

Chang J, Manem J & Beaubien A (1993) Membrane biopro-cesses for the denitrification of drinking water supplies.J. Membr. Sci. 80: 233–239

Cicek N (2003) A review of membrane bioreactors and theirpotential application in the treatment of agricultural waste-water. Can. Biosyst. Eng. 45: 6.37–6.49

Coates JD & Anderson RT (2000) Emerging techniques foranaerobic bioremediation of contaminated environments.Trends Biotechnol. 18: 408–412

Cohen D & Conrad HM (1998) 65,000 GPD fluoride removalmembrane system in Lakeland, California, USA. Desalina-tion 117: 19–35.

Cote P, Bersillon JL & Faup G (1988) Bubble free aerationusing membranes: Process analysis. J. Water Pollut. ControlFed. 60: 1986–1992

Crespo JG, Reis AM, Fonseca AD & Almeida JS (1999) IonExchange Membrane Bioreactor for Water Denitrification.Portuguese National Patent No. 102385 N

Crespo JG & Reis AM (2001) Treatment of Aqueous MediaContaining Electrically Charged Compounds. InternationalPatent PCT-WO 01/40118 A1

Cwirko EH & Carbonell RG (1990) A theoretical analysis ofDonnan dialysis across charged porous membranes.J. Membr. Sci. 48: 155–179

Daub K, Emig G, Chollier M-J, Callant M & Dittmeyer R(1999) Studies on the use of catalytic membranes forreduction of nitrate in drinking water. Chem. Eng. Sci. 54:1577–1582

Delanghe B, Nakamura F, Myoga H & Magara Y (1994)Biological denitrification with ethanol in a membrane biore-actor. Environ. Technol. 15: 61–70

DeZuane J (1997) Handbook of Drinking Water Quality. 2ndedn. John Wiley & Sons, New York

Diawara CK, Lo SM, Rumeau M, Pontie M & Sarr O (2003) Aphenomenological mass transfer approach in nanofiltrationof halide ions for a selective defluorination of brackishdrinking water. J. Membr. Sci. 219: 103–112.

Dieye A, Larchet C, Auclair B & Mar-Diop C (1998)Elimination des fluorures par la dialyse ionique croisee.Eur. Polymer J. 34: 67–75.

Duan J & Gregory J (2003) Coagulation by hydrolysing metalsalts. Adv. Coll. Inter. Sci. 100–102: 475–502

Elmidaoui A, Elhannouni F, Taky M, Chay L, Sahli MAM,Echihabi L & Hafsi M (2002) Optimization of nitrateremoval operation from ground water by electrodialysis.Sep. Purif. Technol. 29: 235–244

Ergas SJ & Reuss AF (2001) Hydrogenotrophic denitrificationof drinking water using a hollow fibre membrane bioreactor.J. Water SRT – Aqua 50: 161–171

Fonseca AD, Crespo JG, Almeida JS & Reis AM (2000)Drinking water denitrification using a novel ion-exchangemembrane bioreactor. Environ. Sci. Technol. 2000 34: 1557–1562

Gander M, Jefferson B & Judd S (2000) Aerobic MBRs fordomestic wastewater treatment: A review with cost consid-erations. Sep. Purif. Technol. 18: 119–130

Garmes H, Persin F, Sandeaux J, Pourcelly G & Mountadar M(2002) Defluoridation of groundwater by a hybrid processcombining adsorption and Donnan dialysis. Desalination145: 87–291

Hafiane A, Lemordant D & Dhahbi M (2000) Removal ofhexavalent chromium by nanofiltration. Desalination 130:305–312

Hagmeyer G & Gimbel R (1998) Modelling the salt rejection ofnanofiltration membranes for ternary ion mixtures and forsingle salts at different pH values. Desalination 117: 247–256

Han B, Runnells T, Zimbron J & Wickramasinghe R (2002)Arsenic removal from drinking water by flocculation andmicrofiltration, Desalination 145: 293–298

Haugen KS, Semmens MJ & Novak PJ (2002) A novel in situtechnology for the treatment of nitrate contaminated ground-water. Water Res. 36: 3497–3506

Hell F, Lahnsteiner J, Frischherz H & Baumgartner G (1998)Experience with full-scale electrodialysis for nitrate andhardness removal. Desalination 117: 173–180

Hichour M, Persin F, Sandeaux J & Gavach C (2000) Fluorideremoval from waters by Donnan dialysis. Sep. Purif. Tech-nol. 18: 1–11

Ho CM, Tseng SK & Chang YJ (2001) Autotrophic denitrifi-cation via a novel membrane-attached biofilm reactor. Lett.Appl. Microbiol. 33: 201–205

Ho WSW & Sirkar KK (1992) Membrane Handbook. VanNostrand Reinhold, New York

Jacangelo JG, Trussell RR & Watson M (1997) Role ofmembrane technology in drinking water treatment in theUnited States. Desalination 113: 119–127

Jackson PE (2001) Determination of inorganic ions in drinkingwater by ion chromatography. TrAC Trends. Anal. Chem.20: 320–329

Kapoor A & Viraraghavan T (1997) Nitrate removal fromdrinking water – review. J. Environ. Eng. 123: 371–380

Kimura K, Nakamura M & Watanabe Y (2002) Nitrateremoval by a combination of elemental sulfur-based denitri-fication and membrane filtration. Water Res. 36: 1758–1766

Lee K-C & Rittmann BE (2000) A novel hollow-fiber mem-brane biofilm reactor for autohydrogenotrophic denitrifica-tion of drinking water. Wat Sci. Technol. 41: 219–226

378

Lee K-C & Rittmann BE (2002) Applying a novel autohydro-genotrophic hollow-fiber membrane biofilm reactor fordenitrification of drinking water. Water Res. 36: 2040–2052

Lee K-C & Rittmann BE (2003) Effects of pH and precipitationon autohydrogenotrophic denitrification using the hollow-fiber membrane-biofilm reactor. Water Res. 37: 1551–1556.

Levenstein R, Hasson D & Semiat R (1996) Utilization of theDonnan effect for improving electrolyte separation withnanofiltration membranes. J. Membr. Sci. 116: 77–92

Lhassani A, Rumeau M, Benjelloun D & Pontie M (2001)Selective demineralization of water by nanofiltration. Appli-cation to the defluorination of brackish water. Water Res. 35:3260–3264

Mansel BO & Schroeder ED (1999) Biological denitrification ina continuous flow membrane reactor. Water Res. 33: 1845–1850

Mansel BO & Schroeder ED (2002) Hydrogenotrophic denitri-fication in a microporous membrane bioreactor. Wat Res. 36:4683–4690

McCleaf PR & Schroeder ED (1995) Denitrification using amembrane immobilized biofilm. J. AWWA 87(3): 77–86

Min B, Evans PJ, Chu AK & Logan BE (2004) Perchlorateremoval in sand and plastic media bioreactors. Water Res.38: 47–60

Nerenberg R, Rittmann BE & Najm I (2002) Perchloratereduction in a hydrogen-based membrane-biofilm reactor. J.AWWA 94(11): 103–114

Nicoll H (2001) Nanofiltration makes surface water drinkable.Filtr. Sep. 38(1): 22–23

Ning RY (2002) Arsenic removal by reverse osmosis. Desali-nation 143: 237–241.

Paugam L, Taha S, Cabon J & Dorange G (2002) Eliminationof nitrate ions in drinking waters by nanofiltration. Desali-nation 152: 271–274

Petrovic M, Gonzalez S & Barcelo D (2003) Analysis andremoval of emerging contaminants in wastewater and drink-ing water. TrAC Trends Anal. Chem. 22: 685–696.

Richardson SD (2003) Disinfection by-products and otheremerging contaminants in drinking water. TrAC TrendsAnal. Chem. 22: 666–684

Ritchie SMC & Bhattacharyya D (2002) Membrane-basedhybrid processes for high water recovery and selectiveinorganic pollutant separation. J. Hazard Mat. 92: 21–32

Rittmann BE, Nerenberg R, Lee K-C, Najm I, Gillogly TE,Lehman GE & Adham SS (2004) The hydrogen-basedhollow-fiber membrane biofilm reactor (HFMBfR) forremoving oxidized contaminants. Water Sci. Technol. 14:127–133

Roggy DK, Novak PJ, Hozalski RM, Clapp LW & SemmensMJ (2002) Membrane gas transfer for groundwater remedi-ation: Chemical and biological fouling. Environ. Eng. Sci. 19:563–574

Roquebert V, Booth S, Cushing RS, Crozes G & Hansen E(2000) Electrodialysis reversal (EDR) and ion exchange aspolishing treatment for perchlorate treatment. Desalination131: 285–291

Salem K, Sandeaux J, Molenat J, Sandeaux R & Gavach C(1995) Elimination of nitrate from drinking water by electro-chemical membrane processes. Desalination 101: 123–131

Sata T (2000) Studies on anion exchange membranes havingpermselectivity for specific anions in electrodialysis-effect ofhydrophilicity of anion exchange membranes on permselec-tivity of anions. J. Membr. Sci. 167: 1–31

Sato Y, Kang M, Kamei T & Magara Y (2002) Performance ofnanofiltration for arsenic removal. Water Res. 36: 3371–3377

Schoeman JJ & Steyn A (2003) Nitrate removal with reverseosmosis in a rural area in South Africa. Desalination 155:15–26

Semmens MJ (1991) Bubbleless aeration. Water Eng. Manage.138(4): 18–19

Semmens MJ & Essila (2001) Modeling biofilms on gas-permeable supports: Flux limitations. J. Environ. Eng. 127:126–133

Simbeck DR (2004) CO2 capture and storage – the essentialbridge to the hydrogen economy. Energy 29: 1633–1641

Smith CW, Di Gregorio D & Talcott RM (1969) The use ofultrafiltration membrane for activated sludge separation.Proceedingds of the 24th Annual Purdue Industrial WasteConference, West Lafayette, Indiana, USA (pp 1300–1310)

Smith AH, Lopipero PA, Bates MN & Steinmaus CM (2002)Arsenic epidemiology and drinking water standards. Science296: 2145–2146

Thanuttamavong M, Yamamoto K, Oh J-I, Choo KH & ChoiSJ (2002) Rejection characteristics of organic and inorganicpollutants by ultra low-pressure nanofiltration of surfacewater for drinking water treatment. Desalination 145: 257–264

Urase T, Oh J-I. & Yamamoto K (1998) Effect of pH onrejection of different species of arsenic by nanofiltration.Desalination 117: 11–18

US EPA (1998) Federal Register 63 (170) FR 44511fUS EPA (2002) Perchlorate environmental contamination:Toxicological review and risk characterization, ExternalReview Draft, NCEA-1-0503, January 16, 2002

Van der Bruggen B, Everaert K, Wilms D & Vandecasteele C(2001) Application of nanofiltration for removal of pesti-cides, nitrate and hardness from ground water: Rejectionproperties and economic evaluation. J. Membr. Sci. 193: 239–248

Van der Bruggen B, Koninckx A & Vandecasteele C (2004)Separation of monovalent and divalent ions from aqueoussolution by electrodialysis and nanofiltration. Water Res. 38:1347–1353

Velizarov, S, Crespo JG & Reis AM (2002) Ion exchangemembrane bioreactor for selective removal of nitrate fromdrinking water: Control of ion fluxes and process perfor-mance. Biotechnol Prog 18: 296–302

Velizarov S., Matos C, Crespo JG & Reis AM (2004) Removalof perchlorate and nitrate from drinking water in an ionexchange membrane bioreactor. Proceedings of the EuropeanSymposium on Environmental Biotechnology ESEB 2004,April 25–28, Oostende, Belgium (pp 99–102)

Velizarov S, Matos C, Sequeura A, Reis AM & Crespo JG(2003b) Removal of trace mono-valent inorganic pollutantsusing the ion exchange membrane bioreactor concept. Pro-ceedings of the Membrane Science and Technology Confer-ence ‘‘PERMEA 2003’’, September 7–11, Tatranske Matliare,Slovakia (Conference CD-ROM)

Velizarov S, Reis AM & Crespo JG (2003a) Removal of tracemono-valent inorganic pollutants in an ion exchange mem-brane bioreactor: Analysis of transport rate in a denitrifica-tion process. J. Membr. Sci. 217: 269–284

Velizarov S, Rodrigues CM, Reis AM & Crespo JG (2000)Mechanism of charged pollutants removal in an ion exchangemembrane bioreactor: Drinking water denitrification. Bio-technol Bioeng 71: 245–254

379

Vrijenhoek EM & Waypa JJ (2000) Arsenic removal fromdrinking water by a ‘‘loose’’ nanofiltration membrane.Desalination 130: 265–277

WHO (2003) Draft guidelines for drinking water quality, 3rdedn. World Health Organization, Geneva. http://www.who.int/water_sanitation_health/GDWQ/

WintgensT,RosenJ,MelinT,BrepolsC,DrenslaK&EngelhardtN (2003) Modelling of a membrane bioreactor system formunicipal wastewater treatment. J. Membr. Sci. 216: 55–65

Wisniewski C, Persin F, Cherif T, Sandeaux R, Grasmick G,Gavach C & Lutin F (2002) Use of a membrane bioreactorfor denitrification of brine from an electrodialysis process.Desalination 149: 331–336

Yoon J, Yoon Y, Amy G, Cho J, Foss D & Kim T-H (2003)Use of surfactant modified ultrafiltration for perchlorate(ClO4

)) removal. Water Res. 37: 2001–2012

380