ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 6

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding autE-mail address: f

journal homepage: www.elsevier.com/locate/watres

Selective removal of dissolved uranium in drinking waterby nanofiltration

A. Favre-Reguillona,�, G. Lebuzitb, D. Muratb, J. Foosb, C. Mansourc, M. Drayed

aLaboratoire de Chimie Organique (UMR CNRS 7084), CNAM, 2 rue Conte, 75003 Paris, FrancebLaboratoire des Sciences Nucleaires, CNAM, 2 rue Conte, 75003 Paris, FrancecLaboratoire d’Electrochimie et de Chimie Analytique (UMR CNRS 7575), ENSCP, 11 rue Pierre et Marie Curie, 75005 Paris, FrancedLaboratoire de Chimie Moleculaire et Environnement (EA 1651), Universite de Savoie, Polytech’Savoie,

73376 Le Bourget du Lac Cedex, France

a r t i c l e i n f o

Article history:

Received 30 March 2007

Received in revised form

28 August 2007

Accepted 29 August 2007

Available online 7 September 2007

Keywords:

Uranium

Nanofiltration

Remediation

Drinking water

Membrane

Rejection mechanism

nt matter & 2007 Elsevie.2007.08.034

hor. Tel.: +33 472448507; favrereg@cnam.fr (A. Fav

a b s t r a c t

A procedure for the selective removal of uranium traces dissolved in drinking water has

been studied. Plate module membrane filtration equipment was operated to evaluate the

performance and selectivity of three different nanofiltration flat-sheet membranes.

Experiments were carried out using various commercial mineral waters with distinct

physicochemical compositions. The membranes were first discriminating by their ability to

reject uranium in the presence of the main cations found in mineral waters, using a

2 mg L�1 (2000 ppb) concentration of uranium. The rejection of U(VI) was dependent on the

uranyl speciation and the ionic strength. Second, removal of uranium traces (0.02 mg L�1,

20 ppb) was performed using the nanofiltration membrane showing the highest selectivity

for uranium toward alkaline and alkaline-earth ions. The results showed a high

performance of the nanofiltration membrane, Osmonics DL, for selective uranium rejection

at low pressure (1 bar), illustrating the advantage of nanofiltration for the selective removal

of uranium from drinking water.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Natural uranium occurs in very small amounts in many soil

and rock types. The average uranium concentration in the

Earth’s crust is about 3 mg/kg. In addition, the formation of

uranium accumulations is a normal geological process,

especially in granites and sedimentary rocks, two widely

occurring rock types.

Assessment of the risk of impact from most radionuclides

is based on the total radiological dose rate. However, for

uranium, there can be a greater risk from chemical toxicity

than radiological toxicity. Chemical toxicity of uranium is

dependent on several environmental parameters (Grenthe

et al., 1992; Sheppard et al., 2005). Although uranium has

r Ltd. All rights reserved.

ax: +33 472431408.re-Reguillon).

oxidation states of (III)–(VI), the two major oxidation states

are (VI) and (IV). The latter is readily oxidized to (VI) as UO2þ2 ,

the most prevalent form of uranium in the environment.

U(VI) readily complexes with carbonate, phosphate or sulfate

ions and in these forms is soluble and easily transported

(Cheng et al., 2004; Zhou and Gu, 2005).

The maximum admissible concentrations (MACs) of ura-

nium for drinking water depend on each public authority

(Reimann and Banks, 2004). Although there is no defined MAC

in western European legislation, the US Environmental

Protection Agency has promulgated a MAC of 0.03 mg L�1

(30 ppb) and the World Health Organization (WHO) has

suggested a guideline value of 0.002 mg L�1 (2 ppb) (Reimann

and Banks, 2004). In recent investigations of drinking water

ARTICLE IN PRESS

Table 1 – Determined composition of the mineral waterafter the addition of uranyl nitrate

Sample Composition (mg L�1) pH

Na+ K+ Mg2+ Ca2+ UO22+

A 11.2 6.2 7.9 12.0 2.0 or

0.002

7.0

B 7.0 1.5 26.3 81.4 2.0 7.2

C 5.7 2.8 43.5 218.0 2.0 7.0

Table 2 – Composition of the mineral water as specifiedby the producers

Sample Composition (mg L�1) Dryresidue

Cl� NO3� SO4

2� HCO3� SiO2

A 13.5 6.3 8.1 71.0 31.7 130.0

B 4.5 3.8 10.0 357.0 13.5 309.0

C 0.0 4.6 336.0 400.2 0.0 841.0

Table 3 – General properties of Osmonics membranes

Membrane Molecular weight cut-off (Da) pI

G10 2500a 3.7b

DL 150–300c –

DK 150–300c 4.0d

a Afonso et al. (2001).b Sbai et al. (2003).c Supplied by GE Water & Process Technologies.d Tanninen and Nystrom (2002).

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 6 1161

quality in the USA and Europe, high concentrations of

uranium (up to 1.2 mg L�1, 1200 ppb) were detected in some

water samples collected from private wells (Hakonson-Hayes

et al., 2002; Orloff et al., 2004). Utilization of water from these

wells has raised concerns of potential radiological and

toxicological risks to human consumers. The alternative

sources of potable water are commercially bottled water or

those after water treatment. The water treatment techniques

that have been used to reduce uranium concentration to less

than the MAC are chemical coagulation–flocculation (Gafvert

et al., 2002), activated carbon (Coleman et al., 2003), ion

exchange (Barbette et al., 2004; Barton et al., 2004; Bryant

et al., 2003; Gu et al., 2004; Vaaramaa et al., 2000), ultrafiltra-

tion assisted by complexation (Kryvoruchko et al., 2004) and

reverse osmosis (Hsiue et al., 1989; Huikuri et al., 1998; Lin

et al., 1987; Raff and Wilken, 1999).

Membrane separation processes have been used for several

years to concentrate or fractionate suspended particles and

dissolved substances. Reverse osmosis (RO), now in wide-

spread use to prepare irrigation water from briny waters, and

ultrafiltration (UF) both constitute a valuable aid for the

fractionation and concentration of colloidal substances.

Nanofiltration (NF) membranes have intermediate molecular

weight cut-offs (MWCOs) between UF and RO membranes. NF

has been found to be very useful in recent years for water and

wastewater treatment. NF membranes have been shown to be

able to remove hardness as well as a fraction of dissolved

salts. Some micropollutants, such as pesticides (Bellona et al.,

2004; Causserand et al., 2005; Kiso et al., 2001), endocrine-

disrupting compounds and pharmaceutically active com-

pounds (Kimura et al., 2003; Zhang et al., 2006), can be

effectively removed by NF membranes.

The purpose of this study was to evaluate the possibilities

of NF for the selective removal of U(VI) from well water. We

first selected the most appropriate membrane that showed

the highest U(VI) rejection and the highest selectivity toward

alkaline and alkaline earth ions. The ability of various

membranes to selectively reject U(VI) was then evaluated

using three different types of mineral water and a U(VI)

concentration of 2 mg L�1 (2000 ppb). Next, the membranes

showing the highest U(VI) selectivity were evaluated with

mineral water containing U(VI) traces (0.02 mg L�1, 20 ppb).

2. Materials and methods

2.1. Mineral water samples

Commercially available mineral water samples, in plastic

bottles, were obtained through the normal distribution

channels. Uranyl nitrate (UO2(NO3)2 � 6H2O) was added to the

water samples to obtain concentrations of 2 or 0.02 mg L�1.

Table 1 lists the cation concentration present in water

samples and Table 2 lists the main anion concentration and

dry residues as specified from the producers.

2.2. Membrane

The three flat membranes, obtained from Osmonics (GE

Water & Process Technologies), were all constituted of

polyamide filtering layers with a membrane surface area of

155 cm2. DK, DL and G10 membranes are polymeric flat thin

film composite membranes in which a polyamide selective

layer is supported on a polysulfone layer (Favre-Reguillon

et al., 2007; Mazzoni and Bandini, 2006; Sbai et al., 2003). The

MWCOs are between 150 and 2500 Da and the isoelectric

points (pI) are close to 4 (Table 3). For all membranes, the

dependence of the volumetric water flux on the transmem-

brane pressure was determined with deionized water. In

general, a pressure increase from 1 to 3 bar leads to a linear

volumetric water flux volume versus transmembrane pres-

sure. The pure water permeability results suggest that the DK

membrane is the most compact membrane, whereas the G10

and DL membranes have the most open structures.

2.3. Nanofiltration experiments

NF experiments were performed with an Osmonics Sepa CF

lab-scale membrane cell and the lab-scale membrane system

as already described (Sorin et al., 2006). The feed is

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 61162

maintained at constant composition during the experiments

by totally recycling the permeate and the feed.

Mineral waters were filtered on membranes in a cross-flow

mode. During membrane filtration, the metal ions in the feed

water are convectively driven to the membrane surface, thus

forming a concentrated polarization boundary layer near the

membrane. This layer involves a reduction in charge density of

the membrane and consequently attenuates the electrostatic

repulsions between the metal ions in solution and the charge of

the membranes (Bian et al., 2000), leading to a decrease in the

rejection (Bacchin et al., 2002; Bhattacharjee et al., 1999; Schaep

et al., 1998). The accumulation at the membrane surface is a

function of the back-transport of the complex into the bulk

solution. To decrease the accumulation at the membrane

surface, the tangential velocity (Jr) at the membrane surface

is maintained at a constant value of 145 mm s�1.

2.4. Rejection measurements

NF tests were carried out by filtering mineral water on the

membrane using a transmembrane pressure of 1–4 bar, a

tangential velocity of 145 mm s�1 and at a constant tempera-

ture of 25 1C. The pH of the solution was constant during the

nanofiltration process. The permeate flux was measured and

samples of permeate and feed were taken 30 min after each

stage of operation. Each experiment was repeated 3 times to

improve the reliability of the results.

Rejection of metal ions was evaluated from the determined

feed and permeate concentration of samples collected during

the course of the experimental tests. The rejection Ri (%) of a

substance i was calculated according

Ri ¼Cf

i � Cpi

Cfi

!� 100, (1)

where Cfi is the concentration of i in the feed and Cp

i is the

concentration of i in the permeate.

The transmission Ti (%) of a solute i is the ability of solute i

to pass through the membrane and can be expressed as

Ti ¼Cp

i

Cfi

� 100. (2)

The membrane selectivity of two solutes i and j, MSi=j, can

be expressed by the ratio of their transmission and thus may

be expressed as the function of the concentrations of i and j in

the permeate and in the feed as

MSi=j ¼Ti

Tj¼

Cpi

Cfi

�Cf

j

Cpj

. (3)

Metal ion concentrations were determined by ICP-AES with

a Spectro D ICP spectrometer. U(VI) concentration at trace

levels was determined by ICP-MS with a FINIGAN-MAT

element mass spectrometer.

3. Results and discussion

3.1. U(VI) speciation in water samples

Uranium is known to easily form stable complexes with both

organic and inorganic ligands in natural aquatic environ-

ments (Krepelova et al., 2006). There is extensive literature on

equilibria involved in the uranyl–carbonate systems (Clark

et al., 1995; Grenthe et al., 1992). These are usually quite

complex in that they consist of several different complexed

ions in rapid equilibria with one another and the hydrolyzed

species (Barton et al., 2004; Grenthe et al., 1992). It is therefore

important to quantitatively predict the reactions that are

likely to occur between uranyl and the anions present in the

aqueous media in order to understand the rejection of

aqueous species by the nanofiltration membranes. This

information is provided by speciation calculations using

chemical thermodynamic data. The speciation of uranyl as

a function of the water composition was evaluated using the

speciation code CHESS (van der Lee, 2002), which is a

speciation model specifically developed to simulate the

equilibrium state of complex aquatic systems (Fig. 1).

In the presence of carbonate and when the uranyl

concentration does not exceed the carbonate concentration,

the monomeric uranyl carbonate species UO2(CO3),

UO2(CO3)22� and UO2(CO3)3

4� are expected to dominate depend-

ing on pH (Grenthe et al., 1992). For water samples, at pH close

to 7, uranyl speciation distribution in the aqueous media

showed that UO2(CO3)22� and UO2(CO3)3

4� should be the major

individual components. For a lower concentration of carbo-

nate (water sample A, Fig. 1), the uranyl dihydroxid specie

UO2(OH)2 should be one of the major individual components.

3.2. Rejection of U(VI) by nanofiltration membranes

The effectiveness of NF membranes in removing uranium in

water was tested on U(VI) artificially enriched water samples.

Uranyl nitrate, UO2(NO3)2, was added to the water samples to

obtain uranium concentration of 2 mg L�1. Membrane perfor-

mance was measured in terms of U(VI) rejection (Fig. 2) and

U(VI)/alkaline or alkaline-earth cation selectivities.

Nanofiltration membranes are electrically charged in aqu-

eous media due to the type of material or the adsorption of

charged species. Hence, size and charge of solutes influence

the extent of rejection by nanofiltration membranes, though

the precise mechanism of rejection will depend on the type of

membrane used (Bellona et al., 2004). Predictably, the highest

cut-off membrane exhibits the lowest solute rejection for the

same water composition. The rejection of U(VI) with water

sample A was found to be 40% with G10, and 95% and 99% for

DL and DK membranes, respectively. The same trend was

observed for the rejection of alkaline and alkaline-earth ions.

The difference in rejection observed between DL and DK

membranes could be explained by a slightly lower mean pore

radius for DK than for DL (Bargeman et al., 2005).

The high rejection of cations may be attributed to the

Donnan exclusion model (Bowen et al., 1997). NF membranes

used in the work show amphoteric behavior. The isoelectric

points (pI) of DK, DL and G10 membranes are close to 4

(Table 3) and thus the electric charge density of the NF

membranes at pH 7, above the pI, is negative. In the Donnan

exclusion model, the rejection of the multicharge anions

by a negatively charged membrane is highest, whereas that of

the bivalent or monovalent cations is the lowest. The higher

rejection of Mg2+ compared with Ca2+ may be due to the

fact that magnesium ions, although smaller, display larger

ARTICLE IN PRESS

A C

0

10

20

30

40

50

60

70

80

90

100G 10

Reje

ction (

%)

Na+

K+

Mg2+

Ca2+ UO2

2+

A C

0

10

20

30

40

50

60

70

80

90

100DL

Reje

ction (

%)

Na+

K+

Mg2+

Ca2+ UO2

2+

A C

0

10

20

30

40

50

60

70

80

90

100DK

Reje

ction (

%)

Na+

K+

Mg2+

Ca2+ UO2

2+

B

B

B

Fig. 2 – Rejection coefficient of sodium, potassium,

magnesium, calcium and uranium as a function of water

samples and membrane; U(VI) ¼ 2 mg L�1 (water

composition, see Table 1 and Table 2). Experimental

conditions: T ¼ 20 1C, DP ¼ 1 bar, Jr ¼ 145 mm s�1.

4 6 8

0

20

40

60

80

100

UO2(OH)3-

UO2(CO3)22-

UO2(CO3)34-UO2(OH)2

UO2CO3

UO2OH-

UO2SO4

UO22+

Ura

nyl specie

s in w

ate

r A

(%

)

pH

4 8

0

20

40

60

80

100

UO2(CO3)34-UO2(OH)2

UO2SO4

UO2OH-

UO2(CO3)22-UO2CO3

UO22+

Ura

nyl specie

s in w

ate

r B

(%

)

pH

4 6 8

0

20

40

60

80

100

UO2(CO3)

Ura

nyl specie

s in w

ate

r C

(%

)

pH

CaUO4

UO2OH+

UO22+

UO2SO4

UO2(OH)2

UO2(CO3)34-

UO2(CO3)22-

5 7

5 6 7

75

Fig. 1 – Aqueous U(VI) speciation distribution as the function

of pH and water composition evaluated using CHESS

(van der Lee, 2002).

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 6 1163

hydrated diameters (0.80 nm) than those of calcium ions

(0.46 nm) (Rios et al., 1996).

An interesting membrane selectivity was observed with the

DL membrane for the three water samples studied. In all

cases, U(VI) rejection was high (490%), whereas the rejection

of the monovalent ions and divalent cations was low, o25%

and o40%, respectively, for water sample A, o10% and o20%,

respectively, for water sample B, and o25% and o50%,

respectively, for water sample C, which has the highest

mineral content. Under those conditions, high U(VI)/Na+ and

U(VI)/Mg2+ membrane selectivity could be obtained (Table 4).

The rejection of cations by NF membranes is dependent on

their valence, but the structure of ions in solution must also

be taken into account. Fig. 1 shows the uranyl speciation of

the different water samples. For DL and DK membranes the

observed rejection was high (490%) whatever the water

samples, but for G10 membrane the observed rejection of

ARTICLE IN PRESS

Table 5 – Observed U(VI) rejection with G10 membranes

Watersample

U(VI)rejection (%)

Dry residue(mg L�1)

MSNaþ

UðVIÞ

A 40 130 1.5

B 21 309 1.2

C 13 841 1.1

Experimental conditions: T ¼ 20 1C, Jr ¼ 145 mm s�1, U(VI) ¼

2 mg L�1, water composition, see Table 1 and Table 2.

Table 4 – Membrane selectivity observed with DLmembrane with an uranium concentration of 2 mg L�1

(water composition, see Table 1)

Membrane selectivity Water sample

A B C

U(VI)/Na+ 16.0 8.5 7.6

U(VI)/Mg2+ 11.8 7.4 4.8

Experimental conditions: T ¼ 20 1C, DP ¼ 1 bar, Jr ¼ 145 mm s�1.

1.0 1.5 2.0 2.5 3.0 3.5 4.00

10

20

30

40

50

60

70

80

90

100

G10

Reje

ction (

%)

ΔP (bar)

Na+ K

+Ca

2+Mg

2+UO2

2+

1.0 1.5 2.0 2.5 3.0 3.5 4.00

10

20

30

40

50

60

70

80

90

100

DL

Reje

ction (

%)

ΔP (bar)

Na+ K

+Mg

2+ Ca2+

UO22+

Fig. 3 – Rejection coefficient of sodium, potassium,

magnesium, calcium and uranium as a function of

transmembrane pressure (DP) for G10 and DL membranes.

Water sample A; experimental conditions: T ¼ 20 1C,

Jr ¼ 145 mm s�1, U(VI) ¼ 2 mg L�1, water composition, see

Table 1 and Table 2.

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 61164

uranyl was lower and thus gave us the opportunity to study

the influence of speciation and ionic strength.

The different rejections of uranyl obtained with the G10

membrane are listed in Table 5. For water samples A and B, at

pH 7.0, up to 70% and 80%, respectively, of uranyl is in the

form of UO2(CO3)22�. The relatively high rejection of uranyl

species compared with alkaline and alkaline-earth observed

with water sample A could be explained by the Donnan

exclusion model. The negatively charged uranyl complexes

are rejected by the negatively charged membrane while

cations are less rejected. As can been seen in Table 5, an

increase in the concentration of salts from 130 to 309 mg L�1

decreased the observed rejection of uranyl by a factor of 2. An

increase in the electrolyte concentration shielded the mem-

brane charges by the counter-ions, causing reduced mem-

brane rejection. The alkaline and alkaline-earth rejections are

somewhat different but are also very slightly dependent on

the feed concentration. The idea of rejection on the basis of

the Donnan exclusion model is in accordance with the two

experimental findings: high rejection of negatively charged

species and rejection depends on the ionic strength.

For water sample C, up to 75% of uranyl is in the form of

(UO2(CO3)2)2� and 20% is in the form of (UO2(CO3)3)4� (Fig. 1).

However, the ionic strength is at its highest and thus

the observed rejection of uranyl using G10 nanofiltration

membrane is at its lowest. Moreover, under these conditions,

no U/Na+ or U/Mg2+ selectivity was observed.

3.3. Influence of process parameters on U(VI) rejection

The influence of the operating parameters on the rejection of

U(VI) was then studied using water sample A with mem-

branes G10 and DL. The filtration tests were carried out with a

transmembrane pressure (DP) varying from 1 to 4 bar at a

constant tangential velocity (Jr) of 145 mm s�1. The effect of

applied pressure on U(VI) rejection is shown in Fig. 3.

The permeate water flux increased linearly as the pressure

increased when the composition of feed water was kept

constant. Fig. 3 shows that the rejection of each component

increased linearly as the pressure increased. The U(VI)

rejection at 1 bar in water sample A was 40%, although at

4 bars the U(VI) rejection rose to 65%. The same increase of

rejection was observed for alkaline and alkaline-earth ca-

tions; the rejection of Mg2+ was 15% at 1 bar and 27% at 4 bar.

The same trend was observed with the DL membrane (Fig. 3),

but transmembrane pressure had a small effect on U(VI)

rejection. The rejection of U(VI) with the DL membrane was

95% at 1 bar, although it was 98% at 4 bar while we observed

an increased in the Mg2+ rejection from 41% at 1 bar to 78% at

4 bar. We can thus consider this linearity as due to an absence

of concentration polarization under our conditions.

ARTICLE IN PRESS

WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 6 1165

An increase of transmembrane pressure slightly increased

the membrane selectivity for both membranes. When the

transmembrane pressure is increasing from 1 to 4 bars,

MSNaþ

UðVIÞ is increased from 1.5 to 2.8 and from 16 to 20 for G10

and DL membrane, respectively (Fig. 3). Although at higher

pressure, higher permeate flux can be obtained, there is a

limit to what operating pressure can be reached. Indeed, if the

pressure is too high, much more energy is consumed and the

concentration polarization boundary layer effect is increased.

Therefore, it is very important to optimize these factors to

obtain the most economic operating conditions in designing

the NF system. Operating a DL membrane at 1 bar was then

further investigated for the removal of U(VI) traces.

3.4. Removal of U(VI) traces by DL membranes

Uranyl nitrate, UO2(NO3)2, was added to water sample A to

obtain a uranium concentration of 0.02 mg L�1 (20 ppb). The

NF tests were carried out on the DL membrane with a

transmembrane pressure of 1 bar, a tangential velocity of

145 mm s�1 and at a temperature of 20 1C. The concentration

of the cations in the permeate and in the feed were

determined by ICP-MS for uranium and by ICP-AES for the

alkaline and alkaline-earth metal ions (Table 6).

As expected, the alkaline and alkaline-earth rejections were

lower than 45% and 25% for the respective cations. Under

these above conditions, the U(VI) trace (20 ppb) rejection was

high (95%) and the membrane selectivity was comparable to

the value observed with U(VI) at 2 mg L�1 (2000 ppb). Under

the same conditions, the U(VI) concentration in the permeate

was lower than the MAC suggested by the WHO, i.e. 2 ppb. The

choice of an membrane with a high selectivity toward the

removal of U(VI) should be emphasized. The ability of

membranes to reject inorganic cations and among then

U(VI) has already been demonstrated (Raff and Wilken,

1999). But our results show that DL membrane, under

appropriate experimental conditions, can selectively removed

U(VI) carbonate complexes with a high selectivity toward

others cations present in the feed (Table 6). Thus, the

composition of the permeate obtained using a DL membrane

Table 6 – Measured composition of the water sample A inthe feed and in the permeate

Cations Initialconc.(mg/L)

Permeateconc.(mg/L)

Rejection(%)

MScationUðVIÞ

Na+ 11.2a 8.7a 22 15.6

K+ 6.2a 4.7a 24 15.2

Mg2+ 7.9a 4.4a 44 11.2

Ca2+ 12.0a 7.3a 39 12.2

UO22+ 0.02b 0.001b 95 _

Experimental conditions: U(VI) ¼ 20 ppb; DL membrane, T ¼ 20 1C,

DP ¼ 1 bar, Jr ¼ 145 mm s�1.a Measured by ICP-AES.b Measured by ICP-MS.

does not require additional treatments to meet drinking

water standards such as hardness.

4. Conclusion

NF membranes are able to selectively reject U(VI) from

mineral water with a relatively high selectivity, despite a

high concentration of alkaline and alkaline-earth cations.

Consequently, nanofiltration equipment can be considered

suitable for reducing the U(VI) concentration in drinking

water to less than the World Health Organization (WHO)

maximum admissible concentration (MAC). The rejection

mechanism for uranyl is predominantly due to charge effects

between uranyl carbonate complexes and the charged

membrane. The general pattern was similar for all three

membranes. The rejection was dependent on the uranyl

speciation and the ionic strength. Furthermore, a mineral

water containing up to 20 ppb of U(VI) can be lowered to

0.8 ppb, which is below the MAC suggested by WHO with the

DL membrane with a transmembrane pressure of 1 bar, a

tangential velocity of 145 mm s�1 and at a temperature of

20 1C. A major advantage of membrane filtration is that

compared with ion exchange processes, it is a continuous

process that does not require further chemical products (for

example, for the regeneration of the adsorbant). Use of such a

system may therefore offer a viable and cost-effective method

for in situ water remediation.

R E F E R E N C E S

Afonso, M.D., Hagmeyer, G., Gimbel, R., 2001. Streaming potentialmeasurements to assess the variation of nanofiltrationmembranes surface charge with the concentration of saltsolutions. Sep. Purif. Technol. 22–23, 529–541.

Bacchin, P., Si-Hassen, D., Starov, V., Clifton, M.J., Aimar, P., 2002.A unifying model for concentration polarization, gel-layerformation and particle deposition in cross-flow membranefiltration of colloidal suspensions. Chem. Eng. Sci. 57, 77–91.

Barbette, F., Rascalou, F., Chollet, H., Babouhot, J.L., Denat, F.,Guilard, R., 2004. Extraction of uranyl ions from aqueoussolutions using silica-gel-bound macrocycles for alpha con-taminated wastewater treatment. Anal. Chim. Acta 502,179–187.

Bargeman, G., Vollenbroek, J.M., Straatsma, J., Schroen, C.G.P.H.,Boom, R.M., 2005. Nanofiltration of multi-component feeds.Interactions between neutral and charged components andtheir effect on retention. J. Membr. Sci. 247, 11–20.

Barton, C.S., Stewart, D.I., Morris, K., Bryant, D.E., 2004. Perfor-mance of three resin-based materials for treating uranium-contaminated groundwater within a PRB. J. Hazard. Mater. 116,191–204.

Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting therejection of organic solutes during NF/RO treatment—aliterature review. Water Res. 38, 2795–2809.

Bhattacharjee, S., Kim, A.S., Elimelech, M., 1999. Concentrationpolarization of interacting solute particles in cross-flowmembrane filtration. J. Colloid Interface Sci. 212, 81–99.

Bian, R., Yamamoto, K., Watanabe, Y., 2000. The effect of shearrate on controlling the concentration polarization and mem-brane fouling. Desalination 131, 225–236.

Bowen, W.R., Mohammad, A.W., Hilal, N., 1997. Characterisationof nanofiltration membranes for predictive purposes—use of

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 1 1 6 0 – 1 1 6 61166

salts, uncharged solutes and atomic force microscopy.J. Membr. Sci. 126, 91–105.

Bryant, D.E., Stewart, D.I., Kee, T.P., Barton, C.S., 2003. Develop-ment of a functionalized polymer-coated silica for the removalof uranium from groundwater. Environ. Sci. Technol. 37,4011–4016.

Causserand, C., Aimar, P., Cravedi, J.P., Singlande, E., 2005.Dichloroaniline retention by nanofiltration membranes.Water Res. 39, 1594–1600.

Cheng, T., Barnett, M.O., Roden, E.E., Zhuang, J., 2004. Effects ofphosphate on uranium(VI) adsorption to goethite-coated sand.Environ. Sci. Technol. 38, 6059–6065.

Clark, D.L., Hobart, D.E., Neu, M.P., 1995. Actinide carbonatecomplexes and their importance in actinide environmentalchemistry. Chem. Rev. 95, 25–48.

Coleman, S.J., Coronado, P.R., Maxwell, R.S., Reynolds, J.G., 2003.Granulated activated carbon modified with hydrophobic silicaaerogel—potential composite materials for the removal ofuranium from aqueous solutions. Environ. Sci. Technol. 37,2286–2290.

Favre-Reguillon, A., Sorin, A., Pellet-Rostaing, S., Bernier, G.,Lemaire, M., 2007. Nanofiltration assisted by complexation: apromising process for the separation of trivalent long livedminor actinides from lanthanides. C. R. Chim., in press,doi:10.1016/j.crci.2007.01.012.

Gafvert, T., Ellmark, C., Holm, E., 2002. Removal of radionuclidesat a waterworks. J. Environ. Radioactivity 63, 105–115.

Grenthe, I., Fuger, J., Konings, R.J.M., Lemire, R.J., Muller, A.B.,Nguyen-Trung, C., Wanner, H., 1992. Chemical Thermody-namics of Uranium. North-Holland, Amsterdam.

Gu, B., Ku, Y.-K., Jardine, P.M., 2004. Sorption and binary exchangeof nitrate, sulfate, and uranium on an anion-exchange resin.Environ. Sci. Technol. 38, 3184–3188.

Hakonson-Hayes, A.C., Fresquez, P.R., Whicker, F.W., 2002. Asses-sing potential risks from exposure to natural uranium in wellwater. J. Environ. Radioactivity 59, 29–40.

Hsiue, G.H., Pung, L.S., Chu, M.L., Shieh, M.C., 1989. Treatment ofuranium effluent by reverse osmosis membrane. Desalination71, 35–44.

Huikuri, P., Salonen, L., Raff, O., 1998. Removal of naturalradionuclides from drinking water by point of entry reverseosmosis. Desalination 119, 235–239.

Kimura, K., Amy, G., Drewes, J.E., Heberer, T., Kim, T.-U.,Watanabe, Y., 2003. Rejection of organic micropollutants(disinfection by-products, endocrine disrupting compounds,and pharmaceutically active compounds) by NF/RO mem-branes. J. Membr. Sci. 227, 113–121.

Kiso, Y., Sugiura, Y., Kitao, T., Nishimura, K., 2001. Effects ofhydrophobicity and molecular size on rejection of aromaticpesticides with nanofiltration membranes. J. Membr. Sci.192, 1–10.

Krepelova, A., Sachs, S., Bernhard, G., 2006. Uranium(VI) sorptiononto kaolinite in the presence and absence of humic acid.Radiochim. Acta 94, 825–833.

Kryvoruchko, A.P., Yurlova, L.Y., Atamanenko, I.D., Kornilovich,B.Y., 2004. Ultrafiltration removal of U(VI) from contaminatedwater. Desalination 162, 229–236.

Lin, K.L., Min Lin, C., Shieh, M.C., 1987. Treatment of uraniumcontaining effluents with reverse osmosis process. Desalina-tion 61, 125–136.

Mazzoni, C., Bandini, S., 2006. On nanofiltration Desal-5 DKperformances with calcium chloride-water solutions. Sep.Purif. Technol. 52, 232–240.

Orloff, K.G., Mistry, K., Charp, P., Metcalf, S., Marino, R., Shelly, T.,Melaro, E., Donohoe, A.M., Jones, R.L., 2004. Human exposureto uranium in groundwater. Environ. Res. 94, 319–326.

Raff, O., Wilken, R.-D., 1999. Removal of dissolved uranium bynanofiltration. Desalination 122, 147–150.

Reimann, C., Banks, D., 2004. Setting action levels for drinkingwater: are we protecting our health or our economy (or ourbacks!)? Sci. Total Environ. 332, 13–21.

Rios, G.M., Joulie, R., Sarrade, S.J., Carles, M., 1996. Investigation ofion separation by microporous nanofiltration membranes.AIChE J. 42, 2521–2528.

Sbai, M., Szymczyk, A., Fievet, P., Sorin, A., Vidonne, A., Pellet-Rostaing, S., Favre-Reguillon, A., Lemaire, M., 2003. Influenceof the membrane pore conductance on tangential streamingpotential. Langmuir 19, 8867–8871.

Schaep, J., Van Der Bruggen, B., Vandecasteele, C., Wilms, D., 1998.Retention mechanisms in nanofiltration. Environ. Sci. Res. 55,117–125.

Sheppard, S.C., Sheppard, M.I., Gallerand, M.-O., Sanipelli, B.,2005. Derivation of ecotoxicity thresholds for uranium.J. Environ. Radioactivity 79, 55–83.

Sorin, A., Favre-Reguillon, A., Pellet-Rostaing, S., Bernier, G.,Lemaire, M., 2006. Polyaminocarboxylic acids rejectionby charged nanofiltration membrane. J. Membr. Sci. 279,446–452.

Tanninen, J., Nystrom, M., 2002. Separation of ions in acidicconditions using NF. Desalination 147, 295–299.

Vaaramaa, K., Pulli, S., Lehto, J., 2000. Effects of pH and uraniumconcentration on the removal of uranium from drinking waterby ion exchange. Radiochim. Acta 88, 845–849.

van der Lee, J., 2002. JCHESS 2.0, Ecole Nationale Superieure desMines de Paris, 35 rue Saint Honore , 77305 FontainebleauCedex, France.

Zhang, Y., Causserand, C., Aimar, P., Cravedi, J.P., 2006. Removal ofbisphenol A by a nanofiltration membrane in view of drinkingwater production. Water Res. 40, 3793–3799.

Zhou, P., Gu, B., 2005. Extraction of oxidized and reduced forms ofuranium from contaminated soils: effects of carbonate con-centration and pH. Environ. Sci. Technol. 39, 4435–4440.